USF Libraries
USF Digital Collections

The causal effect of alcohol consumption on employment status

MISSING IMAGE

Material Information

Title:
The causal effect of alcohol consumption on employment status
Physical Description:
Book
Language:
English
Creator:
Sangchai, Chanvuth
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Labor market
Human capital
Alcoholism
Drinking
Instrumental variable
Dissertations, Academic -- Business Administration -- Doctoral -- USF
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Alcohol consumption may affect labor market outcomes directly through a reduction in productivity and indirectly through human capital accumulation. However, empirical results from previous studies in the economics literature are mixed and inconclusive. While some researchers found negative effects of alcohol use on labor market outcomes, quite a few studies found either positive or insignificant effects. The purpose of this dissertation is to estimate causal effects of alcohol consumption on employment status. It uses three data sets previously unexploited for this purpose and attempts to eliminate any potential estimation problems from previous studies. The results show that previous problematic heavy drinking, i.e. clinically-defined alcohol abuse and/or dependence, has no significant direct effects, but has significant indirect effects on current employment propensity for both genders through human capital components, specifically educational attainment and health status. While general alcohol consumption has only an indirect effect on employment status for females, it has both direct and indirect effects on employment status for males, though the direct effect is very small.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2006.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Chanvuth Sangchai.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 159 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001914392
oclc - 176152802
usfldc doi - E14-SFE0001754
usfldc handle - e14.1754
System ID:
SFS0026072:00001


This item is only available as the following downloads:


Full Text

PAGE 1

The Causal Effect of Alcohol Consumption on Employment Status by Chanvuth Sangchai A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Economics College of Business Administration University of South Florida Co-Major Professor: Jeffrey S. DeSimone, Ph.D. Co-Major Professor: Donald Bellante, Ph.D. Carole A. Green, Ph.D. Gabriel A. Picone, Ph.D. Date of Approval: October 27, 2006 Keywords: labor market, human capital, alcohol ism, drinking, instrumental variable Copyright 2006 Chanvuth Sangchai

PAGE 2

i Table of Contents List of Tables.......................................................................................................................x Abstract....................................................................................................................... .....xiii 1. Introduction......................................................................................................................1 2. Literature Review.............................................................................................................5 2.1. Earnings............................................................................................................6 2.2. Labor Supply.....................................................................................................8 3. Theoretical Background and Methodology...................................................................11 4. Data Descriptions........................................................................................................... 15 4.1. Variable Selections a nd Sample Restrictions.................................................15 4.1.1. Dependent Variables........................................................................15 4.1.2. Alcohol Variables............................................................................15 4.1.3. Instrumental Variables.....................................................................16 4.1.4. Exogenous Variables.......................................................................17 4.1.5. Sample Restrictions.........................................................................18 4.2. Data.................................................................................................................20 4.2.1. NESARC..........................................................................................21 4.2.1.1. Dependent Variables.........................................................22 4.2.1.2. Alcohol Variables.............................................................22 4.2.1.3. Instrumental Variables......................................................22 4.2.1.4. Exogenous Variables........................................................23

PAGE 3

ii 4.2.1.5. Sample Restrictions..........................................................25 4.2.1.6. Sample Descriptive Statistics............................................26 4.2.2. NSDUH............................................................................................26 4.2.2.1. Dependent Variables.........................................................28 4.2.2.2. Alcohol Variables.............................................................28 4.2.2.3. Instrumental Variables......................................................29 4.2.2.4. Exogenous Variables........................................................29 4.2.2.5. Sample Restrictions..........................................................30 4.2.2.6. Sample Descriptive Statistics............................................30 4.2.3. NELS:88..........................................................................................32 4.2.3.1. Dependent Variables.........................................................33 4.2.3.2. Alcohol Variables.............................................................33 4.2.3.3. Instrumental Variables......................................................33 4.2.3.4. Exogenous Variables........................................................34 4.2.3.5. Sample Restrictions..........................................................35 4.2.3.6. Sample Descriptive Statistics............................................35 5. Estimation Results.........................................................................................................37 5.1. Results from NESARC...................................................................................37 5.1.1. Alcohol Abuse and/or Dependence.................................................37 5.1.2. Frequencies of Alcohol Consumption.............................................47 5.2. Results from NSDUH.....................................................................................56 5.2.1. Alcohol Abuse and/or Dependence.................................................56 5.2.2. Frequencies of Alcohol Consumption.............................................61

PAGE 4

iii 5.3. Results from NELS:88....................................................................................69 6. Conclusions, Caveats, Policy Implications, and Future Works.....................................77 References..........................................................................................................................83 Appendix A: Complete Second Stage Regression Results using NESARC......................85 Appendix A-1: Current Employment (Full Time/Part Time) versus Alcohol Abuse and/or Dependence prior to the Last 12 Months: the Male Sample.....................................................................................................86 Appendix A-2: Current Employment (Full Time/Part Time) versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Male Sample...............................................................................................................88 Appendix A-3: Currently Working Full Time (35+ Hours a Week) versus Alcohol Abuse and/or Dependence prior to the Last 12 Months: the Male Sample.....................................................................................................90 Appendix A-4: Currently Working Full Time (35+ Hours a Week) versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Male Sample...............................................................................................................92 Appendix A-5: Current Employment (Full Time/Part Time) versus Alcohol Abuse and/or Dependence prior to the Last 12 Months: the Female Sample..................................................................................................94 Appendix A-6: Current Employment (Full Time/Part Time) versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Female Sample...............................................................................................................96

PAGE 5

iv Appendix A-7: Currently Working Full Time (35+ Hours a Week) versus Alcohol Abuse and/or Dependence prior to the Last 12 Months: the Female Sample..................................................................................................98 Appendix A-8: Currently Working Full Time (35+ Hours a Week) versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Female Sample.............................................................................................................100 Appendix A-9: Current Employment (Full Time/Part Time) versus the Number of Drinks Consumed in th e Last 12 Months: the Male Sample........102 Appendix A-10: Current Employment (Full Time/Part Time) versus the Number of Days Drank 5+ Drinks in the Last 12 Months: the Male Sample.............................................................................................................104 Appendix A-11: Current Employment (Full Time/Part Time) versus the Number of Days Drunk in the Last 12 Months: the Male Sample..................106 Appendix A-12: Current Employment (Full Time/Part Time) versus the Average Daily Volume of Ethanol Consumed in the Last 12 Months: the Male Sample..............................................................................................108 Appendix A-13: Currently Working Full Time (35+ Hours a Week) versus the Number of Drinks Consumed in the Last 12 Months: the Male Sample.............................................................................................................110 Appendix A-14: Currently Working Full Time (35+ Hours a Week) versus the Number of Days Drank 5+ Drinks in the Last 12 Months: the Male Sample.............................................................................................................112

PAGE 6

v Appendix A-15: Currently Working Full Time (35+ Hours a Week) versus the Number of Days Drunk in the Last 12 Months: the Male Sample............114 Appendix A-16: Currently Working Full Time (35+ Hours a Week) versus the Average Daily Volume of Ethanol Consumed in the Last 12 Months: the Male Sample...............................................................................116 Appendix A-17: Current Employment (Full Time/Part Time) versus the Number of Drinks Consumed in the Last 12 Months: the Female Sample.............................................................................................................118 Appendix A-18: Current Employment (Full Time/Part Time) versus the Number of Days Drank 5+ Drinks in the Last 12 Months: the Female Sample.............................................................................................................120 Appendix A-19: Current Employment (Full Time/Part Time) versus the Number of Days Drunk in the Last 12 Months: the Female Sample..............122 Appendix A-20: Current Employment (Full Time/Part Time) versus the Average Daily Volume of Ethanol Consumed in the Last 12 Months: The Female Sample.........................................................................................124 Appendix A-21: Currently Working Full Time (35+ Hours a Week) versus the Number of Drinks Consumed in the Last 12 Months: the Female Sample.............................................................................................................126 Appendix A-22: Currently Working Full Time (35+ Hours a Week) versus the Number of Days Drank 5+ Dr inks in the Last 12 Months: the Female Sample................................................................................................128

PAGE 7

vi Appendix A-23: Currently Working Full Time (35+ Hours a Week) versus the Number of Days Drunk in the Last 12 Months: the Female Sample........130 Appendix A-24: Currently Working Full Time (35+ Hours a Week) versus the Average Daily Volume of Ethanol Consumed in the Last 12 Months: the Female Sample............................................................................132 Appendix B: Complete Second Stage Regression Results using NSDUH......................134 Appendix B-1: Employment (Full Time /Part Time) Last Week versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Male Sample.............................................................................................................135 Appendix B-2: Employment (Full Time /Part Time) Last Week versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Female Sample.............................................................................................................136 Appendix B-3: Working Full Time Last Week versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Male Sample........................137 Appendix B-4: Working Full Time Last Week versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Female Sample.....................138 Appendix B-5: Employment (Full Time/Part Time) Last Week versus the Number of Days Drinki ng Alcohol in the Last 12 Months: the Male Sample.............................................................................................................139 Appendix B-6: Employment (Full Time/Part Time) Last Week versus the Number of Days Drinking Alcohol in the Last Month: the Male Sample......140

PAGE 8

vii Appendix B-7: Employment (Full Time/Part Time) Last Week versus the Number of Days Drank 5+ Alcoholic Drinks in the Last Month: the Male Sample...................................................................................................141 Appendix B-8: Working Full Time Last Week versus the Number of Days Drinking Alcohol in the Last 12 Months: the Male Sample...........................142 Appendix B-9: Working Full Time Last Week versus the Number of Days Drinking Alcohol in the Last Month: the Male Sample..................................143 Appendix B-10: Working Full Time Last Week versus the Number of Days Drank 5+ Alcoholic Drinks in the Last Month: the Male Sample.........144 Appendix B-11: Employment (Full Time/Part Time) Last Week versus the Number of Days Drinki ng Alcohol in the Last 12 Months: the Female Sample.............................................................................................................145 Appendix B-12: Employment (Full Time/P art Time) Last Week versus the Number of Days Drinking Alcohol in the Last Month: the Female Sample.............................................................................................................146 Appendix B-13: Employment (Full Time/Part Time) Last Week versus the Number of Days Drank 5+ Alcoholic Drinks in the Last Month: the Female Sample................................................................................................147 Appendix B-14: Working Full Time Last Week versus the Number of Days Drinking Alcohol in the Last 12 Months: the Female Sample..............148 Appendix B-15: Working Full Time Last Week versus the Number of Days Drinking Alcohol in the La st Month: The Female Sample....................149

PAGE 9

viii Appendix B-16: Working Full Time Last Week versus the Number of Days Drank 5+ Alcoholic Drinks in the Last Month: the Female Sample.............................................................................................................150 Appendix C: Complete Second Stage Re gression Results using NELS:88....................151 Appendix C-1: Current Employment (Full Time/Part Time) versus the Number of Occasions Drank Alcohol in the Last 30 Days: the Male Sample.............................................................................................................152 Appendix C-2: Current Employment (Full Time/Part Time) versus the Number of Occasions Drank 5+ Alcoholic Drinks in a Row over the Last Two Weeks: the Male Sample................................................................153 Appendix C-3: Currently Working Full Time versus the Number of Occasions Drank Alcohol in the Last 30 Days: the Male Sample..................154 Appendix C-4: Currently Working Full Time versus the Number of Occasions Drank 5+ Alcoholic Drin ks in a Row over the Last Two Weeks: the Male Sample.................................................................................155 Appendix C-5: Current Employment (Full Time/Part Time) versus the Number of Occasions Drank Alcohol in the Last 30 Days: the Female Sample.............................................................................................................156 Appendix C-6: Current Employment (Full Time/Part Time) versus the Number of Occasions Drank 5+ Alcoholic Drinks in a Row over the Last Two Weeks: the Female Sample.............................................................157 Appendix C-7: Currently Working Full Time versus the Number of Occasions Drank Alcohol in the Last 30 Days: the Female Sample...............158

PAGE 10

ix Appendix C-8: Currently Working Full Time versus the Number of Occasions Drank 5+ Alcoholic Drin ks in a Row over the Last Two Weeks: the Female Sample.............................................................................159 About the Author...................................................................................................End Page

PAGE 11

x List of Tables Table 1 Alcohol prevalence rates from NHSDA surveys.........................................2 Table 2 Sample descriptive statistics from NESARC.............................................27 Table 3 Sample descriptive statistics from NSDUH...............................................31 Table 4 Sample descriptive statistics from NELS:88.............................................36 Table 5 Regression estimates of binary alcohol measures on the probability of current employment (full time/part time): the male sample from NESARC....................................................................................................38 Table 6 Regression estimates of binary alcohol measures on the probability of currently working full time (35+ hours a week): the male sample from NESARC...........................................................................................41 Table 7 Regression estimates of binary alcohol measures on the probability of current employment (full time /part time): the female sample from NESARC...........................................................................................43 Table 8 Regression estimates of binary alcohol measures on the probability of currently working full time (35+ hours a week): the female sample from NESARC...............................................................................46 Table 9 Regression estimates of continuous alcohol measures on the probability of current employment (full time/part time): the male sample from NESARC...............................................................................48

PAGE 12

xi Table 10 Regression estimates of continuous alcohol measures on the probability of currently working fu ll time (35+ hours a week): the male sample from NESARC......................................................................51 Table 11 Regression estimates of continuous alcohol measures on the probability of current employment (f ull time/part time): the female sample from NESARC...............................................................................52 Table 12 Regression estimates of continuous alcohol measures on the probability of currently working fu ll time (35+ hours a week): the female sample from NESARC...................................................................53 Table 13 Regression estimates of binary alcohol measures on the probability of employment (full time/part time) last week using NSDUH..................57 Table 14 Regression estimates of binary alcohol measures on the probability of working full time last week using NSDUH...........................................60 Table 15 Regression estimates of continuous alcohol measures on the probability of employment (full time/part time) last week: the male sample from NSDUH.................................................................................62 Table 16 Regression estimates of continuous alcohol measures on the probability of working full time last week: the male sample from NSDUH......................................................................................................65 Table 17 Regression estimates of continuous alcohol measures on the probability of employment (full time/part time) last week: the female sample from NSDUH.....................................................................67

PAGE 13

xii Table 18 Regression estimates of continuous alcohol measures on the probability of working full time last week: the female sample from NSDUH......................................................................................................70 Table 19 Regression estimates of continuous alcohol measures on the probability of current employment (full time/part time): the male sample from NELS:88...............................................................................71 Table 20 Regression estimates of continuous alcohol measures on the probability of currently working fu ll time: the male sample from NELS:88....................................................................................................73 Table 21 Regression estimates of continuous alcohol measures on the probability of current employment (f ull time/part time): the female sample from NELS:88...............................................................................74 Table 22 Regression estimates of continuous alcohol measures on the probability of currently working full time: the female sample from NELS:88....................................................................................................76

PAGE 14

xiii The Causal Effect of Alcohol C onsumption on Employment Status Chanvuth Sangchai ABSTRACT Alcohol consumption may affect labo r market outcomes directly through a reduction in productivity and indirectly through human capital accumulation. However, empirical results from previous studies in the economics literature are mixed and inconclusive. While some re searchers found negative effect s of alcohol use on labor market outcomes, quite a few studies found either positive or insignificant effects. The purpose of this dissertation is to estimate causal effects of alcohol consumption on employment status. It uses three data sets previously unexploited for this purpose and attempts to eliminate any potential estima tion problems from previous studies. The results show that previous problematic heavy drinking, i.e. clini cally-defined alcohol abuse and/or dependence, has no significant direct effects, but has significant indirect effects on current employment propensity for both genders through human capital components, specifically educational attainment and health status. While general alcohol consumption has only an indirect effect on employment status for females, it has both direct and indirect effects on em ployment status for males, though the direct effect is very small.

PAGE 15

1. Introduction Many researchers have attributed tr emendous economic costs to alcohol consumption, and the U.S. federal government has used various public health policies to prevent and treat alcohol abuse. Nonetheless, alcohol prevalence rates in the U.S. have persistently been very high. As Table 1 show s, data from the National Household Survey on Drug Abuse (NHSDA) from the Substance Abuse and Mental Health Data Archive (SAMHDA) reveal that alcohol prevalence rates for past year and past month use ranged from 62 percent and 46 percent, re spectively, during 1979. The new versions of this survey, the 2002 and 2003 National Survey on Drug Use and Health (NSDUH), reveal similar rates of 66.0 percen t and 65.3 percent for past year use and 51.0 percent and 50.4 percent for past month us e. The 2002 and 2003 NSDUH also show that nearly 8 percent of the U.S. population exhi bited symptoms of al cohol abuse and/or dependence in the past year, based on criteria from the Diagnostic a nd Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), published by the American Psychiatric Association (APA). According to medical researchers, harmfu l effects of alcohol intoxication include decreased thinking and reasoni ng performance, diminished concentration, impaired shortterm memory, and reduced coordination. Prol onged alcohol consumption can have even more serious consequences, such as health problems, unplanned pregnancy, illegal drug use, accidents, crime and suicide. 1

PAGE 16

Labor and health economists are also inte rested in the consequences of alcohol consumption on labor market outcomes such as employment status, hours worked, and wages. Alcohol use could negatively aff ect labor market outcomes directly through a reduction in productivity and indirectly by decreasing human capital accumulation. In contrast, despite the conventional wisdom about the social costs of alcohol consumption, some medical research has show n that moderate alcohol use ma y be beneficial to health. If so, and productivity is positively related to health, moderate drinking could improve labor market outcomes. Indeed, empirical results on this relationship are mixed and inconclusive, suggesting the n eed for further investigation. Year Past year use (%) Past month use (%) 1979 72.9 60.9 1982 68.0 55.0 1985 72.9 58.3 1988 68.1 53.4 1990 66.0 51.2 1991 68.0 50.9 1992 64.7 47.8 1993 66.5 49.6 1994-A 66.9 52.6 1994-B 66.9 53.9 1995 65.4 52.2 1996 64.9 51.0 1997 64.1 51.4 1998 64.0 51.7 1999 62.2 46.4 2000 61.9 46.6 2001 63.6 48.3 Table 1. Alcohol prevalence rates from NHSDA surveysStatistics are calculated from online analysis on the Inter-University Consortium for Political and Social Research (ICPSR) web site using sample weights. The purpose of this study is to estimate the causal effect of alcohol consumption on employment status. Specifi cally, it investigates whether previous problematic heavy drinking, i.e. clinical ly-defined alcohol abuse and/or dependence, or more frequent alcohol use lowers the probability of current employment. The focus is on attempting to 2

PAGE 17

eliminate potential estimation problems found in previous studies in order to generate estimates that can be reliably interpreted as cau sal effects. In particular, an instrumental variables (IV) regression approach will be applied to control for potential endogeneity that is inherent in the natu re of the relationship between alcohol consumption and labor market outcomes. IV estimates will be co mpared to single-equation estimates using ordinary least squares (OLS) to test whethe r alcohol use is exogenously determined with respect to employment. Overidentification tests will be performed to establish the validity of the instru mental variables. The study will use survey data from the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) the National Survey on Drug Use and Health (NSDUH), and the National Educa tion Longitudinal Study of 1988 (NELS:88) that have not previously been used to study the relationship between alcohol use and labor market outcomes. There are several a dvantages in using th ese data sets. In particular, they contain information on vari ous measures of empl oyment status and alcohol consumption. Self-administered answ er sheets for the alcohol use questions will help minimize underreporting. The data also contain many variab les that have the potential to be highly correla ted with drinking but uncorrela ted with current employment, which would allow their specification as instrumental variables in the estimation procedure. Finally, the richness of the da ta enables controlling for many demographic factors that are likely related to both drinking and employment. 3 Various strategies, solutions and sugges tions from previous studies will be implemented. First, this study will recogni ze that effects of alcohol consumption on employment might vary by age and gender. Ma le and female samples will be analyzed

PAGE 18

separately, and only respondents of prime labor force age will be included so that potential effects on school enro llment and retirement at either end of the age distribution are filtered out. Second, the study will us e alternative measures of drinking and employment as well as different data sets in order to examine the robustness of the results. Finally, different model specificat ions will be used to further investigate robustness and to distinguish between direct and indirect effects of alcohol consumption on labor market outcomes. The remainder of this study is organized as follows. Section 2 summaries the results of previous relevant studies. S ection 3 documents the theory that motivates studying this topic as well as the empirical me thods used. Section 4 describes the data sets, sample restrictions, and the lists of variables included in the analysis. Section 5 provides detailed estimation results. Section 6 offers conclusions, caveats, suggestions for future research, and policy implications. 4

PAGE 19

2. Literature Review In the past two decades, a growing body of literature has examined the relationship between alcohol use and labor market outcomes. Medical research has shown that alcohol consumption can cause se rious health problems. Since poor health can reduce worker productivity, the conventional wisdom is th at alcohol consumption is detrimental to labor market outcomes. Ho wever, the majority of empirical results contradict this conventional wisdom. Some researchers have argued that moderate alcohol use may positively affect labor market outcomes, paralleling medical re search showing that moderate drinking is beneficial to health. Drinking in modera tion can also be relaxing, which might also improve productivity. Meanwhile, social drinki ng possibly enhances social capital, i.e. the ability to network among other employed in dividuals, which could also improve labor market outcomes. 1 It has also been argued that the unexp ected positive correlation may arise from income effects running in the opposite directi on. An increase in earnings may lead to an increase in alcohol consumpti on if alcohol is a normal good. Another possible mechanism generating be neficial effects of drinking on labor market outcomes is unobserved heterogeneit y. An unobserved family or individual specific effect that simultaneously affects bot h the alcohol consumption and labor market decisions, and is positively associated with labor market outcomes, will lead to an 5 1 See Durlauf (2002) for pros pective roles of social capital in socioeconomic outcomes.

PAGE 20

upwardly biased estimate. For instance, aggr essive or type A pe rsonality individuals might be more likely to both drink and be employed full time. Studies of labor market effects of alcohol use can be divided into two categories based on the specific labor market outcomes examined: (1) earnings, i.e. income and wage, and (2) labor supply, i.e. employment status and hours of work. Mixed results have been obtained for both categories of studies. 2.1. Earnings A majority of the literature on labor market effects of drinking has concentrated on the income or wage of workers, where fre quencies of drinking or number of drinks are the commonly used measures of drinking. Berger and Leigh (1988), the first empirical study in this area, used the 1972 Quality of Employment Survey to investigate the effect of alcohol use on wages by estimating separate earnings equa tions for non-drinkers and drinkers who worked at le ast 20 hours per week as part of a sample selection model. They found that both moderate and heavy drin kers earn higher wages than non-drinkers. Employing a similar model, Bryant et al. ( 1993) studied the effect of alcohol use on earnings of young white males in the National Longitudinal Survey of Youth (NLSY) and found that drinkers earned more and received higher wages than non-drinkers. 6 Later, to explain these unexpected results, some researchers proposed the existence of an inverse U-shaped relationship between alcohol consumption and wages: light to moderate alcohol use increases h ealth and thus produc tivity, but at some moderate level of drinking additional consump tion is detrimental. French and Zarkin (1995) used data from four worksites to investigate this hypothesis by adding quadratic

PAGE 21

and cubic alcohol use terms to a wage equati on, showing that modera te drinkers indeed have higher wages than both non-drinkers a nd heavy drinkers. Heien (1996) obtained similar results using 1979 and 1984 National Ho usehold Survey on Alcohol Use data, as did Hamilton and Hamilton (1997) for earni ngs of prime-age males using the 1985 Canadian General Social Survey, Barrett (2002) using the 1989 Australian National Health Survey, and Auld (2005) using the 1985 and 1991 cycles of the Canadian General Social Survey. 7 Yet, several analyses have yielded different results. Zarkin et al. (1998a) used a sample of prime-age workers from the 1991 and 1992 NHSDA and found a significant positive relationship of alcohol use on wages for men that had no turning point at which the effect became negative. Using a pr ime-age sample from the 1979 Current Population Survey (CPS), Dave and Kaestner (2002) estimated the reduced-form equation of the effect of alcohol taxes on wa ges. They found significant positive effects of alcohol taxes on the natural log of wages for females, but insignificant positive effects for males, suggesting that heavy alcohol c onsumption reduces wages for females but not males. But because these estimates are impl ausibly large and inconsistent with their results for employment and work hours, they concluded that there is no systematic relationship between alcohol consumption and labor market outcomes. Employing a methodology similar to that used by Dave and Kaestner (2002), Cook and Peters (2005) used a sample of full-time workers from the 1979 National Longitudinal Survey of Youth (NLSY) to estimate a reduced-form equation by replacing the drinking variables with alcohol prices. They found evidence of a pos itive association between alcohol price and earnings for both genders, but the estimate fo r males is not statistically significant.

PAGE 22

Hence, the results suggest that there is a negative association between drinking and earnings for females only. Another possible explanation for positive ear nings effects of alcohol consumption is unmeasured individual characteristics that po sitively influence wages. This problem is called unobserved time invariant heterogeneity. Two studies of earni ngs addressed this problem using different panel data sets from the National Longitudinal Survey of Youth (NLSY). Kenkel and Ribar (1994) applied an IV approach to address this problem and found negative effects of alcohol use on ear nings for both genders where the magnitude of the estimates is larger compared to OL S estimates. Peters (2004) addressed this problem by applying a fixed effect model and found positive effects of alcohol use on wages using both OLS and individual fixed e ffect approaches. However, the estimates from the fixed effect model are smaller in magnitude and statistically insignificant. Both studies suggest that unobserv ed heterogeneity might be the cause of the unexpected positive drinking effects found by other researchers. 2.2. Labor Supply 8 Earnings studies have ignored the effect of alcohol consumption on employment by analyzing samples consisting only of workers. Other researchers have thus examined effects on employment and, more generally, hours worked. Many labor supply studies defined a binary indicator of alcohol abus e and/or dependence based on criteria from different versions of the Diagnostic and Sta tistical Manual of Ment al Disorders (DSM). Depending on the study, this variable is referre d to as representing alcohol abuse and/or dependence, problem drinking, or alcoholism.

PAGE 23

Using multiple site data from the Epidemiologic Catchment Area (ECA) survey, Mullahy and Sindelar (1991) f ound negative and statistically significant effects of previous alcoholism (based on DSM-IIIR) on the probability of working full time, for both genders, using logit regressions. These effects were smaller, but still significant, after controlling for human capital measures. Using the same data to investigate the effect of previous alcoholism on employment (full or part time), Mullahy and Sindelar (1993) found similar results. Both studies tr eated the alcoholism variable as exogenous. 9 While the results of these two studies s upport the conventional wisdom that labor market effects of drinking are negative, ot her studies estimated either insignificant or mixed effects. In the 1979-1990 NLSY, Ke nkel and Ribar (1994) found that in IV regressions, alcohol abuse and alcohol depende nce (based on DSM-III) have insignificant negative effects on hours of work for men, but have a positive and si gnificant effect for women. However, the results were not robust to alternative model specifications. Using prime-age males and females in the Alc ohol Supplement of the 1988 National Health Interview Survey (NHIS), Mullahy and Sinde lar (1996) applied an IV approach and found that for both genders, problem drinki ng (based on DSM-IIIR) lowered employment and increased unemployment, though the results are not statistically significant. Zarkin et al. (1998b) investigated the effect of alcohol use on pa st month work hours among young male workers, finding an insignificant posit ive effect in the 1991 NHSDA data but a significant negative effect in the 1992 NHSDA data. Thus, they concluded that the relationship is unstabl e. Feng et al. (2001) used a random sample of prime working age respondents from a four-wave l ongitudinal population-based surv ey of six southern U.S. states. In their bivariate probit model, problem drinking (based on DSM-IV) has an

PAGE 24

insignificant negative effect on employment for women but a significant positive effect for men. Two recent studies revealed significant negative effects of alcoholism on the probability of working. Using the same data and variable specification as did Mullahy and Sindelar (1996) but restrict ing the sample to males, Terza (2002) applied a non-linear multinomial logit model that accounted for the possible endogeneity of drinking and a non-linear relationship, a nd obtained results that are simila r but significant and twice as large. MacDonald and Shields (2004) also found significant negativ e effects of problem drinking on the probability of working when allowing for potential unobserved heterogeneity using bivariate probit models for working age males in the 1996 Health Survey of England (HSE). They de fined problem drinking using CAGE, an internationally used assessment instrument for identifying problems with alcohol that, like DSM, is based on the observed psychol ogical and physical symptoms of alcohol. Earnings studies are prone to sample selection bias because they examine only workers. This raises the po ssibility that the negative effect of alcohol use is underestimated. In addition, the instrumental variables employed in this study are more likely to be endogenous with respect to earnings than employment. Therefore, this study will focus on labor supply, specifi cally employment status. 10

PAGE 25

3. Theoretical Background and Methodology The model used in this study is based on human capital theory. Labor market outcomes have a variety of determinants, su ch as demographic and geographic factors, education, health status, social status, and consumption decisions. By accumulating human capital such as education, training, a nd work experience, individuals increase labor market productivity and thus the probabil ity of employment. Health capital is also an importance component of human capital because, ceteris paribus, healthier workers are more productive. Alcohol consumption can lower employment probability both directly, through a reduction in productivity, and indirectly, by decreasing human capital accumulation (Mullahy and Sindelar 1989). Othe r exogenous factors such as age, race, marital status, the number of children, and local labor supply and demand also determine whether an individual participates in the labor force and is employed. Therefore, consider a model of the re lationship between employment and alcohol consumption of the form E i = 0 + 1 X 1i + + k X ki + A A i + i i = 1, n, (1) where E is employment status, A is al cohol consumption, the Xs are observable explanatory variables such as demographic fa ctors, education, health status, marital status, and geographic controls, k is the number of exogenous variables, is the error term (which contains unobservabl e determinants of employment), s are parameters to be estimated, i indexes individuals, and n is the number of observations. 11

PAGE 26

This study attempts to obtain an econo metrically consistent estimate of the alcohol use coefficient, A in equation (1). For ordinary least squares (OLS) to yield this, the assumption of zero correl ation between the error term ( ) and alcohol consumption (A), i.e. zero conditional mean, must hold (Greene 2002). However, zero conditional mean is particularly unlikely to ho ld in equation (1). This is because of a variety of factors, including the model specification, the natu re of survey data, and the likely bi-directional relationship between alcohol consumption and employment. The violation of this assumption is calle d the endogeneity problem and can arise for three distinct reasons. First, unobse rved determinants of employment are by definition part of the error term. If any of these unobserved determinants are also correlated with alcohol use, the error term will be correlated with alcohol use. This is called the omitted variable problem. Second, when actual alcohol use is measured with error, observed alcohol use is correlated with the measurement error, which is subsumed into the error term. This is called the measurement error problem. Alcohol consumption in particular might be ei ther underreported or overre ported, depending on whether and which others are present during the interview. Third, if the alcohol consumption decision also responds to changes in employment, i.e. through the income eff ect if alcohol is a normal good, then changes in the error term will circulate to alcohol use through employment. This is called the simultaneity problem. 12 The most popular econometric solution to the problem of endogeneity, including measurement error in alcohol use as long as it is not systematically related to employment, is the instrumental variables (IV) approach, which produces consistent estimates in the presence of an endogenous explanatory variable. This approach

PAGE 27

requires one or more instrumental variables th at are not correlated with the error term ( ), but have a high partial correlation with alcohol use even after effects through other exogenous variables have been netted out (Wooldridge 2002). Typically, the IV approach is estimated using two-stage least squares (2SLS). In the first stage regression, the following equation is estimated using OLS and fitted values ( i ) are obtained: A i = 0 + 1 X 1i + + k X ki + 1 Z 1i + + j Z ji + i i = 1, n, (2) where the Zs are the instrumental variables, the Xs are the explanatory variables from equation (1), j is the number of instrumental variables, the s and s are parameters to be estimated, and is the error term. In the second st age, equation (3), which is identical to equation (1) with i substituted for A i is estimated by OLS: E i = 0 + 1 X 1i + + k X ki + A i + i i = 1, n. (3) The standard errors obtained from this regression are incorrect, but are easily corrected (Wooldridge 2002). 13 A minimal condition for identification of equation (3) is the rank condition, which can be tested using a standard pa rtial F test of the null hypothesis 1 = = j = 0 in equation (2), i.e. the first stag e regression. This reduces to a t-test if there is only one instrumental variable, i.e. if j = 1. If this hypothesis is rejected, equation (3) passes the rank condition and is identifie d. Identification is exact in the case of only one instrumental variable. Otherwise, the equation is overidentif ied, and a test of overidentifying restrictions s hould be carried out to determ ine whether the instrumental variables are correlated with the residuals, i.e. sample error terms, from equation (3). The test statistic is nR 2 where R 2 is the multiple coefficient of determination obtained by

PAGE 28

regressing the residuals from equation (3) on the Xs and the Zs using OLS. Under the null hypothesis of overidentifying restriction validity along with homoskedastic errors, this statistic is distributed as chi-square, w ith degrees of freedom equal to one less than the number of instrumental variables. Failure to reject the null hypothesis implies that 2SLS will produce consistent causal effect estimates. The weakness of even a properly specified 2SLS model is that it is inefficient relative to OLS in the situation where al cohol use is actually exogenous, i.e. zero conditional mean is satisfied, in which case OLS produces consistent estimates. Thus, the Hausman (1978) test for exogeneity will be conducted. The idea of th is test is that if alcohol use and the error term of (1) are uncorrelated, the estimates obtained from OLS and 2SLS should differ only by sampling error. The Hausman t statistic is equal to the difference between the 2SLS and OLS estimates of A divided by the square root of the difference in the estimated variances of A under 2SLS and OLS. If the null hypothesis of exogeneity is rejected, 2SLS estimates should be used to make inferences about causal effects. If not, OLS estimates are preferab le due to their smaller standard errors. In sum, IV, estimated by 2SLS, produces a consistent estimate of the causal effect of alcohol use on employment in the presen ce of endogeneity, if valid instrumental variables can be obtained. Give n the potential endogeneity of al cohol use in this context, 2SLS will be used in this study. 14

PAGE 29

4. Data Descriptions 4.1. Variable Selections a nd Sample Restrictions 4.1.1. Dependent Variables Different labor market outcomes, and different definitions of the same labor market outcomes, can be affected differentia lly by the same variab les. Wage studies concentrate on workers and thus ignore effects on employment and labor force participation. Studies of income or earni ngs estimate combined effects on the wage and labor supply, so different sample inclusi on criteria regarding wo rk status and hours worked can lead to different conclusions. This study concentrates on employment status. Most studies have grouped individuals who are unemployed and those w ho are out of labor force into the same category, although alcohol consumption might affect labor force participation differently than it does employment among those in the la bor force. Carefully defining employment and using alternative measures to ex amine robustness are thus important. 4.1.2. Alcohol Variables Properly defining drinking status is also important, sinc e light and heavy drinking might have different effects on labor market outcomes. The variety of alcohol use measures in the different data sets studied allow for examination of many different past year drinking variables. For maintaining the possibility of causality, it is useful for the 15

PAGE 30

period in which drinking status is measured to have preceded that for labor market status. This study will use both binary indicators a nd numerical measures of alcohol use. 4.1.3. Instrumental Variables Estimates from the many studies that ha ve treated drinking as exogenous and use a single equation method reflect the partial correlation between alcohol consumption and labor market outcomes, holding constant other variables in th e regression, but not necessarily the causal effect of drinking on labor market outco mes. Mullahy and Sindelar (1991, 1993), French and Zarkin (1995), Heien ( 1996), Zarkin et al. (1998b) and Feng et al. (2001) each acknowledged the endogeneity prob lem, but applied a single equation approach because of data limitations. 16 Previous IV studies have us ed a variety of instruments as attempted sources of exogenous variation with which to identify drinking behavior. Kenkel and Ribar (1994) used as instruments the percentage of the st ates population residing in dry counties, the average beer price, parents alcoholism, and relatives alcoholism. Instruments specified by Mullahy and Sindelar (1996) included an indicat or of whether an individual lived with alcoholic relatives or parents, the state-level excise tax on beer, the state-level excise tax on cigarettes, and the state-level apparent alcohol consumption. Zarkin et al. (1998a, 1998b) used respondents assessments of risk fr om using alcohol as instruments. Barrett (2002) specified as instruments a dummy va riable indicating whether the individual smoked at 18 years of age (to capture attit udes toward risk) and a set of variables measuring drinking patterns in the locality of residence (to capture so cial influences on drinking behaviors). Macdonald and Shields (2004) used respondent non-chronic health

PAGE 31

conditions as well as information about whether parents and partners smoked as instruments. However, these studies still fail to reach consistent conclusions, leading to the question of instrument validity. It is pa rticularly important to perform tests of overidentification to en sure that the instrumental variab les are exogenous with respect to employment and the resulting IV estimates are thus consistent Mullahy and Sindelar (1996) and Zarkin et al. (1998a, 1998b) are the only studies in this area that included the tests of instrument validity. Mullahy and Sindelar (1996) found th at their instruments marginally pass overidentification tests, impl ying that their instruments can be validly excluded from the labor market outcome e quations. However, Zarkin et al. (1998a, 1998b) found that their instruments did not pass ove ridentification tests, i.e. could not be validly excluded from the labor market outco me equations, thus preventing the authors from obtaining definitive conclusions. 4.1.4. Exogenous Variables Alcohol consumption has both direct and indirect effects on employment. Alcohol consumption can directly increase absenteeism and lower productivity. It can also affect productivity indirectly by reduc ing the accumulation of human capital through deleterious effects on schooling, health, and work experience. It is important to estimate regressions with and without these human cap ital measures in order to distinguish the direct effect from the overall effect that in cludes the indirect effect through human capital accumulation. 17

PAGE 32

To investigate the direct and indirect effects of alcohol consumption, four specifications of the set of explanatory f actors included in equations (2) and (3) are defined, corresponding to specifications labe led Model 1. In Model 1, these include only demographic and geographic factors th at are reasonably exogenous. Under this specification, the estimated alcohol use coefficient, A will reveal the overall effect of alcohol use on employment. In Model 2, the explanatory variables include those from Model 1, along with the number of children and marital status. Model 3 includes the variables in Model 2 plus edu cational attainment. Model 4 ad ds health status indicators to the variables from Model 3. The estimated coefficient of alcohol use in Model 4 will reveal its direct effect on employment, cont rolling for indirect effects through the main types of human capital that determine employ ment and might also be altered by drinking. 4.1.5. Sample Restrictions 18 Previous studies reveal the importance of stratifying analysis samples based on gender and age. Various studies, includi ng Mullahy and Sindelar (1991, 1996), Kenkel and Ribar (1994), Zarkin et al (1998a), Feng et al. (2001) an d Dave and Kaestner (2002) have found that labor market e ffects of alcohol use vary by gender. Because men tend to have higher tolerance levels for alcohol than do women a nd alcohol can cause medical problems to progress more rapidly in women than in men, the National Institute on Alcohol Abuse and Alcoholism (NIAAA) sugge sts a lower drinking limit for women than for men. Meanwhile, men tend to drink more heavily than women and are more likely to develop symptoms of alcoholism. Moreover, ge nder differences in la bor market behavior over the course of the life cycle, driven in large part by child b earing among women, are

PAGE 33

well known. Therefore, this study will analyze males and females separately. Samples will include only prime age respondents, because labor market effects of alcohol use are likely to vary with age. The drinking behavior of young people is qualitatively different from that of older people. All else equal, the level of tolerance to alcohol decreases with age. In addition, alcohol can harmfully interact with many medical conditions that are common in olde r people. The NIAAA recommends that older people limit their alcohol consumption to one drink per day. Perhaps more important are life-cycle differences in labor market activity. Mullahy and Sindelar (1993) argued that this is responsible for changes in the relationship between alcohol consumption and labor market outcomes over the lifetime of an individual. For instance, alcohol may cau se an adolescent drinker to drop out of school and enter the labor force early, thus leading to more work experience and higher earnings in early adulthood rela tive to non-drinkers. But this relationship reverses itself during prime work ages, as returns to schooling ultimately dominate those from experience. Older drinkers will therefore ha ve accumulated fewer financial assets than non-users, forcing them to retir e later. Thus, even if dr inking reduces employment for individuals of prime work ages drinkers might be more likel y to work in early adulthood and as normal retirement age approaches. Indeed, Hamilton and Hamilton (1997) found that heavy drinkers have flatter age-earni ngs profiles than abstainers and moderate drinkers, implying that heavy alcohol use lo wers educational attainment and increases work propensity among young adults, and precludes early retirement later in life. Failure to recognize these life cycle differenc es may yield misleading conclusions. 19

PAGE 34

Most researchers have restricted their samples to prime-age individuals, between the ages of 25 years old, to eliminate e ffects of alcohol use that operate through school enrollment and retirement. In contra st, Kenkel and Ribar (1994) and Zarkin et al. (1998b) used samples of young adults, aged 14 and 18 respectively, and obtained weak results. This is predicted if negativ e productivity effects are masked because some of these effects act to increase earning s by reducing school enrollment and thus lengthening labor market expe rience among young adults. Moreover, respondents who report being out of the labor force, including homemakers, students and retirees, are omitted from analysis samples. If drinking has harmful effects on productivity and human capital accumulation, it might raise the likelihood of unemployment while simultaneously reducing the likelihood of being out of the labor force. Thus, categorizing homema kers, students and retirees as non-workers might produce misleading results. Finally, respondents for whom informati on on any of the included variables is missing are also excluded from the analysis samples. 4.2. Data This study uses three cross secti onal data sources: the 2001 National Epidemiologic Survey on Alcohol and Related Conditions (NESARC), the 2002 and 2003 waves of the National Survey on Drug Use and Health (NSDUH), and the National Education Longitudinal Study of 1988 (NELS:88). 20

PAGE 35

4.2.1 NESARC The National Epidemiologic Survey on Alcohol and Related Conditions (NESARC), conducted by the NIAAA, is a survey of U.S. non-institutionalized households with individuals ag ed 18 and older. Non-inst itutional group quarters housing units such as boarding and rooming houses non-transient hotels and motels, shelters, facilities for housing workers, college reside nces, and group homes were also surveyed. The 2001 wave used here includes 43,093 re spondents and is av ailable online at www.niaaa.census.gov NESARC is designed to be a l ongitudinal survey but the second wave of interviews is still ongoi ng and thus not yet available. 21 There are two particular advantages to using these data. First, information on alcohol consumption is very rich. Not onl y does the data set contain information on drinking status, quantity, fre quency, and types of beverages, but it also contains an estimate of the average daily volume of ethanol intake in the past year. To collect precise information on the quantity of alcohol cons umed by respondents, interviewers used flashcards containing categorical res ponse options, accompanied by life-sized photographs of common glasses, with and without ice, with lines for various fill levels that indicated the corresponding number of ounces. This data set also contains several indicators for symptoms of alcohol abuse and/or dependen ce (based on DSM-IV) in the last 12 months and prior to the last 12 m onths, which will allow comparison with other studies that use similar dia gnostic criteria. Second, the data set contains abundant information on family alcohol consumption hist ory. Measures of problem drinking status of parents, siblings and other family me mbers can plausibly serve as instrumental variables in the econometric analysis.

PAGE 36

4.2.1.1. Dependent Variables This study examines two binary indicators of current employment status, one for any full-time or part-time employment and a nother for working full time. The latter is used to examine the robustness of the results from the former variable. The effect of drinking on full-time employment is expected to be larger than th at on any employment, since one manifestation of a deleterious produc tivity impact could be a switch from a full to part time job. 4.2.1.2. Alcohol Variables Two binary indicators and four numerical measures of alcohol use are studied. The indicators reflect whethe r the respondent had symptoms of alcohol abuse and/or dependence (based on DSM-IV); one pertains to only the last 12 months and the other relates to the period prior to the last 12 m onths. The numerical variables all cover the previous year. They are the number of alcoholic drinks consumed in the last 12 months, the number of days on which the respondent had fi ve or more alcoholic drinks in the last 12 months, the number of days on which the respondent felt intoxicated in the last 12 months, and the average daily volume of etha nol consumed in the last 12 months. The number of days intoxicated is useful becau se drinking quantities and frequencies ignore differences across individuals in th e level of tolerance to alcohol. 4.2.1.3. Instrumental Variables 22 A strength of the NESARC is the inform ation it contains on problem drinking among family members. Indicators of drinking problems among family members are

PAGE 37

candidates to serve as instrumental variables. Through hereditary or environmental factors, drinking among parent s, siblings, grandparents a nd other family members is likely to be highly correlated with respondent drinking, even after controlling for the influence of other exogenous determinants of alcohol use. The medical literature has observed that alcoholism is passed from one generation to the next. Shuckit (1999) showed that first-degree relatives of alcohol ics are three to four times more likely to develop alcoholism than first-degree relatives of non-alcoholics. However, previous drinking problems among family members are plausibly not further related to respondent employment status, holding constant respondent and other family background characteristics. Tests of instrument strength and overi dentifying restrictions were used to establish which combinations of these measures are the most empirically defensible to specify as instruments. Ultimately, two indicators of problem drinking among family members were selected: whether the respondent s blood/natural father and/or mother was ever an alcoholic or problem drinker and whether any of the respondents full brothers and/or sisters was ever an al coholic or problem drinker. 4.2.1.4. Exogenous Variables 23 The set of exogenous variable s in Model 1 includes age, age squared, four race dummy variables, two Metropolitan Statistical Area (MSA) residence dummy variables, state unemployment rates (from th e Bureau of Labor Statistics ), and four indicators for family history of major depression and/or an tisocial personality diso rder. The four race dummy variables consist of White Black, American Indian/Alaska Native and

PAGE 38

Asian/Pacific Islander with Hispanic or Latino serving as the omitted category. Two Metropolitan Statistical Area (MSA) residence dummy variables indicate whether a respondent living in an MSA re sides inside or outside the central city. An indicator variable for whether a respondent is not livi ng in an MSA is the omitted category for the residence dummy variables. Four binary indicators for family history of major depression and/or antisocial personality diso rder include an indicator of whether the respondents father/mother were ever depresse d; an indicator of wh ether the respondents brother/sister were ev er depressed; an indi cator of whether the re spondents father/mother ever had behavior problems; and an indicator of whether the responde nts brother/sister ever had behavior problems. These family me ntal health variables are included to control for the possibility that respondent employm ent is correlated with problem drinking among family members through a correlation between family member problem drinking and mental health. State unemployment rates ar e included to capture variations in labor market conditions across states. 24 Model 2 contains the same set of exoge nous variables as Model 1, but adds the number of children and five dummy variables for marital status, Married Cohabitating Widowed Divorced and Separated, with Never married as the omitted category. Model 3 includes the Model 2 variables, along w ith five dummy variables for educational attainment, Completed graduate degree Had some graduate studies Completed college Had some college education and Completed high school with Not completed high school as the omitted category. Finally, Model 4 contains the variables in Model 3, plus four dummy variables for health status, Excellent Very good Good, and Fair with Poor as the omitted category.

PAGE 39

The expected signs of the regression coe fficients in the employment equation (3) are positive for age and negative for the squa re of age, since employment tends to increase with age at a decreasing rate. Th e expected sign for geographic controls is positive, if the additional labor demand in bigger cities is greater than the additional labor supply. According to human capital theory, la bor market outcomes are positively related to educational attainment and health status, making their expected signs positive. Marital status and the number of ch ildren are expected to have positive signs for men and negative signs for women: based on traditi onal gender roles, married men will have incentive to financially support their spous es and children, while married women will have less incentive to work if married and a higher opportunity cost of working as the number of children rises. The coefficient of the state unemployment rate, as a proxy for labor demand, is expected to be negative. Th e coefficients of family history of mental health disorders will also be negative if th ese conditions adversely affect employment and are positively correlated across family members. 4.2.1.5. Sample Restrictions After the sample restrictions discusse d in section 4.1.5 are applied, the final NESARC analysis samples contain 8,673 males and 9,355 females who are 26 years old. These lower and upper age limits are chos en in order to minimize any effects of drinking that operate through sc hool enrollment and retirement while maintaining a fairly broad definition of prime working age. 25

PAGE 40

4.2.1.6. Sample Descriptive Statistics Table 2 summarizes descriptive statistic s for the NESARC sample. Average ages for both genders are about 40 years old and the majority of each sample is white. About 86 percent of the male sample and 72 percent of the female sample currently works full time. About half of the respondents live in an MSA outside the central city, with roughly two-thirds of the remainder living in central cities. About 13 percent of males and 11 percent of females did not complete high sc hool, while about 17 percent of males and 15 percent of females completed four years of college. Very few respondents report poor health. Past alcohol abuse a nd/or dependence is about twi ce as prevalent among males as among females, while average alcohol consumpti on is two to four times greater for males depending on the specific measure. Alcoho lism or problem drinking is reported about one-quarter of the time for a parent a nd one-fifth of the time for a sibling. 4.2.2. NSDUH The National Survey on Drug Use and Health (NSDUH) is a multistage area probability sample of the U.S. civilian non-institutionalized population aged 12 and above. The 2002 NSDUH is the 22 nd in the series, formerly titled the National Household Survey on Drug Abuse (NHSDA) which started in 1971. The primary purpose is to measure the prevalence and correl ates of drug use in the U.S. The surveys are funded by the Substance Abuse and Me ntal Health Services Administration (SAMHSA) and conducted by RTI Interna tional, and can be obtained from www.icpsr.umich.edu Starting in 2002, each respondent to complete a survey was given 26

PAGE 41

Table 2. Sample descriptive statistics from NESARC Variable MeanMinMaxMeanMinMax Exogenous variables Age 40.36265540.322655 White 0.58010.5301 Black 0.16010.2301 American Indian/Alaska Native 0.02010.0201 Asian/Pacific Islander 0.04010.0301 Hispanic or Latino* 0.21010.1901 Living in MSA in central city 0.33010.3601 Living in MSA outside central city 0.50010.4801 Not living in MSA* 0.17010.1601 Father/mother ever depressed 0.20010.2801 Brother/sister ever depressed 0.16010.2301 Father/mother had behavior problems 0.08010.1001 Brother/sister had behavior problems 0.11010.1401 Number of children 1.700151.85015 Married 0.60010.5001 Cohabitating 0.04010.0301 Widowed 0.01010.0201 Divorced 0.13010.1801 Separated 0.03010.0501 Never married* 0.20010.2101 Completed graduate degree 0.10010.1001 Had some graduate studies 0.04010.0401 Completed college 0.17010.1501 Had some college education 0.33010.3601 Completed high school 0.23010.2301 Not completed high school* 0.13010.1101 Excellent health 0.35010.3301 Very good health 0.33010.3101 Good health 0.22010.2401 Fair health 0.07010.0901 Poor health* 0.02010.0301 Dependent variables Current employment (full time/part time) 0.89010.8601 Currently working full time (35+ hours a week) 0.86010.7201 Alcohol variables Alcohol abuse and/or dependence prior to the last 12 months 0.43010.2301 Alcohol abuse and/or dependence in last 12 months 0.13010.0601 Number of drinks consumed in the last 12 months 257.1701095091.87010950 Number of days drank 5+ drinks in the last 12 months 22.2403654.920365 Number of days drunk in the last 12 months 7.8203653.580365 Average daily volume of ethanol consumed in the last 12 months 0.600139.200.20076.82 Instrumental variables Father/mother was ever an alcoholic/problem drinker 0.20010.2701 Brother/sister was ever an alcoholic/problem drinker 0.18010.2201 Omitted dummy variables Male (n=8,673) Female (n=9,355) 27

PAGE 42

an incentive payment of $30, resu lting in a substantial improve ment in the response rate. Because of this, as well as other impr ovements in data collection quality control procedures implemented in 2002, NSDUH data ar e not necessarily co mparable with the NHSDA data that preceded them. The pub lic use file contains data on 54,079 respondents for 2002 and 55,230 respondents for 2003. The data are not longitudinal, but the two cross sections are pooled for this analysis. Like NESARC, the NSDUH data include a variety of demographic variables, alcohol consumption measures, and employment measures, as well as several potential instrumental variables. Another advantage of NSDUH is that it used self-administered questionnaires to collect the al cohol consumption data. Zark in et al. (1998 b) suggested that this will help minimize systematic misreporting compared to questionnaires administered by interviewers. 4.2.2.1. Dependent Variables Two binary indicators for past week em ployment status, one for any employment and the other for working full time, are used. 4.2.2.2. Alcohol Variables A variety of alcohol use measures are re ported in the NSDUH. The ones studied here are a binary indicator of whether the respondent had symptoms of alcohol abuse and/or dependence (based on DSM-IV) in the pa st year, along with the number of days alcohol was consumed in the last year, the num ber of days alcohol was consumed in the 28

PAGE 43

last month, and the number of days the respondent consumed five or more alcoholic drinks in the last month. 4.2.2.3. Instrumental Variables Examples of potential instrumental vari ables in the NSDUH include the number of friends who drink at all and who get dr unk at least once a wee k, along with indicators for serious arguments in the respondents household, whether the respondents religious beliefs are an important part of his/her life, whether these beliefs influence his/her decisions, whether it is importa nt that his/her friends sh are these beliefs, and the respondents opinion about the risk of using alcohol. The intuition behind using these variables as instruments is not necessarily as compelling as for the family alcohol consumption history variables in the NESARC. However, tests of instrument strength and overidentification were used to establis h empirically valid combinations. The two chosen were a binary indicator of whether the responde nt thinks there is no or slight risk, rather than moderate or great risk, from consuming 4 alcoho lic drinks nearly every day and a binary indicator of whether the respondents religious belie fs are an important, rather than an unimportant part of his/her life. 4.2.2.4. Exogenous Variables 29 Similar to the model used to analyze th e NESARC, the set of exogenous variables in Model 1 includes two age dummy variable s, three race dummy variables, and two MSA residence dummy variables. The two age indicators are Age between 30 and 34 years old and Age between 35 and 49 years old with Age between 26 and 29 years old

PAGE 44

as the omitted category. The three race indicators are White Black and Hispanic with Other races as the omitted category. The two MSA residence indicators are for Living in an MSA with one million or more persons and Living in an MSA with fewer than 1 million persons with Not living in an MSA as the omitted category. Model 2 contains the Model 1 variables, the number of children, a nd three dummy variable s for marital status, Married Widowed and Divorced or separated with Never married as the omitted category. Model 3 adds three dummy va riables for educational attainment, Completed undergraduate/graduate study Had some college education and Completed high school with Not completed high school as the omitted category. Model 4 further contains four dummy variables for health status, Excellent Very good Good and Fair, with Poor as the omitted category. The expected signs of the co efficients are analogous to those discussed in Section 4.2.1.4. 4.2.2.5. Sample Restrictions Because only categorical information on age is available in the NSDUH and 50 64 year olds are grouped into the same category, the age range is restricted to 26 years old. The pooled 2002 and 2003 NSDUH samples consist of 12,046 men and 11,779 women. 4.2.2.6. Sample Descriptive Statistics 30 Table 3 summarizes sample descript ive statistics from the pooled 2002 and 2003 NSDUH data set. The majority of th e respondents are white and 35 years old. The largest percentage live in an MSA w ith population over 1 million and have very

PAGE 45

Table 3. Sample descriptive statistics from NSDUH Variable MeanMinMaxMeanMinMax Exogenous variables Age between 26 and 29 years old* 0.18010.1701 Age between 30 and 34 years old 0.23010.2201 Age between 35 and 49 years old 0.59010.6101 White 0.71010.7001 Black 0.10010.1301 Hispanic 0.13010.1101 Other races* 0.06010.0601 Living in a MSA with 1 million or more persons 0.38010.3701 Living in a MSA with fewer than 1 million persons 0.37010.3701 Not living in MSA* 0.25010.2601 Number of children aged <18 in household 1.05031.1603 Married 0.62010.6001 Widowed 0.00010.0101 Divorced or separated 0.13010.1901 Never married* 0.24010.2101 Completed undergraduate/graduate stud y 0.29010.3101 Had some college education 0.24010.2901 Completed high school 0.33010.3001 Not completed high school* 0.14010.1001 Excellent health 0.28010.3001 Very good health 0.41010.4001 Good health 0.25010.2401 Fair health 0.05010.0601 Poor health* 0.01010.0101 Dependent variables Employment (full time/part time) last week 0.94010.9301 Working full time last week 0.89010.7301 Alcohol variables Alcohol abuse and/or dependence in the last 12 months 0.13010.0601 Number of days drinking alcohol in the last 12 months 82.07036548.360365 Number of days drinking alcohol in the last month 6.380303.60030 Number of days drank 5+ alcoholic drinks in the last month 2.130300.74030 Instrumental variables Perceived risk (=1 if there is a slight risk or no risk 0.10010.0401 when people have 4-5 drinks of an alcoholic beverage nearly every day, =0 else) Religious influence (=1 if his/her religious beliefs are 0.71010.7901 an important part of his/her life, =0 else) *O mitted dummy variables Male (n=12,046) Female (n=11,779) 31

PAGE 46

good health. Eighty-nine percent of males a nd seventy-three percent of females were employed full time the previous week. As in the NESARC, alcohol prevalence is higher among males than among females. Around 10 pe rcent of the male sample, but only 4 percent of the female sample, th inks that there is no more th an a slight risk of physical and other harm from having four or five dr inks nearly every day. Roughly 71 percent of males and 79 percent of females report that relig ious beliefs are an important part of their lives. 4.2.3. NELS:88 The National Education Longitudinal St udy of 1988 (NELS:88), from the U.S. Department of Education's National Center for Education Statistics (NCES), is a longitudinal survey consisting of a base year and four follow-ups. Information on obtaining the data is available at http://nces.ed.gov/surveys/nels88/ In 1988, the first survey was conducted using a nationally representative sample of eighth-graders. These respondents were then re-surveyed in 1990, 1992, 1994 and 2000. The base year and first and second follow-up surveys each consist of about 27,000 respondents. The third follow-up survey c onsists of approximately 15,000 respondents. The fourth follow-up samples 12,144 respondents who are by then ages 2528. 32 Data from the fourth follow-up interview in 2000 will be used in this study since the majority of the respondents had already completed postsecondary education, started or even changed careers, and started to form families. However, some information from the first and second follow-ups will also be used. Relative to the NESARC and the NSDUH, the disadvantages of the NELS:88 are that it contains very few exogenous

PAGE 47

variables to serve as controls and the sample size is relatively small. However, an advantage is that the longitudinal aspect of the data provides pot ential instrumental variables. The most relevant of these are measures of previous alcohol use, which is likely to be correlated with current alcohol us e but has no obvious reason to be correlated with current employment, at least once other human capital measures that might be both affected by previous drinking and determinants of current employment are held constant. 4.2.3.1. Dependent Variables As with the NESARC and the NSDUH, two binary indicators for current employment status, any current employment and current full-time employment, are specified. 4.2.3.2. Alcohol Variables There are only two measures of current alcohol use reporte d in the 2000 NELS:88 survey. These are the number of occasions th at the respondent drank alcohol in the last 30 days, and drank five or more alcoholic dri nks in a row, over the previous two weeks. Both are used in the analysis. 4.2.3.3. Instrumental Variables 33 Examples of potential instruments in NELS:88 include relig ious beliefs, the frequency of participation in religious servi ces, the perspective of friends about alcohol use, and alcohol use during the 10 th and 12 th grade. Based on the same types of specification tests used for the NESARC and the NSDUH, namely tests for joint

PAGE 48

significance and overidentificati on, two variables were chosen as instruments, indicators of whether the respondent drank on 20 or more occasions in the last 30 days during 10th grade and whether the respondent drank 5 or mo re drinks in a row on at least 6 occasions in the last two weeks during 12th grade. Information on both of these measures is reported in categorical form, which is conve rted to binary indicators for use in the analysis. The cutoff thresholds for these variables represent the 95 th and 99 th percentiles of the distributions for males and females, respectively. 4.2.3.4. Exogenous Variables The set of exogenous variable s in Model 1 includes age, age squared, and four race dummy variables. 2 The four race indicators consist of White Black Native American/Alaska Native, and Asian/Pacific Islander with Hispanic as the omitted category. Model 2 contains the same set of e xogenous variables as Model 1, plus the number of children and stepchildren in the household, and five indicators for marital status, Married Cohabitating Widowed Separated and Divorced with Never married as the omitted category. Model 3 includes th e variables in Model 2 plus five dummy variables for educational attainment, Ph.D. or Professional Doctorate Masters degree Bachelor degree Had some college education and Completed high school with Not completed high school as the omitted category. No indicators of health status are available, so Model 4 cannot be estimated. The expected coefficien t signs are again the same as those described in Section 4.2.1.4. 34 2 Note that there are no MSA residence dummy variables available in NELS:88.

PAGE 49

4.2.3.5. Sample Restrictions After excluding respondents who report school enrollment or keeping house, the NELS:88 samples consist of 3,876 men and 4,638 women. 4.2.3.6. Sample Descriptive Statistics Table 4 summarizes the NELS:88 sample descriptive statistics. About three quarters of the sample is white. Ninety-two percent of males and eighty-three percent of females are currently working full time or part time. Less than one percent of respondents had completed a Ph.D. or professi onal doctorate degree, but only about three percent did not complete high school. The number of times the respondent drank alcohol in the last 30 days is about six for males but only three for females, while the number of times the respondent drank five or more drinks in a row over the last two weeks is close to one for males but only 0.30 for females. Around 2 percent of male respondents and 1 percent of female respondents drank on 20 or more occasions in the last 30 days when they were in 10 th grade. Finally, about 7 percent of male respondents and 2 percent of female respondents drank 5 or more drinks in a row at least 6 times in the last two weeks when they were in 12 th grade. 35

PAGE 50

Table 4. Sample descriptive statistics from NELS:88 Variable MeanMinMaxMeanMinMax Exogenous variables Age 26.39252826.272528 White 0.75010.7101 Black 0.07010.0901 Native American/Alaska Native 0.01010.0101 Asian/Pacific Islander 0.07010.0701 Hispanic* 0.11010.1101 Household number of children/stepchildren 0.40070.6906 Married 0.34010.4501 Cohabitating 0.01010.0101 Widowed 0.00000.0001 Separated 0.01010.0101 Divorced 0.04010.0501 Never married* 0.60010.4801 Ph.D. or Professional Doctorate 0.00010.0001 Master degree 0.01010.0101 Bachelor degree 0.33010.3701 Had some college education 0.15010.1901 Completed high school 0.48010.4001 Not completed high school* 0.03010.0301 Dependent variables Current employment (full time/part time) 0.92010.8301 Currently working full time 0.85010.7101 Alcohol variables Number of occasions drank alcohol in the last 30 days 5.880302.92030 Number of occasions drank 5+ alcoholic drinks in a row 0.960100.30010 over the last two weeks Instrumental variables Drinking at 10th grade (=1 if the respondent drank 0.02010.0101 on 20+ occasions in the last 30 days during 10th grade, =0 else) Drinking at 12th grade (=1 if the respondent drank 0.07010.0201 5+ drinks in a row at least 6 times in the last two weeks during 12th grade, =0 else) Omitted dummy variables Male (n=3,876) Female (n=4,638) 36

PAGE 51

5. Estimation Results 5.1. Results from NESARC 5.1.1. Alcohol Abuse and/or Dependence Table 5 summarizes the 2SLS and OLS regr ession estimates of the effects of past year and prior alcohol abuse and/or dependen ce on the probability of current full or part time employment for males. In the first stage regressions, the t tests for individual significance and the F tests for joint significance show th at both instrumental variables are highly significant and positively relate d to alcohol abuse and/or dependence, indicating that the models are overidentified. The tests of ove ridentification fail to reject the null hypothesis that the instrumental variables are uncor related with the residuals in the employment equations. This suggests that the identification restrictions are valid, and therefore 2SLS will produce consistent estima tes of the causal effect of alcohol abuse and/or dependence on employment. For all four model specifications, coefficien ts of the instruments in the first stage regressions imply that a respondent with an al coholic father/mother or brother/sister is about 11 percentage points more likely to develop symptoms of alcohol abuse and/or dependence prior to the last 12 months, and thr ee to five percentage points more likely to develop symptoms of alcohol abuse and/or de pendence in the last 12 months. The 2SLS estimates in Model 1 imply reductions in the probability of current employment of 10.2 percentage points from alcohol abuse and/or dependence occurring prior to the past year 37

PAGE 52

38 Table 5. Re g ression estimates of binar y alcohol measures on the p robabilit y of current em p lo y ment (full time/ p art time): the male sam p le from NESARC Alcohol variables -0.1022**-0.0169**-0.0864*-0.0141**-0.0413-0.0144**-0.0122-0.0050 (0.0496)(0.0069)(0.0509)(0.0069)(0.0506)(0.0068)(0.0483)(0.0064) Alcoholic father/mother (1 st stage) 0.1163*** 0.1131*** 0.1124*** 0.1089*** (0.0136) (0.0136) (0.0137) (0.0136) Alcoholic brother/sister (1 st stage) 0.1199*** 0.1159*** 0.1156*** 0.1134*** (0.0146) (0.0146) (0.0146) (0.0146) F-statistic (IV's in the 1 st stage) 87.55*** 81.58*** 80.43*** 76.91*** Overidentification test (p-value) 0.8376 0.8941 0.8441 0.6718 Hausman Statistic -1.7364* -1.4347 -0.5356 -0.1503 -0.2666**-0.0229**-0.2368*-0.0095 -0.1186-0.0088 -0.0470-0.0052 (0.1328)(0.0100)(0.1439)(0.0100)(0.1422)(0.0099)(0.1357)(0.0092) Alcoholic father/mother (1 st stage) 0.0545*** 0.0508*** 0.0502*** 0.0487*** (0.0095) (0.0095) (0.0095) (0.0095) Alcoholic brother/sister (1 st stage) 0.0329*** 0.0287*** 0.0279*** 0.0271*** (0.0102) (0.0101) (0.0102) (0.0101) F-statistic (IV's in the 1 st stage) 26.42*** 22.21*** 21.41*** 20.16*** Overidentification test (p-value) 0.8256 0.7997 0.9985 0.7265 Hausman Statistic -1.8409* -1.5841 -0.7747 -0.3090 **Statistically significant at 5%. *Statistically significant at 10%. Male (n=8,673) Model 1 (Total Effect) 2SLSOLS Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. Model 1 contains general exogenous variables. Model 4 contains those variables in Model 3 plus health status indicators. See Appendix A1and A2 for full estimation results. Alcohol abuse and/or de p p endence rior to the last 12 months Alcohol abuse and/or de p endence in the last 12 months Standard errors are in parentheses. ***Statistically significant at 1%.

PAGE 53

and 26.7 percentage points from past year al cohol abuse and/or dependence. All else equal, the likelihood of current employment for an individual with prior and past year alcohol abuse and/or depende nce problems is, on average, 11 percent (0.10/0.89) and 30 percent (0.27/0.89) less than for an individual without prior and past year alcohol abuse and/or dependence problems. 3 Both of these estimates are statistically significant at the five percent level. Effects of past year alcohol abuse and/or dependence are larger than those of prior alcohol abuse and/or dependen ce in both the 2SLS and OLS models. This is expected, since individuals who previously abused or had been dependent on alcohol might have stopped drinking in the past year and thus have fewer limitations preventing them from working currently. These estimates represent the total effe ct of alcohol abuse and/or dependence on employment propensities, including indirect effects through human capital variables not held constant in Model 1. The OLS estimates are negative, but much smaller than those from 2SLS. This implies that bias dampening the negative effects of alcohol use, which arises from reverse causation (i.e. the income effect runni ng from employment to alcohol use) and measurement error, outweighs bias in the opposite direction from unobserved heterogeneity. In Model 1, Hausman tests reject the null hypothesis that alcohol use is exogenous, meaning that the alcohol vari ables are endogenous with respect to employment and the 2SLS estimates should thus be used to make inferences about causal effects of past alcohol abuse and/or de pendence on the probability of current employment. In Model 2, which adds the number of child ren and marital status as explanatory 39 3 To obtain the percentage effect, the coefficient is divided by the sample mean of the dependent variable.

PAGE 54

variables, the effects become slightly smaller and less si gnificant. The effects then become much smaller and lose significance in Models 3 and 4, which also control for educational attainment and health status, respectively. This implies that there is no significant direct effect of past alcohol abuse and/or dependence on employment propensity. In addition, Hausman statistics are no longer statistically significant. Therefore, the results from Table 5 suggest that, for males, past alcohol abuse and/or dependence does not dir ectly affect current employme nt, but indirectly affects current employment through human capital com ponents, specifically health status and educational attainment. In pa rticular, the estimates in M odel 4 are much smaller than those in Model 3, implying that the indirect e ffects through health capit al are substantial. This is not surprising, since health status aff ects the capacity to work as well as the ability to accumulate other human capital components. Complete second stage regression results in Appendices A1 and A2 show that the educational attainment and health status co efficients are positive and statistically significant in Model 4, supporting the above conclusion. Appendices A1 and A2 also show that the signs of the coefficients of the other covariates are as hypothesized in Section 4. 40 Table 6 is identical to Table 5, except that the dependent variab le is an indicator of working full time. Since the first stage re gressions are exactly the same as those in Table 5, the only difference between the ta bles is in the second stage drinking coefficients. The 2SLS estimates in Model 1 are similar to those in Table 5. They imply that prior and past year alcohol abuse and/or depende nce significantly reduce the probability of currently working full time by 12.3 and 30.4 percentage points,

PAGE 55

41 Table 6. Re g ression estimates of binar y alcohol measures on the p robabilit y of currentl y workin g full time (35+ hours a week): the male sam p le from NESARC Alcohol variables -0.1226**-0.0138*-0.1080*-0.0107 -0.0521-0.0113 -0.0210-0.0012 (0.0564)(0.0079)(0.0576)(0.0078)(0.0571)(0.0077)(0.0554)(0.0073) Alcoholic father/mother (1 st stage) 0.1163*** 0.1131*** 0.1124*** 0.1089*** (0.0136) (0.0136) (0.0137) (0.0136) Alcoholic brother/sister (1 st stage) 0.1199*** 0.1159*** 0.1156*** 0.1134*** (0.0146) (0.0146) (0.0146) (0.0146) F-statistic (IV's in the 1 st stage) 87.55*** 81.58*** 80.43*** 76.91*** Overidentification test (p-value) 0.7544 0.6572 0.6996 0.8144 Hausman Statistic -1.9479* -1.7024* -0.7224 -0.3611 -0.3038**-0.0351***-0.2744*-0.0188*-0.1271-0.0182 -0.0485-0.0140 (0.1502)(0.0113)(0.1625)(0.0113)(0.1601)(0.0112)(0.1556)(0.0106) Alcoholic father/mother (1 st stage) 0.0545*** 0.0508*** 0.0502*** 0.0487*** (0.0095) (0.0095) (0.0095) (0.0095) Alcoholic brother/sister (1 st stage) 0.0329*** 0.0287*** 0.0279*** 0.0271*** (0.0102) (0.0101) (0.0102) (0.0101) F-statistic (IV's in the 1 st stage) 26.42*** 22.21*** 21.41*** 20.16*** Overidentification test (p-value) 0.4588 0.3985 0.5561 0.7492 Hausman Statistic -1.7934* -1.5764 -0.6817 -0.2226 **Statistically significant at 5%. *Statistically significant at 10%. Alcohol abuse and/or de p p endence rior to the last 12 months Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. Model 4 contains those variables in Model 3 plus health status indicators. 2SLSOLS 2SLS Standard errors are in parentheses. ***Statistically significant at 1%. Alcohol abuse and/or de p endence in the last 12 months See Appendix A3and A4 for full estimation results. Male (n=8,673) Model 1 (Total Effect) 2SLSOLS 2SLSOLS OLS Model 3 Model 4 (Direct Effect) Model 2

PAGE 56

respectively. In percentage terms, the likelihood of current ly working full time for an individual with prior and past year alc ohol abuse and/or dependence problems is, on average, 14 percent (0.12/0.86) and 35 percen t (0.30/0.86) less than for an individual without prior and past year alcohol abuse and/or dependence problems. The Hausman statistics in Model 1 are stat istically significant, suggesting that the alcohol variables are endogenous with respect to labor market outco mes, and the 2SLS estimates in Model 1 should be used to make inferences about cau sal effects of past alcohol abuse and/or dependence on the probability of currently working full time. As expected, employment effects in Tabl e 6 are larger than those in Table 5, because problem drinking can cause workers to drop out of full-time jobs and work part time instead. As in Table 5, the magnitudes of the OLS estimates are smaller than those from 2SLS. After human capital variables are added into th e model, the 2SLS estimates became smaller and insignificant, suggesting th e same conclusion as that from Table 5. Complete second stage regression results in Appendices A3 and A4 show that the educational attainment and health status co efficients are again positive and statistically significant in Model 4. 42 Table 7 summarizes the 2SLS and OLS regr ession results of the effect of prior and past year alcohol abuse and/or dependen ce on the probability of current employment for females. Similar to the male sample, the t tests for individual significance and F tests for joint significance of the instrumental variables, together with the tests of overidentification, indicate that 2SLS produces consistent causal effect estimates for all four models. Both instruments are positivel y correlated with alcohol abuse and/or dependence in the first stage regressions, and their coefficients slightly decrease as

PAGE 57

43 Table 7. Re g ression estimates of binar y alcohol measures on the p robabilit y of current em p lo y ment (full time/ p art time): the female sam p le from NESARC Alcohol variables -0.2342***-0.0084 -0.1454*-0.0061 -0.0948-0.0103 0.0006-0.0074 (0.0777)(0.0089)(0.0746)(0.0088)(0.0734)(0.0088)(0.0685)(0.0082) Alcoholic father/mother (1 st stage) 0.0861*** 0.0866*** 0.0865*** 0.0858*** (0.0101) (0.0101) (0.0101) (0.0101) Alcoholic brother/sister (1 st stage) 0.0615*** 0.0637*** 0.0642*** 0.0638*** (0.0110) (0.0110) (0.0110) (0.0110) F-statistic (IV's in the 1 st stage) 66.08*** 67.93*** 68.01*** 67.05*** Overidentification test (p-value) 0.7767 0.9969 0.7349 0.9985 Hausman Statistic -2.9251*** -1.8793* -1.1599 0.1176 -0.7488***-0.0243 -0.4936*-0.0149 -0.3504-0.0122 0.0020-0.0150 (0.2794)(0.0155)(0.2692)(0.0154)(0.2678)(0.0153)(0.2426)(0.0142) Alcoholic father/mother (1 st stage) 0.0302*** 0.0293*** 0.0288*** 0.0289*** (0.0058) (0.0058) (0.0058) (0.0058) Alcoholic brother/sister (1 st stage) 0.0107* 0.0108* 0.0101 0.0101 (0.0064) (0.0063) (0.0064) (0.0064) F-statistic (IV's in the 1 st stage) 17.82*** 16.97*** 16.04*** 16.10*** Overidentification test (p-value) 0.3826 0.6812 0.9813 0.9969 Hausman Statistic -2.5972*** -1.7809* -1.2650 0.0701 **Statistically significant at 5%. *Statistically significant at 10%. Standard errors are in parentheses. Model 2 Model 3 Alcohol abuse and/or de p p endence rior to the last 12 months Alcohol abuse and/or de p endence in the last 12 months 2SLSOLS 2SLS 2SLSOLS Model 4 (Direct Effect) ***Statistically significant at 1%. Model 1 (Total Effect) OLS 2SLS Female (n=9,355) OLS See Appendix A5and A6 for full estimation results. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. Model 4 contains those variables in Model 3 plus health status indicators.

PAGE 58

human capital covariates are a dded. An individual with eith er an alcoholic father/mother or brother/sister is six to nine percentage points more li kely to develop symptoms of alcohol abuse and/or dependence prior to the last 12 months, and one to three percentage points more likely to develop symptoms of alcohol abuse and/or dependence in the last 12 months. The 2SLS estimates in Model 1, which re present the total employment effect of alcohol abuse and/or dependence for females, show that prior and past year alcohol abuse and/or dependence significan tly reduce the probability of current employment by 23.4 and 74.9 percentage points, resp ectively. Once again, the eff ects of past year alcohol abuse and/or dependence are larger than thos e of prior alcohol abuse and/or dependence using both 2SLS and OLS. However, the 2SLS effects seem implausibly large, as they imply reductions in current employment of 27 percent (0.23/0.86) from alcohol problems prior to the past year and 87 percent (0.75/0.86) from past y ear alcohol problems. Their standard errors are also large, meaning that statistical power is inadequate to identify effects of reasonable size. OLS estimates are also negative, but ar e statistically insignificant and much smaller than are those from 2SLS. Hausma n tests are significant in Model 1, again suggesting that the alcohol variables are e ndogenous with respect to the probability of current employment and 2SLS estimates should be used to infer causal effects of past alcohol abuse and/or dependence. 44 After adding other covariates, the eff ects become smaller and insignificant as shown in Models 2 and 3 for both 2SLS and OLS methods. The estimates even become positive, though statistically insignificant, in Model 4. In addition, the Hausman

PAGE 59

statistics are no longer statistic ally significant. Thus, these results for females suggest the same conclusion as that for males in Table 5, namely negative employment effects of problem drinking that occur entirely through human capital accumulation. Table 8 summarizes 2SLS and OLS regressi on results for the e ffect of prior and past year alcohol abuse and/or dependence on the probability of working full time for females. Once again, the first stage regressi ons are the same as those in the preceding table. The 2SLS estimates in Model 1 imply th at prior and past year alcohol abuse and/or dependence significantly reduce the probability of currently working full time by 35.4 and 120.8 percentage points, respectively. The full-time effects in Table 8 are larger than those for any employment in Table 7, and the effect of past year alcohol abuse and/or dependence is again much higher than that of alcohol abuse and/or dependence prior to the last 12 months. However, it is clear, particularly for the past ye ar measure, that the 2SLS estimates are again implausibly large. The accompanying standard errors continue to be extremely large, relative to the fu ll-time employment rate of 72 percent. For example, to be statistically significant, th e 2SLS estimate for past year drinking would have to imply a reduction in full-time employ ment propensity of n early 100 percent. 45 The 2SLS estimates become slightly smaller and insignificant when the human capital variables are added to the regression, implying that prior and past year alcohol abuse and/or dependence have no significant di rect effects on the pr obability of working full time for females. Moreover, complete s econd stage regression re sults in Appendices A7 and A8 show that all the estimated human capital variable coefficients in Model 4 are statistically significant and positively affect employment propensity for both the 2SLS and OLS approaches, implying that prior and past year alcohol abuse and/or dependence

PAGE 60

46 Table 8. Re g ression estimates of binar y alcohol measures on the p robabilit y of currentl y workin g full time (35+ hours a week): the female sam p le from NESARC Alcohol variables -0.3544***0.0190 -0.2490**0.0110 -0.1727*0.0057 -0.08210.0090 (0.1040)(0.0117)(0.0990)(0.0116)(0.0967)(0.0114)(0.0932)(0.0111) Alcoholic father/mother (1 st stage) 0.0861*** 0.0866*** 0.0865*** 0.0858*** (0.0101) (0.0101) (0.0101) (0.0101) Alcoholic brother/sister (1 st stage) 0.0615*** 0.0637*** 0.0642*** 0.0638*** (0.0110) (0.0110) (0.0110) (0.0110) F-statistic (IV's in the 1 st stage) 66.08*** 67.93*** 68.01*** 67.05*** Overidentification test (p-value) 0.5066 0.2411 0.1225 0.1762 Hausman Statistic -3.6144*** -2.6440*** -1.8584* -0.9839 -1.2083***0.0138 -0.9339**0.0014 -0.7258**0.0063 -0.38810.0039 (0.3889)(0.0204)(0.3721)(0.0202)(0.3651)(0.0200)(0.3362)(0.0193) Alcoholic father/mother (1 st stage) 0.0302*** 0.0293*** 0.0288*** 0.0289*** (0.0058) (0.0058) (0.0058) (0.0058) Alcoholic brother/sister (1 st stage) 0.0107* 0.0108* 0.0101 0.0101 (0.0064) (0.0063) (0.0064) (0.0064) F-statistic (IV's in the 1 st stage) 17.82*** 16.97*** 16.04*** 16.10*** Overidentification test (p-value) 0.9289 0.5858 0.3067 0.2777 Hausman Statistic -3.1470*** -2.5174** -2.0085** -1.1678 **Statistically significant at 5%. *Statistically significant at 10%. Model 1 (Total Effect) Model 4 (Direct Effect) Standard errors are in parentheses. Alcohol abuse and/or de p p endence rior to the last 12 months 2SLS 2SLS OLS 2SLS OLS Alcohol abuse and/or de p endence in the last 12 months OLS OLS 2SLS Model 3 ***Statistically significant at 1%. Female (n=9,355) Model 2 See Appendix A7and A8 for full estimation results. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. Model 4 contains those variables in Model 3 plus health status indicators.

PAGE 61

have significant indirect effects on the pr obability of working full time for females through human capital components. OLS estimat es are positive in al l models, suggesting that income effects cause OLS to underestim ate the negative effect of problem drinking on the probability of working full time for females. 5.1.2. Frequencies of Alcohol Consumption Table 9 summarizes the results from th e current employment regressions for males in which the four continuous alcohol measures are used. The highly significant t and F test statistics in the first stage regres sions, combined with high overidentification test p -values, show that the instrumental variables are appropriate for making inferences about causal effects of alcohol consumption. 47 As before, estimates in Model 1 represent to tal effects. Both 2SLS and OLS estimates show that alcohol consumption negatively and significantly affects the probability of current employment. 2SLS effects are larger than those from OLS, but the Hausman test rejects exogeneity of alcohol use for only one of the four drinking variables, and never does so in Models 2. Thus the relevant causal effect interpreta tions are those from OLS. Drinking one alcoholic beverage each week in the previous 12 months (52 weeks) reduces the probability of current employ ment by 0.2 percentage points (-0.000031 x 52), which is equivalent to 0.2 percent (-0 .000031 x 52 / 0.89). Drinking five or more alcoholic beverages once each week in the pr evious 12 months reduces the probability of current employment by 1.6 percentage poi nts (-0.000303 x 52), or 1.8 percent (-0.000303 x 52 / 0.89). Getting drunk once a week in the last 12 months reduces the probability of current employment by 2.1 percentage points (-0.000402 x 52), or 2.3 percent

PAGE 62

48 Alcohol variables -0.000176*-0.000031***-0.000157-0.000025***-0.000086-0.000021***-0.000046-0.000018*** (0.0001)(0.0000) (0.0001)(0.0000) (0.0001)(0.0000) (0.0001)(0.0000) Alcoholic father/mother (1 st stage) 89.9339*** 81.8730*** 77.0466*** 74.4399*** (16.1563) (16.0734) (16.0680) (16.0509) Alcoholic brother/sister (1 st stage) 24.6531 15.3125 9.8275 8.5371 (17.3067) (17.2308) (17.2255) (17.2032) F-statistic (IV's in the 1 st stage) 18.84*** 14.81*** 12.65*** 11.74*** Overidentification test (p-value) 0.5047 0.4893 0.7912 0.8274 Hausman Statistic -1.5722 -1.2824 -0.6031 -0.2714 -0.001325**-0.000303***-0.001216-0.000236***-0.000670-0.000188***-0.000297-0.000157*** (0.0007)(0.0001) (0.0007)(0.0001) (0.0008)(0.0001) (0.0008)(0.0000) Alcoholic father/mother (1 st stage) 11.2847*** 10.2260*** 9.4532*** 9.1054*** (1.8583) (1.8459) (1.8414) (1.8387) Alcoholic brother/sister (1 st stage) 5.9464*** 4.7028** 3.7983* 3.6102* (1.9907) (1.9788) (1.9740) (1.9707) F-statistic (IV's in the 1 st stage) 27.62*** 21.55*** 17.56*** 16.26*** Overidentification test (p-value) 0.7495 0.7127 0.9299 0.7556 Hausman Statistic -1.5631 -1.3252 -0.5997 -0.1798 -0.003925*-0.000402***-0.003409-0.000338***-0.001703-0.000316***-0.000596-0.000250*** (0.0020)(0.0001) (0.0021)(0.0001) (0.0021)(0.0001) (0.0019)(0.0001) Alcoholic father/mother (1 st stage) 3.4518*** 3.2780*** 3.2010*** 3.1327*** (1.0809) (1.0811) (1.0825) (1.0826) Alcoholic brother/sister (1 st stage) 2.6419** 2.4407** 2.3568** 2.3166** (1.1579) (1.1589) (1.1604) (1.1603) F-statistic (IV's in the 1 st stage) 9.54*** 8.38*** 7.87*** 7.56*** Overidentification test (p-value) 0.9686 0.9354 0.9330 0.7005 Hausman Statistic -1.7250* -1.4454 -0.6720 -0.1787 -0.077235*-0.009342***-0.069738-0.007761***-0.038149-0.006925***-0.017712-0.004498*** (0.0417)(0.0015) (0.0458)(0.0015) (0.0469)(0.0015) (0.0448)(0.0014) Alcoholic father/mother (1 st stage) 0.1977*** 0.1807*** 0.1686*** 0.1628*** (0.0617) (0.0616) (0.0616) (0.0616) Alcoholic brother/sister (1 st stage) 0.0917 0.0739 0.0593 0.0535 (0.0661) (0.0660) (0.0661) (0.0660) F-statistic (IV's in the 1 st stage) 7.27*** 5.79*** 4.79*** 4.39** Overidentification test (p-value) 0.7100 0.6900 0.9075 0.7715 Hausman Statistic -1.6312 -1.3549 -0.6660 -0.2949 See Appendix A9-A12 for full estimation results. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marit al status indicators. Model 3 contains those variables in Model 2 plus educational atta inment indicators. Model 4 contains those variables in Model 3 plus health status indicators. Table 9. Regression estimates of continuous alcohol measures on the probability of current employment (full time/part time): th e male sample from NESARC Number of days drank 5+ drinks in the last 12 months 2SLS OLS OLS 2SLS OLS 2SLS OLS Standard errors are in parentheses. ***Statistically significant at 1%. **Statistically significant at 5%. *Statistically significant at 10%. Number of drinks consumed in the last 12 months Number of days drunk in the last 12 months Average daily volume of ethanol consumed in the last 12 months 2SLS Male (n=8,673) Model 1 (Total Effect) Model 2 Model 3 Model 4 (Direct Effect)

PAGE 63

(-0.000402 x 52 / 0.89). As expected, the estimates b ecome larger, in percentage terms, as drinking becomes heavier, assuming that many male respondents require more than five drinks in one sitting to consider themselves drunk. Since alcohol tolerance levels vary across individuals, the number of days drunk in the last 12 months might well be the most relevant of these three drinking measures. Drinking an additional ounce of ethanol daily in the previous 12 months reduces the probability of current employment by 0.9 pe rcentage points, or one percent. A 12ounce bottle of beer, five ounce glass of wi ne, and a 1.5 shot of 80-proof liquor each contain about 0.6 ounces of ethanol. Therefor e, an additional one of these standard alcoholic drinks each day would reduce the probability of current employment by 0.6 percent (-0.009342 x 0.6 / 0.89). This is a sm aller effect than implied by drinking measures above, but does not take into account the extent to which consumption is spread out over time. For instance, getting dr unk once/week and not drinking at all the remainder of the week seems likely to be mo re damaging to productivity than having one drink per day, given that the medical literature has found health bene fits to the latter, even though the number of drinks consumed weekly might be the same in each scenario. 49 As before, the negative effects of dr inking diminish in size substantially upon adding the human capital variables in Models 2, 3 and 4. However, the OLS effects are still statistically significant in all four vari ations of Model 4. The complete second stage regression results in Appendices A9 to A12 again show that all human capital components significantly and positively affect the probability of current employment for males. Overall, the implica tion is that alcohol consumpti on has both direct and indirect effects on current employment fo r males, though both are small.

PAGE 64

Table 10 summarizes the 2SLS and OLS regr ession results of the effects of the four numerical alcohol consumption measures on the probability of working full time for males. The pattern of results is quite similar to that from Table 9, with each having identical first stage regressions. Effects on working full time are again larger than those on any current employment. 2SLS effects are much larger than ar e those of OLS, but also have large standard errors. Thus Hausman tests show no statistical difference between 2SLS and OLS except for in one Model 1 specification. OLS estimates are therefore preferable to 2SLS estimates in ma king inferences about th e causal effect of alcohol consumption on the probability of working full time. Drinking one alcoholic bevera ge each week in the previ ous 12 months reduces the probability of working full time by 0.2 percentage points (-0.000044 x 52), or 0.3 percent (-0.000044 x 52 / 0.86). Drinking five or more alcoholic beverages once each week in the previous 12 months reduces the probability of working full time by 2.0 percentage points (-0.000375 x 52), or 2.3 percent (0.000375 x 52 / 0.86). Getting drunk once a week in the last 12 months reduces the proba bility of working full time by 2.8 percentage points (-0.000533 x 52) or 3.2 percent (-0.000533 x 52 / 0.86). As in Table 9, the estimates grow in magnitude as the intensity of drinking increases. Meanwhile, drinking an additional standard alcoholic beverage, i.e. 0.6 ounces of ethanol, daily in the previous 12 months reduces the probability of working full time by 0.7 percentage points (-0.011950 x 0.6), or 0.8 pe rcent (-0.011950 x 0.6 / 0.86). Tables 11 and 12 summarize the 2SLS and OL S regression results for the effect of the four numerical drinking measures on the probability of current employment and on 50

PAGE 65

51 Alcohol variables -0.000189*-0.000044***-0.000164-0.000036***-0.000070-0.000031***-0.000023-0.000028*** (0.0001)(0.0000) (0.0001)(0.0000) (0.0001)(0.0000) (0.0001)(0.0000) Alcoholic father/mother (1 st stage) 89.9339*** 81.8730*** 77.0466*** 74.4399*** (16.1563) (16.0734) (16.0680) (16.0509) Alcoholic brother/sister (1 st stage) 24.6531 15.3125 9.8275 8.5371 (17.3067) (17.2308) (17.2255) (17.2032) F-statistic (IV's in the 1 st stage) 18.84*** 14.81*** 12.65*** 11.74*** Overidentification test (p-value) 0.2359 0.1996 0.4165 0.6859 Hausman Statistic -1.3933 -1.1001 -0.3144 0.0440 -0.001491**-0.000375***-0.001376-0.000295***-0.000663-0.000237***-0.000246-0.000203*** (0.0007)(0.0001) (0.0008)(0.0001) (0.0009)(0.0001) (0.0009)(0.0001) Alcoholic father/mother (1 st stage) 11.2847*** 10.2260*** 9.4532*** 9.1054*** (1.8583) (1.8459) (1.8414) (1.8387) Alcoholic brother/sister (1 st stage) 5.9464*** 4.7028** 3.7983* 3.6102* (1.9907) (1.9788) (1.9740) (1.9707) F-statistic (IV's in the 1 st stage) 27.62*** 21.55*** 17.56*** 16.26*** Overidentification test (p-value) 0.3974 0.3336 0.5034 0.7252 Hausman Statistic -1.5084 -1.2948 -0.4710 -0.0489 -0.004573**-0.000533***-0.004089*-0.000453***-0.001964-0.000425***-0.000764-0.000357*** (0.0023)(0.0001) (0.0024)(0.0001) (0.0023)(0.0001) (0.0022)(0.0001) Alcoholic father/mother (1 st stage) 3.4518*** 3.2780*** 3.2010*** 3.1327*** (1.0809) (1.0811) (1.0825) (1.0826) Alcoholic brother/sister (1 st stage) 2.6419** 2.4407** 2.3568** 2.3166** (1.1579) (1.1589) (1.1604) (1.1603) F-statistic (IV's in the 1 st stage) 9.54*** 8.38*** 7.87*** 7.56*** Overidentification test (p-value) 0.5966 0.5226 0.6176 0.7757 Hausman Statistic -1.7417* -1.5054 -0.6619 -0.1831 -0.085961*-0.011950***-0.077880-0.009969***-0.036625-0.008936***-0.013234-0.006506*** (0.0469)(0.0017) (0.0515)(0.0017) (0.0524)(0.0017) (0.0512)(0.0016) Alcoholic father/mother (1 st stage) 0.1977*** 0.1807*** 0.1686*** 0.1628*** (0.0617) (0.0616) (0.0616) (0.0616) Alcoholic brother/sister (1 st stage) 0.0917 0.0739 0.0593 0.0535 (0.0661) (0.0660) (0.0661) (0.0660) F-statistic (IV's in the 1 st stage) 7.27*** 5.79*** 4.79*** 4.39** Overidentification test (p-value) 0.3881 0.3354 0.4915 0.7154 Hausman Statistic -1.5799 -1.3205 -0.5286 -0.1314 See Appendix A13-A16 for full estimation results. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marit al status indicators. Model 3 contains those variables in Model 2 plus educational atta inment indicators. Model 4 contains those variables in Model 3 plus health status indicators. Number of drinks consumed in the last 12 months Number of days drank 5+ drinks in the last 12 months Number of days drunk in the last 12 months Average daily volume of ethanol consumed in the last 12 months OLS Standard errors are in parentheses. ***Statistically significant at 1%. **Statistically significant at 5%. *Statistically significant at 10%. 2SLS OLS 2SLS OLS Table 10. Regression estimates of continuous alcohol measures on the probability of currently working full time (35+ hours a we ek): the male sample from NESARC 2SLS OLS Male (n=8,673) Model 1 (Total Effect) Model 2 Model 3 Model 4 (Direct Effect) 2SLS

PAGE 66

52 Alcohol variables -0.000596***-0.000015-0.000362*-0.000012-0.000229-0.000008 0.000002-0.000016 (0.0002)(0.0000) (0.0002)(0.0000) (0.0002)(0.0000) (0.0002)(0.0000) Alcoholic father/mother (1 st stage) 26.8696*** 26.9459*** 26.5282*** 26.8724*** (7.1163) (7.1137) (7.1189) (7.1210) Alcoholic brother/sister (1 st stage) 32.4049*** 33.3923*** 32.6577*** 33.4474*** (7.7555) (7.7585) (7.7736) (7.7724) F-statistic (IV's in the 1 st stage) 20.60*** 21.20*** 20.27*** 21.05*** Overidentification test (p-value) 0.7580 0.6982 0.5617 0.9999 Hausman Statistic -2.7480*** -1.7982* -1.1454 0.1007 -0.005564***-0.000304**-0.003533*-0.000250**-0.002301-0.000174 0.000016-0.000165 (0.0020)(0.0001) (0.0019)(0.0001) (0.0019)(0.0001) (0.0018)(0.0001) Alcoholic father/mother (1 st stage) 2.8946*** 2.7897*** 2.6909*** 2.6585*** (0.7489) (0.7491) (0.7488) (0.7494) Alcoholic brother/sister (1 st stage) 3.4595*** 3.4038*** 3.2268*** 3.2475*** (0.8162) (0.8170) (0.8177) (0.8180) F-statistic (IV's in the 1 st stage) 21.37*** 20.15*** 18.32*** 18.22*** Overidentification test (p-value) 0.7635 0.7064 0.5702 0.9999 Hausman Statistic -2.6880*** -1.7329* -1.1040 0.1020 -0.016028**-0.000187-0.010464-0.000174-0.006657-0.000151 0.000043-0.000150 (0.0081)(0.0001) (0.0068)(0.0001) (0.0059)(0.0001) (0.0049)(0.0001) Alcoholic father/mother (1 st stage) 1.1429* 1.0877 1.0838 1.0823 (0.6771) (0.6783) (0.6788) (0.6795) Alcoholic brother/sister (1 st stage) 1.0601 1.0221 1.0304 1.0431 (0.7379) (0.7397) (0.7412) (0.7416) F-statistic (IV's in the 1 st stage) 3.19** 2.89* 2.88* 2.90* Overidentification test (p-value) 0.9912 0.8827 0.6804 0.9992 Hausman Statistic -1.9558* -1.5057 -1.0960 0.0394 -0.206406***-0.007928**-0.123943*-0.007412**-0.075313-0.006248*0.000545-0.007025** (0.0790)(0.0034) (0.0704)(0.0034) (0.0685)(0.0034) (0.0617)(0.0031) Alcoholic father/mother (1 st stage) 0.0623** 0.0626** 0.0617** 0.0614** (0.0264) (0.0264) (0.0265) (0.0265) Alcoholic brother/sister (1 st stage) 0.1042*** 0.1067*** 0.1048*** 0.1070*** (0.0288) (0.0288) (0.0289) (0.0289) F-statistic (IV's in the 1 st stage) 11.94*** 12.25*** 11.78*** 12.06*** Overidentification test (p-value) 0.5322 0.5485 0.4722 0.9991 Hausman Statistic -2.5135** -1.6572* -1.0096 0.1229 See Appendix A17-A20 for full estimation results. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marit al status indicators. Model 3 contains those variables in Model 2 plus educational atta inment indicators. Model 4 contains those variables in Model 3 plus health status indicators. Number of days drank 5+ drinks in the last 12 months Number of days drunk in the last 12 months Average daily volume of ethanol consumed in the last 12 months ***Statistically significant at 1%. **Statistically significant at 5%. *Statistically significant at 10%. Model 1 (Total Effect) Table 11. Regression estimates of continuous alcohol measures on the probability of current employment (full time/part time): t he female sample from NESARC Number of drinks consumed in the last 12 months OLS 2SLS OLS 2SLS OLS 2SLS Model 2 Model 3 Model 4 (Direct Effect) Standard errors are in parentheses. OLS 2SLS Female (n=9,355)

PAGE 67

53 Alcohol variables -0.000850***0.000002-0.000561**-0.000007-0.000364-0.000002-0.000140-0.000009 (0.0003)(0.0000) (0.0003)(0.0000) (0.0003)(0.0000) (0.0002)(0.0000) Alcoholic father/mother (1 st stage) 26.8696*** 26.9459*** 26.5282*** 26.8724*** (7.1163) (7.1137) (7.1189) (7.1210) Alcoholic brother/sister (1 st stage) 32.4049*** 33.3923*** 32.6577*** 33.4474*** (7.7555) (7.7585) (7.7736) (7.7724) F-statistic (IV's in the 1 st stage) 20.60*** 21.20*** 20.27*** 21.05*** Overidentification test (p-value) 0.2128 0.1080 0.0653 0.1330 Hausman Statistic -2.9981*** -2.1358** -1.4236 -0.5541 -0.007938***-0.000076-0.005489**-0.000109-0.003686-0.000001-0.0014530.000011 (0.0026)(0.0002) (0.0025)(0.0002) (0.0026)(0.0002) (0.0024)(0.0002) Alcoholic father/mother (1 st stage) 2.8946*** 2.7897*** 2.6909*** 2.6585*** (0.7489) (0.7491) (0.7488) (0.7494) Alcoholic brother/sister (1 st stage) 3.4595*** 3.4038*** 3.2268*** 3.2475*** (0.8162) (0.8170) (0.8177) (0.8180) F-statistic (IV's in the 1 st stage) 21.37*** 20.15*** 18.32*** 18.22*** Overidentification test (p-value) 0.2139 0.1118 0.0685 0.1348 Hausman Statistic -2.9847*** -2.1244** -1.4444 -0.6076 -0.023566**-0.000057-0.017139*-0.000072-0.011404-0.000040-0.004938-0.000039 (0.0115)(0.0002) (0.0100)(0.0002) (0.0084)(0.0002) (0.0069)(0.0002) Alcoholic father/mother (1 st stage) 1.1429* 1.0877 1.0838 1.0823 (0.6771) (0.6783) (0.6788) (0.6795) Alcoholic brother/sister (1 st stage) 1.0601 1.0221 1.0304 1.0431 (0.7379) (0.7397) (0.7412) (0.7416) F-statistic (IV's in the 1 st stage) 3.19** 2.89* 2.88* 2.90* Overidentification test (p-value) 0.5378 0.3127 0.1535 0.1687 Hausman Statistic -2.0463** -1.7148* -1.3574 -0.7054 -0.282608***-0.004227-0.178972*-0.005630-0.106929-0.004103-0.033081-0.004792 (0.1057)(0.0045) (0.0937)(0.0044) (0.0901)(0.0044) (0.0838)(0.0042) Alcoholic father/mother (1 st stage) 0.0623** 0.0626** 0.0617** 0.0614** (0.0264) (0.0264) (0.0265) (0.0265) Alcoholic brother/sister (1 st stage) 0.1042*** 0.1067*** 0.1048*** 0.1070*** (0.0288) (0.0288) (0.0289) (0.0289) F-statistic (IV's in the 1 st stage) 11.94*** 12.25*** 11.78*** 12.06*** Overidentification test (p-value) 0.1251 0.0680 0.0455 0.1169 Hausman Statistic -2.6351*** -1.8529* -1.1423 -0.3380 Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marit al status indicators. Model 3 contains those variables in Model 2 plus educational atta inment indicators. Model 4 contains those variables in Model 3 plus health status indicators. Number of days drunk in the last 12 months Average daily volume of ethanol consumed in the last 12 months 2SLS Number of drinks consumed in the last 12 months Female (n=9,355) Model 1 (Total Effect) Standard errors are in parentheses. ***Statistically significant at 1%. **Statistically significant at 5%. *Statistically significant at 10%. See Appendix A21-A24 for full estimation results. 2SLS OLS Model 2 Model 3 Model 4 (Direct Effect) OLS OLS 2SLS OLS 2SLS Number of days drank 5+ drinks in the last 12 months Table 12. Regression estimates of continuous alcohol measures on the probability of currently working full time (35+ hours a we ek): the female sample from NESARC

PAGE 68

the probability of working full time, respectively, for females. Except for the number of days getting drunk, the statis tics in the first stage re gression and the tests of overidentification for other alcohol variables show that th e instrumental variables are adequate. Full-time effects in Table 12 are agai n larger than employment effects in Table 11. Both the 2SLS and OLS estimates in Model 1 show negative e ffects of all four alcohol measures, but the OLS estimates are less significant and much smaller than those of the 2SLS method. Hausman tests for exogeneity in Models 1 and 2 suggest that the alcohol variables are endogenous with respect to employment, but exogeneity cannot be rejected in Models 3 and 4. Therefore, the total causal effects can be interpreted from the 2SLS estimates in Model 1 while the direct causal effects can be interpreted from the OLS estimates in Model 4. However, the OLS estimates in Model 4 are very sma ll and statistically insignificant, implying that there is no di rect effect of alcohol consumption on employment propensity. The total effects from Model 1 can be interpreted as follows. Drinking one alcoholic bevera ge each week in the previ ous 12 months reduces the probability of current employment and th e probability of working full time by 3.1 percentage points (-0.000596 x 52) an d 4.4 percentage points (-0.000850 x 52), respectively. Drinking five or more alcoholic beverages once each week in the previous 12 months reduces the probability of current employment and the probability of working full time by 28.9 percentage points (-0.005564 x 52) and 41.2 percentage points 54 (-0.007938 x 52), respectively. Getting drunk once a week in the last 12 months reduces the probability of current employment and the probability of working full time by 83.3 percentage points (-0.016028 x 52) and 122.5 percentage points (-0.023566 x 52),

PAGE 69

respectively. Similar to in Tables 9 and 10, the estimates become larger as drinking becomes heavier. Drinking an additional 0.6 ounce of ethanol daily in the previous 12 months reduces the probability of current employment and th e probability of working full time by 12.4 percentage points (-0.206406 x 0.6) and 16.9 percentage points (-0.282608 x 0.6), respectively. Comparing the total effects from Tabl es 9 and 10 and Tables 11 and 12, the effects of alcohol consumption for females are much larger than those for males. Substantively, this could be because alcohol consumption tends to more quickly intoxicate females and causes medical problems to progress more rapidly in females than males, as mentioned in Section 4.1.5. Howeve r, this comparison is of limited usefulness because it is between much smaller OLS estimates in Tables 9 and 10 and 2SLS estimates in Tables 11 and 12. Using in stead the OLS estimates in Table 11, for example, would yield the conclu sion that effects on males are slightly larger than are those on females. Similar to the results of the male sample, the effects become smaller and insignificant in both 2SLS and OLS appro aches after controlling for human capital covariates. The complete second stage regr ession results in Appendices A17 to A24 show that all human capital co ntrols significantly affect employment status. The alcohol and human capital estimates and their significan ce levels imply that alcohol consumption affects employment status indirectly but not directly. All other c ovariate coefficients, again, have the signs that we re hypothesized in Section 4. 55

PAGE 70

5.2. Results from NSDUH 5.2.1. Alcohol Abuse and/or Dependence Table 13 summarizes the 2SLS and OLS regr ession results for the effect of past year alcohol abuse and/or dependence on the probability of employment (full time/part time) last week for both males and female s. In the first stage regressions, both instrumental variables are individually a nd jointly significant in all models. The overidentification statistics show that the null hypotheses of overidentification can not be rejected, again supporting the instrumentation strategy. From the first stage regressions in each model, the signs of the estimates for both instrumental variables are as expected. Alcohol abuse and/or de pendence is positively related to perceiving no more than slight risk from consuming 4-5 alcoholic drinks nearly every day, but negatively related to religious beliefs being important. A male (female) who thinks that there is no more than s light risk undertaken from consuming 4-5 alcoholic drinks nearly every da y is about 8 to 10 percentage points (6 percentage points) more likely to develop alcohol abuse and/ or dependence in the last 12 months. Respondents of both genders who think their religious beliefs are im portant are about 2 percentage points less likely to develop alcohol abuse and/or dependence. These estimated effects decrease slightly as human capital covariates are added into the models. 56 For the male sample in Model 1, past year alcohol abuse and/or dependence has a statistically significant negative total effect on the probability of past week employment using both estimation approaches. Past year alcohol abuse and/or dependence lowers the probability of employment by 19.3 percentage points or 20.5 percent (-0.1929 / 0.94) using 2SLS, but by only 3.3 percentage poi nts or 3.5 percent (-0.0325 / 0.94) using

PAGE 71

57 Table 13. Re g ression estimates of binar y alcohol measures on the p robabilit y of em p lo y ment (full time/ p art time) last week usin g NSDUH Male (n=12,046) Alcohol variables -0.1929***-0.0325***-0.1377*-0.0236***-0.0764-0.0207***-0.0689-0.0188*** (0.0677)(0.0063)(0.0743)(0.0063)(0.0788)(0.0063)(0.0793)(0.0063) Perceived risk (1 st stage) 0.1006*** 0.0923*** 0.0862*** 0.0857*** (0.0102) (0.0102) (0.0102) (0.0102) Religious influence (1 st stage) -0.0216*** -0.0151** -0.0149** -0.0145** (0.0068) (0.0068) (0.0068) (0.0067) F-statistic (IV's in the 1 st stage) 54.81*** 44.43*** 38.58*** 38.11*** Overidentification test (p-value) 0.7126 0.8943 0.9855 0.9817 Hausman Statistic -2.3815** -1.5412 -0.7088 -0.6333 Female (n=11,779) Alcohol variables -0.6529***-0.0276***-0.6759***-0.0212**-0.5304***-0.0157 -0.5102***-0.0116 (0.1733)(0.0096)(0.1922)(0.0097)(0.1903)(0.0096)(0.1906)(0.0096) Perceived risk (1 st stage) 0.0622*** 0.0603*** 0.0562*** 0.0556*** (0.0118) (0.0117) (0.0117) (0.0117) Religious influence (1 st stage) -0.0239*** -0.0196*** -0.0194*** -0.0192*** (0.0054) (0.0054) (0.0054) (0.0054) F-statistic (IV's in the 1 st stage) 24.71*** 20.77*** 18.69*** 18.34*** Overidentification test (p-value) 0.5177 0.5783 0.8219 0.7769 Hausman Statistic -3.6145*** -3.4117*** -2.7081*** -2.6197*** **Statistically significant at 5%. *Statistically significant at 10%. Model 3 contains those variables in Model 2 plus educational attainment indicators. Model 4 (Direct Effect) 2SLSOLS Model 4 (Direct Effect) 2SLSOLS Model 3 2SLSOLS OLS 2SLS Standard errors are in parentheses. ***Statistically significant at 1%. Model 1 (Total Effect) Model 2 Alcohol abuse and/or de p endence in the last 12 months OLS 2SLSOLS 2SLS 2SLSOLS 2SLSOLS Alcohol abuse and/or de p endence in the last 12 months Model 1 (Total Effect) Model 2 Model 3 Model 4 contains those variables in Model 3 plus health status indicators. See Appendix B1and B2 for full estimation results. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators.

PAGE 72

OLS. The Hausman test statistic shows that there is a significant difference between the two estimates, which means past year alcoho l abuse and/or dependence should be treated as an endogenous variable. Therefore, the 2SLS estimate from Model 1 should be used to draw an inference about th e total causal effect of alc ohol abuse and/or dependence on the probability of employment last week. Adding marital status and ch ildren as explanatory vari ables in Model 2 reduces the size and significance of the 2SLS estimate, and adding educational attainment causes the 2SLS estimate to lose significance. The OLS estimate also becomes smaller upon adding other covariates into the equation in Models 2, but remains significant. Because the Hausman statistic shows an insignificant difference between the 2SLS and OLS estimates in Models 2, but the former are al ways negative and larger in magnitude, the latter can be used to draw inferences about the direct effect of alcohol abuse and/or dependence on the probability of employment last week. Past year alcohol abuse and/or dependence directly lowers the probability of employment last week by 1.9 percentage points or 2.0 percent (-0.0188 / 0.94). 58 Literal interpretation of the results, i.e. relying on 2SLS for the total effect, implies that most of the effect of past year alcohol abuse and/or dependence on the probability of past week employment among males is an indirect effect through human capital accumulation. However, using the c onservative OLS estimate for the total effect, nearly 60 percent of the total effect would be considered direct rather than indirect. Regardless, complete second stage regres sion results in Appendix B1 support the conclusion that human capital effects are impor tant, since all the coefficients of the human capital components are statis tically significant and positive.

PAGE 73

The 2SLS estimates for females are statistically significant and larger than for males. The OLS estimates are much smaller than 2SLS estimates and statistically significant in Model 1 and 2, but not in Mode ls 3 and 4. However, the Hausman test statistics show that there is a significant difference between 2SLS and OLS, which means that OLS is inconsistent and 2SLS should be used for interpretation. Therefore, the results imply that there is a large signifi cant direct effect and a relatively smaller significant indirect effect of past year alcohol abuse and/or dependence on the probability of employment last week for the female sample. For a randomly chosen female in the population, other variables held constant, the likelihood of em ployment last week for an individual with alcohol abuse and/or dependence problems in th e past year is, on average, 65.3 percentage points (Model 1) and 51.0 percen tage points (Model 4) less than for an individual without alcohol abus e and/or dependence problems in the past year. Complete second stage regression results in Appe ndix B2 also support this conclusion. Table 14 summarizes similar results for th e effect of past year alcohol abuse and/or dependence on the proba bility of working full time last week. Once again, the first stage regressions are the same as in Table 13. For both gende rs, overidentification statistics show that the instrumental variab les are valid in all m odels, albeit weakly in Model 1 with p-values between 0.05 and 0.1. As expected, 2SLS estimates are larger than those in Table 13. For males, in Model 1 past year alcohol a buse and/or dependence lowers the probability of working full time last week by 32.3 percentage points or 36.3 percent (-0.3232 / 0.89) using 2SLS, but by onl y 3.8 percentage point s or 4.3 percent 59 (-0.0380 / 0.89) using OLS. After adding human capital covariates, the 2SLS effects become smaller and insignificant, and in Mode ls 3 and 4 the Hausman test statistics

PAGE 74

60 Male (n=12,046) Alcohol variables -0.3232***-0.0380***-0.2157**-0.0228***-0.1672-0.0199**-0.1464-0.0167** (0.0925)(0.0084)(0.0999)(0.0084)(0.1062)(0.0084)(0.1063)(0.0084) Perceived risk (1 st stage) 0.1006*** 0.0923*** 0.0862*** 0.0857*** (0.0102) (0.0102) (0.0102) (0.0102) Religious influence (1 st stage) -0.0216*** -0.0151** -0.0149** -0.0145** (0.0068) (0.0068) (0.0068) (0.0067) F-statistic (IV's in the 1 st stage) 54.81*** 44.43*** 38.58*** 38.11*** Overidentification test (p-value) 0.0764 0.2629 0.2300 0.2134 Hausman Statistic -3.0967*** -1.9380* -1.3906 -1.2240 Female (n=11,779) Alcohol variables -0.21910.0205 -0.4763-0.0070 -0.3147-0.0012 -0.27590.0030 (0.2649)(0.0170)(0.2942)(0.0169)(0.3042)(0.0169)(0.3059)(0.0169) Perceived risk (1 st stage) 0.0622*** 0.0603*** 0.0562*** 0.0556*** (0.0118) (0.0117) (0.0117) (0.0117) Religious influence (1 st stage) -0.0239*** -0.0196*** -0.0194*** -0.0192*** (0.0054) (0.0054) (0.0054) (0.0054) F-statistic (IV's in the 1 st stage) 24.71*** 20.77*** 18.69*** 18.34*** Overidentification test (p-value) 0.0534 0.2791 0.4000 0.3509 Hausman Statistic -0.9064 -1.5978 -1.0321 -0.9132 **Statistically significant at 5%. *Statistically significant at 10%. OLS 2SLS Alcohol abuse and/or de p endence in the last 12 months Standard errors are in parentheses. ***Statistically significant at 1%. Model 1 contains general exogenous variables. Model 4 (Direct Effect) 2SLSOLS Model 4 (Direct Effect) OLS Model 1 (Total Effect) Model 2 Model 3 Model 1 (Total Effect) Model 2 Model 3 2SLSOLS 2SLS 2SLSOLS 2SLSOLS 2SLSOLS OLS 2SLS Table 14. Re g ression estimates of binar y alcohol measures on the p robabilit y of workin g full time last week usin g NSDUH Alcohol abuse and/or de p endence in the last 12 months Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. Model 4 contains those variables in Model 3 plus health status indicators. See Appendix B3 and B4 for full estimation results.

PAGE 75

show that there are no significant differen ces between the 2SLS and OLS estimates. The OLS effects become smaller as human capital covariates are added in Models 2, but remain significant. This again implies that alcohol abuse and/or dependence has a small direct effect on the pr obability of working full time, but a large indirect effect through human capital components. Comple te second stage regr ession results in Appendix B3 that show positive and significant human capital variable coefficients also support this conclusion. For females, the results for all models s how that the effects from 2SLS are large, yet insignificant due to high standard errors Because the Hausman statistic shows an insignificant difference between the 2SLS and OLS estimates in every model, the OLS estimates should be used to draw inferences about the causal effect of past year alcohol abuse and/or dependence on the probability of working full time last week. However, in all four models the OLS estimates are very sm all and not statistically different from zero. Therefore, it can be concluded that there is no si gnificant direct or indi rect effect of past year alcohol abuse and/or dependence on the pr obability of working full time last week for females. 5.2.2. Frequencies of Alcohol Consumption 61 Table 15 summarizes the 2SLS and OLS regr ession results for the effects of the three numerical alcohol measures on the probabi lity of employment in the past week for males. All test statistics in the first st age regressions show th at both instrumental variables are adequate for making inference about causal effects. The signs of the estimates for both instrumental variables from the first stage regressions are as expected.

PAGE 76

62 Table 15. Re g ression estimates of continuous alcohol measures on the p robabilit y of em p lo y ment (full time/ p art time) last week: the male sam p le from NSDUH Male (n=12,046) Alcohol variables -0.00033***-0.00004*-0.00022*-0.00002-0.00011-0.00002-0.00010-0.00001 (0.0001)(0.0000)(0.0001)(0.0000)(0.0001)(0.0000)(0.0001)(0.0000) Perceived risk (1 st stage) 57.7730*** 56.4162*** 56.6514*** 56.5771*** (2.8461) (2.8403) (2.8601) (2.8611) Religious influence (1 st stage) -13.6000*** -12.3000*** -12.3000*** -12.3000*** (1.8876) (1.8862) (1.8865) (1.8868) F-statistic (IV's in the 1 st stage) 238.98*** 224.05*** 222.76*** 222.17*** Overidentification test (p-value) 0.7644 0.7859 0.9399 0.9362 Hausman Statistic -2.6086*** -1.7393* -0.8532 -0.7603 -0.00382***-0.00041-0.00246*-0.00016-0.00127-0.00018-0.00113-0.00017 (0.0013)(0.0003)(0.0013)(0.0003)(0.0013)(0.0003)(0.0013)(0.0003) Perceived risk (1 st stage) 4.9686*** 4.8644*** 4.9561*** 4.9530*** (0.2421) (0.2417) (0.2433) (0.2434) Religious influence (1 st stage) -1.3412*** -1.2430*** -1.2473*** -1.2500*** (0.1605) (0.1605) (0.1605) (0.1605) F-statistic (IV's in the 1 st stage) 253.97*** 239.20*** 244.26*** 243.98*** Overidentification test (p-value) 0.8665 0.7162 0.9045 0.9044 Hausman Statistic -2.6599*** -1.7533* -0.8443 -0.7466 -0.00616***-0.00224***-0.00413*-0.00166***-0.00227-0.00127***-0.00204-0.00116*** (0.0021)(0.0004)(0.0022)(0.0004)(0.0023)(0.0004)(0.0023)(0.0004) Perceived risk (1 st stage) 3.1747*** 3.0771*** 2.8979*** 2.8913*** (0.1420) (0.1415) (0.1418) (0.1417) Religious influence (1 st stage) -0.5855*** -0.5100*** -0.5011*** -0.4971*** (0.0942) (0.0940) (0.0936) (0.0935) F-statistic (IV's in the 1 st stage) 276.17*** 256.33*** 227.74*** 226.84*** Overidentification test (p-value) 0.6196 0.8878 0.9859 0.9789 Hausman Statistic -1.8866* -1.1435 -0.4384 -0.3817 **Statistically significant at 5%. *Statistically significant at 10%. Model 3 contains those variables in Model 2 plus educational attainment indicators. Model 4 contains those variables in Model 3 plus health status indicators. OLS 2SLSOLS 2SLSOLS ***Statistically significant at 1%. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 4 (Direct Effect) 2SLS 2SLSOLS Number of da y s drinkin g alcohol in the last 12 months Number of da y s drank 5+ alcohol drinks in the last month Number of da y s drinkin g alcohol in the last month Standard errors are in parentheses. Model 1 (Total Effect) Model 2 Model 3 See Appendix B5-B7 for full estimation results.

PAGE 77

Alcohol consumption frequency is positively related to perceived lack of risk and negatively related to religious influence. For Model 1, an individual who thinks that there is no more than slight risk from consuming 4-5 alcoholic drinks nearly ever y day, drank 58 more days in the last 12 months, five more days in the last month, a nd heavily, three more days in the last month than other individuals. Those who think their religious belief s are an important part of their lives drank 14 fewer days in the last 12 months, 2 fewer days in the last month, and 1 fewer day heavily in the last month than t hose who think otherwise. For Models 2 to 4, the coefficients of both instruments become s lightly smaller as other covariates are added into the model. For Model 1, the 2SLS estimates are st atistically significan t, showing negative total effects of alcohol cons umption on the probability of employment last week. The interpretations are as follows. If the respondent drank alcohol one additional day a week in the last 12 months, th e probability of employment last week would fall by 1.7 percentage points (-0.00033 x 52) or 1.8 pe rcent (-0.00033 x 52 / 0.94). If the respondent drank alcohol one additional day per week in the last month (four weeks), the probability of employment last week would decrease by 1.5 percentage points (-0.00382 x 4) or 1.6 percent (-0.00382 x 4 / 0.94). If the respondent drank five or more alcoholic beverages one additional day a week in the last mont h, the probability of employment last week would be reduced by 2.5 percentage points (-0.00616 x 4) or 2.6 percent (-0.00616 x 4 / 0.94). 63 Using OLS, the estimates are less statisti cally significant and are smaller than using 2SLS. Hausman statistics suggest that there are significant differences between

PAGE 78

2SLS and OLS estimates, implying that these alcohol measures are endogenous and thus 2SLS is preferred to OL S in these specifications. For both 2SLS and OLS estimates in Models 2 to 4, the estimates are smaller than those in Model 1 and are less statistically significant, except for the OLS estimates of the number of days in which five or more alc ohol drinks were consumed in the last month, which are significant in every model. Theref ore, it can be concluded that the number of days drinking alcohol has si gnificant indirect effects on current employment, but no significant direct effects. However, there ar e small direct and indirect effects of number of days on which the respondent drank heavily (five or more drinks) on the probability of employment last week. If the respondent drank five or more alcoholic drinks one additional day a week in the last month, the probability of employment last week would directly be reduced by 0. 5 percent (-0.00116 x 4 / 0.94). Similar to the results presented in Ta ble 15, Table 16 summarizes the 2SLS and OLS regression results of the effect of th ree alcohol measures on the probability of working full time for males. The first stage regressions are the same as in Table 15, and the tests of overiden tification yield identic al conclusions about the validity of the instrumental variables. For the 2SLS appro ach, the estimates are bigger than those in Table 15, as expected. The Hausman statistics in Model 1 are stat istically significant, suggesting that the alcohol variables are endoge nous with respect to employment and the 2SLS estimates in Model 1 should be used to make inferences about causal total effects of past alcohol consumption on the probability of working full time. 64 The 2SLS estimates in Model 1 can be inte rpreted as follows. Drinking alcohol one additional day a week in the last 12 months reduces the probability of working full time

PAGE 79

65 Male (n=12,046) Alcohol variables -0.00056***-0.00002-0.00036**0.00002-0.00026*0.00003-0.000230.00003 (0.0002)(0.0000)(0.0002)(0.0000)(0.0002)(0.0000)(0.0002)(0.0000) Perceived risk (1 st stage) 57.7730*** 56.4162*** 56.6514*** 56.5771*** (2.8461) (2.8403) (2.8601) (2.8611) Religious influence (1 st stage) -13.6000*** -12.3000*** -12.3000*** -12.3000*** (1.8876) (1.8862) (1.8865) (1.8868) F-statistic (IV's in the 1 st stage) 238.98*** 224.05*** 222.76*** 222.17*** Overidentification test (p-value) 0.0830 0.3313 0.2648 0.2465 Hausman Statistic -3.6173*** -2.4583** -1.8450* -1.6685* -0.00656***-0.00042-0.00414**-0.00001-0.00301*-0.00002-0.002660.00000 (0.0018)(0.0004)(0.0018)(0.0003)(0.0018)(0.0003)(0.0018)(0.0003) Perceived risk (1 st stage) 4.9686*** 4.8644*** 4.9561*** 4.9530*** (0.2421) (0.2417) (0.2433) (0.2434) Religious influence (1 st stage) -1.3412*** -1.2430*** -1.2473*** -1.2500*** (0.1605) (0.1605) (0.1605) (0.1605) F-statistic (IV's in the 1 st stage) 253.97*** 239.20*** 244.26*** 243.98*** Overidentification test (p-value) 0.1165 0.3906 0.2991 0.2754 Hausman Statistic -3.5606*** -2.3582** -1.7265* -1.5463 -0.01016***-0.00237***-0.00648**-0.00140**-0.00497-0.00102*-0.00435-0.00084 (0.0029)(0.0006)(0.0029)(0.0006)(0.0031)(0.0006)(0.0031)(0.0006) Perceived risk (1 st stage) 3.1747*** 3.0771*** 2.8979*** 2.8913*** (0.1420) (0.1415) (0.1418) (0.1417) Religious influence (1 st stage) -0.5855*** -0.5100*** -0.5011*** -0.4971*** (0.0942) (0.0940) (0.0936) (0.0935) F-statistic (IV's in the 1 st stage) 276.17*** 256.33*** 227.74*** 226.84*** Overidentification test (p-value) 0.0467 0.2573 0.2247 0.2115 Hausman Statistic -2.7838*** -1.7636* -1.2857 -1.1444 **Statistically significant at 5%. *Statistically significant at 10%. 2SLSOLS 2SLSOLS 2SLSOLS Model 1 (Total Effect) Model 2 Model 3 Table 16. Re g ression estimates of continuous alcohol measures on the p robabilit y of workin g full time last week: the male sam p le from NSDUH Model 4 (Direct Effect) 2SLSOLS Number of da y s drinkin g alcohol in the last 12 months Number of da y s drank 5+ alcohol drinks in the last month Number of da y s drinkin g alcohol in the last month Standard errors are in parentheses. ***Statistically significant at 1%. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. Model 4 contains those variables in Model 3 plus health status indicators. See Appendix B8-B10 for full estimation results.

PAGE 80

by 2.9 percentage points (-0.00056 x 52) or 3.3 percent (-0.00056 x 52 / 0.89). Drinking alcohol one additional day a w eek in the last month reduces the probability of working full time by 2.6 percentage points (-0.00656 x 4) or 2.9 percent (-0.00656 x 4 / 0.89). Drinking five or more alcoholic drinks one additional day a week in the last month reduces the probability of working full time by 4.1 percentage points (-0.01016 x 4) or 4.6 percent (-0.01016 x 4 / 0.89). However, after other covariates are added, the estimates become smaller and insignificant. The OLS estimates even become positive, albeit insignificant, in Model 4. Therefore, it can be concluded that there ar e no significant direct effects, but significant indirect effects of alcohol consum ption on working full time for males. Table 17 summarizes the 2SLS and OLS regr ession results of the effect of the three alcohol measures on the pr obability of employment last week for females. Similar to those of the male sample in Table 15, a ll first stage statistics and the tests of overidentification show that both instrumental variables ar e adequate for the female sample. The coefficients on both instrument s in the first stage regressions have the expected signs. For Model 1, an individual w ho thinks that there is no more than slight risk from consuming 4-5 alcoholic drinks near ly every day drank 43 more days in the last 12 months, four more days in the last month, and two more days heavily in the last month than others. An individual who thinks that his/her religious beliefs are an important part of his/her life drank 14 fewer days in the la st 12 months, one less day in the last month, and one less day heavily in the last month th an an individual who thinks otherwise. The coefficients of both instruments are similar in other model specifications. 66 The 2SLS estimates from all models indi cate significant nega tive effects of

PAGE 81

67 Table 17. Re g ression estimates of continuous alcohol measures on the p robabilit y of em p lo y ment (full time/ p art time) last week: the female sam p le from NSDUH Female (n=11,779) Alcohol variables -0.00102***-0.00009***-0.00099***-0.00007**-0.00073***-0.00008**-0.00069***-0.00009*** (0.0002)(0.0000)(0.0002)(0.0000)(0.0002)(0.0000)(0.0002)(0.0000) Perceived risk (1 st stage) 43.0742*** 42.7496*** 43.8184*** 43.7135*** (3.5306) (3.5137) (3.5246) (3.5217) Religious influence (1 st stage) -13.6000*** -12.4000*** -12.4000*** -12.5000*** (1.6353) (1.6309) (1.6304) (1.6293) F-statistic (IV's in the 1 st stage) 113.87*** 107.39*** 110.43*** 110.77*** Overidentification test (p-value) 0.7277 0.6641 0.9815 0.9503 Hausman Statistic -3.9536*** -3.7757*** -2.7603*** -2.5996*** -0.01208***-0.00093**-0.01164***-0.00075*-0.00842***-0.00097**-0.00803***-0.00106*** (0.0028)(0.0004)(0.0029)(0.0004)(0.0027)(0.0004)(0.0027)(0.0004) Perceived risk (1 st stage) 3.6151*** 3.5956*** 3.7605*** 3.7488*** (0.2832) (0.2822) (0.2825) (0.2822) Religious influence (1 st stage) -1.1740*** -1.0906*** -1.0852*** -1.0918*** (0.1312) (0.1310) (0.1307) (0.1306) F-statistic (IV's in the 1 st stage) 126.91*** 120.81*** 128.12*** 128.26*** Overidentification test (p-value) 0.6873 0.6098 0.9961 0.9308 Hausman Statistic -4.0056*** -3.8147*** -2.7479*** -2.5840*** -0.02724***-0.00436***-0.02638***-0.00382***-0.02004***-0.00298***-0.01923***-0.00268*** (0.0063)(0.0009)(0.0064)(0.0009)(0.0066)(0.0009)(0.0066)(0.0009) Perceived risk (1 st stage) 1.7803*** 1.7629*** 1.6914*** 1.6865*** (0.1230) (0.1224) (0.1226) (0.1226) Religious influence (1 st stage) -0.3893*** -0.3502*** -0.3494*** -0.3484*** (0.0570) (0.0568) (0.0567) (0.0567) F-statistic (IV's in the 1 st stage) 132.84*** 126.92*** 118.04*** 117.48*** Overidentification test (p-value) 0.7596 0.8621 0.6873 0.7520 Hausman Statistic -3.6795*** -3.5353*** -2.6200*** -2.5451** **Statistically significant at 5%. *Statistically significant at 10%. OLS Model 4 (Direct Effect) 2SLS OLS 2SLS OLS Number of da y s drinkin g alcohol in the last 12 months Number of da y s drank 5+ alcohol drinks in the last month Number of da y s drinkin g alcohol in the last month Standard errors are in parentheses. ***Statistically significant at 1%. Model 1 contains general exogenous variables. OLS 2SLS Model 1 (Total Effect) Model 2 Model 3 2SLS Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. Model 4 contains those variables in Model 3 plus health status indicators. See Appendix B11-B13 for full estimation results.

PAGE 82

alcohol consumption on the probability of em ployment. In Model 1, the total effects of alcohol are three to four times larger than those of the male sample in Table 15 and can be interpreted as follows. If the respondent drank alcohol one additional day a week in the last 12 months, the probabi lity of past week employment is reduced by 5.3 percentage points (-0.00102 x 52) or 5.7 percent (-0.00102 x 52 / 0.93). If the respondent drank alcohol one additional day a week in the la st month, the probability of past week employment is decreased by 4.8 percenta ge points (-0.01208 x 4) or 5.2 percent (-0.01208 x 4 / 0.93). If the respondent drank five or more alcoholic drinks one additional day a week in the last month, the probability of past week employment falls by 10.9 percentage points (-0.02724 x 4) or 11.7 percent (-0.02724 x 4 / 0.93). As expected, the OLS estimates are statistically significant but smaller than the 2SLS estimates due to the simultaneity probl em, thereby causing a positive bias in OLS model. The Hausman statistics imply that OL S is inconsistent, so 2SLS should be used to make inferences in all models. 68 For Models 2-4, both 2SLS and OLS effect s become smaller but remain negative and significantly different from zero, suggesting that there are sma ll direct effects of alcohol consumption on the probability of em ployment for females. The interpretations of the direct effects are as follows. If th e respondent drank alcohol one additional day a week in the last 12 months, the probabil ity of past week employment falls by 3.6 percentage points (-0.00069 x 52) or 3.9 pe rcent (-0.00069 x 52 / 0.93). If the respondent drank one additional day a week in the last month, the probability of past week employment is reduced by 3.2 percentage points (-0.00803 x 4) or 3.5 percent (-0.00803 x 4 / 0.93). If the respondent drank five or more alcoholic beverages one additional day

PAGE 83

a week in the last month, the probability of past week employment last week decreases by 7.7 percentage points (-0.01923 x 4) or 8.3 percent (-0.01923 x 4 / 0.93). Table 18 summarizes the 2SLS and OLS regr ession results of the effect of the three alcohol measures on the probability of working full time in the last week for females. Both 2SLS and OLS estimates are statistically insignificant in most models, implying that there are no effects of alcohol consumption on the probability of working full time in the last week for females. 5.3. Results from NELS:88 Table 19 summarizes the 2SLS and OLS regr ession results of the effects of the two continuous alcohol measures on the probability of current employment for males. The coefficients of both instruments, indicato rs of whether the respondent drank on 20 or more occasions in the last 30 days during 10th grade and five or more drinks in a row on at least six occasions in the last two weeks during 12th grade, have the expected sign in each first stage regression. The t and F statistics in the first stage regressions also imply that both instruments are individually and jointly significant. Moreover, the p-values for the overidentification tests suggest that th e instruments are not correlated with the employment equation residuals. Theref ore, both instruments appear valid. 69 The coefficients of the instruments in Mode l 1 can be interpreted as follows. On average, respondents repor ting heavy drinking in 10 th or 12 th grades drank on two additional instances in the last 30 days, and drank heavily on one additional instance over the last two weeks. These effects change onl y slightly when other covariates are added into the model.

PAGE 84

70 Female (n=11,779) Alcohol variables -0.000410.00005-0.00072*-0.00005-0.00046-0.00005-0.00040-0.00006 (0.0004)(0.0001)(0.0004)(0.0001)(0.0004)(0.0001)(0.0004)(0.0001) Perceived risk (1 st stage) 43.0742*** 42.7496*** 43.8184*** 43.7135*** (3.5306) (3.5137) (3.5246) (3.5217) Religious influence (1 st stage) -13.6000*** -12.4000*** -12.4000*** -12.5000*** (1.6353) (1.6309) (1.6304) (1.6293) F-statistic (IV's in the 1 st stage) 113.87*** 107.39*** 110.43*** 110.77*** Overidentification test (p-value) 0.0628 0.3045 0.4494 0.3886 Hausman Statistic -1.1298 -1.6217 -1.0005 -0.8430 -0.004720.00053-0.00833*-0.00057-0.00529-0.00080-0.00465-0.00090 (0.0048)(0.0007)(0.0049)(0.0007)(0.0048)(0.0007)(0.0047)(0.0007) Perceived risk (1 st stage) 3.6151*** 3.5956*** 3.7605*** 3.7488*** (0.2832) (0.2822) (0.2825) (0.2822) Religious influence (1 st stage) -1.1740*** -1.0906*** -1.0852*** -1.0918*** (0.1312) (0.1310) (0.1307) (0.1306) F-statistic (IV's in the 1 st stage) 126.91*** 120.81*** 128.12*** 128.26*** Overidentification test (p-value) 0.0610 0.2895 0.4431 0.3839 Hausman Statistic -1.0998 -1.5954 -0.9537 -0.7985 -0.01379-0.00037-0.02044*-0.00263*-0.01374-0.00175-0.01244-0.00145 (0.0109)(0.0016)(0.0111)(0.0016)(0.0114)(0.0016)(0.0114)(0.0016) Perceived risk (1 st stage) 1.7803*** 1.7629*** 1.6914*** 1.6865*** (0.1230) (0.1224) (0.1226) (0.1226) Religious influence (1 st stage) -0.3893*** -0.3502*** -0.3494*** -0.3484*** (0.0570) (0.0568) (0.0567) (0.0567) F-statistic (IV's in the 1 st stage) 132.84*** 126.92*** 118.04*** 117.48*** Overidentification test (p-value) 0.0914 0.4469 0.5401 0.4665 Hausman Statistic -1.2501 -1.6282 -1.0596 -0.9711 **Statistically significant at 5%. *Statistically significant at 10%. OLS Model 4 (Direct Effect) 2SLS Number of da y s drinkin g alcohol in the last 12 months Model 3 Number of da y s drank 5+ alcohol drinks in the last month Number of da y s drinkin g alcohol in the last month OLS Model 1 (Total Effect) Model 2 Standard errors are in parentheses. Table 18. Re full time last week: the female sam p g ression estimates of continuous alcohol measures on the p robabilit y of workin g le from NSDUH 2SLSOLS 2SLSOLS 2SLS Model 4 contains those variables in Model 3 plus health status indicators. See Appendix B14-B16 for full estimation results. ***Statistically significant at 1%. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators.

PAGE 85

71 Table 19. Re g ression estimates of continuous alcohol measures on the p robabilit y of current em p lo y ment (full time/ p art time): the male sam p le from NELS:88 Alcohol variables Number of occasions drank alcohol in the last 30 da y s -0.010680.00122*-0.010270.00162**-0.011430.00171*** (0.0072)(0.0006) (0.0073)(0.0006) (0.0071)(0.0006) Drank for 20+ occasions in the last 30 days 2.0560*** 2.0540*** 2.1251*** during 10th grade (1 st stage) (0.7435) (0.7360) (0.7371) Drank 5+ drinks in a row at least 6 times i n 2.0247*** 1.9960*** 2.0657*** the last two weeks during 12th grade (1 st stage) (0.4504) (0.4457) (0.4468) F-statistic (IV's in the 1 st stage) 16.51*** 16.49*** 17.57*** Overidentification test (p-value) 0.1731 0.1638 0.2097 Hausman Statistic -1.6552* -1.6400 -1.8552* -0.031980.00546**-0.030880.00702***-0.03625*0.00692*** (0.0208)(0.0025) (0.0211)(0.0025) (0.0218)(0.0025) Drank for 20+ occasions in the last 30 days 0.7342*** 0.7242*** 0.7112*** during 10th grade (1 st stage) (0.1877) (0.1859) (0.1860) Drank 5+ drinks in a row at least 6 times i n 0.6775*** 0.6659*** 0.6502*** the last two weeks during 12th grade (1 st stage) (0.1137) (0.1126) (0.1128) F-statistic (IV's in the 1 st stage) 30.22*** 29.81*** 28.41*** Overidentification test (p-value) 0.1785 0.1671 0.2171 Hausman Statistic -1.8124* -1.8072* -1.9932** Standard errors are in parentheses. ***Statistically significant at 1%. **Statistically significant at 5%. *Statistically significant at 10%. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. See Appendix C1 and C2 for full estimation results. Male (n=3,876) Model 1 (Total Effect) Model 2 Model 3 OLS 2SLS OLS 2SLS OLS 2SLS Number of occasions drank 5+ alcoholic drinks in a row over the last two weeks

PAGE 86

All of the 2SLS estimates in Table 19 im ply that alcohol consumption negatively affects the probability of current employment for males. However, these estimates are very small and insignificantly different from zero. By contrast, all of the OLS estimates in Table 19 are positive and statistically diffe rent from zero. Nonetheless, the Hausman test statistics in most models imply that these alcohol measures are endogenous and therefore the 2SLS estimates s hould be used to draw infere nces about causal effects. Therefore, it can be concluded that there is no significant direct or indirect effects of alcohol consumption on the probability of current employment for males. Table 20 summarizes similar results of the 2SLS and OLS regressions of currently working full time on the two continuous alcoho l measures for males. The only difference between the results of Table 19 and Table 20 is that the Hausman test statistics in every model imply that these alcohol measures are not endogenous. The OLS estimates are similar to those in the previous literature in signifying positive e ffects of drinking on full time employment. As before, the effect becomes more positive when human capital measures, here family structure and schooling but not health, are held constant. Thus, indirect effects of drinking again appear to be negative, but direct effects have the unexpected positive sign that many other researchers have similarly uncovered. 72 Table 21 summarizes the 2SLS and OLS re gression results of the effects of the two continuous alcohol measures on the probabi lity of current employment for females. All of the first stage statistics and the overiden tification tests imply that both instrumental variables are adequate to identify the effects of alcohol consumption on female employment propensities. The coefficients of the instruments in Model 1 can be interpreted as follows. On average, re spondents indicating hea vy 10th grade drinking

PAGE 87

73 Alcohol variables Number of occasions drank alcohol in the last 30 da y s -0.004390.00119 -0.003390.00229***-0.003540.00232*** (0.0092)(0.0008) (0.0092)(0.0008) (0.0090)(0.0008) Drank for 20+ occasions in the last 30 days 2.0560*** 2.0540*** 2.1251*** during 10th grade (1 st stage) (0.7435) (0.7360) (0.7371) Drank 5+ drinks in a row at least 6 times i n 2.0247*** 1.9960*** 2.0657*** the last two weeks during 12th grade (1 st stage) (0.4504) (0.4457) (0.4468) F-statistic (IV's in the 1 st stage) 16.51*** 16.49*** 17.57*** Overidentification test (p-value) 0.2677 0.2447 0.2689 Hausman Statistic -0.6075 -0.6174 -0.6569 -0.013610.00908***-0.010660.01339***-0.011750.01351*** (0.0270)(0.0033) (0.0272)(0.0033) (0.0279)(0.0033) Drank for 20+ occasions in the last 30 days 0.7342*** 0.7242*** 0.7112*** during 10th grade (1 st stage) (0.1877) (0.1859) (0.1860) Drank 5+ drinks in a row at least 6 times i n 0.6775*** 0.6659*** 0.6502*** the last two weeks during 12th grade (1 st stage) (0.1137) (0.1126) (0.1128) F-statistic (IV's in the 1 st stage) 30.22*** 29.81*** 28.41*** Overidentification test (p-value) 0.2728 0.2479 0.2732 Hausman Statistic -0.8469 -0.8911 -0.9116 Standard errors are in parentheses. ***Statistically significant at 1%. **Statistically significant at 5%. *Statistically significant at 10%. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. See Appendix C3 and C4 for full estimation results. Model 2 Model 3 2SLS OLS Table 20. Re full time: the male sam p g g ression estimates of continuous alcohol measures on the p robabilit y of currentl y workin le from NELS:88 Number of occasions drank 5+ alcoholic drinks in a row over the last two weeks 2SLS OLS 2SLS OLS Male (n=3,876) Model 1 (Total Effect)

PAGE 88

74 Table 21. Re g ression estimates of continuous alcohol measures on the p robabilit y of current em p lo y ment (full time/ p art time): the female sam p le from NELS:88 Alcohol variables Number of occasions drank alcohol in the last 30 da y s -0.007660.00648***-0.001320.00128 0.000850.00133 (0.0117)(0.0013) (0.0115)(0.0013) (0.0111)(0.0013) Drank for 20+ occasions in the last 30 days 1.0037 1.4743* 1.6438* during 10th grade (1 st stage) (0.8849) (0.8457) (0.8415) Drank 5+ drinks in a row at least 6 times i n 3.1559*** 3.0406*** 3.1276*** the last two weeks during 12th grade (1 st stage) (0.4354) (0.4162) (0.4140) F-statistic (IV's in the 1 st stage) 27.64*** 29.22*** 31.58*** Overidentification test (p-value) 0.4558 0.1304 0.1097 Hausman Statistic -1.2108 -0.2274 -0.0442 -0.027090.02186***0.006510.00593 0.015790.00655 (0.0541)(0.0063) (0.0530)(0.0062) (0.0533)(0.0062) Drank for 20+ occasions in the last 30 days 0.4806*** 0.5286*** 0.5246*** during 10th grade (1 st stage) (0.1777) (0.1745) (0.1747) Drank 5+ drinks in a row at least 6 times i n 0.6397*** 0.6186*** 0.6147*** the last two weeks during 12th grade (1 st stage) (0.0874) (0.0859) (0.0860) F-statistic (IV's in the 1 st stage) 32.00*** 32.20*** 31.67*** Overidentification test (p-value) 0.3892 0.1304 0.1156 Hausman Statistic -0.9116 0.0111 0.1745 Standard errors are in parentheses. ***Statistically significant at 1%. **Statistically significant at 5%. *Statistically significant at 10%. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. See Appendix C5 and C6 for full estimation results. Number of occasions drank 5+ alcoholic drinks in a row over the last two weeks Female (n=4,638) Model 3 Model 1 (Total Effect) Model 2 2SLS OLS 2SLS OLS 2SLS OLS

PAGE 89

used alcohol on one additional instance in the last 30 days and drank heavily on one additional instance over the la st four weeks. Respondent s indicating heavy 12th grade drinking consumed alcohol on three additional instances in the last 30 days and drank heavily on one additional instance over the la st two weeks. The coefficients of both instruments in the other models are similar to those in Model 1. Using 2SLS, there is no evidence of a stat istically significant effect of alcohol consumption on female employment. Moreover, the Hausman test statistics in every model imply that these alcohol measures ar e not endogenous. Using OLS, the estimates are positive, very small in all models, and st atistically insignificant when human capital controls are added. Therefore, it can be concluded that there is only a positive total effect of alcohol consumption on current employme nt for females, which contradicts the conventional wisdom about the e ffect of alcohol consumption. Table 22 summarizes similar results of the 2SLS and OLS regressions of currently working full time on the two continuous alc ohol measures for females. The Hausman statistics imply that alcohol measures are endogenous in Model 1, suggesting that 2SLS estimates in Model 1 should be used to in fer about the causal total effect of alcohol consumption on the probability of currently working full time for females. However, these 2SLS estimates in Model 1 are not sta tistically different from zero, implying that there is no significant total effect of alcohol consumption on the proba bility of currently working full time for females. 75

PAGE 90

76 Alcohol variables Number of occasions drank alcohol in the last 30 da y s -0.020060.00759***-0.013990.00138 -0.011420.00117 (0.0145)(0.0015) (0.0140)(0.0016) (0.0135)(0.0016) Drank for 20+ occasions in the last 30 days 1.0037 1.4743* 1.6438* during 10th grade (1 st stage) (0.8849) (0.8457) (0.8415) Drank 5+ drinks in a row at least 6 times i n 3.1559*** 3.0406*** 3.1276*** the last two weeks during 12th grade (1 st stage) (0.4354) (0.4162) (0.4140) F-statistic (IV's in the 1 st stage) 27.64*** 29.22*** 31.58*** Overidentification test (p-value) 0.5412 0.8834 0.8299 Hausman Statistic -1.9150* -1.1016 -0.9381 -0.098970.03382***-0.062310.01528**-0.052500.01598** (0.0671)(0.0076) (0.0648)(0.0075) (0.0651)(0.0075) Drank for 20+ occasions in the last 30 days 0.4806*** 0.5286*** 0.5246*** during 10th grade (1 st stage) (0.1777) (0.1745) (0.1747) Drank 5+ drinks in a row at least 6 times i n 0.6397*** 0.6186*** 0.6147*** the last two weeks during 12th grade (1 st stage) (0.0874) (0.0859) (0.0860) F-statistic (IV's in the 1 st stage) 32.00*** 32.20*** 31.67*** Overidentification test (p-value) 0.7363 0.7636 0.7420 Hausman Statistic -1.9924** -1.2046 -1.0588 Standard errors are in parentheses. ***Statistically significant at 1%. **Statistically significant at 5%. *Statistically significant at 10%. Model 1 contains general exogenous variables. Model 2 contains those variables in Model 1 plus the number of children and marital status indicators. Model 3 contains those variables in Model 2 plus educational attainment indicators. See Appendix C7 and C8 for full estimation results. Model 3 2SLS Female (n=4,638) Model 1 (Total Effect) Model 2 Number of occasions drank 5+ alcoholic drinks in a row over the last two weeks OLS 2SLS OLS OLS 2SLS Table 22. Re full time: the female sam p g g ression estimates of continuous alcohol measures on the p robabilit y of currentl y workin le from NELS:88

PAGE 91

6. Conclusions, Caveats, Policy Implications, and Future Works Although conventional wisdom suggests that alcohol consumption would adversely affect labor market outcomes di rectly through a reducti on in productivity and indirectly by decreasing human capital accumulation, many economists have obtained estimates of this relationship that are contrary to expectations. This study attempted to estimate causal effects of alcohol consump tion on employment status and test the hypothesis that past alcohol use directly and indirectly leads to lower employment propensities by addressing potential estimati on problems found in previous studies. Not only are several sample restrictions applied to the data sets to obtain more precise estimates, but the instrumental variables me thod is also employed to control for the potential endogeneity problem in order to obta in estimates that can be interpreted as causal effects. The results from three different data sets can be summarized as follows. Using NESARC, the effects of drinking are identified using indicators of whether at least one of the respondent s parents and at least one of the respondents siblings was ever an alcoholic or problem drinker. Estim ates for both genders show that past alcohol abuse and/or dependence has no significant dir ect effects on current employment but has indirect negative effects through marriage, fe rtility, education and health. For males, alcohol abuse and/or dependence prior to and during the last 12 months indirectly lowers the probability of current employment by 10.2 and 26.7 percentage points, respectively. The negative indirect effects for females ar e larger: 23.4 percentage points for alcohol 77

PAGE 92

abuse and/or dependence prior to the past ye ar and 74.9 percentage points for past year abuse and/or dependence. For frequencies of recent alcohol cons umption, estimates imply that alcohol consumption has only an indirect effect on current employment propensity for females, but has both direct and indire ct effects on current employ ment propensity for males, though the direct effect is relatively small. For males, having five or more alcoholic drinks one additional day a week in the past year directly and totally lowers the probability of current employment by 0.8 and 1.6 percentage points, respectively. Getting drunk one additional day a week in the past year directly and totally lowers the probability of current employment by 1.3 and 2.1 percentage points, respectively. For the female sample, the indirect eff ects are much bigger than those of males. Drinking five or more alcoholic drinks one additional day a week in the past year indirectly lowers the probability of current employment by 28.9 percentage points. Getting drunk one additional day a week in the past year indi rectly lowers the probability of current employment by 83.3 percentage points. The results and conclusions from NESARC for alcohol and/or dependence effects are partially comparable to those found by Mullahy and Sindelar (1991), Mullahy and Sindelar (1993), and Mullahy and Sindelar (1996), who also studi ed the effects of alcohol abuse and/or dependence on employment. Sim ilar to these studies, this analysis finds that there is only an indirect effect of past alcohol abus e and/or dependence on current employment. 78 Using NSDUH, drinking effect s are identified by indica tors of the respondents perspective about the risk of using alc ohol and whether the respondents religious

PAGE 93

beliefs are an important part of his/her life For males, alcohol abuse and/or dependence has small but significant negative direct empl oyment effects and has significant negative indirect effects through human capital covariates. Alcohol abuse and/or dependence in the past year directly reduces the probab ility of past week employment by about 1.9 percentage points, which is about a tenth of th e size of the total effect. For the number of days alcohol was consumed in the last year and in the last month, there are only negative indirect effects on employment. Drinking alcohol one addition al day a week in the past year indirectly reduces the probability of past week employment by 1.7 percentage points. However, there are small direct and indirect effects of nu mber of days on which the respondent drank heavily on the probability of employment last week. For females, there is a larg e negative total and direct e ffect of past year alcohol abuse and/or dependence on employment. Alco hol abuse and/or dependence in the past year directly and totally reduces the proba bility of past week employment by 51.0 and 65.3 percentage points, respectively. Sim ilarly, for all measures of consumption frequency, there are significant negative tota l and direct effects on female employment. Drinking alcohol one additional day a week in the past year directly and totally reduces the probability of past week employment by 3.6 and 5.3 percentage points, respectively. Drinking 5 or more alcohol drinks one additi onal day a week in the past month directly and totally reduces the probability of past week employment by 7.7 and 10.9 percentage points, respectively. 79 Using NELS:88, current drinking is identi fied using indicators of alcohol use when the respondents were in 10 th and 12 th grades. For males, the results do not show a statistically significant total e ffect of alcohol use on employ ment. However, the results

PAGE 94

for females show that there is a small positiv e total effect of alc ohol use on employment, contradicting the conventional wisdom about the negative effect of alcohol use on labor market outcomes. Results from each data set provide different conclusions. For various reasons, the results obtained from NESARC might be preferred to those from the NSDUH and NELS:88. First, both of th e latter data sets contain fe wer control variables than NESARC, potentially causing the omitted variable problem. Second, the instrumental variables used in these two data sets are neither theoretically nor empirically superior to those used in NESARC, though the diagnostic tests imply that they are statistically adequate. Finally, NESARC offers more pr ecise measures of alcohol consumption, reducing the measurement error. Based on the results from NESARC, it can be concluded that previous problematic heavy dr inking, i.e. clinically-d efined alcohol abuse and/or dependence or more fr equent alcohol use, does lowe r the probability of current employment for both genders through hu man capital components, specifically educational attainment and health status. 80 The results from this study suggest some explanations for the conflicting results of previous studies about the effect of alcohol consumpti on on employment. First, the inclusion of human capital components ma kes drinking effects less likely to be significant. Second, different tolerances to alcohol might cause unstable results for frequency measures. Third, controlling for th e endogeneity of alcohol use is difficult. Even though the instruments used here are extremely strong predictors of drinking, standard errors of 2SLS estimates were ofte n too large to statisti cally distinguish them from OLS estimates even though 2SLS estimates were sometimes much larger in

PAGE 95

magnitude from a practical perspective. Since the results suggest that heavy al cohol consumption indirectly causes adverse labor market outcomes through human capital components, the true benefits to programs that reduce alcohol use, particularly heavy drinking and alcohol abuse and/or dependence, exceed those related just to crime, drunk driving, and other behaviors that are negatively affected by drinking. Any h ealth policy tool that limits human capitalimpairing alcohol consumption will not only reduce alcohol-related social problems but will also improve labor market outcomes. Such policy tools include information campaigns directed towards teenagers a nd students, a common period during which drinking is initiated; further minimum lega l drinking age increases although these are unlikely, given that the current age is 21; st ricter penalties on unde raged drinking and for any person who provides alcohol to an underage d drinker; and increa sed excise taxes on alcohol at the federal or stat e level. The 2SLS approaches in this study imply that any policy that prevents alcohol problems among parents and siblings, and to a lesser extent increases the risk awareness and reduces drinking during high school, will reduce alcohol problems, and thus increase employment, among working-age adults. 81 There are several suggestions for future research. First, one might perform crossvalidation by applying the same methodology an d model specification to the second wave of NESARC, which should be available in th e near future. Second, the cross-sectional nature of the data in this study allows for i nvestigation only of short-run effects and does not allow for changes in dri nking patterns to be observe d, but the medical literature suggests that alcohol is likely to have long term health effects. Therefore, one might try to obtain panel data to inves tigate the long te rm effects of alcohol use on labor market

PAGE 96

outcomes using a similar approach to that of this study. Third, one might try to apply a different methodology or model specification to the same data used here to ensure the robustness of the results. More specificall y, one could try a non-lin ear specification to investigate a possible non-lin ear relationship between alcohol consumption and labor market outcomes. Finally, one might try to obt ain other instrumental variables to address the endogeneity problem that is inherent in the nature of the relationship between alcohol consumption and labor market outcomes. 82

PAGE 97

References Auld, Christopher M. 2005. Smoking, Drinking, and Income. Journal of Human Resources 40, no. 2: 505-518. Barrett, Garry F. 2002. The Effect of Alcohol Consumption on Earnings. Economic Record 78, no. 240:79-96. Berger, Mark C., and J. P. Leigh. 1988. The Effect of Alcohol Use on Wages. Applied Economics 20, no. 10:1343-1351. Bryant, Richard R., V. A. Samaranayake, and Allen Wilhite. 1993. The Influence of Current and Past Alcohol Use on Earni ngs: Three Approaches to Estimation. Journal of Applied Behavioral Science 29, 9-31. Cook, Philip J., and Bethany L. Peters. 2005. The Myth of the Drinkers Bonus. National Bureau of Economic Research Inc. NBER Working Papers:11902. Dave, Dhaval, and Robert Kaestner. 2002. Al cohol Taxes and Labor Market Outcomes. Journal of Health Economics 21, no. 3:357-371. Durlauf, Steven N. 2002. On the Empirics of Social Capital. The Economic Journal, no. 112: 459-479. Feng, Weiwei, Wei Zhou, J.S. Butler, Brenda M. Booth, and Michael T. French. 2001. The Impact of Problem Drinking on Employment. Health Economics 10, no. 6:509-521. French, Michael T., and Gary A. Zarkin. 1995. Is Moderate Alcohol Use Related to Wages? Evidence from Four Worksites. Journal of Health Economics 14, 319344. Greene, William H. 2002. Econometric Analysis 5th ed. Prentice Hall. Hamilton, Vivian, and Barton H. Hamilton. 1997. Alcohol and Earnings: Does Drinking Yield a Wage Premium? Canadian Journal of Economics 30, no. 1:135-151. Hausman, Jerry A. 1978. Specification Tests in Econometrics. Econometrica 46, no. 6:1251-1271. 83

PAGE 98

Heien, Dale M. 1996. Do Drinkers Earn Less? Southern Economic Journal 63, no. 1:6068. Kenkel, Donald S., and David C. Ribar. 1994. Alcohol Consumption and Young Adults' Socioeconomic Status. Brookings Papers on Economic Activity. Microeconomics 1994, 119-175. MacDonald Ziggy, and Michael A. Shields. 2004. Does Problem Drinking Affect Employment? Evidence from England. Health Economics 13, 139-155. Mullahy, John, and Jody L. Sindelar. 1989. Li fe-Cycle Effects of Alcoholism on Education, Earnings, and Occupation. Inquiry 26, no. 2:272-82. Mullahy, John, and Jody L. Sindelar. 1996. Em ployment, Unemployment, and Problem Drinking. Journal of Health Economics 15, no. 4:409-434. Mullahy, John, and Jody L. Sindelar. 1993. Alcoholism, Work, and Income. Journal of Labor Economics 11, no. 3:494-520. Mullahy, John, and Jody L. Sindelar. 1991. Gender Differences in Labor Market Effects of Alcoholism. The American Economic Review 81, no. 2, Papers and Proceedings of the Hundred and Thir d Annual Meeting of the American Economic Association:161-165. Peters, Bethany L. 2004. Is There a Wa ge Bonus from Drinking? Unobserved Heterogeneity Examined. Applied Economics 36, no. 20:2299-2315. Shuckit, Marc A. (1999). New findi ngs on the genetics of alcoholism. Journal of the American Medical Association 281 (20) 1875-1876. Terza, Joseph V. 2002. Alcohol Abuse and Employment: A Second Look. Journal of Applied Econometrics 17, no. 4:393-404. Wooldridge, Jeffrey M. 2002. Econometric Analysis of Cross Section and Panel Data. MIT Press. Zarkin, Gary A., Michael T. French, Thom as A. Mroz, and Jeremy W. Bray. 1998a. Alcohol Use and Wages: New Results fr om the National Household Survey on Drug Abuse. Journal of Health Economics 17, no. 1:53-68. Zarkin, Gary A., Thomas A. Mroz, Jeremy W. Bray, and Michael T. French. 1998b. The Relationship between Drug Use and Labor Supply for Young Men. Labour Economics 5, no. 4:385-409. 84

PAGE 99

Appendix A: Complete Second Stage Regression Results Using NESARC 85

PAGE 100

Appendix A-1: Current Employment (F ull Time/Part Time) versus Alcohol Abuse and/or Dependence prior to th e Last 12 Months: the Male Sample 86 Male (n=8,673) Variables -0.1022**-0.0169**-0.0864*-0.0141**-0.0413-0.0144**-0.0122-0.0050 (0.0496)(0.0069)(0.0509)(0.0069)(0.0506)(0.0068)(0.0483)(0.0064) Age 0.0101**0.0089**0.00570.0048 0.00470.0043 0.00350.0034 (0.0044)(0.0043)(0.0043)(0.0043)(0.0043)(0.0042)(0.0040)(0.0039) Age squared -0.0002***-0.0001***-0.0001**-0.0001*-0.0001*-0.0001*-0.0001-0.0001 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0000)(0.0000) White 0.0365***0.0216**0.0393***0.0265***0.0026-0.0021 0.0001-0.0012 (0.0122)(0.0087)(0.0125)(0.0088)(0.0125)(0.0090)(0.0117)(0.0084) Black -0.0505***-0.0517***-0.0334***-0.0343***-0.0526***-0.0527***-0.0350***-0.0350*** (0.0112)(0.0111)(0.0111)(0.0111)(0.0111)(0.0111)(0.0104)(0.0104) American Indian/Alaska Native 0.0125-0.0011 0.01740.0059 -0.0078-0.0119 0.01490.0139 (0.0287)(0.0273)(0.0284)(0.0271)(0.0280)(0.0269)(0.0260)(0.0251) Asian/Pacific Islander 0.00500.0141 0.00090.0078 -0.0330*-0.0306 -0.0215-0.0209 (0.0197)(0.0188)(0.0194)(0.0187)(0.0194)(0.0188)(0.0181)(0.0175) Living in MSA in central city 0.00270.0047 0.01600.0176*0.00480.0053 -0.0058-0.0057 (0.0102)(0.0100)(0.0102)(0.0100)(0.0101)(0.0100)(0.0094)(0.0094) Living in MSA outside central cit y 0.0378***0.0392***0.0375***0.0384***0.0238**0.0241***0.00750.0076 (0.0094)(0.0093)(0.0093)(0.0092)(0.0093)(0.0092)(0.0086)(0.0086) State unemployment rates -0.0084**-0.0077**-0.0078**-0.0071**-0.0066*-0.0063*-0.0047-0.0046 (0.0036)(0.0036)(0.0036)(0.0035)(0.0035)(0.0035)(0.0033)(0.0033) Father/mother ever depressed 0.0033-0.0052 0.0044-0.0030 -0.0060-0.0087 -0.0035-0.0042 (0.0106)(0.0093)(0.0106)(0.0092)(0.0105)(0.0091)(0.0097)(0.0085) Brother/sister ever depressed 0.00790.0034 0.00950.0058 0.00340.0020 0.01470.0143 (0.0108)(0.0103)(0.0106)(0.0103)(0.0105)(0.0102)(0.0098)(0.0095) Father/mother had behavior problems -0.0111-0.0242*-0.0119-0.0226*-0.0139-0.0178 -0.0062-0.0072 (0.0155)(0.0135)(0.0154)(0.0133)(0.0152)(0.0132)(0.0141)(0.0123) Brother/sister had behavior problems -0.0213-0.0313***-0.0254**-0.0335***-0.0267**-0.0297***-0.0288**-0.0295*** (0.0130)(0.0116)(0.0129)(0.0115)(0.0127)(0.0114)(0.0118)(0.0106) Number of childre n -0.0048**-0.0052**-0.0002-0.0004 0.00120.0012 (0.0024)(0.0024)(0.0024)(0.0024)(0.0022)(0.0022) Married 0.1064***0.1073***0.1010***0.1013***0.0787***0.0787*** (0.0097)(0.0096)(0.0095)(0.0095)(0.0089)(0.0089) Cohabitating 0.0598***0.0540***0.0630***0.0609***0.0520***0.0514*** (0.0191)(0.0186)(0.0189)(0.0185)(0.0176)(0.0172) Widowed -0.0275-0.0298 -0.0254-0.0262 0.00450.0043 (0.0430)(0.0427)(0.0424)(0.0423)(0.0394)(0.0394) Divorced 0.0425***0.0373***0.0398***0.0380***0.0355***0.0350*** (0.0130)(0.0124)(0.0128)(0.0123)(0.0119)(0.0115) Separated 0.01390.0096 0.01530.0138 0.00580.0054 (0.0211)(0.0207)(0.0208)(0.0206)(0.0194)(0.0191) Model 1 (Total Effect) 2SLSOLS Alcohol abuse and/or dependence prior to the last 12 months Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 101

Appendix A-1: (Continued) 87 Completed graduate degree 0.1541***0.1545***0.0706***0.0704*** (0.0144)(0.0144)(0.0138)(0.0137) Had some graduate studies 0.1558***0.1552***0.0847***0.0843*** (0.0198)(0.0197)(0.0187)(0.0185) Completed college 0.1297***0.1292***0.0593***0.0589*** (0.0128)(0.0127)(0.0123)(0.0121) Had some college educatio n 0.1004***0.0989***0.0473***0.0468*** (0.0115)(0.0111)(0.0110)(0.0105) Completed high school 0.0875***0.0868***0.0460***0.0458*** (0.0117)(0.0116)(0.0110)(0.0109) Excellent healt h 0.6725***0.6731*** (0.0208)(0.0204) Very good healt h 0.6553***0.6554*** (0.0203)(0.0203) Good healt h 0.6191***0.6190*** (0.0205)(0.0205) Fair healt h 0.4393***0.4388*** (0.0225)(0.0222) Constant 0.8086***0.8035***0.8221***0.8165***0.7612***0.7596***0.1676**0.1669** (0.0871)(0.0863)(0.0864)(0.0858)(0.0854)(0.0852)(0.0817)(0.0816) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 102

Appendix A-2: Current Employment (F ull Time/Part Time) versus Alcohol Abuse and/or Dependence in the La st 12 Months: the Male Sample 88 Male (n=8,673) Variables -0.2666**-0.0229**-0.2368*-0.0095 -0.1186-0.0088 -0.0470-0.0052 (0.1328)(0.0100)(0.1439)(0.0100)(0.1422)(0.0099)(0.1357)(0.0092) Age 0.00580.0084**0.00240.0045 0.00300.0041 0.00290.0033 (0.0046)(0.0043)(0.0046)(0.0043)(0.0045)(0.0042)(0.0042)(0.0039) Age squared -0.0001**-0.0001***-0.0001-0.0001*-0.0001-0.0001*-0.0001-0.0001 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0000)(0.0000) White 0.0272***0.0194**0.0310***0.0243***-0.0012-0.0043 -0.0007-0.0019 (0.0099)(0.0086)(0.0099)(0.0087)(0.0099)(0.0089)(0.0092)(0.0083) Black -0.0512***-0.0519***-0.0367***-0.0345***-0.0543***-0.0528***-0.0357***-0.0351*** (0.0114)(0.0111)(0.0115)(0.0111)(0.0114)(0.0111)(0.0106)(0.0104) American Indian/Alaska Native 0.0029-0.0032 0.00860.0039 -0.0119-0.0139 0.01390.0133 (0.0284)(0.0273)(0.0280)(0.0271)(0.0272)(0.0269)(0.0252)(0.0251) Asian/Pacific Islander 0.00220.0148 -0.00060.0088 -0.0336*-0.0297 -0.0221-0.0206 (0.0206)(0.0188)(0.0201)(0.0187)(0.0196)(0.0188)(0.0182)(0.0175) Living in MSA in central city 0.00350.0050 0.01490.0178*0.00440.0056 -0.0060-0.0056 (0.0104)(0.0100)(0.0105)(0.0100)(0.0102)(0.0100)(0.0094)(0.0094) Living in MSA outside central cit y 0.0342***0.0390***0.0348***0.0384***0.0226**0.0242***0.00700.0075 (0.0099)(0.0093)(0.0098)(0.0092)(0.0095)(0.0092)(0.0088)(0.0086) State unemployment rates -0.0098**-0.0077**-0.0091**-0.0070**-0.0073*-0.0062*-0.0050-0.0046 (0.0038)(0.0036)(0.0039)(0.0035)(0.0038)(0.0035)(0.0035)(0.0033) Father/mother ever depressed -0.0007-0.0064 0.0010-0.0042 -0.0073-0.0100 -0.0036-0.0046 (0.0100)(0.0093)(0.0100)(0.0092)(0.0098)(0.0091)(0.0090)(0.0085) Brother/sister ever depressed 0.00600.0028 0.00760.0052 0.00260.0013 0.01460.0141 (0.0108)(0.0103)(0.0107)(0.0103)(0.0104)(0.0102)(0.0096)(0.0095) Father/mother had behavior problems -0.0126-0.0256*-0.0132-0.0242*-0.0143-0.0196 -0.0057-0.0077 (0.0156)(0.0134)(0.0154)(0.0133)(0.0150)(0.0132)(0.0139)(0.0123) Brother/sister had behavior problems -0.0212-0.0322***-0.0248*-0.0347***-0.0263**-0.0310***-0.0282**-0.0299*** (0.0134)(0.0116)(0.0133)(0.0115)(0.0129)(0.0114)(0.0119)(0.0106) Number of children -0.0041-0.0053**0.0001-0.0004 0.00130.0011 (0.0026)(0.0024)(0.0025)(0.0024)(0.0023)(0.0022) Married 0.0928***0.1069***0.0942***0.1009***0.0759***0.0784*** (0.0133)(0.0096)(0.0130)(0.0095)(0.0120)(0.0089) Cohabitating 0.0572***0.0531***0.0618***0.0599***0.0518***0.0511*** (0.0193)(0.0186)(0.0187)(0.0184)(0.0173)(0.0172) Widowed -0.0227-0.0299 -0.0229-0.0264 0.00560.0043 (0.0442)(0.0427)(0.0428)(0.0423)(0.0397)(0.0394) Divorced 0.0458***0.0367***0.0415***0.0373***0.0365***0.0349*** (0.0140)(0.0124)(0.0136)(0.0123)(0.0126)(0.0115) Separated 0.02530.0094 0.02110.0136 0.00830.0054 (0.0236)(0.0208)(0.0229)(0.0206)(0.0214)(0.0192) Model 2 Model 3 Model 4 (Direct Effect) Alcohol abuse and/or dependence in the last 12 months Model 1 (Total Effect) 2SLSOLS 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 103

Appendix A-2: (Continued) 89 Completed graduate degree 0.1522***0.1545***0.0700***0.0703*** (0.0148)(0.0144)(0.0138)(0.0137) Had some graduate studies 0.1556***0.1549***0.0848***0.0842*** (0.0199)(0.0197)(0.0187)(0.0185) Completed college 0.1298***0.1289***0.0595***0.0588*** (0.0129)(0.0127)(0.0123)(0.0121) Had some college educatio n 0.1005***0.0983***0.0475***0.0465*** (0.0116)(0.0111)(0.0110)(0.0105) Completed high school 0.0885***0.0866***0.0465***0.0457*** (0.0120)(0.0117)(0.0112)(0.0109) Excellent healt h 0.6732***0.6735*** (0.0204)(0.0204) Very good healt h 0.6564***0.6555*** (0.0205)(0.0203) Good healt h 0.6207***0.6192*** (0.0211)(0.0205) Fair healt h 0.4406***0.4386*** (0.0231)(0.0222) Constant 0.9166***0.8123***0.9182***0.8195***0.8095***0.7626***0.1860*0.1687** (0.1057)(0.0864)(0.1082)(0.0859)(0.1051)(0.0854)(0.0993)(0.0817) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 104

Appendix A-3: Currently Working Full Time (35+ Hours a Week) versus Alcohol Abuse and/or Dependence prior to the Last 12 Months: the Male Sample 90 Male (n=8,673) Variables -0.1226**-0.0138*-0.1080*-0.0107 -0.0521-0.0113 -0.0210-0.0012 (0.0564)(0.0079)(0.0576)(0.0078)(0.0571)(0.0077)(0.0554)(0.0073) Age 0.0180***0.0165***0.0114**0.0101**0.0100**0.0095**0.0087*0.0085* (0.0049)(0.0048)(0.0049)(0.0048)(0.0048)(0.0048)(0.0046)(0.0045) Age squared -0.0003***-0.0002***-0.0002***-0.0002***-0.0002***-0.0002***-0.0001**-0.0001** (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0532***0.0342***0.0585***0.0413***0.01200.0050 0.00860.0052 (0.0139)(0.0099)(0.0142)(0.0099)(0.0141)(0.0102)(0.0135)(0.0096) Black -0.0431***-0.0447***-0.0208*-0.0219*-0.0458***-0.0460***-0.0285**-0.0285** (0.0127)(0.0126)(0.0126)(0.0125)(0.0126)(0.0126)(0.0119)(0.0119) American Indian/Alaska Native 0.0022-0.0151 0.0101-0.0053 -0.0223-0.0285 -0.0001-0.0030 (0.0326)(0.0310)(0.0322)(0.0306)(0.0316)(0.0304)(0.0299)(0.0288) Asian/Pacific Islander -0.00410.0076 -0.00870.0006 -0.0510**-0.0474**-0.0394*-0.0376* (0.0223)(0.0213)(0.0220)(0.0211)(0.0219)(0.0213)(0.0208)(0.0201) Living in MSA in central city -0.00250.0000 0.01660.0188*0.00310.0039 -0.0077-0.0073 (0.0116)(0.0114)(0.0115)(0.0114)(0.0114)(0.0113)(0.0108)(0.0107) Living in MSA outside central cit y 0.0470***0.0488***0.0481***0.0493***0.0313***0.0318***0.01480.0150 (0.0107)(0.0105)(0.0105)(0.0104)(0.0104)(0.0104)(0.0099)(0.0099) State unemployment rates -0.0097**-0.0088**-0.0089**-0.0080**-0.0074*-0.0070*-0.0056-0.0054 (0.0041)(0.0040)(0.0041)(0.0040)(0.0040)(0.0039)(0.0038)(0.0037) Father/mother ever depressed 0.0079-0.0030 0.01020.0003 -0.0024-0.0066 0.0002-0.0018 (0.0120)(0.0105)(0.0120)(0.0104)(0.0118)(0.0103)(0.0112)(0.0098) Brother/sister ever depressed -0.0029-0.0087 -0.0004-0.0054 -0.0079-0.0101 0.00320.0022 (0.0122)(0.0117)(0.0121)(0.0116)(0.0119)(0.0115)(0.0112)(0.0109) Father/mother had behavior problems -0.0079-0.0246 -0.0088-0.0232 -0.0112-0.0172 -0.0037-0.0066 (0.0177)(0.0153)(0.0174)(0.0151)(0.0171)(0.0149)(0.0162)(0.0141) Brother/sister had behavior problems -0.0188-0.0315**-0.0239-0.0349***-0.0257*-0.0302**-0.0276**-0.0298** (0.0148)(0.0131)(0.0146)(0.0130)(0.0143)(0.0128)(0.0135)(0.0122) Number of childre n -0.0046*-0.0052*0.00120.0009 0.00270.0026 (0.0027)(0.0027)(0.0027)(0.0027)(0.0026)(0.0026) Married 0.1452***0.1464***0.1384***0.1389***0.1159***0.1160*** (0.0109)(0.0108)(0.0108)(0.0107)(0.0102)(0.0102) Cohabitating 0.0639***0.0562***0.0678***0.0646***0.0570***0.0554*** (0.0217)(0.0210)(0.0213)(0.0208)(0.0202)(0.0197) Widowed -0.0108-0.0139 -0.0085-0.0098 0.02030.0197 (0.0487)(0.0482)(0.0478)(0.0477)(0.0452)(0.0452) Divorced 0.0767***0.0698***0.0731***0.0703***0.0686***0.0672*** (0.0147)(0.0140)(0.0144)(0.0139)(0.0137)(0.0132) Separated 0.0469**0.0411*0.0486**0.0463**0.0389*0.0377* (0.0239)(0.0234)(0.0235)(0.0232)(0.0222)(0.0220) Model 1 (Total Effect) 2SLSOLS Alcohol abuse and/or dependence prior to the last 12 months Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 105

Appendix A-3: (Continued) 91 Completed graduate degree 0.1932***0.1937***0.1080***0.1075*** (0.0163)(0.0163)(0.0158)(0.0158) Had some graduate studies 0.1957***0.1946***0.1232***0.1221*** (0.0223)(0.0222)(0.0215)(0.0212) Completed college 0.1663***0.1654***0.0947***0.0937*** (0.0144)(0.0144)(0.0141)(0.0138) Had some college educatio n 0.1299***0.1276***0.0762***0.0749*** (0.0130)(0.0126)(0.0126)(0.0120) Completed high school 0.1181***0.1171***0.0766***0.0760*** (0.0132)(0.0131)(0.0126)(0.0125) Excellent healt h 0.6551***0.6568*** (0.0238)(0.0234) Very good healt h 0.6398***0.6399*** (0.0233)(0.0233) Good healt h 0.5911***0.5908*** (0.0235)(0.0235) Fair healt h 0.4146***0.4131*** (0.0258)(0.0255) Constant 0.6252***0.6187***0.6573***0.6497***0.5792***0.5769***0.00530.0036 (0.0990)(0.0978)(0.0979)(0.0969)(0.0963)(0.0961)(0.0938)(0.0937) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 106

Appendix A-4: Currently Working Full Time (35+ Hours a Week) versus Alcohol Abuse and/or Dependence in th e Last 12 Months: the Male Sample 92 Male (n=8,673) Variables -0.3038**-0.0351***-0.2744*-0.0188*-0.1271-0.0182 -0.0485-0.0140 (0.1502)(0.0113)(0.1625)(0.0113)(0.1601)(0.0112)(0.1556)(0.0106) Age 0.0131**0.0160***0.00740.0098**0.00810.0091*0.0080*0.0083* (0.0052)(0.0048)(0.0052)(0.0048)(0.0050)(0.0048)(0.0048)(0.0045) Age squared -0.0002***-0.0002***-0.0002**-0.0002***-0.0002***-0.0002***-0.0001**-0.0001** (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0415***0.0329***0.0475***0.0400***0.00660.0035 0.00640.0054 (0.0112)(0.0098)(0.0112)(0.0098)(0.0111)(0.0101)(0.0105)(0.0096) Black -0.0441***-0.0448***-0.0246*-0.0222*-0.0477***-0.0463***-0.0292**-0.0287** (0.0130)(0.0125)(0.0130)(0.0125)(0.0128)(0.0126)(0.0121)(0.0119) American Indian/Alaska Native -0.0097-0.0165 -0.0014-0.0067 -0.0279-0.0299 -0.0024-0.0030 (0.0322)(0.0310)(0.0317)(0.0306)(0.0307)(0.0304)(0.0289)(0.0288) Asian/Pacific Islander -0.00660.0072 -0.00980.0008 -0.0509**-0.0470**-0.0392*-0.0380* (0.0233)(0.0213)(0.0227)(0.0211)(0.0221)(0.0212)(0.0209)(0.0201) Living in MSA in central city -0.00160.0001 0.01560.0188*0.00290.0040 -0.0077-0.0074 (0.0118)(0.0114)(0.0119)(0.0114)(0.0115)(0.0113)(0.0108)(0.0107) Living in MSA outside central cit y 0.0430***0.0484***0.0451***0.0491***0.0301***0.0316***0.01440.0148 (0.0113)(0.0105)(0.0110)(0.0104)(0.0107)(0.0104)(0.0101)(0.0099) State unemployment rates -0.0112***-0.0089**-0.0104**-0.0081**-0.0081*-0.0070*-0.0058-0.0055 (0.0043)(0.0040)(0.0044)(0.0040)(0.0042)(0.0039)(0.0040)(0.0037) Father/mother ever depressed 0.0026-0.0036 0.0055-0.0003 -0.0047-0.0073 -0.0008-0.0016 (0.0114)(0.0105)(0.0113)(0.0104)(0.0110)(0.0103)(0.0103)(0.0097) Brother/sister ever depressed -0.0054-0.0089 -0.0030-0.0057 -0.0092-0.0104 0.00270.0023 (0.0123)(0.0117)(0.0120)(0.0116)(0.0117)(0.0115)(0.0110)(0.0109) Father/mother had behavior problems -0.0106-0.0249 -0.0115-0.0239 -0.0127-0.0180 -0.0045-0.0061 (0.0176)(0.0152)(0.0174)(0.0150)(0.0168)(0.0149)(0.0159)(0.0141) Brother/sister had behavior problems -0.0194-0.0315**-0.0242-0.0353***-0.0261*-0.0307**-0.0279**-0.0293** (0.0151)(0.0131)(0.0151)(0.0130)(0.0146)(0.0128)(0.0137)(0.0121) Number of childre n -0.0038-0.0052*0.00150.0009 0.00280.0026 (0.0029)(0.0027)(0.0028)(0.0027)(0.0026)(0.0026) Married 0.1295***0.1453***0.1313***0.1379***0.1131***0.1152*** (0.0150)(0.0108)(0.0146)(0.0107)(0.0137)(0.0102) Cohabitating 0.0603***0.0557***0.0658***0.0640***0.0561***0.0555*** (0.0218)(0.0210)(0.0211)(0.0208)(0.0199)(0.0197) Widowed -0.0056-0.0136 -0.0062-0.0096 0.02120.0201 (0.0499)(0.0482)(0.0482)(0.0477)(0.0455)(0.0452) Divorced 0.0800***0.0698***0.0744***0.0702***0.0690***0.0677*** (0.0158)(0.0140)(0.0153)(0.0139)(0.0145)(0.0132) Separated 0.0597**0.0418*0.0543**0.0469**0.0410*0.0386* (0.0267)(0.0234)(0.0258)(0.0232)(0.0245)(0.0220) Model 1 (Total Effect) 2SLSOLS Alcohol abuse and/or dependence in the last 12 months Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 107

Appendix A-4: (Continued) 93 Completed graduate degree 0.1912***0.1935***0.1071***0.1074*** (0.0167)(0.0163)(0.0158)(0.0158) Had some graduate studies 0.1952***0.1945***0.1229***0.1223*** (0.0224)(0.0222)(0.0214)(0.0212) Completed college 0.1662***0.1653***0.0945***0.0939*** (0.0145)(0.0144)(0.0141)(0.0138) Had some college educatio n 0.1295***0.1273***0.0759***0.0751*** (0.0130)(0.0126)(0.0126)(0.0120) Completed high school 0.1190***0.1171***0.0769***0.0762*** (0.0135)(0.0131)(0.0129)(0.0125) Excellent healt h 0.6565***0.6568*** (0.0234)(0.0234) Very good healt h 0.6409***0.6402*** (0.0235)(0.0233) Good healt h 0.5925***0.5913*** (0.0242)(0.0235) Fair healt h 0.4153***0.4137*** (0.0265)(0.0255) Constant 0.7479***0.6329***0.7681***0.6571***0.6305***0.5840***0.02370.0093 (0.1197)(0.0979)(0.1222)(0.0971)(0.1184)(0.0962)(0.1139)(0.0938) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 108

Appendix A-5: Current Employment (F ull Time/Part Time) versus Alcohol Abuse and/or Dependence prior to the Last 12 Months: the Female Sample 94 Female (n=9,355) Variables -0.2342***-0.0084 -0.1454*-0.0061 -0.0948-0.0103 0.0006-0.0074 (0.0777)(0.0089)(0.0746)(0.0088)(0.0734)(0.0088)(0.0685)(0.0082) Age 0.0111**0.0072 0.0098**0.0070 0.0083*0.0067 0.0076*0.0078* (0.0048)(0.0044)(0.0047)(0.0044)(0.0046)(0.0044)(0.0043)(0.0041) Age squared -0.0002***-0.0001**-0.0001**-0.0001**-0.0001**-0.0001**-0.0001*-0.0001* (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0953***0.0573***0.0649***0.0420***0.0274*0.0141 0.01200.0132 (0.0165)(0.0099)(0.0158)(0.0099)(0.0152)(0.0100)(0.0142)(0.0093) Black -0.0115-0.0112 0.00220.0040 -0.0188*-0.0172 -0.0079-0.0081 (0.0113)(0.0109)(0.0112)(0.0110)(0.0112)(0.0111)(0.0104)(0.0103) American Indian/Alaska Native 0.0245-0.0191 0.0077-0.0191 -0.0221-0.0377 -0.0153-0.0138 (0.0336)(0.0291)(0.0325)(0.0289)(0.0318)(0.0286)(0.0294)(0.0266) Asian/Pacific Islander 0.0722***0.0774***0.0538**0.0570***0.02220.0246 0.02400.0238 (0.0226)(0.0218)(0.0220)(0.0217)(0.0218)(0.0216)(0.0202)(0.0201) Living in MSA in central city 0.01100.0089 0.00750.0076 0.00040.0008 -0.0086-0.0086 (0.0112)(0.0109)(0.0110)(0.0108)(0.0108)(0.0108)(0.0100)(0.0100) Living in MSA outside central cit y 0.0396***0.0397***0.0279***0.0284***0.0170*0.0177*0.00070.0007 (0.0106)(0.0102)(0.0103)(0.0102)(0.0102)(0.0102)(0.0095)(0.0095) State unemployment rates -0.0087**-0.0057 -0.0078*-0.0058 -0.0068-0.0055 -0.0025-0.0026 (0.0044)(0.0042)(0.0043)(0.0041)(0.0042)(0.0041)(0.0039)(0.0038) Father/mother ever depressed 0.0059-0.0165*-0.0031-0.0166*-0.0148-0.0227**-0.0130-0.0123 (0.0121)(0.0091)(0.0116)(0.0090)(0.0112)(0.0089)(0.0104)(0.0083) Brother/sister ever depressed 0.0035-0.0020 0.0025-0.0008 -0.0004-0.0024 0.00460.0048 (0.0103)(0.0098)(0.0100)(0.0097)(0.0098)(0.0096)(0.0091)(0.0089) Father/mother had behavior problems -0.0343**-0.0510***-0.0324**-0.0425***-0.0319**-0.0380***-0.0128-0.0122 (0.0144)(0.0128)(0.0140)(0.0127)(0.0137)(0.0126)(0.0127)(0.0118) Brother/sister had behavior problems -0.0088-0.0343***-0.0155-0.0313***-0.0200-0.0294***-0.0213*-0.0204** (0.0146)(0.0114)(0.0142)(0.0113)(0.0139)(0.0112)(0.0128)(0.0104) Number of childre n -0.0246***-0.0234***-0.0144***-0.0138***-0.0094***-0.0095*** (0.0025)(0.0024)(0.0025)(0.0025)(0.0024)(0.0023) Married 0.0724***0.0793***0.0663***0.0707***0.0438***0.0434*** (0.0108)(0.0100)(0.0107)(0.0099)(0.0099)(0.0093) Cohabitating 0.02190.0114 0.02560.0193 -0.0053-0.0047 (0.0215)(0.0205)(0.0212)(0.0203)(0.0196)(0.0190) Widowed -0.0539**-0.0548**-0.0473*-0.0480**-0.0234-0.0234 (0.0247)(0.0244)(0.0242)(0.0241)(0.0225)(0.0225) Divorced 0.01350.0152 0.01100.0123 0.01030.0102 (0.0121)(0.0119)(0.0119)(0.0118)(0.0111)(0.0110) Separated 0.00520.0044 0.01300.0125 0.01780.0179 (0.0174)(0.0172)(0.0171)(0.0170)(0.0159)(0.0158) Alcohol abuse and/or dependence prior to the last 12 months OLS Model 1 (Total Effect) 2SLS OLS Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLS Model 2 Model 3

PAGE 109

Appendix A-5: (Continued) 95 Completed graduate degree 0.1958***0.1934***0.1102***0.1105*** (0.0164)(0.0162)(0.0156)(0.0154) Had some graduate studies 0.1987***0.1922***0.1109***0.1115*** (0.0211)(0.0202)(0.0198)(0.0190) Completed college 0.1810***0.1772***0.1023***0.1027*** (0.0152)(0.0148)(0.0144)(0.0140) Had some college educatio n 0.1410***0.1359***0.0802***0.0807*** (0.0135)(0.0126)(0.0127)(0.0119) Completed high school 0.1153***0.1144***0.0670***0.0671*** (0.0134)(0.0133)(0.0125)(0.0125) Excellent healt h 0.6235***0.6232*** (0.0193)(0.0191) Very good healt h 0.6176***0.6176*** (0.0190)(0.0190) Good healt h 0.5831***0.5831*** (0.0191)(0.0191) Fair healt h 0.3530***0.3529*** (0.0209)(0.0209) Constant 0.7382***0.7727***0.7481***0.7713***0.6492***0.6643***0.07890.0776 (0.0951)(0.0913)(0.0925)(0.0905)(0.0915)(0.0901)(0.0866)(0.0859) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 110

Appendix A-6: Current Employment (F ull Time/Part Time) versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Female Sample 96 Female (n=9,355) Variables -0.7488***-0.0243 -0.4936*-0.0149 -0.3504-0.0122 0.0020-0.0150 (0.2794)(0.0155)(0.2692)(0.0154)(0.2678)(0.0153)(0.2426)(0.0142) Age 0.00610.0071 0.00700.0069 0.00650.0065 0.0076*0.0076* (0.0050)(0.0044)(0.0047)(0.0044)(0.0045)(0.0044)(0.0041)(0.0041) Age squared -0.0001**-0.0001**-0.0001**-0.0001**-0.0001**-0.0001*-0.0001*-0.0001* (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0878***0.0569***0.0613***0.0417***0.0274*0.0130 0.01200.0127 (0.0161)(0.0098)(0.0151)(0.0098)(0.0152)(0.0099)(0.0138)(0.0092) Black -0.0058-0.0110 0.00270.0040 -0.0178-0.0170 -0.0079-0.0080 (0.0123)(0.0109)(0.0116)(0.0110)(0.0114)(0.0111)(0.0103)(0.0103) American Indian/Alaska Native 0.0252-0.0192 0.0102-0.0194 -0.0180-0.0389 -0.0153-0.0143 (0.0366)(0.0291)(0.0345)(0.0288)(0.0336)(0.0286)(0.0305)(0.0266) Asian/Pacific Islander 0.0836***0.0777***0.0623***0.0573***0.02970.0251 0.02400.0242 (0.0243)(0.0218)(0.0229)(0.0217)(0.0225)(0.0216)(0.0204)(0.0201) Living in MSA in central city 0.01890.0091 0.01110.0078 0.00360.0009 -0.0086-0.0084 (0.0126)(0.0109)(0.0115)(0.0108)(0.0113)(0.0108)(0.0102)(0.0100) Living in MSA outside central cit y 0.0401***0.0397***0.0282***0.0284***0.0179*0.0178*0.00070.0007 (0.0114)(0.0102)(0.0107)(0.0102)(0.0104)(0.0102)(0.0095)(0.0095) State unemployment rates -0.0110**-0.0058 -0.0095**-0.0059 -0.0080*-0.0055 -0.0025-0.0026 (0.0050)(0.0042)(0.0048)(0.0041)(0.0047)(0.0041)(0.0042)(0.0038) Father/mother ever depressed -0.0071-0.0170*-0.0110-0.0170*-0.0192**-0.0235***-0.0130-0.0128 (0.0107)(0.0090)(0.0100)(0.0089)(0.0097)(0.0089)(0.0088)(0.0083) Brother/sister ever depressed 0.0180-0.0016 0.0121-0.0005 0.0068-0.0023 0.00450.0050 (0.0132)(0.0098)(0.0124)(0.0097)(0.0122)(0.0096)(0.0110)(0.0089) Father/mother had behavior problems -0.0361**-0.0511***-0.0337**-0.0427***-0.0324**-0.0385***-0.0128-0.0125 (0.0154)(0.0128)(0.0143)(0.0127)(0.0138)(0.0126)(0.0125)(0.0117) Brother/sister had behavior problems -0.0159-0.0346***-0.0194-0.0316***-0.0219*-0.0303***-0.0212*-0.0208** (0.0145)(0.0113)(0.0136)(0.0112)(0.0132)(0.0111)(0.0120)(0.0104) Number of childre n -0.0247***-0.0234***-0.0150***-0.0138***-0.0094***-0.0095*** (0.0026)(0.0024)(0.0027)(0.0025)(0.0025)(0.0023) Married 0.0602***0.0790***0.0575***0.0707***0.0438***0.0432*** (0.0149)(0.0100)(0.0146)(0.0100)(0.0133)(0.0093) Cohabitating 0.01310.0110 0.01970.0186 -0.0053-0.0052 (0.0216)(0.0205)(0.0209)(0.0203)(0.0190)(0.0189) Widowed -0.0513**-0.0548**-0.0458*-0.0480**-0.0234-0.0233 (0.0256)(0.0244)(0.0248)(0.0241)(0.0225)(0.0225) Divorced 0.01620.0153 0.01290.0125 0.01030.0103 (0.0125)(0.0119)(0.0121)(0.0118)(0.0110)(0.0110) Separated 0.00600.0044 0.01320.0124 0.01780.0178 (0.0181)(0.0172)(0.0175)(0.0170)(0.0159)(0.0158) Model 1 (Total Effect) 2SLSOLS Alcohol abuse and/or dependence in the last 12 months Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 111

Appendix A-6: (Continued) 97 Completed graduate degree 0.1848***0.1928***0.1103***0.1099*** (0.0178)(0.0162)(0.0164)(0.0154) Had some graduate studies 0.1909***0.1914***0.1109***0.1109*** (0.0207)(0.0202)(0.0190)(0.0190) Completed college 0.1749***0.1766***0.1023***0.1023*** (0.0152)(0.0147)(0.0140)(0.0140) Had some college educatio n 0.1358***0.1352***0.0802***0.0802*** (0.0130)(0.0126)(0.0119)(0.0119) Completed high school 0.1133***0.1143***0.0670***0.0669*** (0.0137)(0.0133)(0.0125)(0.0125) Excellent healt h 0.6235***0.6235*** (0.0191)(0.0191) Very good healt h 0.6175***0.6177*** (0.0191)(0.0190) Good healt h 0.5831***0.5832*** (0.0191)(0.0191) Fair healt h 0.3530***0.3529*** (0.0209)(0.0209) Constant 0.8674***0.7770***0.8333***0.7741***0.7117***0.6677***0.07860.0807 (0.1072)(0.0913)(0.1007)(0.0905)(0.0988)(0.0901)(0.0913)(0.0859) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 112

Appendix A-7: Currently Working Full Time (35+ Hours a Week) versus Alcohol Abuse and/or Dependence prior to the Last 12 Months: the Female Sample 98 Female (n=9,355) Variables -0.3544***0.0190 -0.2490**0.0110 -0.1727*0.0057 -0.08210.0090 (0.1040)(0.0117)(0.0990)(0.0116)(0.0967)(0.0114)(0.0932)(0.0111) Age 0.0029-0.0035 0.00580.0005 0.00380.0004 0.00310.0014 (0.0064)(0.0058)(0.0063)(0.0058)(0.0061)(0.0057)(0.0058)(0.0055) Age squared -0.00010.0000 -0.00010.0000 -0.00010.0000 0.00000.0000 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.1062***0.0434***0.0583***0.0156 0.0068-0.0211 -0.0082-0.0226* (0.0221)(0.0129)(0.0209)(0.0129)(0.0200)(0.0131)(0.0194)(0.0126) Black 0.0359**0.0365**0.0261*0.0294**-0.00020.0031 0.01030.0122 (0.0151)(0.0144)(0.0149)(0.0145)(0.0148)(0.0145)(0.0142)(0.0140) American Indian/Alaska Native 0.0370-0.0351 0.0189-0.0312 -0.0210-0.0540 -0.0147-0.0314 (0.0450)(0.0382)(0.0431)(0.0378)(0.0418)(0.0374)(0.0400)(0.0361) Asian/Pacific Islander 0.0593*0.0679**0.03860.0446 -0.0056-0.0004 -0.0038-0.0012 (0.0303)(0.0286)(0.0292)(0.0284)(0.0288)(0.0283)(0.0275)(0.0273) Living in MSA in central city 0.0424***0.0389***0.01690.0172 0.00590.0067 -0.0025-0.0021 (0.0150)(0.0143)(0.0146)(0.0142)(0.0143)(0.0141)(0.0137)(0.0136) Living in MSA outside central cit y 0.0423***0.0424***0.02060.0216 0.00510.0065 -0.0103-0.0097 (0.0142)(0.0135)(0.0137)(0.0134)(0.0135)(0.0133)(0.0129)(0.0128) State unemployment rates -0.00190.0032 -0.00180.0018 -0.00050.0021 0.00350.0049 (0.0059)(0.0055)(0.0057)(0.0054)(0.0056)(0.0053)(0.0054)(0.0052) Father/mother ever depressed 0.0166-0.0206*0.0026-0.0227*-0.0146-0.0313***-0.0129-0.0213* (0.0162)(0.0119)(0.0154)(0.0117)(0.0148)(0.0116)(0.0141)(0.0112) Brother/sister ever depressed 0.01250.0034 0.01000.0039 0.00510.0009 0.01000.0080 (0.0138)(0.0128)(0.0132)(0.0127)(0.0129)(0.0125)(0.0123)(0.0121) Father/mother had behavior problems -0.0280-0.0556***-0.0294-0.0482***-0.0291-0.0419**-0.0112-0.0175 (0.0193)(0.0168)(0.0185)(0.0167)(0.0180)(0.0165)(0.0173)(0.0159) Brother/sister had behavior problems -0.0035-0.0455***-0.0117-0.0411***-0.0181-0.0380***-0.0195-0.0296** (0.0196)(0.0149)(0.0188)(0.0147)(0.0183)(0.0146)(0.0175)(0.0141) Number of childre n -0.0479***-0.0456***-0.0336***-0.0324***-0.0289***-0.0282*** (0.0033)(0.0031)(0.0033)(0.0032)(0.0032)(0.0031) Married -0.00630.0065 -0.0139-0.0046 -0.0354***-0.0309** (0.0143)(0.0131)(0.0141)(0.0130)(0.0135)(0.0126) Cohabitating 0.02730.0077 0.03180.0185 0.0026-0.0042 (0.0286)(0.0269)(0.0279)(0.0266)(0.0267)(0.0257) Widowed -0.0695**-0.0713**-0.0604*-0.0620**-0.0375-0.0382 (0.0327)(0.0319)(0.0319)(0.0315)(0.0306)(0.0305) Divorced 0.0505***0.0537***0.0484***0.0511***0.0475***0.0488*** (0.0161)(0.0156)(0.0157)(0.0155)(0.0150)(0.0149) Separated 0.03320.0317 0.0445**0.0433*0.0487**0.0481** (0.0231)(0.0225)(0.0226)(0.0223)(0.0216)(0.0215) Model 1 (Total Effect) 2SLSOLS Alcohol abuse and/or dependence prior to the last 12 months Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS Model 2 Model 3

PAGE 113

Appendix A-7: (Continued) 99 Completed graduate degree 0.2862***0.2812***0.2044***0.2011*** (0.0217)(0.0212)(0.0212)(0.0209) Had some graduate studies 0.2696***0.2558***0.1861***0.1784*** (0.0278)(0.0264)(0.0270)(0.0257) Completed college 0.2286***0.2205***0.1535***0.1487*** (0.0200)(0.0193)(0.0196)(0.0189) Had some college educatio n 0.1803***0.1694***0.1226***0.1167*** (0.0177)(0.0165)(0.0173)(0.0161) Completed high school 0.1419***0.1402***0.0963***0.0953*** (0.0177)(0.0174)(0.0170)(0.0169) Excellent healt h 0.5695***0.5734*** (0.0263)(0.0259) Very good healt h 0.5590***0.5591*** (0.0258)(0.0258) Good healt h 0.5254***0.5255*** (0.0259)(0.0258) Fair healt h 0.3014***0.3021*** (0.0284)(0.0283) Constant 0.6921***0.7492***0.7088***0.7520***0.5812***0.6130***0.06620.0810 (0.1272)(0.1198)(0.1227)(0.1185)(0.1206)(0.1178)(0.1179)(0.1165) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 114

Appendix A-8: Currently Working Full Time (35+ Hours a Week) versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Female Sample 100 Female (n=9,355) Variables -1.2083***0.0138 -0.9339**0.0014 -0.7258**0.0063 -0.38810.0039 (0.3889)(0.0204)(0.3721)(0.0202)(0.3651)(0.0200)(0.3362)(0.0193) Age -0.0048-0.0032 0.00100.0007 0.00050.0005 0.00160.0016 (0.0069)(0.0058)(0.0064)(0.0058)(0.0061)(0.0057)(0.0057)(0.0055) Age squared 0.00000.0000 0.00000.0000 0.00000.0000 0.00000.0000 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0981***0.0460***0.0557***0.0174 0.0105-0.0205 -0.0047-0.0213* (0.0224)(0.0128)(0.0208)(0.0128)(0.0208)(0.0130)(0.0192)(0.0125) Black 0.0451***0.0363**0.0268*0.0293**0.00130.0030 0.01110.0120 (0.0171)(0.0144)(0.0161)(0.0145)(0.0155)(0.0145)(0.0143)(0.0140) American Indian/Alaska Native 0.0427-0.0323 0.0286-0.0291 -0.0081-0.0533 -0.0058-0.0300 (0.0509)(0.0382)(0.0477)(0.0377)(0.0459)(0.0374)(0.0423)(0.0361) Asian/Pacific Islander 0.0772**0.0673**0.0540*0.0443 0.0092-0.0007 0.0038-0.0015 (0.0339)(0.0286)(0.0317)(0.0284)(0.0306)(0.0283)(0.0282)(0.0273) Living in MSA in central city 0.0553***0.0389***0.02370.0172 0.01240.0066 0.0009-0.0022 (0.0176)(0.0143)(0.0160)(0.0142)(0.0153)(0.0141)(0.0142)(0.0136) Living in MSA outside central cit y 0.0431***0.0424***0.02110.0216 0.00670.0064 -0.0096-0.0098 (0.0158)(0.0135)(0.0148)(0.0134)(0.0142)(0.0133)(0.0131)(0.0128) State unemployment rates -0.00580.0030 -0.00530.0017 -0.00350.0021 0.00180.0048 (0.0070)(0.0055)(0.0066)(0.0054)(0.0063)(0.0053)(0.0059)(0.0052) Father/mother ever depressed -0.0021-0.0189 -0.0099-0.0217*-0.0215-0.0309***-0.0156-0.0205* (0.0149)(0.0118)(0.0138)(0.0117)(0.0132)(0.0116)(0.0122)(0.0112) Brother/sister ever depressed 0.0365**0.0035 0.0288*0.0041 0.02050.0009 0.01850.0081 (0.0184)(0.0128)(0.0171)(0.0127)(0.0166)(0.0126)(0.0153)(0.0121) Father/mother had behavior problems -0.0292-0.0545***-0.0299-0.0475***-0.0283-0.0416**-0.0099-0.0170 (0.0214)(0.0168)(0.0197)(0.0166)(0.0188)(0.0165)(0.0174)(0.0159) Brother/sister had behavior problems -0.0122-0.0437***-0.0161-0.0399***-0.0194-0.0375***-0.0190-0.0287** (0.0202)(0.0149)(0.0189)(0.0147)(0.0180)(0.0145)(0.0166)(0.0141) Number of childre n -0.0484***-0.0457***-0.0351***-0.0324***-0.0297***-0.0283*** (0.0036)(0.0031)(0.0037)(0.0032)(0.0034)(0.0031) Married -0.03070.0060 -0.0333*-0.0047 -0.0465**-0.0312** (0.0206)(0.0131)(0.0199)(0.0130)(0.0184)(0.0126) Cohabitating 0.01260.0085 0.02140.0189 -0.0023-0.0036 (0.0298)(0.0269)(0.0285)(0.0266)(0.0263)(0.0257) Widowed -0.0646*-0.0713**-0.0570*-0.0619**-0.0353-0.0381 (0.0354)(0.0319)(0.0338)(0.0315)(0.0312)(0.0305) Divorced 0.0554***0.0536***0.0518***0.0510***0.0492***0.0487*** (0.0173)(0.0156)(0.0165)(0.0155)(0.0153)(0.0149) Separated 0.03490.0318 0.0450*0.0433*0.0491**0.0482** (0.0250)(0.0225)(0.0238)(0.0223)(0.0220)(0.0215) 2SLSOLS Model 1 (Total Effect) Alcohol abuse and/or dependence in the last 12 months Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 115

Appendix A-8: (Continued) 101 Completed graduate degree 0.2641***0.2815***0.1921***0.2015*** (0.0243)(0.0212)(0.0228)(0.0209) Had some graduate studies 0.2553***0.2563***0.1785***0.1792*** (0.0282)(0.0264)(0.0263)(0.0257) Completed college 0.2171***0.2208***0.1471***0.1491*** (0.0207)(0.0193)(0.0194)(0.0189) Had some college educatio n 0.1710***0.1697***0.1177***0.1172*** (0.0177)(0.0165)(0.0165)(0.0161) Completed high school 0.1381***0.1402***0.0940***0.0954*** (0.0187)(0.0174)(0.0174)(0.0169) Excellent healt h 0.5722***0.5730*** (0.0265)(0.0259) Very good healt h 0.5621***0.5591*** (0.0264)(0.0258) Good healt h 0.5271***0.5255*** (0.0264)(0.0258) Fair healt h 0.3010***0.3021*** (0.0289)(0.0283) Constant 0.8970***0.7445***0.8656***0.7500***0.7065***0.6111***0.12900.0790 (0.1492)(0.1199)(0.1392)(0.1185)(0.1347)(0.1179)(0.1265)(0.1165) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 116

Appendix A-9: Current Employment (Full Time/Part Time) versus the Number of Drinks Consumed in th e Last 12 Months: the Male Sample 102 Male (n=8,673) Variables -0.00018*-0.00003***-0.00016-0.00003***-0.00009-0.00002***-0.00005-0.00002*** (0.0001)(0.0000)(0.0001)(0.0000)(0.0001)(0.0000)(0.0001)(0.0000) Age 0.0087**0.0087**0.00480.0046 0.00420.0042 0.00340.0033 (0.0044)(0.0043)(0.0044)(0.0043)(0.0043)(0.0042)(0.0039)(0.0039) Age squared -0.0001***-0.0001***-0.0001*-0.0001*-0.0001*-0.0001*-0.0001-0.0001 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0000)(0.0000) White 0.0290***0.0205**0.0330***0.0254***0.0020-0.0030 0.0015-0.0007 (0.0104)(0.0086)(0.0107)(0.0087)(0.0122)(0.0089)(0.0115)(0.0083) Black -0.0482***-0.0512***-0.0337***-0.0343***-0.0518***-0.0525***-0.0346***-0.0348*** (0.0116)(0.0111)(0.0114)(0.0110)(0.0113)(0.0111)(0.0104)(0.0104) American Indian/Alaska Native 0.0056-0.0021 0.01100.0048 -0.0095-0.0130 0.01540.0140 (0.0286)(0.0273)(0.0282)(0.0271)(0.0277)(0.0269)(0.0256)(0.0250) Asian/Pacific Islander -0.00130.0129 -0.00290.0073 -0.0328*-0.0302 -0.0224-0.0212 (0.0214)(0.0187)(0.0208)(0.0187)(0.0195)(0.0188)(0.0181)(0.0175) Living in MSA in central city 0.00390.0049 0.01570.0176*0.00560.0056 -0.0055-0.0056 (0.0104)(0.0100)(0.0104)(0.0100)(0.0101)(0.0100)(0.0094)(0.0093) Living in MSA outside central cit y 0.0345***0.0386***0.0353***0.0380***0.0237**0.0241***0.00740.0075 (0.0099)(0.0093)(0.0097)(0.0092)(0.0093)(0.0092)(0.0086)(0.0086) State unemployment rates -0.0088**-0.0077**-0.0083**-0.0071**-0.0069*-0.0063*-0.0050-0.0048 (0.0037)(0.0035)(0.0037)(0.0035)(0.0037)(0.0035)(0.0034)(0.0033) Father/mother ever depressed -0.0097-0.0074 -0.0068-0.0048 -0.0110-0.0104 -0.0052-0.0049 (0.0097)(0.0092)(0.0096)(0.0092)(0.0092)(0.0091)(0.0086)(0.0085) Brother/sister ever depressed 0.00420.0028 0.00590.0052 0.00200.0014 0.01440.0142 (0.0107)(0.0103)(0.0106)(0.0102)(0.0103)(0.0102)(0.0095)(0.0095) Father/mother had behavior problems -0.0175-0.0251*-0.0179-0.0236*-0.0166-0.0192 -0.0062-0.0073 (0.0147)(0.0134)(0.0144)(0.0133)(0.0140)(0.0132)(0.0129)(0.0123) Brother/sister had behavior problems -0.0316***-0.0329***-0.0345***-0.0350***-0.0314***-0.0313***-0.0302***-0.0301*** (0.0120)(0.0115)(0.0118)(0.0114)(0.0114)(0.0114)(0.0106)(0.0106) Number of childre n -0.0034-0.0050**0.0003-0.0003 0.00150.0013 (0.0028)(0.0024)(0.0026)(0.0024)(0.0024)(0.0022) Married 0.0932***0.1052***0.0941***0.0997***0.0750***0.0773*** (0.0136)(0.0096)(0.0133)(0.0095)(0.0123)(0.0089) Cohabitating 0.0658***0.0550***0.0663***0.0613***0.0545***0.0523*** (0.0209)(0.0186)(0.0203)(0.0184)(0.0189)(0.0172) Widowed -0.0311-0.0303 -0.0276-0.0269 0.00360.0039 (0.0439)(0.0426)(0.0426)(0.0423)(0.0394)(0.0394) Divorced 0.0514***0.0387***0.0449***0.0389***0.0389***0.0363*** (0.0162)(0.0124)(0.0159)(0.0123)(0.0150)(0.0115) Separated 0.02190.0108 0.01980.0146 0.00880.0065 (0.0230)(0.0207)(0.0224)(0.0206)(0.0209)(0.0191) 2SLSOLS Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS 2SLSOLS Number of drinks consumed in the last 12 months Model 1 (Total Effect) 2SLSOLS

PAGE 117

Appendix A-9: (Continued) 103 Completed graduate degree 0.1435***0.1520***0.0655***0.0684*** (0.0203)(0.0144)(0.0176)(0.0137) Had some graduate studies 0.1447***0.1524***0.0795***0.0823*** (0.0236)(0.0197)(0.0212)(0.0185) Completed college 0.1220***0.1272***0.0558***0.0576*** (0.0155)(0.0127)(0.0138)(0.0121) Had some college educatio n 0.0957***0.0975***0.0455***0.0461*** (0.0116)(0.0111)(0.0107)(0.0105) Completed high school 0.0850***0.0861***0.0451***0.0454*** (0.0119)(0.0116)(0.0110)(0.0109) Excellent healt h 0.6734***0.6735*** (0.0204)(0.0204) Very good healt h 0.6576***0.6562*** (0.0209)(0.0203) Good healt h 0.6219***0.6201*** (0.0215)(0.0205) Fair healt h 0.4440***0.4406*** (0.0256)(0.0222) Constant 0.8514***0.8112***0.8623***0.8227***0.7869***0.7655***0.1799**0.1717** (0.0927)(0.0862)(0.0934)(0.0857)(0.0928)(0.0852)(0.0872)(0.0816) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 118

Appendix A-10: Current Employment (Full Time/Part Time) versus the Number of Days Drank 5+ Drinks in the Last 12 Months: the Male Sample 104 Male (n=8,673) Variables -0.00133**-0.00030***-0.00122-0.00024***-0.00067-0.00019***-0.00030-0.00016*** (0.0007)(0.0001)(0.0007)(0.0001)(0.0008)(0.0001)(0.0008)(0.0000) Age 0.0086**0.0087**0.00460.0046 0.00410.0041 0.00330.0033 (0.0044)(0.0043)(0.0043)(0.0043)(0.0042)(0.0042)(0.0039)(0.0039) Age squared -0.0001***-0.0001***-0.0001*-0.0001*-0.0001*-0.0001*-0.0001-0.0001 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0000)(0.0000) White 0.0239***0.0199**0.0291***0.0250***0.0003-0.0032 0.0001-0.0009 (0.0092)(0.0086)(0.0094)(0.0087)(0.0107)(0.0089)(0.0100)(0.0083) Black -0.0536***-0.0523***-0.0387***-0.0353***-0.0544***-0.0532***-0.0358***-0.0354*** (0.0113)(0.0110)(0.0116)(0.0110)(0.0114)(0.0111)(0.0106)(0.0104) American Indian/Alaska Native -0.0008-0.0031 0.00550.0041 -0.0122-0.0136 0.01380.0135 (0.0279)(0.0273)(0.0276)(0.0270)(0.0271)(0.0269)(0.0251)(0.0250) Asian/Pacific Islander 0.00050.0124 -0.00130.0072 -0.0311-0.0298 -0.0213-0.0209 (0.0206)(0.0187)(0.0201)(0.0187)(0.0190)(0.0188)(0.0177)(0.0175) Living in MSA in central city 0.00320.0047 0.01520.0174*0.00570.0057 -0.0055-0.0055 (0.0103)(0.0100)(0.0104)(0.0100)(0.0101)(0.0100)(0.0094)(0.0093) Living in MSA outside central cit y 0.0345***0.0384***0.0354***0.0379***0.0241***0.0242***0.00760.0076 (0.0098)(0.0093)(0.0096)(0.0092)(0.0093)(0.0092)(0.0086)(0.0086) State unemployment rates -0.0085**-0.0077**-0.0081**-0.0072**-0.0069*-0.0064*-0.0049-0.0048 (0.0037)(0.0035)(0.0037)(0.0035)(0.0036)(0.0035)(0.0034)(0.0033) Father/mother ever depressed -0.0095-0.0075 -0.0067-0.0048 -0.0108-0.0103 -0.0050-0.0049 (0.0095)(0.0092)(0.0095)(0.0092)(0.0092)(0.0091)(0.0085)(0.0085) Brother/sister ever depressed 0.00330.0027 0.00510.0051 0.00170.0014 0.01420.0142 (0.0106)(0.0103)(0.0105)(0.0102)(0.0102)(0.0102)(0.0095)(0.0095) Father/mother had behavior problems -0.0227-0.0258*-0.0224-0.0242*-0.0192-0.0198 -0.0077-0.0078 (0.0138)(0.0134)(0.0136)(0.0133)(0.0133)(0.0132)(0.0123)(0.0123) Brother/sister had behavior problems -0.0302**-0.0325***-0.0331***-0.0347***-0.0307***-0.0312***-0.0299***-0.0300*** (0.0119)(0.0115)(0.0117)(0.0114)(0.0114)(0.0114)(0.0106)(0.0106) Number of childre n -0.0025-0.0048**0.0006-0.0002 0.00160.0014 (0.0030)(0.0024)(0.0027)(0.0024)(0.0025)(0.0022) Married 0.0914***0.1043***0.0932***0.0992***0.0752***0.0769*** (0.0138)(0.0096)(0.0138)(0.0095)(0.0127)(0.0089) Cohabitating 0.0611***0.0545***0.0635***0.0608***0.0526***0.0519*** (0.0196)(0.0186)(0.0191)(0.0184)(0.0177)(0.0172) Widowed -0.0289-0.0299 -0.0265-0.0266 0.00420.0042 (0.0435)(0.0426)(0.0425)(0.0423)(0.0394)(0.0394) Divorced 0.0477***0.0385***0.0428***0.0386***0.0372***0.0360*** (0.0144)(0.0124)(0.0142)(0.0123)(0.0133)(0.0115) Separated 0.02420.0118 0.02090.0152 0.00870.0070 (0.0231)(0.0207)(0.0228)(0.0206)(0.0213)(0.0191) 2SLSOLS Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS 2SLSOLS Number of days drank 5+ drinks in the last 12 months Model 1 (Total Effect) 2SLSOLS

PAGE 119

Appendix A-10: (Continued) 105 Completed graduate degree 0.1409***0.1508***0.0652***0.0676*** (0.0220)(0.0144)(0.0192)(0.0137) Had some graduate studies 0.1440***0.1518***0.0800***0.0819*** (0.0237)(0.0197)(0.0213)(0.0185) Completed college 0.1191***0.1261***0.0550***0.0567*** (0.0174)(0.0128)(0.0155)(0.0121) Had some college educatio n 0.0946***0.0971***0.0452***0.0458*** (0.0120)(0.0111)(0.0110)(0.0105) Completed high school 0.0848***0.0860***0.0451***0.0453*** (0.0119)(0.0116)(0.0110)(0.0109) Excellent healt h 0.6727***0.6731*** (0.0205)(0.0204) Very good healt h 0.6567***0.6561*** (0.0206)(0.0203) Good healt h 0.6209***0.6200*** (0.0210)(0.0205) Fair healt h 0.4419***0.4403*** (0.0240)(0.0222) Constant 0.8468***0.8126***0.8612***0.8243***0.7874***0.7668***0.1786**0.1729** (0.0908)(0.0861)(0.0919)(0.0857)(0.0922)(0.0852)(0.0875)(0.0816) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 120

Appendix A-11: Current Employment (Full Time/Part Time) versus the Number of Days Drunk in the Last 12 Months: the Male Sample 106 Male (n=8,673) Variables -0.00393*-0.00040***-0.00341-0.00034***-0.00170-0.00032***-0.00060-0.00025*** (0.0020)(0.0001)(0.0021)(0.0001)(0.0021)(0.0001)(0.0019)(0.0001) Age 0.00700.0085**0.00360.0045 0.00370.0041 0.00320.0033 (0.0047)(0.0043)(0.0046)(0.0043)(0.0043)(0.0042)(0.0040)(0.0039) Age squared -0.0001**-0.0001***-0.0001-0.0001*-0.0001*-0.0001*-0.0001-0.0001 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0000)(0.0000) White 0.0251**0.0193**0.0282***0.0244***-0.0013-0.0040 -0.0009-0.0015 (0.0099)(0.0086)(0.0096)(0.0087)(0.0099)(0.0089)(0.0091)(0.0083) Black -0.0499***-0.0517***-0.0354***-0.0345***-0.0523***-0.0526***-0.0349***-0.0350*** (0.0121)(0.0111)(0.0118)(0.0110)(0.0113)(0.0111)(0.0104)(0.0104) American Indian/Alaska Native 0.0130-0.0021 0.01670.0050 -0.0067-0.0127 0.01550.0142 (0.0310)(0.0273)(0.0300)(0.0271)(0.0287)(0.0269)(0.0262)(0.0250) Asian/Pacific Islander -0.00750.0135 -0.01000.0073 -0.0378*-0.0309 -0.0234-0.0217 (0.0238)(0.0188)(0.0233)(0.0187)(0.0217)(0.0188)(0.0200)(0.0175) Living in MSA in central city 0.00760.0054 0.0181*0.0179*0.00630.0058 -0.0053-0.0055 (0.0110)(0.0100)(0.0107)(0.0100)(0.0102)(0.0100)(0.0094)(0.0093) Living in MSA outside central cit y 0.0345***0.0390***0.0346***0.0382***0.0230**0.0240***0.00720.0074 (0.0104)(0.0093)(0.0101)(0.0092)(0.0095)(0.0092)(0.0087)(0.0086) State unemployment rates -0.0086**-0.0076**-0.0080**-0.0070**-0.0067*-0.0063*-0.0048-0.0047 (0.0039)(0.0035)(0.0038)(0.0035)(0.0036)(0.0035)(0.0033)(0.0033) Father/mother ever depressed -0.0050-0.0068 -0.0030-0.0043 -0.0094-0.0100 -0.0044-0.0046 (0.0101)(0.0092)(0.0098)(0.0092)(0.0093)(0.0091)(0.0085)(0.0085) Brother/sister ever depressed -0.00460.0018 -0.00160.0045 -0.00210.0006 0.01280.0136 (0.0119)(0.0103)(0.0117)(0.0102)(0.0111)(0.0102)(0.0103)(0.0095) Father/mother had behavior problems -0.0107-0.0251*-0.0113-0.0233*-0.0136-0.0188 -0.0058-0.0070 (0.0168)(0.0134)(0.0165)(0.0133)(0.0155)(0.0132)(0.0142)(0.0123) Brother/sister had behavior problems -0.0358***-0.0335***-0.0374***-0.0353***-0.0326***-0.0316***-0.0305***-0.0303*** (0.0126)(0.0115)(0.0123)(0.0114)(0.0116)(0.0114)(0.0107)(0.0106) Number of childre n -0.0051**-0.0053**-0.0006-0.0005 0.00110.0011 (0.0025)(0.0024)(0.0024)(0.0024)(0.0022)(0.0022) Married 0.0929***0.1060***0.0943***0.1001***0.0764***0.0778*** (0.0137)(0.0096)(0.0130)(0.0095)(0.0117)(0.0089) Cohabitating 0.0543***0.0531***0.0601***0.0598***0.0512***0.0511*** (0.0199)(0.0186)(0.0187)(0.0184)(0.0172)(0.0172) Widowed -0.0548-0.0326 -0.0392-0.0290 -0.00040.0022 (0.0481)(0.0426)(0.0455)(0.0423)(0.0420)(0.0394) Divorced 0.0430***0.0370***0.0403***0.0376***0.0358***0.0351*** (0.0139)(0.0124)(0.0131)(0.0123)(0.0121)(0.0115) Separated 0.01590.0095 0.01640.0136 0.00630.0056 (0.0226)(0.0207)(0.0213)(0.0206)(0.0196)(0.0191) Model 3 Model 4 (Direct Effect) 2SLSOLS Model 2 Model 1 (Total Effect) 2SLSOLS 2SLSOLS 2SLSOLS Number of days drunk in the last 12 months

PAGE 121

Appendix A-11: (Continued) 107 Completed graduate degree 0.1519***0.1542***0.0700***0.0702*** (0.0150)(0.0144)(0.0138)(0.0137) Had some graduate studies 0.1520***0.1543***0.0836***0.0839*** (0.0203)(0.0197)(0.0186)(0.0185) Completed college 0.1209***0.1274***0.0564***0.0577*** (0.0161)(0.0127)(0.0142)(0.0121) Had some college educatio n 0.0937***0.0973***0.0452***0.0459*** (0.0125)(0.0111)(0.0112)(0.0105) Completed high school 0.0832***0.0859***0.0447***0.0453*** (0.0125)(0.0116)(0.0113)(0.0109) Excellent healt h 0.6720***0.6729*** (0.0210)(0.0204) Very good healt h 0.6546***0.6551*** (0.0204)(0.0203) Good healt h 0.6182***0.6186*** (0.0206)(0.0205) Fair healt h 0.4404***0.4392*** (0.0231)(0.0222) Constant 0.8850***0.8109***0.8865***0.8224***0.7974***0.7660***0.1806*0.1724** (0.1032)(0.0862)(0.1018)(0.0857)(0.0983)(0.0852)(0.0936)(0.0816) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 122

Appendix A-12: Current Employment (Full Time/Part Time) versus the Average Daily Volume of Ethanol Consumed in the Last 12 Months: the Male Sample 108 Male (n=8,673) Variables -0.07724*-0.00934***-0.06974-0.00776***-0.03815-0.00693***-0.01771-0.00450*** (0.0417)(0.0015)(0.0458)(0.0015)(0.0469)(0.0015)(0.0448)(0.0014) Age 0.0079*0.0086**0.00470.0046 0.00420.0042 0.00330.0033 (0.0047)(0.0043)(0.0046)(0.0043)(0.0043)(0.0042)(0.0040)(0.0039) Age squared -0.0001**-0.0001***-0.0001*-0.0001*-0.0001*-0.0001*-0.0001-0.0001 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0000)(0.0000) White 0.0246**0.0194**0.0279***0.0244***0.0000-0.0037 0.0000-0.0015 (0.0100)(0.0086)(0.0098)(0.0087)(0.0107)(0.0089)(0.0098)(0.0083) Black -0.0458***-0.0512***-0.0334***-0.0343***-0.0509***-0.0524***-0.0345***-0.0349*** (0.0127)(0.0110)(0.0121)(0.0110)(0.0116)(0.0111)(0.0105)(0.0104) American Indian/Alaska Native 0.0061-0.0026 0.01050.0045 -0.0090-0.0132 0.01500.0136 (0.0306)(0.0273)(0.0298)(0.0270)(0.0282)(0.0269)(0.0256)(0.0250) Asian/Pacific Islander -0.00470.0134 -0.00660.0074 -0.0342*-0.0302 -0.0228-0.0210 (0.0235)(0.0187)(0.0228)(0.0187)(0.0202)(0.0188)(0.0186)(0.0175) Living in MSA in central city 0.00820.0055 0.01770.0179*0.00690.0059 -0.0048-0.0054 (0.0112)(0.0100)(0.0109)(0.0100)(0.0104)(0.0100)(0.0096)(0.0093) Living in MSA outside central cit y 0.0330***0.0387***0.0336***0.0380***0.0230**0.0240***0.00730.0075 (0.0108)(0.0093)(0.0106)(0.0092)(0.0096)(0.0092)(0.0087)(0.0086) State unemployment rates -0.0076*-0.0075**-0.0073*-0.0070**-0.0064*-0.0062*-0.0048-0.0046 (0.0039)(0.0035)(0.0038)(0.0035)(0.0036)(0.0035)(0.0033)(0.0033) Father/mother ever depressed -0.0079-0.0071 -0.0056-0.0045 -0.0102-0.0102 -0.0049-0.0047 (0.0102)(0.0092)(0.0100)(0.0092)(0.0093)(0.0091)(0.0085)(0.0085) Brother/sister ever depressed 0.00170.0024 0.00350.0049 0.00070.0011 0.01360.0140 (0.0114)(0.0103)(0.0112)(0.0102)(0.0104)(0.0102)(0.0096)(0.0095) Father/mother had behavior problems -0.0113-0.0249*-0.0120-0.0232*-0.0135-0.0188 -0.0051-0.0072 (0.0170)(0.0134)(0.0167)(0.0133)(0.0157)(0.0132)(0.0142)(0.0123) Brother/sister had behavior problems -0.0339***-0.0333***-0.0360***-0.0352***-0.0322***-0.0315***-0.0306***-0.0302*** (0.0128)(0.0115)(0.0125)(0.0114)(0.0117)(0.0114)(0.0107)(0.0106) Number of childre n -0.0040-0.0052**-0.0002-0.0004 0.00120.0011 (0.0027)(0.0024)(0.0025)(0.0024)(0.0022)(0.0022) Married 0.0825***0.1047***0.0884***0.0991***0.0731***0.0773*** (0.0194)(0.0096)(0.0188)(0.0095)(0.0168)(0.0089) Cohabitating 0.0569***0.0534***0.0612***0.0600***0.0518***0.0512*** (0.0204)(0.0186)(0.0190)(0.0184)(0.0174)(0.0172) Widowed -0.0402-0.0313 -0.0326-0.0277 0.00100.0033 (0.0469)(0.0426)(0.0439)(0.0423)(0.0403)(0.0394) Divorced 0.0439***0.0372***0.0409***0.0377***0.0365***0.0351*** (0.0144)(0.0124)(0.0135)(0.0123)(0.0124)(0.0115) Separated 0.02140.0102 0.01940.0141 0.00820.0059 (0.0241)(0.0207)(0.0225)(0.0206)(0.0208)(0.0191) 2SLS Model 4 (Direct Effect) 2SLSOLS 2SLSOLS Model 1 (Total Effect) Model 2 Model 3 Average daily volume of ethanol consumed in the last 12 months OLS 2SLSOLS

PAGE 123

Appendix A-12: (Continued) 109 Completed graduate degree 0.1399***0.1520***0.0653***0.0690*** (0.0234)(0.0144)(0.0187)(0.0137) Had some graduate studies 0.1411***0.1523***0.0792***0.0828*** (0.0263)(0.0197)(0.0223)(0.0185) Completed college 0.1181***0.1269***0.0552***0.0578*** (0.0185)(0.0127)(0.0151)(0.0121) Had some college educatio n 0.0911***0.0968***0.0441***0.0458*** (0.0143)(0.0111)(0.0121)(0.0105) Completed high school 0.0811***0.0855***0.0438***0.0452*** (0.0136)(0.0116)(0.0119)(0.0109) Excellent healt h 0.6637***0.6710*** (0.0322)(0.0204) Very good healt h 0.6478***0.6535*** (0.0281)(0.0203) Good healt h 0.6118***0.6172*** (0.0274)(0.0205) Fair healt h 0.4353***0.4376*** (0.0237)(0.0222) Constant 0.8597***0.8094***0.8663***0.8210***0.7911***0.7647***0.1891*0.1723** (0.1001)(0.0861)(0.0992)(0.0857)(0.0958)(0.0852)(0.1000)(0.0816) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 124

Appendix A-13: Currently Working Full Time (35+ Hours a Week) versus the Number of Drinks Consumed in the Last 12 Months: the Male Sample 110 Male (n=8,673) Variables -0.00019*-0.00004***-0.00016-0.00004***-0.00007-0.00003***-0.00002-0.00003*** (0.0001)(0.0000)(0.0001)(0.0000)(0.0001)(0.0000)(0.0001)(0.0000) Age 0.0164***0.0164***0.0102**0.0101**0.0094**0.0093**0.0085*0.0085* (0.0050)(0.0048)(0.0049)(0.0048)(0.0048)(0.0048)(0.0045)(0.0045) Age squared -0.0002***-0.0002***-0.0002***-0.0002***-0.0002***-0.0002***-0.0001**-0.0001** (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0428***0.0344***0.0488***0.0415***0.00830.0054 0.00670.0071 (0.0117)(0.0097)(0.0120)(0.0098)(0.0137)(0.0101)(0.0132)(0.0095) Black -0.0409***-0.0439***-0.0212*-0.0218*-0.0453***-0.0457***-0.0283**-0.0283** (0.0130)(0.0125)(0.0127)(0.0125)(0.0126)(0.0125)(0.0119)(0.0119) American Indian/Alaska Native -0.0073-0.0150 0.0005-0.0054 -0.0265-0.0286 -0.0021-0.0019 (0.0322)(0.0309)(0.0317)(0.0306)(0.0311)(0.0303)(0.0293)(0.0287) Asian/Pacific Islander -0.00940.0047 -0.0110-0.0012 -0.0492**-0.0476**-0.0384*-0.0386* (0.0241)(0.0212)(0.0233)(0.0211)(0.0218)(0.0212)(0.0207)(0.0201) Living in MSA in central city -0.00100.0000 0.01670.0185 0.00410.0041 -0.0073-0.0073 (0.0117)(0.0114)(0.0117)(0.0113)(0.0113)(0.0113)(0.0107)(0.0107) Living in MSA outside central cit y 0.0437***0.0478***0.0461***0.0487***0.0314***0.0317***0.01490.0149 (0.0112)(0.0105)(0.0109)(0.0104)(0.0105)(0.0104)(0.0099)(0.0099) State unemployment rates -0.0100**-0.0090**-0.0093**-0.0082**-0.0075*-0.0071*-0.0056-0.0056 (0.0042)(0.0040)(0.0042)(0.0040)(0.0041)(0.0039)(0.0039)(0.0037) Father/mother ever depressed -0.0074-0.0051 -0.0033-0.0013 -0.0083-0.0080 -0.0021-0.0022 (0.0109)(0.0105)(0.0107)(0.0103)(0.0103)(0.0103)(0.0098)(0.0097) Brother/sister ever depressed -0.0077-0.0090 -0.0051-0.0057 -0.0100-0.0104 0.00230.0024 (0.0121)(0.0117)(0.0118)(0.0116)(0.0115)(0.0115)(0.0109)(0.0109) Father/mother had behavior problems -0.0168-0.0244 -0.0178-0.0232 -0.0161-0.0176 -0.0059-0.0057 (0.0165)(0.0152)(0.0161)(0.0150)(0.0156)(0.0149)(0.0148)(0.0141) Brother/sister had behavior problems -0.0314**-0.0327**-0.0354***-0.0359***-0.0315**-0.0315**-0.0300**-0.0300** (0.0134)(0.0131)(0.0132)(0.0129)(0.0128)(0.0128)(0.0121)(0.0121) Number of childre n -0.0032-0.0048*0.00150.0011 0.00280.0028 (0.0031)(0.0027)(0.0029)(0.0027)(0.0027)(0.0026) Married 0.1316***0.1432***0.1331***0.1364***0.1142***0.1137*** (0.0152)(0.0108)(0.0150)(0.0107)(0.0140)(0.0102) Cohabitating 0.0688***0.0583***0.0689***0.0660***0.0570***0.0574*** (0.0234)(0.0210)(0.0228)(0.0208)(0.0217)(0.0197) Widowed -0.0151-0.0144 -0.0109-0.0105 0.01940.0194 (0.0491)(0.0481)(0.0478)(0.0477)(0.0452)(0.0451) Divorced 0.0847***0.0725***0.0759***0.0724***0.0692***0.0697*** (0.0181)(0.0140)(0.0179)(0.0139)(0.0172)(0.0131) Separated 0.0542**0.0435*0.0511**0.0481**0.0395*0.0399* (0.0258)(0.0234)(0.0251)(0.0232)(0.0240)(0.0219) Model 1 (Total Effect) 2SLSOLS Number of drinks consumed in the last 12 months Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 125

Appendix A-13: (Continued) 111 Completed graduate degree 0.1848***0.1898***0.1051***0.1045*** (0.0227)(0.0163)(0.0202)(0.0158) Had some graduate studies 0.1862***0.1907***0.1198***0.1193*** (0.0265)(0.0222)(0.0243)(0.0212) Completed college 0.1596***0.1627***0.0922***0.0919*** (0.0174)(0.0144)(0.0158)(0.0138) Had some college educatio n 0.1250***0.1261***0.0743***0.0742*** (0.0130)(0.0125)(0.0122)(0.0120) Completed high school 0.1156***0.1163***0.0757***0.0756*** (0.0133)(0.0131)(0.0126)(0.0125) Excellent healt h 0.6568***0.6568*** (0.0234)(0.0234) Very good healt h 0.6410***0.6413*** (0.0240)(0.0232) Good healt h 0.5922***0.5926*** (0.0246)(0.0235) Fair healt h 0.4158***0.4165*** (0.0293)(0.0255) Constant 0.6702***0.6301***0.6977***0.6597***0.5990***0.5865***0.01020.0117 (0.1043)(0.0976)(0.1047)(0.0968)(0.1041)(0.0960)(0.0999)(0.0936) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 126

Appendix A-14: Currently Working Full Time (35+ Hours a Week) versus the Number of Days Drank 5+ Drinks in the Last 12 Months: the Male Sample 112 Male (n=8,673) Variables -0.00149**-0.00038***-0.00138-0.00030***-0.00066-0.00024***-0.00025-0.00020*** (0.0007)(0.0001)(0.0008)(0.0001)(0.0009)(0.0001)(0.0009)(0.0001) Age 0.0163***0.0163***0.0100**0.0100**0.0093*0.0093*0.0085*0.0085* (0.0049)(0.0048)(0.0049)(0.0048)(0.0048)(0.0048)(0.0045)(0.0045) Age squared -0.0002***-0.0002***-0.0002***-0.0002***-0.0002***-0.0002***-0.0001**-0.0001** (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0377***0.0333***0.0452***0.0407***0.00780.0047 0.00680.0064 (0.0104)(0.0097)(0.0106)(0.0098)(0.0121)(0.0101)(0.0115)(0.0096) Black -0.0468***-0.0454***-0.0268**-0.0230*-0.0477***-0.0466***-0.0292**-0.0291** (0.0128)(0.0125)(0.0131)(0.0125)(0.0128)(0.0125)(0.0122)(0.0119) American Indian/Alaska Native -0.0140-0.0165 -0.0050-0.0066 -0.0284-0.0296 -0.0027-0.0028 (0.0316)(0.0309)(0.0312)(0.0306)(0.0305)(0.0303)(0.0288)(0.0287) Asian/Pacific Islander -0.00830.0047 -0.0102-0.0009 -0.0481**-0.0470**-0.0382*-0.0380* (0.0233)(0.0212)(0.0227)(0.0211)(0.0214)(0.0212)(0.0203)(0.0201) Living in MSA in central city -0.0018-0.0002 0.01590.0183 0.00420.0042 -0.0072-0.0072 (0.0117)(0.0114)(0.0117)(0.0113)(0.0114)(0.0113)(0.0107)(0.0107) Living in MSA outside central cit y 0.0435***0.0477***0.0458***0.0486***0.0317***0.0318***0.01500.0150 (0.0111)(0.0105)(0.0108)(0.0104)(0.0104)(0.0104)(0.0099)(0.0099) State unemployment rates -0.0098**-0.0089**-0.0092**-0.0082**-0.0076*-0.0071*-0.0056-0.0056 (0.0041)(0.0040)(0.0041)(0.0040)(0.0041)(0.0039)(0.0039)(0.0037) Father/mother ever depressed -0.0072-0.0051 -0.0033-0.0013 -0.0083-0.0079 -0.0022-0.0021 (0.0108)(0.0105)(0.0107)(0.0103)(0.0103)(0.0103)(0.0098)(0.0097) Brother/sister ever depressed -0.0085-0.0092 -0.0059-0.0059 -0.0102-0.0105 0.00230.0022 (0.0120)(0.0117)(0.0118)(0.0116)(0.0115)(0.0115)(0.0109)(0.0109) Father/mother had behavior problems -0.0222-0.0256*-0.0222-0.0242 -0.0181-0.0186 -0.0065-0.0066 (0.0157)(0.0152)(0.0154)(0.0150)(0.0150)(0.0149)(0.0141)(0.0141) Brother/sister had behavior problems -0.0297**-0.0322**-0.0338**-0.0356***-0.0309**-0.0313**-0.0298**-0.0298** (0.0134)(0.0131)(0.0133)(0.0129)(0.0129)(0.0128)(0.0121)(0.0121) Number of childre n -0.0021-0.0046*0.00190.0012 0.00300.0029 (0.0033)(0.0027)(0.0031)(0.0027)(0.0029)(0.0026) Married 0.1284***0.1426***0.1309***0.1361***0.1131***0.1136*** (0.0156)(0.0108)(0.0155)(0.0107)(0.0146)(0.0102) Cohabitating 0.0646***0.0573***0.0673***0.0650***0.0567***0.0564*** (0.0221)(0.0210)(0.0215)(0.0208)(0.0203)(0.0197) Widowed -0.0127-0.0139 -0.0100-0.0101 0.01980.0197 (0.0491)(0.0481)(0.0478)(0.0477)(0.0452)(0.0452) Divorced 0.0818***0.0718***0.0753***0.0716***0.0693***0.0689*** (0.0163)(0.0140)(0.0160)(0.0139)(0.0153)(0.0132) Separated 0.0580**0.0443*0.0535**0.0484**0.0406*0.0401* (0.0261)(0.0234)(0.0256)(0.0232)(0.0244)(0.0220) Model 1 (Total Effect) 2SLSOLS Number of days drank 5+ drinks in the last 12 months Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 127

Appendix A-14: (Continued) 113 Completed graduate degree 0.1802***0.1890***0.1032***0.1040*** (0.0248)(0.0163)(0.0221)(0.0158) Had some graduate studies 0.1837***0.1905***0.1187***0.1193*** (0.0266)(0.0222)(0.0244)(0.0212) Completed college 0.1555***0.1617***0.0906***0.0911*** (0.0196)(0.0144)(0.0177)(0.0139) Had some college educatio n 0.1235***0.1257***0.0737***0.0739*** (0.0135)(0.0125)(0.0126)(0.0120) Completed high school 0.1151***0.1162***0.0755***0.0755*** (0.0134)(0.0131)(0.0126)(0.0125) Excellent healt h 0.6562***0.6563*** (0.0235)(0.0234) Very good healt h 0.6410***0.6408*** (0.0236)(0.0233) Good healt h 0.5923***0.5921*** (0.0241)(0.0235) Fair healt h 0.4159***0.4154*** (0.0275)(0.0255) Constant 0.6677***0.6304***0.7008***0.6600***0.6045***0.5864***0.01350.0118 (0.1027)(0.0976)(0.1037)(0.0968)(0.1038)(0.0961)(0.1004)(0.0936) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 128

Appendix A-15: Currently Working Full Time (35+ Hours a Week) versus the Number of Days Drunk in the Last 12 Months: the Male Sample 114 Male (n=8,673) Variables -0.00457**-0.00053***-0.00409*-0.00045***-0.00196-0.00043***-0.00076-0.00036*** (0.0023)(0.0001)(0.0024)(0.0001)(0.0023)(0.0001)(0.0022)(0.0001) Age 0.0144***0.0161***0.0089*0.0099**0.0088*0.0092*0.0083*0.0084* (0.0054)(0.0048)(0.0052)(0.0048)(0.0049)(0.0048)(0.0046)(0.0045) Age squared -0.0002***-0.0002***-0.0002***-0.0002***-0.0002***-0.0002***-0.0001**-0.0001** (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0393***0.0327***0.0444***0.0400***0.00680.0038 0.00650.0057 (0.0113)(0.0097)(0.0109)(0.0098)(0.0111)(0.0101)(0.0105)(0.0095) Black -0.0425***-0.0446***-0.0232*-0.0221*-0.0456***-0.0459***-0.0285**-0.0285** (0.0137)(0.0125)(0.0134)(0.0125)(0.0127)(0.0125)(0.0119)(0.0119) American Indian/Alaska Native 0.0023-0.0151 0.0085-0.0053 -0.0217-0.0284 -0.0002-0.0018 (0.0352)(0.0309)(0.0341)(0.0306)(0.0324)(0.0303)(0.0301)(0.0287) Asian/Pacific Islander -0.01830.0059 -0.0214-0.0009 -0.0561**-0.0485**-0.0413*-0.0392* (0.0270)(0.0213)(0.0264)(0.0211)(0.0244)(0.0212)(0.0230)(0.0201) Living in MSA in central city 0.00320.0007 0.01920.0190*0.00490.0043 -0.0069-0.0071 (0.0125)(0.0114)(0.0122)(0.0113)(0.0115)(0.0113)(0.0108)(0.0107) Living in MSA outside central cit y 0.0433***0.0484***0.0447***0.0489***0.0304***0.0316***0.01450.0148 (0.0118)(0.0105)(0.0115)(0.0104)(0.0107)(0.0104)(0.0100)(0.0099) State unemployment rates -0.0099**-0.0088**-0.0092**-0.0080**-0.0076*-0.0070*-0.0056-0.0055 (0.0044)(0.0040)(0.0043)(0.0040)(0.0041)(0.0039)(0.0038)(0.0037) Father/mother ever depressed -0.0022-0.0042 0.0009-0.0006 -0.0068-0.0075 -0.0016-0.0017 (0.0115)(0.0105)(0.0112)(0.0103)(0.0105)(0.0103)(0.0098)(0.0097) Brother/sister ever depressed -0.0178-0.0104 -0.0139-0.0068 -0.0145-0.0115 0.00060.0014 (0.0135)(0.0117)(0.0133)(0.0116)(0.0125)(0.0115)(0.0118)(0.0109) Father/mother had behavior problems -0.0080-0.0246 -0.0088-0.0230 -0.0115-0.0173 -0.0040-0.0054 (0.0191)(0.0152)(0.0187)(0.0150)(0.0175)(0.0149)(0.0163)(0.0141) Brother/sister had behavior problems -0.0360**-0.0334**-0.0388***-0.0364***-0.0329**-0.0318**-0.0305**-0.0302** (0.0143)(0.0131)(0.0140)(0.0129)(0.0131)(0.0128)(0.0123)(0.0121) Number of childre n -0.0050*-0.0052*0.00080.0008 0.00250.0025 (0.0029)(0.0027)(0.0027)(0.0027)(0.0026)(0.0026) Married 0.1290***0.1446***0.1308***0.1373***0.1130***0.1146*** (0.0156)(0.0108)(0.0146)(0.0107)(0.0134)(0.0102) Cohabitating 0.0570**0.0555***0.0640***0.0637***0.0555***0.0554*** (0.0226)(0.0210)(0.0211)(0.0208)(0.0197)(0.0197) Widowed -0.0437-0.0175 -0.0247-0.0133 0.01390.0170 (0.0546)(0.0482)(0.0513)(0.0477)(0.0482)(0.0452) Divorced 0.0770***0.0699***0.0734***0.0704***0.0686***0.0678*** (0.0158)(0.0140)(0.0148)(0.0139)(0.0139)(0.0131) Separated 0.0491*0.0415*0.0496**0.0465**0.0392*0.0384* (0.0257)(0.0234)(0.0240)(0.0232)(0.0225)(0.0219) Model 1 (Total Effect) 2SLSOLS Number of days drunk in the last 12 months Model 4 (Direct Effect) OLS 2SLSOLS Model 2 Model 3 2SLS 2SLSOLS

PAGE 129

Appendix A-15: (Continued) 115 Completed graduate degree 0.1906***0.1932***0.1071***0.1073*** (0.0169)(0.0162)(0.0158)(0.0157) Had some graduate studies 0.1911***0.1936***0.1215***0.1218*** (0.0229)(0.0222)(0.0213)(0.0212) Completed college 0.1560***0.1632***0.0907***0.0923*** (0.0182)(0.0144)(0.0163)(0.0138) Had some college educatio n 0.1219***0.1259***0.0732***0.0741*** (0.0141)(0.0125)(0.0129)(0.0120) Completed high school 0.1131***0.1160***0.0748***0.0754*** (0.0140)(0.0131)(0.0130)(0.0125) Excellent healt h 0.6549***0.6560*** (0.0241)(0.0234) Very good healt h 0.6389***0.6394*** (0.0235)(0.0233) Good healt h 0.5898***0.5903*** (0.0237)(0.0235) Fair healt h 0.4156***0.4142*** (0.0265)(0.0255) Constant 0.7140***0.6291***0.7342***0.6583***0.6208***0.5859***0.02160.0119 (0.1172)(0.0977)(0.1157)(0.0969)(0.1108)(0.0961)(0.1074)(0.0936) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 130

Appendix A-16: Currently Working Full Time (35+ Hours a Week) versus the Average Daily Volume of Ethanol Consumed in the Last 12 Months: the Male Sample 116 Male (n=8,673) Variables -0.08596*-0.01195***-0.07788-0.00997***-0.03663-0.00894***-0.01323-0.00651*** (0.0469)(0.0017)(0.0515)(0.0017)(0.0524)(0.0017)(0.0512)(0.0016) Age 0.0155***0.0162***0.0101*0.0100**0.0094*0.0093*0.0085*0.0085* (0.0053)(0.0048)(0.0052)(0.0048)(0.0048)(0.0048)(0.0045)(0.0045) Age squared -0.0002***-0.0002***-0.0002***-0.0002***-0.0002***-0.0002***-0.0001**-0.0001** (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0384***0.0327***0.0438***0.0400***0.00740.0041 0.00650.0057 (0.0113)(0.0097)(0.0110)(0.0098)(0.0120)(0.0101)(0.0112)(0.0095) Black -0.0380***-0.0439***-0.0208-0.0218*-0.0443***-0.0456***-0.0281**-0.0283** (0.0143)(0.0125)(0.0136)(0.0125)(0.0130)(0.0125)(0.0120)(0.0119) American Indian/Alaska Native -0.0064-0.0158 0.0006-0.0061 -0.0254-0.0291 -0.0018-0.0025 (0.0345)(0.0309)(0.0335)(0.0305)(0.0315)(0.0303)(0.0292)(0.0287) Asian/Pacific Islander -0.01400.0058 -0.0161-0.0007 -0.0510**-0.0475**-0.0393*-0.0383* (0.0265)(0.0212)(0.0257)(0.0211)(0.0225)(0.0212)(0.0213)(0.0201) Living in MSA in central city 0.00370.0008 0.01880.0190*0.00530.0044 -0.0067-0.0070 (0.0126)(0.0114)(0.0123)(0.0113)(0.0116)(0.0113)(0.0110)(0.0107) Living in MSA outside central cit y 0.0418***0.0481***0.0438***0.0487***0.0307***0.0316***0.01480.0149 (0.0122)(0.0105)(0.0119)(0.0104)(0.0107)(0.0104)(0.0099)(0.0099) State unemployment rates -0.0088**-0.0087**-0.0083*-0.0079**-0.0071*-0.0069*-0.0055-0.0054 (0.0044)(0.0040)(0.0043)(0.0040)(0.0040)(0.0039)(0.0038)(0.0037) Father/mother ever depressed -0.0055-0.0046 -0.0021-0.0009 -0.0077-0.0077 -0.0020-0.0019 (0.0115)(0.0105)(0.0113)(0.0103)(0.0104)(0.0103)(0.0097)(0.0097) Brother/sister ever depressed -0.0104-0.0095 -0.0077-0.0061 -0.0111-0.0108 0.00180.0020 (0.0129)(0.0117)(0.0126)(0.0116)(0.0116)(0.0115)(0.0109)(0.0109) Father/mother had behavior problems -0.0095-0.0244 -0.0106-0.0230 -0.0126-0.0173 -0.0047-0.0057 (0.0191)(0.0152)(0.0188)(0.0150)(0.0175)(0.0149)(0.0163)(0.0141) Brother/sister had behavior problems -0.0338**-0.0332**-0.0370***-0.0362***-0.0323**-0.0317**-0.0303**-0.0301** (0.0144)(0.0131)(0.0140)(0.0129)(0.0130)(0.0128)(0.0122)(0.0121) Number of childre n -0.0038-0.0051*0.00110.0009 0.00260.0026 (0.0031)(0.0027)(0.0028)(0.0027)(0.0026)(0.0026) Married 0.1186***0.1429***0.1265***0.1360***0.1118***0.1139*** (0.0218)(0.0108)(0.0210)(0.0107)(0.0192)(0.0102) Cohabitating 0.0598***0.0559***0.0651***0.0640***0.0559***0.0556*** (0.0229)(0.0210)(0.0212)(0.0208)(0.0198)(0.0197) Widowed -0.0254-0.0157 -0.0158-0.0116 0.01730.0185 (0.0527)(0.0481)(0.0490)(0.0476)(0.0461)(0.0451) Divorced 0.0774***0.0701***0.0733***0.0704***0.0685***0.0678*** (0.0162)(0.0140)(0.0151)(0.0139)(0.0142)(0.0131) Separated 0.0546**0.0423*0.0518**0.0471**0.0400*0.0388* (0.0270)(0.0234)(0.0251)(0.0232)(0.0238)(0.0219) Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS Model 1 (Total Effect) Model 2 Model 3 Average daily volume of ethanol consumed in the last 12 months 2SLSOLS

PAGE 131

Appendix A-16: (Continued) 117 Completed graduate degree 0.1796***0.1904***0.1037***0.1056*** (0.0262)(0.0163)(0.0214)(0.0158) Had some graduate studies 0.1812***0.1911***0.1184***0.1203*** (0.0294)(0.0222)(0.0255)(0.0212) Completed college 0.1549***0.1627***0.0910***0.0923*** (0.0207)(0.0144)(0.0172)(0.0138) Had some college educatio n 0.1203***0.1253***0.0730***0.0739*** (0.0159)(0.0125)(0.0138)(0.0120) Completed high school 0.1117***0.1156***0.0746***0.0753*** (0.0152)(0.0131)(0.0136)(0.0125) Excellent healt h 0.6496***0.6533*** (0.0367)(0.0234) Very good healt h 0.6342***0.6371*** (0.0321)(0.0233) Good healt h 0.5854***0.5882*** (0.0313)(0.0235) Fair healt h 0.4107***0.4119*** (0.0271)(0.0255) Constant 0.6816***0.6267***0.7058***0.6562***0.6073***0.5838***0.02040.0118 (0.1127)(0.0976)(0.1115)(0.0968)(0.1070)(0.0960)(0.1143)(0.0936) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 132

Appendix A-17: Current Employment (Full Time/Part Time) versus the Number of Drinks Consumed in the Last 12 Months: the Female Sample 118 Female (n=9,355) Variables -0.00060***-0.00002-0.00036*-0.00001-0.00023-0.00001 0.00000-0.00002 (0.0002)(0.0000)(0.0002)(0.0000)(0.0002)(0.0000)(0.0002)(0.0000) Age 0.0087*0.0071 0.0083*0.0069 0.00740.0065 0.0076*0.0077* (0.0050)(0.0044)(0.0047)(0.0044)(0.0045)(0.0044)(0.0041)(0.0041) Age squared -0.0001**-0.0001**-0.0001**-0.0001**-0.0001**-0.0001*-0.0001*-0.0001* (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0923***0.0568***0.0626***0.0418***0.0270*0.0130 0.01200.0131 (0.0168)(0.0098)(0.0154)(0.0098)(0.0158)(0.0099)(0.0142)(0.0093) Black 0.0073-0.0107 0.01320.0043 -0.0107-0.0168 -0.0080-0.0075 (0.0138)(0.0109)(0.0125)(0.0110)(0.0124)(0.0111)(0.0114)(0.0103) American Indian/Alaska Native 0.0074-0.0200 -0.0032-0.0197 -0.0283-0.0392 -0.0153-0.0144 (0.0337)(0.0291)(0.0314)(0.0288)(0.0306)(0.0286)(0.0281)(0.0266) Asian/Pacific Islander 0.0764***0.0775***0.0567**0.0572***0.02580.0249 0.02400.0241 (0.0241)(0.0218)(0.0226)(0.0217)(0.0220)(0.0216)(0.0201)(0.0201) Living in MSA in central city 0.0299**0.0093 0.01820.0080 0.00780.0011 -0.0086-0.0081 (0.0142)(0.0109)(0.0126)(0.0109)(0.0124)(0.0108)(0.0113)(0.0100) Living in MSA outside central cit y 0.0521***0.0400***0.0351***0.0287***0.0224**0.0179*0.00070.0010 (0.0122)(0.0103)(0.0112)(0.0102)(0.0110)(0.0102)(0.0100)(0.0095) State unemployment rates -0.0069-0.0056 -0.0067-0.0058 -0.0060-0.0054 -0.0025-0.0025 (0.0046)(0.0042)(0.0043)(0.0041)(0.0042)(0.0041)(0.0038)(0.0038) Father/mother ever depressed -0.0133-0.0173*-0.0152-0.0172*-0.0223**-0.0236***-0.0130-0.0128 (0.0101)(0.0090)(0.0094)(0.0089)(0.0091)(0.0089)(0.0084)(0.0083) Brother/sister ever depressed -0.0008-0.0022 -0.0001-0.0009 -0.0021-0.0026 0.00460.0046 (0.0108)(0.0098)(0.0101)(0.0097)(0.0098)(0.0096)(0.0089)(0.0089) Father/mother had behavior problems -0.0353**-0.0512***-0.0334**-0.0426***-0.0329**-0.0385***-0.0128-0.0123 (0.0153)(0.0128)(0.0142)(0.0127)(0.0137)(0.0126)(0.0126)(0.0117) Brother/sister had behavior problems -0.0233*-0.0349***-0.0246**-0.0317***-0.0260**-0.0304***-0.0212*-0.0209** (0.0132)(0.0113)(0.0123)(0.0112)(0.0119)(0.0111)(0.0110)(0.0104) Number of childre n -0.0252***-0.0234***-0.0152***-0.0138***-0.0094***-0.0095*** (0.0027)(0.0024)(0.0028)(0.0025)(0.0025)(0.0023) Married 0.0705***0.0793***0.0656***0.0710***0.0438***0.0433*** (0.0115)(0.0100)(0.0111)(0.0099)(0.0103)(0.0093) Cohabitating 0.02620.0114 0.02790.0189 -0.0053-0.0046 (0.0229)(0.0205)(0.0221)(0.0203)(0.0202)(0.0189) Widowed -0.0577**-0.0550**-0.0501**-0.0482**-0.0234-0.0235 (0.0254)(0.0244)(0.0246)(0.0241)(0.0225)(0.0225) Divorced 0.01230.0152 0.01050.0124 0.01030.0101 (0.0125)(0.0119)(0.0121)(0.0118)(0.0111)(0.0110) Separated 0.01540.0047 0.01910.0127 0.01780.0183 (0.0188)(0.0172)(0.0182)(0.0170)(0.0167)(0.0158) OLS Number of drinks consumed in the last 12 months Model 1 (Total Effect) 2SLSOLS Model 2 Model 3 2SLSOLS Model 4 (Direct Effect) 2SLSOLS 2SLS

PAGE 133

Appendix A-17: (Continued) 119 Completed graduate degree 0.1868***0.1929***0.1103***0.1097*** (0.0173)(0.0162)(0.0165)(0.0154) Had some graduate studies 0.1897***0.1914***0.1109***0.1107*** (0.0206)(0.0202)(0.0191)(0.0190) Completed college 0.1709***0.1765***0.1024***0.1018*** (0.0158)(0.0147)(0.0150)(0.0140) Had some college educatio n 0.1324***0.1351***0.0802***0.0800*** (0.0131)(0.0126)(0.0122)(0.0119) Completed high school 0.1115***0.1142***0.0670***0.0667*** (0.0138)(0.0133)(0.0128)(0.0125) Excellent healt h 0.6235***0.6240*** (0.0199)(0.0191) Very good healt h 0.6175***0.6181*** (0.0197)(0.0190) Good healt h 0.5831***0.5832*** (0.0191)(0.0191) Fair healt h 0.3529***0.3533*** (0.0212)(0.0209) Constant 0.7699***0.7739***0.7682***0.7722***0.6667***0.6661***0.07880.0784 (0.1010)(0.0912)(0.0942)(0.0905)(0.0916)(0.0901)(0.0860)(0.0859) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 134

Appendix A-18: Current Employment (Full Time/Part Time) versus the Number of Days Drank 5+ Drinks in the Last 12 Months: the Female Sample 120 Female (n=9,355) Variables -0.00556***-0.00030**-0.00353*-0.00025**-0.00230-0.00017 0.00002-0.00017 (0.0020)(0.0001)(0.0019)(0.0001)(0.0019)(0.0001)(0.0018)(0.0001) Age 0.00700.0071 0.00700.0069 0.00650.0065 0.0076*0.0076* (0.0049)(0.0044)(0.0046)(0.0044)(0.0045)(0.0044)(0.0041)(0.0041) Age squared -0.0001**-0.0001**-0.0001**-0.0001**-0.0001**-0.0001*-0.0001*-0.0001* (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0711***0.0567***0.0513***0.0418***0.0208*0.0131 0.01200.0127 (0.0120)(0.0097)(0.0115)(0.0098)(0.0123)(0.0099)(0.0112)(0.0092) Black -0.0031-0.0108 0.00720.0043 -0.0141-0.0168 -0.0080-0.0078 (0.0123)(0.0109)(0.0116)(0.0110)(0.0115)(0.0111)(0.0105)(0.0103) American Indian/Alaska Native 0.0129-0.0189 0.0010-0.0188 -0.0248-0.0385 -0.0153-0.0141 (0.0340)(0.0291)(0.0321)(0.0288)(0.0316)(0.0286)(0.0290)(0.0266) Asian/Pacific Islander 0.0733***0.0773***0.0562**0.0571***0.02680.0250 0.02400.0242 (0.0240)(0.0218)(0.0225)(0.0217)(0.0220)(0.0216)(0.0202)(0.0201) Living in MSA in central city 0.01730.0093 0.01160.0079 0.00430.0011 -0.0086-0.0083 (0.0123)(0.0109)(0.0115)(0.0108)(0.0113)(0.0108)(0.0104)(0.0100) Living in MSA outside central cit y 0.0403***0.0397***0.0289***0.0285***0.0191*0.0179*0.00070.0008 (0.0112)(0.0102)(0.0106)(0.0102)(0.0104)(0.0102)(0.0095)(0.0095) State unemployment rates -0.0061-0.0056 -0.0062-0.0058 -0.0056-0.0054 -0.0025-0.0025 (0.0046)(0.0042)(0.0043)(0.0041)(0.0042)(0.0041)(0.0038)(0.0038) Father/mother ever depressed -0.0157-0.0173*-0.0165*-0.0172*-0.0228**-0.0236***-0.0130-0.0129 (0.0099)(0.0090)(0.0093)(0.0089)(0.0090)(0.0089)(0.0083)(0.0083) Brother/sister ever depressed -0.0008-0.0022 -0.0001-0.0009 -0.0019-0.0026 0.00460.0046 (0.0107)(0.0098)(0.0101)(0.0097)(0.0098)(0.0096)(0.0089)(0.0089) Father/mother had behavior problems -0.0365**-0.0508***-0.0342**-0.0423***-0.0333**-0.0383***-0.0128-0.0124 (0.0150)(0.0128)(0.0140)(0.0127)(0.0136)(0.0126)(0.0124)(0.0117) Brother/sister had behavior problems -0.0204-0.0344***-0.0226*-0.0313***-0.0247**-0.0301***-0.0212*-0.0208** (0.0135)(0.0113)(0.0127)(0.0112)(0.0123)(0.0111)(0.0113)(0.0104) Number of childre n -0.0234***-0.0233***-0.0144***-0.0138***-0.0094***-0.0095*** (0.0025)(0.0024)(0.0026)(0.0025)(0.0024)(0.0023) Married 0.0700***0.0789***0.0653***0.0707***0.0438***0.0433*** (0.0116)(0.0100)(0.0112)(0.0099)(0.0103)(0.0093) Cohabitating 0.02500.0119 0.02740.0192 -0.0053-0.0046 (0.0226)(0.0205)(0.0220)(0.0203)(0.0202)(0.0189) Widowed -0.0600**-0.0552**-0.0518**-0.0484**-0.0234-0.0237 (0.0255)(0.0243)(0.0247)(0.0241)(0.0227)(0.0225) Divorced 0.01460.0153 0.01180.0124 0.01030.0102 (0.0124)(0.0119)(0.0120)(0.0118)(0.0110)(0.0110) Separated 0.01730.0053 0.02030.0130 0.01780.0184 (0.0192)(0.0172)(0.0185)(0.0170)(0.0170)(0.0158) 2SLSOLS OLS Number of days drank 5+ drinks in the last 12 months Model 1 (Total Effect) 2SLSOLS Model 2 Model 3 2SLSOLS Model 4 (Direct Effect) 2SLS

PAGE 135

Appendix A-18: (Continued) 121 Completed graduate degree 0.1814***0.1922***0.1103***0.1094*** (0.0192)(0.0162)(0.0177)(0.0154) Had some graduate studies 0.1829***0.1908***0.1110***0.1103*** (0.0217)(0.0202)(0.0200)(0.0190) Completed college 0.1655***0.1759***0.1024***0.1016*** (0.0177)(0.0148)(0.0163)(0.0140) Had some college educatio n 0.1299***0.1348***0.0803***0.0798*** (0.0136)(0.0126)(0.0125)(0.0119) Completed high school 0.1127***0.1142***0.0670***0.0669*** (0.0136)(0.0133)(0.0126)(0.0125) Excellent healt h 0.6235***0.6235*** (0.0191)(0.0191) Very good healt h 0.6175***0.6178*** (0.0191)(0.0190) Good healt h 0.5831***0.5832*** (0.0191)(0.0191) Fair healt h 0.3529***0.3532*** (0.0211)(0.0209) Constant 0.8068***0.7758***0.7936***0.7738***0.6840***0.6675***0.07870.0800 (0.1007)(0.0912)(0.0948)(0.0905)(0.0929)(0.0901)(0.0868)(0.0859) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 136

Appendix A-19: Current Employment (Full Time/Part Time) versus the Number of Days Drunk in the Last 12 Months: the Female Sample 122 Female (n=9,355) Variables -0.01603**-0.00019-0.01046-0.00017-0.00666-0.00015 0.00004-0.00015 (0.0081)(0.0001)(0.0068)(0.0001)(0.0059)(0.0001)(0.0049)(0.0001) Age 0.00720.0071 0.00640.0069 0.00630.0065 0.0076*0.0076* (0.0070)(0.0044)(0.0057)(0.0044)(0.0049)(0.0044)(0.0041)(0.0041) Age squared -0.0001-0.0001**-0.0001-0.0001**-0.0001*-0.0001*-0.0001*-0.0001* (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0848***0.0562***0.0608***0.0414***0.02620.0128 0.01200.0124 (0.0212)(0.0097)(0.0180)(0.0098)(0.0165)(0.0099)(0.0136)(0.0092) Black 0.0058-0.0110 0.01540.0042 -0.0091-0.0168 -0.0080-0.0078 (0.0193)(0.0109)(0.0160)(0.0110)(0.0143)(0.0111)(0.0119)(0.0103) American Indian/Alaska Native -0.0362-0.0209 -0.0303-0.0205 -0.0450-0.0398 -0.0151-0.0153 (0.0467)(0.0291)(0.0375)(0.0288)(0.0325)(0.0286)(0.0269)(0.0266) Asian/Pacific Islander 0.0808**0.0776***0.0618**0.0573***0.02890.0250 0.02400.0241 (0.0345)(0.0218)(0.0280)(0.0217)(0.0245)(0.0216)(0.0203)(0.0201) Living in MSA in central city 0.0384*0.0092 0.02650.0080 0.01340.0011 -0.0086-0.0083 (0.0228)(0.0109)(0.0186)(0.0108)(0.0165)(0.0108)(0.0136)(0.0100) Living in MSA outside central cit y 0.0538***0.0398***0.0378***0.0286***0.0243*0.0179*0.00070.0009 (0.0177)(0.0102)(0.0144)(0.0102)(0.0128)(0.0102)(0.0106)(0.0095) State unemployment rates -0.0044-0.0056 -0.0050-0.0057 -0.0048-0.0054 -0.0025-0.0025 (0.0066)(0.0042)(0.0053)(0.0041)(0.0046)(0.0041)(0.0038)(0.0038) Father/mother ever depressed -0.0282*-0.0175*-0.0245**-0.0173*-0.0278***-0.0238***-0.0129-0.0130 (0.0153)(0.0090)(0.0124)(0.0089)(0.0106)(0.0089)(0.0088)(0.0083) Brother/sister ever depressed 0.0078-0.0021 0.0059-0.0008 0.0018-0.0025 0.00460.0047 (0.0163)(0.0098)(0.0132)(0.0097)(0.0115)(0.0096)(0.0095)(0.0089) Father/mother had behavior problems -0.0289-0.0513***-0.0292-0.0427***-0.0298*-0.0385***-0.0128-0.0125 (0.0233)(0.0128)(0.0186)(0.0127)(0.0162)(0.0126)(0.0135)(0.0117) Brother/sister had behavior problems -0.0343*-0.0352***-0.0317**-0.0320***-0.0303**-0.0306***-0.0212**-0.0212** (0.0179)(0.0113)(0.0144)(0.0112)(0.0125)(0.0111)(0.0104)(0.0104) Number of childre n -0.0233***-0.0233***-0.0141***-0.0137***-0.0094***-0.0095*** (0.0031)(0.0024)(0.0028)(0.0025)(0.0023)(0.0023) Married 0.0804***0.0796***0.0721***0.0712***0.0437***0.0438*** (0.0129)(0.0100)(0.0112)(0.0099)(0.0093)(0.0093) Cohabitating 0.02980.0112 0.03060.0188 -0.0053-0.0050 (0.0291)(0.0205)(0.0253)(0.0203)(0.0210)(0.0189) Widowed -0.0724**-0.0552**-0.0596**-0.0484**-0.0233-0.0237 (0.0333)(0.0243)(0.0290)(0.0241)(0.0240)(0.0225) Divorced 0.02730.0155 0.02040.0126 0.01020.0105 (0.0172)(0.0119)(0.0151)(0.0118)(0.0125)(0.0110) Separated 0.03330.0048 0.03060.0128 0.01770.0182 (0.0290)(0.0172)(0.0251)(0.0170)(0.0207)(0.0158) Model 1 (Total Effect) 2SLS 2SLS OLS 2SLSOLS OLS Model 4 (Direct Effect) OLS 2SLS Number of days drunk in the last 12 months Model 2 Model 3

PAGE 137

Appendix A-19: (Continued) 123 Completed graduate degree 0.1835***0.1929***0.1103***0.1100*** (0.0202)(0.0162)(0.0171)(0.0154) Had some graduate studies 0.1855***0.1913***0.1109***0.1108*** (0.0233)(0.0202)(0.0195)(0.0190) Completed college 0.1721***0.1766***0.1024***0.1022*** (0.0171)(0.0147)(0.0144)(0.0140) Had some college educatio n 0.1264***0.1350***0.0803***0.0800*** (0.0162)(0.0126)(0.0136)(0.0119) Completed high school 0.1138***0.1143***0.0670***0.0670*** (0.0150)(0.0133)(0.0125)(0.0125) Excellent healt h 0.6235***0.6235*** (0.0191)(0.0191) Very good healt h 0.6176***0.6176*** (0.0190)(0.0190) Good healt h 0.5831***0.5831*** (0.0191)(0.0191) Fair healt h 0.3530***0.3530*** (0.0209)(0.0209) Constant 0.8110***0.7744***0.8024***0.7728***0.6875***0.6666***0.07870.0793 (0.1456)(0.0912)(0.1178)(0.0905)(0.1031)(0.0901)(0.0873)(0.0859) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 138

Appendix A-20: Current Employment (Full Time/Part Time) versus the Average Daily Volume of Ethanol Consumed in the Last 12 Months: The Female Sample 124 Female (n=9,355) Variables -0.20641***-0.00793**-0.12394*-0.00741**-0.07531-0.00625*0.00055-0.00703** (0.0790)(0.0034)(0.0704)(0.0034)(0.0685)(0.0034)(0.0617)(0.0031) Age 0.0096*0.0072 0.0088*0.0070 0.0077*0.0066 0.0076*0.0077* (0.0053)(0.0044)(0.0048)(0.0044)(0.0046)(0.0044)(0.0042)(0.0041) Age squared -0.0002**-0.0001**-0.0001**-0.0001**-0.0001**-0.0001**-0.0001*-0.0001* (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0833***0.0569***0.0572***0.0420***0.0236*0.0134 0.01200.0131 (0.0155)(0.0097)(0.0138)(0.0098)(0.0143)(0.0099)(0.0128)(0.0092) Black 0.0095-0.0104 0.01430.0047 -0.0098-0.0164 -0.0080-0.0073 (0.0150)(0.0109)(0.0131)(0.0110)(0.0131)(0.0111)(0.0119)(0.0103) American Indian/Alaska Native 0.0035-0.0198 -0.0059-0.0194 -0.0299-0.0388 -0.0153-0.0142 (0.0351)(0.0291)(0.0316)(0.0288)(0.0305)(0.0286)(0.0279)(0.0266) Asian/Pacific Islander 0.0807***0.0777***0.0592**0.0573***0.02740.0251 0.02400.0243 (0.0255)(0.0218)(0.0230)(0.0217)(0.0222)(0.0216)(0.0202)(0.0201) Living in MSA in central city 0.0318**0.0097 0.01950.0084 0.00820.0014 -0.0086-0.0079 (0.0154)(0.0109)(0.0133)(0.0108)(0.0129)(0.0108)(0.0117)(0.0100) Living in MSA outside central cit y 0.0526***0.0402***0.0356***0.0289***0.0225**0.0182*0.00070.0011 (0.0129)(0.0102)(0.0116)(0.0102)(0.0112)(0.0102)(0.0102)(0.0095) State unemployment rates -0.0060-0.0056 -0.0061-0.0058 -0.0056-0.0054 -0.0025-0.0025 (0.0049)(0.0042)(0.0044)(0.0041)(0.0042)(0.0041)(0.0038)(0.0038) Father/mother ever depressed -0.0138-0.0172*-0.0154-0.0171*-0.0224**-0.0236***-0.0130-0.0128 (0.0106)(0.0090)(0.0095)(0.0089)(0.0091)(0.0089)(0.0083)(0.0083) Brother/sister ever depressed -0.0025-0.0023 -0.0012-0.0009 -0.0028-0.0026 0.00460.0046 (0.0114)(0.0098)(0.0103)(0.0097)(0.0098)(0.0096)(0.0089)(0.0089) Father/mother had behavior problems -0.0392**-0.0511***-0.0357**-0.0425***-0.0345**-0.0384***-0.0128-0.0123 (0.0157)(0.0128)(0.0141)(0.0127)(0.0134)(0.0126)(0.0123)(0.0117) Brother/sister had behavior problems -0.0231*-0.0347***-0.0245*-0.0315***-0.0261**-0.0302***-0.0212*-0.0208** (0.0140)(0.0113)(0.0126)(0.0112)(0.0121)(0.0111)(0.0110)(0.0104) Number of childre n -0.0245***-0.0234***-0.0148***-0.0138***-0.0094***-0.0095*** (0.0026)(0.0024)(0.0027)(0.0025)(0.0025)(0.0023) Married 0.0692***0.0790***0.0652***0.0707***0.0438***0.0432*** (0.0122)(0.0100)(0.0115)(0.0099)(0.0106)(0.0093) Cohabitating 0.02200.0116 0.02500.0191 -0.0053-0.0046 (0.0227)(0.0205)(0.0216)(0.0203)(0.0197)(0.0189) Widowed -0.0621**-0.0553**-0.0527**-0.0485**-0.0234-0.0238 (0.0262)(0.0243)(0.0250)(0.0241)(0.0228)(0.0225) Divorced 0.00820.0149 0.00830.0121 0.01030.0099 (0.0133)(0.0119)(0.0127)(0.0118)(0.0115)(0.0110) Separated 0.00860.0046 0.01470.0126 0.01780.0180 (0.0184)(0.0172)(0.0175)(0.0170)(0.0160)(0.0158) 2SLS Model 1 (Total Effect) Model 2 Model 3 OLS Average daily volume of ethanol consumed in the last 12 months Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 139

Appendix A-20: (Continued) 125 Completed graduate degree 0.1847***0.1924***0.1103***0.1094*** (0.0183)(0.0162)(0.0171)(0.0154) Had some graduate studies 0.1884***0.1912***0.1109***0.1106*** (0.0208)(0.0202)(0.0192)(0.0190) Completed college 0.1681***0.1760***0.1024***0.1015*** (0.0170)(0.0147)(0.0159)(0.0140) Had some college educatio n 0.1285***0.1347***0.0803***0.0796*** (0.0143)(0.0126)(0.0132)(0.0119) Completed high school 0.1087***0.1139***0.0670***0.0664*** (0.0146)(0.0133)(0.0134)(0.0125) Excellent healt h 0.6234***0.6242*** (0.0201)(0.0191) Very good healt h 0.6175***0.6182*** (0.0198)(0.0190) Good healt h 0.5831***0.5834*** (0.0192)(0.0191) Fair healt h 0.3529***0.3539*** (0.0224)(0.0209) Constant 0.7362***0.7725***0.7479***0.7708***0.6565***0.6653***0.07890.0773 (0.1074)(0.0912)(0.0970)(0.0905)(0.0925)(0.0901)(0.0869)(0.0859) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 140

Appendix A-21: Currently Working Full Time (35+ Hours a Week) versus the Number of Drinks Consumed in th e Last 12 Months: the Female Sample 126 Female (n=9,355) Variables -0.00085***0.00000-0.00056**-0.00001-0.000360.00000-0.00014-0.00001 (0.0003)(0.0000)(0.0003)(0.0000)(0.0003)(0.0000)(0.0002)(0.0000) Age -0.0008-0.0032 0.00300.0008 0.00190.0005 0.00210.0016 (0.0067)(0.0058)(0.0062)(0.0058)(0.0060)(0.0057)(0.0056)(0.0055) Age squared 0.00000.0000 0.00000.0000 0.00000.0000 0.00000.0000 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0984***0.0464***0.0508**0.0179 0.0027-0.0201 -0.0125-0.0206 (0.0226)(0.0128)(0.0206)(0.0128)(0.0209)(0.0130)(0.0193)(0.0126) Black 0.0628***0.0364**0.0434***0.0295**0.01300.0031 0.01590.0123 (0.0185)(0.0144)(0.0166)(0.0145)(0.0164)(0.0145)(0.0155)(0.0140) American Indian/Alaska Native 0.0086-0.0316 -0.0026-0.0287 -0.0349-0.0528 -0.0226-0.0293 (0.0452)(0.0382)(0.0418)(0.0377)(0.0403)(0.0374)(0.0382)(0.0361) Asian/Pacific Islander 0.0658**0.0675**0.04360.0443 0.0007-0.0006 -0.0009-0.0014 (0.0324)(0.0286)(0.0300)(0.0284)(0.0290)(0.0283)(0.0274)(0.0273) Living in MSA in central city 0.0691***0.0390***0.0336**0.0174 0.01770.0067 0.0019-0.0019 (0.0190)(0.0143)(0.0168)(0.0142)(0.0164)(0.0141)(0.0153)(0.0136) Living in MSA outside central cit y 0.0601***0.0423***0.0319**0.0217 0.01370.0065 -0.0072-0.0096 (0.0163)(0.0135)(0.0149)(0.0134)(0.0145)(0.0133)(0.0136)(0.0128) State unemployment rates 0.00100.0029 0.00020.0017 0.00100.0020 0.00440.0047 (0.0062)(0.0055)(0.0058)(0.0054)(0.0055)(0.0053)(0.0052)(0.0052) Father/mother ever depressed -0.0129-0.0187 -0.0185-0.0216*-0.0285**-0.0308***-0.0195*-0.0204* (0.0135)(0.0118)(0.0125)(0.0117)(0.0120)(0.0116)(0.0114)(0.0112) Brother/sister ever depressed 0.00600.0039 0.00540.0041 0.00190.0011 0.00860.0082 (0.0145)(0.0128)(0.0134)(0.0127)(0.0129)(0.0125)(0.0122)(0.0121) Father/mother had behavior problems -0.0310-0.0543***-0.0326*-0.0472***-0.0322*-0.0414**-0.0132-0.0167 (0.0205)(0.0168)(0.0189)(0.0166)(0.0181)(0.0165)(0.0172)(0.0159) Brother/sister had behavior problems -0.0265-0.0434***-0.0284*-0.0398***-0.0301*-0.0373**-0.0257*-0.0284** (0.0178)(0.0149)(0.0164)(0.0147)(0.0157)(0.0145)(0.0149)(0.0140) Number of childre n -0.0486***-0.0458***-0.0348***-0.0324***-0.0291***-0.0284*** (0.0036)(0.0031)(0.0037)(0.0032)(0.0034)(0.0031) Married -0.00810.0058 -0.0137-0.0050 -0.0350**-0.0316** (0.0153)(0.0131)(0.0147)(0.0130)(0.0141)(0.0126) Cohabitating 0.03210.0088 0.03380.0190 0.0021-0.0032 (0.0305)(0.0269)(0.0292)(0.0266)(0.0275)(0.0257) Widowed -0.0757**-0.0713**-0.0650**-0.0619**-0.0391-0.0382 (0.0338)(0.0319)(0.0324)(0.0315)(0.0306)(0.0305) Divorced 0.0490***0.0535***0.0479***0.0510***0.0475***0.0486*** (0.0167)(0.0156)(0.0160)(0.0155)(0.0151)(0.0149) Separated 0.0488*0.0320 0.0539**0.0434*0.0523**0.0485** (0.0251)(0.0225)(0.0240)(0.0223)(0.0227)(0.0215) Model 1 (Total Effect) 2SLSOLS 2SLSOLS Number of drinks consumed in the last 12 months OLS 2SLSOLS 2SLS Model 4 (Direct Effect) Model 2 Model 3

PAGE 141

Appendix A-21: (Continued) 127 Completed graduate degree 0.2714***0.2813***0.1966***0.2011*** (0.0228)(0.0212)(0.0225)(0.0209) Had some graduate studies 0.2536***0.2563***0.1773***0.1790*** (0.0272)(0.0264)(0.0260)(0.0257) Completed college 0.2116***0.2207***0.1447***0.1488*** (0.0208)(0.0193)(0.0204)(0.0189) Had some college educatio n 0.1652***0.1697***0.1150***0.1171*** (0.0172)(0.0165)(0.0166)(0.0161) Completed high school 0.1357***0.1402***0.0932***0.0952*** (0.0182)(0.0174)(0.0174)(0.0169) Excellent healt h 0.5775***0.5733*** (0.0271)(0.0259) Very good healt h 0.5634***0.5594*** (0.0269)(0.0258) Good healt h 0.5266***0.5256*** (0.0260)(0.0258) Fair healt h 0.3051***0.3022*** (0.0288)(0.0283) Constant 0.7405***0.7463***0.7438***0.7501***0.6130***0.6120***0.07630.0793 (0.1356)(0.1199)(0.1254)(0.1185)(0.1209)(0.1178)(0.1170)(0.1165) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 142

Appendix A-22: Currently Working Full Time (35+ Hours a Week) versus the Number of Days Drank 5+ Drinks in the Last 12 Months: the Female Sample 128 Female (n=9,355) Variables -0.00794***-0.00008-0.00549**-0.00011-0.003690.00000-0.001450.00001 (0.0026)(0.0002)(0.0025)(0.0002)(0.0026)(0.0002)(0.0024)(0.0002) Age -0.0034-0.0032 0.00090.0008 0.00060.0005 0.00160.0016 (0.0066)(0.0058)(0.0062)(0.0058)(0.0059)(0.0057)(0.0056)(0.0055) Age squared 0.00000.0000 0.00000.0000 0.00000.0000 0.00000.0000 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0683***0.0468***0.0333**0.0178 -0.0069-0.0202 -0.0159-0.0212* (0.0161)(0.0128)(0.0154)(0.0128)(0.0162)(0.0130)(0.0153)(0.0125) Black 0.0479***0.0365**0.0342**0.0294**0.00760.0030 0.01370.0120 (0.0166)(0.0144)(0.0155)(0.0145)(0.0152)(0.0145)(0.0144)(0.0140) American Indian/Alaska Native 0.0164-0.0310 0.0041-0.0284 -0.0291-0.0529 -0.0204-0.0299 (0.0458)(0.0382)(0.0428)(0.0377)(0.0419)(0.0374)(0.0395)(0.0361) Asian/Pacific Islander 0.0614*0.0674**0.04270.0443 0.0024-0.0006 -0.0003-0.0015 (0.0323)(0.0286)(0.0301)(0.0284)(0.0292)(0.0283)(0.0275)(0.0273) Living in MSA in central city 0.0512***0.0392***0.02330.0173 0.01210.0066 0.0000-0.0022 (0.0165)(0.0143)(0.0153)(0.0142)(0.0150)(0.0141)(0.0141)(0.0136) Living in MSA outside central cit y 0.0433***0.0424***0.02220.0216 0.00850.0064 -0.0089-0.0098 (0.0151)(0.0135)(0.0142)(0.0134)(0.0137)(0.0133)(0.0130)(0.0128) State unemployment rates 0.00220.0029 0.00110.0017 0.00160.0020 0.00460.0047 (0.0061)(0.0055)(0.0057)(0.0054)(0.0055)(0.0053)(0.0052)(0.0052) Father/mother ever depressed -0.0164-0.0186 -0.0206*-0.0217*-0.0294**-0.0308***-0.0200*-0.0205* (0.0133)(0.0118)(0.0124)(0.0117)(0.0120)(0.0116)(0.0113)(0.0112) Brother/sister ever depressed 0.00590.0039 0.00550.0042 0.00220.0011 0.00860.0082 (0.0144)(0.0128)(0.0135)(0.0127)(0.0129)(0.0125)(0.0122)(0.0121) Father/mother had behavior problems -0.0327-0.0540***-0.0338*-0.0472***-0.0328*-0.0414**-0.0136-0.0169 (0.0202)(0.0168)(0.0187)(0.0166)(0.0180)(0.0165)(0.0169)(0.0159) Brother/sister had behavior problems -0.0222-0.0432***-0.0254-0.0396***-0.0279*-0.0374**-0.0249-0.0286** (0.0181)(0.0149)(0.0170)(0.0147)(0.0163)(0.0145)(0.0154)(0.0140) Number of childre n -0.0458***-0.0457***-0.0335***-0.0324***-0.0287***-0.0283*** (0.0033)(0.0031)(0.0034)(0.0032)(0.0032)(0.0031) Married -0.00890.0057 -0.0143-0.0049 -0.0349**-0.0313** (0.0155)(0.0131)(0.0149)(0.0130)(0.0140)(0.0126) Cohabitating 0.03040.0089 0.03310.0189 0.0021-0.0036 (0.0303)(0.0269)(0.0291)(0.0266)(0.0275)(0.0257) Widowed -0.0792**-0.0714**-0.0678**-0.0619**-0.0405-0.0381 (0.0340)(0.0319)(0.0327)(0.0315)(0.0309)(0.0305) Divorced 0.0524***0.0536***0.0500***0.0510***0.0484***0.0487*** (0.0166)(0.0156)(0.0159)(0.0155)(0.0150)(0.0149) Separated 0.0519**0.0322 0.0560**0.0434*0.0532**0.0482** (0.0256)(0.0225)(0.0245)(0.0223)(0.0232)(0.0215) Model 1 (Total Effect) 2SLSOLS Model 3 Model 4 (Direct Effect) Model 2 2SLSOLS Number of days drank 5+ drinks in the last 12 months 2SLSOLS 2SLSOLS

PAGE 143

Appendix A-22: (Continued) 129 Completed graduate degree 0.2626***0.2813***0.1942***0.2014*** (0.0254)(0.0212)(0.0241)(0.0209) Had some graduate studies 0.2426***0.2563***0.1740***0.1792*** (0.0288)(0.0264)(0.0273)(0.0257) Completed college 0.2029***0.2208***0.1423***0.1492*** (0.0234)(0.0193)(0.0222)(0.0189) Had some college educatio n 0.1612***0.1697***0.1140***0.1173*** (0.0180)(0.0165)(0.0171)(0.0161) Completed high school 0.1376***0.1402***0.0944***0.0954*** (0.0181)(0.0174)(0.0171)(0.0169) Excellent healt h 0.5729***0.5730*** (0.0261)(0.0259) Very good healt h 0.5609***0.5591*** (0.0261)(0.0258) Good healt h 0.5261***0.5255*** (0.0260)(0.0258) Fair healt h 0.3045***0.3020*** (0.0287)(0.0283) Constant 0.7932***0.7467***0.7832***0.7508***0.6407***0.6120***0.08990.0794 (0.1356)(0.1199)(0.1266)(0.1185)(0.1230)(0.1178)(0.1183)(0.1165) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 144

Appendix A-23: Currently Working Full Time (35+ Hours a Week) versus the Number of Days Drunk in the Last 12 Months: the Female Sample 130 Female (n=9,355) Variables -0.02357**-0.00006-0.01714*-0.00007-0.01140-0.00004-0.00494-0.00004 (0.0115)(0.0002)(0.0100)(0.0002)(0.0084)(0.0002)(0.0069)(0.0002) Age -0.0030-0.0032 0.00000.0007 0.00010.0005 0.00140.0016 (0.0100)(0.0058)(0.0083)(0.0058)(0.0070)(0.0057)(0.0058)(0.0055) Age squared 0.00000.0000 0.00000.0000 0.00000.0000 0.00000.0000 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0891***0.0467***0.0498*0.0176 0.0031-0.0201 -0.0112-0.0211* (0.0301)(0.0128)(0.0262)(0.0128)(0.0233)(0.0129)(0.0192)(0.0125) Black 0.0614**0.0365**0.0480**0.0294**0.01660.0031 0.01790.0121 (0.0274)(0.0144)(0.0233)(0.0145)(0.0202)(0.0145)(0.0168)(0.0140) American Indian/Alaska Native -0.0542-0.0315 -0.0454-0.0291 -0.0621-0.0529 -0.0337-0.0298 (0.0662)(0.0382)(0.0546)(0.0377)(0.0458)(0.0373)(0.0381)(0.0361) Asian/Pacific Islander 0.07220.0675**0.05180.0443 0.0062-0.0006 0.0015-0.0014 (0.0490)(0.0286)(0.0407)(0.0284)(0.0346)(0.0283)(0.0288)(0.0273) Living in MSA in central city 0.0826**0.0391***0.0482*0.0173 0.02820.0067 0.0071-0.0021 (0.0323)(0.0143)(0.0271)(0.0142)(0.0233)(0.0141)(0.0193)(0.0136) Living in MSA outside central cit y 0.0632**0.0424***0.0368*0.0216 0.01770.0065 -0.0050-0.0097 (0.0251)(0.0135)(0.0210)(0.0134)(0.0181)(0.0133)(0.0150)(0.0128) State unemployment rates 0.00460.0029 0.00290.0017 0.00300.0020 0.00520.0047 (0.0094)(0.0055)(0.0077)(0.0054)(0.0065)(0.0053)(0.0054)(0.0052) Father/mother ever depressed -0.0346-0.0187 -0.0336*-0.0217*-0.0377**-0.0308***-0.0234*-0.0205* (0.0217)(0.0118)(0.0181)(0.0117)(0.0149)(0.0116)(0.0124)(0.0112) Brother/sister ever depressed 0.01860.0039 0.01530.0042 0.00870.0011 0.01150.0082 (0.0231)(0.0128)(0.0192)(0.0127)(0.0162)(0.0125)(0.0135)(0.0121) Father/mother had behavior problems -0.0208-0.0541***-0.0249-0.0473***-0.0261-0.0414**-0.0103-0.0169 (0.0330)(0.0168)(0.0271)(0.0166)(0.0229)(0.0165)(0.0191)(0.0159) Brother/sister had behavior problems -0.0420*-0.0434***-0.0396*-0.0399***-0.0369**-0.0374**-0.0283*-0.0286** (0.0254)(0.0149)(0.0210)(0.0147)(0.0176)(0.0145)(0.0147)(0.0140) Number of childre n -0.0457***-0.0457***-0.0331***-0.0324***-0.0286***-0.0283*** (0.0045)(0.0031)(0.0039)(0.0032)(0.0033)(0.0031) Married 0.00730.0060 -0.0034-0.0049 -0.0307**-0.0313** (0.0187)(0.0131)(0.0158)(0.0130)(0.0132)(0.0126) Cohabitating 0.03950.0086 0.03970.0190 0.0055-0.0035 (0.0424)(0.0269)(0.0357)(0.0266)(0.0297)(0.0257) Widowed -0.1000**-0.0714**-0.0815**-0.0620**-0.0466-0.0382 (0.0485)(0.0319)(0.0408)(0.0315)(0.0340)(0.0305) Divorced 0.0733***0.0537***0.0646***0.0510***0.0546***0.0488*** (0.0250)(0.0156)(0.0212)(0.0155)(0.0177)(0.0149) Separated 0.0791*0.0320 0.0745**0.0435*0.0617**0.0483** (0.0423)(0.0225)(0.0354)(0.0223)(0.0294)(0.0215) 2SLS OLS OLS 2SLS 2SLS OLS Number of days drunk in the last 12 months Model 3 Model 1 (Total Effect) Model 2 OLS 2SLS Model 4 (Direct Effect)

PAGE 145

Appendix A-23: (Continued) 131 Completed graduate degree 0.2649***0.2813***0.1940***0.2013*** (0.0284)(0.0212)(0.0242)(0.0209) Had some graduate studies 0.2461***0.2563***0.1745***0.1791*** (0.0329)(0.0264)(0.0277)(0.0257) Completed college 0.2129***0.2208***0.1455***0.1491*** (0.0241)(0.0193)(0.0204)(0.0189) Had some college educatio n 0.1545***0.1697***0.1106***0.1172*** (0.0229)(0.0165)(0.0193)(0.0161) Completed high school 0.1393***0.1402***0.0949***0.0954*** (0.0212)(0.0174)(0.0177)(0.0169) Excellent healt h 0.5738***0.5730*** (0.0271)(0.0259) Very good healt h 0.5598***0.5591*** (0.0270)(0.0258) Good healt h 0.5243***0.5255*** (0.0271)(0.0258) Fair healt h 0.3032***0.3021*** (0.0296)(0.0283) Constant 0.8007***0.7464***0.7995***0.7504***0.6487***0.6121***0.09510.0796 (0.2065)(0.1199)(0.1715)(0.1185)(0.1454)(0.1178)(0.1237)(0.1165) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 146

Appendix A-24: Currently Working Full Time (35+ Hours a Week) versus the Average Daily Volume of Ethanol Consumed in the Last 12 Months: the Female Sample 132 Female (n=9,355) Variables -0.28261***-0.00423-0.17897*-0.00563-0.10693-0.00410-0.03308-0.00479 (0.1057)(0.0045)(0.0937)(0.0044)(0.0901)(0.0044)(0.0838)(0.0042) Age 0.0003-0.0031 0.00360.0008 0.00220.0006 0.00210.0016 (0.0071)(0.0058)(0.0064)(0.0058)(0.0061)(0.0057)(0.0057)(0.0055) Age squared 0.00000.0000 0.00000.0000 0.00000.0000 0.00000.0000 (0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001)(0.0001) White 0.0841***0.0471***0.0408**0.0182 -0.0045-0.0196 -0.0164-0.0205 (0.0207)(0.0128)(0.0184)(0.0128)(0.0188)(0.0130)(0.0174)(0.0125) Black 0.0647***0.0369**0.0441**0.0298**0.01320.0034 0.01520.0125 (0.0201)(0.0144)(0.0174)(0.0145)(0.0172)(0.0145)(0.0162)(0.0140) American Indian/Alaska Native 0.0016-0.0310 -0.0082-0.0284 -0.0390-0.0524 -0.0254-0.0291 (0.0470)(0.0382)(0.0421)(0.0377)(0.0402)(0.0374)(0.0379)(0.0361) Asian/Pacific Islander 0.0718**0.0675**0.04730.0444 0.0030-0.0005 -0.0003-0.0013 (0.0341)(0.0286)(0.0306)(0.0284)(0.0292)(0.0283)(0.0275)(0.0273) Living in MSA in central city 0.0705***0.0395***0.0343*0.0177 0.01710.0070 0.0010-0.0017 (0.0206)(0.0143)(0.0177)(0.0142)(0.0170)(0.0141)(0.0159)(0.0136) Living in MSA outside central cit y 0.0601***0.0427***0.0318**0.0219 0.01310.0067 -0.0078-0.0095 (0.0173)(0.0135)(0.0154)(0.0134)(0.0148)(0.0133)(0.0138)(0.0128) State unemployment rates 0.00230.0029 0.00110.0017 0.00170.0020 0.00470.0047 (0.0065)(0.0055)(0.0058)(0.0054)(0.0055)(0.0053)(0.0052)(0.0052) Father/mother ever depressed -0.0137-0.0186 -0.0191-0.0216*-0.0289**-0.0307***-0.0198*-0.0204* (0.0142)(0.0118)(0.0127)(0.0117)(0.0120)(0.0116)(0.0113)(0.0112) Brother/sister ever depressed 0.00350.0039 0.00370.0041 0.00080.0010 0.00810.0082 (0.0152)(0.0128)(0.0137)(0.0127)(0.0129)(0.0125)(0.0121)(0.0121) Father/mother had behavior problems -0.0372*-0.0540***-0.0370**-0.0471***-0.0354**-0.0412**-0.0150-0.0166 (0.0210)(0.0168)(0.0187)(0.0166)(0.0177)(0.0165)(0.0167)(0.0159) Brother/sister had behavior problems -0.0268-0.0431***-0.0291*-0.0396***-0.0310*-0.0371**-0.0265*-0.0283** (0.0187)(0.0149)(0.0168)(0.0147)(0.0159)(0.0145)(0.0150)(0.0140) Number of childre n -0.0474***-0.0458***-0.0340***-0.0325***-0.0288***-0.0284*** (0.0035)(0.0031)(0.0036)(0.0032)(0.0033)(0.0031) Married -0.00900.0055 -0.0134-0.0053 -0.0340**-0.0317** (0.0162)(0.0131)(0.0151)(0.0130)(0.0143)(0.0126) Cohabitating 0.02460.0090 0.02800.0193 -0.0007-0.0031 (0.0302)(0.0269)(0.0284)(0.0266)(0.0267)(0.0257) Widowed -0.0816**-0.0716**-0.0684**-0.0622**-0.0401-0.0384 (0.0348)(0.0319)(0.0329)(0.0315)(0.0310)(0.0305) Divorced 0.0433**0.0533***0.0451***0.0508***0.0469***0.0484*** (0.0177)(0.0156)(0.0167)(0.0155)(0.0157)(0.0149) Separated 0.03790.0320 0.0466**0.0435*0.0492**0.0484** (0.0245)(0.0225)(0.0231)(0.0223)(0.0217)(0.0215) OLS 2SLSOLS Model 1 (Total Effect) OLS 2SLSOLS 2SLS Average daily volume of ethanol consumed in the last 12 months Model 4 (Direct Effect) 2SLS Model 2 Model 3

PAGE 147

Appendix A-24: (Continued) 133 Completed graduate degree 0.2694***0.2809***0.1974***0.2008*** (0.0240)(0.0212)(0.0232)(0.0209) Had some graduate studies 0.2520***0.2561***0.1776***0.1789*** (0.0274)(0.0264)(0.0261)(0.0257) Completed college 0.2086***0.2203***0.1451***0.1485*** (0.0223)(0.0193)(0.0216)(0.0189) Had some college educatio n 0.1601***0.1693***0.1142***0.1168*** (0.0188)(0.0165)(0.0179)(0.0161) Completed high school 0.1323***0.1399***0.0928***0.0950*** (0.0192)(0.0174)(0.0182)(0.0169) Excellent healt h 0.5763***0.5735*** (0.0273)(0.0259) Very good healt h 0.5621***0.5595*** (0.0269)(0.0258) Good healt h 0.5269***0.5257*** (0.0262)(0.0258) Fair healt h 0.3064***0.3027*** (0.0304)(0.0283) Constant 0.6946***0.7455***0.7150***0.7490***0.5984***0.6114***0.07240.0785 (0.1436)(0.1198)(0.1291)(0.1185)(0.1218)(0.1178)(0.1181)(0.1165) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10%

PAGE 148

Appendix B: Complete Second Stage Regression Results Using NSDUH 134

PAGE 149

Appendix B-1: Employment (Full Time /Part Time) Last Week versus Alcohol Abuse and/or Dependence in th e Last 12 Months: the Male Sample 135 Variables -0.1929***-0.0325***-0.1377*-0.0236***-0.0764-0.0207***-0.0689-0.0188*** (0.0677)(0.0063)(0.0743)(0.0063)(0.0788)(0.0063)(0.0793)(0.0063) Age between 30 and 34 years old 0.01150.0174***0.00530.0080 0.00610.0073 0.00700.0082 (0.0073)(0.0067)(0.0070)(0.0067)(0.0069)(0.0067)(0.0069)(0.0067) Age between 35 and 49 years old -0.00320.0094 -0.0126*-0.0060 -0.0102-0.0071 -0.0076-0.0046 (0.0079)(0.0058)(0.0074)(0.0060)(0.0074)(0.0060)(0.0076)(0.0060) White 0.0668***0.0659***0.0637***0.0627***0.0629***0.0624***0.0620***0.0615*** (0.0090)(0.0088)(0.0089)(0.0087)(0.0087)(0.0087)(0.0088)(0.0087) Black 0.00520.0024 0.01120.0107 0.01730.0175 0.01660.0168 (0.0111)(0.0108)(0.0109)(0.0107)(0.0108)(0.0108)(0.0108)(0.0108) Hispanic 0.0615***0.0581***0.0575***0.0545***0.0703***0.0698***0.0699***0.0694*** (0.0106)(0.0103)(0.0105)(0.0102)(0.0104)(0.0104)(0.0104)(0.0104) Living in a MSA with >1 million persons 0.00060.0025 0.00460.0065 -0.0017-0.0013 -0.0027-0.0024 (0.0057)(0.0055)(0.0058)(0.0055)(0.0057)(0.0056)(0.0057)(0.0056) Living in a MSA with <1 million persons 0.00470.0049 0.00680.0072 0.00190.0018 0.00120.0010 (0.0056)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055) Number of children aged <18 in household 0.00040.0012 0.00190.0024 0.00180.0022 (0.0022)(0.0022)(0.0023)(0.0022)(0.0022)(0.0022) Married 0.0504***0.0595***0.0532***0.0575***0.0530***0.0568*** (0.0083)(0.0058)(0.0084)(0.0058)(0.0084)(0.0058) Widowed 0.01480.0233 0.02200.0264 0.02410.0281 (0.0336)(0.0327)(0.0333)(0.0326)(0.0332)(0.0325) Divorced or separated 0.0256***0.0250***0.0278***0.0277***0.0286***0.0286*** (0.0076)(0.0074)(0.0075)(0.0074)(0.0074)(0.0074) Completed undergraduate/graduate stud y 0.0523***0.0559***0.0449***0.0473*** (0.0090)(0.0073)(0.0084)(0.0075) Had some college educatio n 0.0354***0.0366***0.0301***0.0307*** (0.0076)(0.0073)(0.0075)(0.0074) Completed high school 0.0236***0.0251***0.0192***0.0202*** (0.0073)(0.0069)(0.0072)(0.0070) Excellent healt h 0.1185***0.1228*** (0.0300)(0.0291) Very good healt h 0.1138***0.1171*** (0.0295)(0.0290) Good healt h 0.1033***0.1051*** (0.0293)(0.0291) Fair healt h 0.0851***0.0859*** (0.0304)(0.0303) Constant 0.9093***0.8798***0.8735***0.8476***0.8312***0.8169***0.7242***0.7089*** (0.0163)(0.0103)(0.0199)(0.0107)(0.0234)(0.0119)(0.0392)(0.0308) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Model 4 (Direct Effect) 2SLSOLS Model 1 (Total Effect) Model 2 Model 3 OLS 2SLSOLS 2SLS Alcohol abuse and/or dependence in the last 12 months Male (n=12,046) 2SLSOLS

PAGE 150

Appendix B-2: Employment (Full Time/Part Time) Last Week versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Female Sample 136 Variables -0.6529***-0.0276***-0.6759***-0.0212**-0.5304***-0.0157 -0.5102***-0.0116 (0.1733)(0.0096)(0.1922)(0.0097)(0.1903)(0.0096)(0.1906)(0.0096) Age between 30 and 34 years old -0.00850.0004 -0.0085-0.0045 -0.0065-0.0032 -0.0052-0.0015 (0.0090)(0.0074)(0.0089)(0.0075)(0.0084)(0.0075)(0.0084)(0.0074) Age between 35 and 49 years old -0.01130.0076 -0.01280.0026 -0.00600.0067 -0.00250.0109* (0.0090)(0.0063)(0.0089)(0.0066)(0.0087)(0.0065)(0.0088)(0.0065) White 0.0429***0.0342***0.0428***0.0314***0.0396***0.0306***0.0338***0.0233** (0.0115)(0.0096)(0.0118)(0.0096)(0.0111)(0.0095)(0.0113)(0.0095) Black -0.0125-0.0052 -0.0144-0.0006 -0.00640.0056 -0.00680.0042 (0.0133)(0.0113)(0.0140)(0.0113)(0.0133)(0.0113)(0.0132)(0.0112) Hispanic -0.0205-0.0071 -0.0214-0.0090 0.00130.0152 0.00200.0149 (0.0140)(0.0116)(0.0141)(0.0116)(0.0140)(0.0116)(0.0138)(0.0116) Living in a MSA with >1 million persons -0.0017-0.0128**-0.0015-0.0104*-0.0107-0.0192***-0.0123-0.0210*** (0.0077)(0.0061)(0.0077)(0.0061)(0.0075)(0.0061)(0.0076)(0.0061) Living in a MSA with <1 million persons 0.0001-0.0024 0.0001-0.0012 -0.0055-0.0076 -0.0057-0.0079 (0.0069)(0.0059)(0.0070)(0.0059)(0.0066)(0.0059)(0.0066)(0.0059) Number of children aged <18 in household -0.00190.0025 0.00170.0058**0.00150.0052** (0.0030)(0.0023)(0.0030)(0.0023)(0.0029)(0.0023) Married -0.00360.0308***0.00020.0264***-0.00130.0231*** (0.0126)(0.0065)(0.0121)(0.0064)(0.0117)(0.0064) Widowed -0.0542**-0.0434*-0.0452*-0.0353 -0.0442*-0.0341 (0.0264)(0.0222)(0.0249)(0.0221)(0.0247)(0.0220) Divorced or separated 0.0203**0.0125 0.0213**0.0159**0.0210**0.0158** (0.0094)(0.0077)(0.0088)(0.0077)(0.0087)(0.0077) Completed undergraduate/graduate stud y 0.0890***0.1079***0.0761***0.0898*** (0.0119)(0.0087)(0.0111)(0.0089) Had some college educatio n 0.0836***0.0991***0.0740***0.0862*** (0.0111)(0.0086)(0.0106)(0.0086) Completed high school 0.0530***0.0645***0.0455***0.0544*** (0.0104)(0.0085)(0.0100)(0.0085) Excellent healt h 0.02920.0697** (0.0393)(0.0326) Very good healt h 0.03070.0614* (0.0379)(0.0325) Good healt h 0.00890.0388 (0.0379)(0.0326) Fair healt h -0.0436-0.0275 (0.0378)(0.0336) Constant 0.9556***0.9122***0.9595***0.8931***0.8759***0.8095***0.8664***0.7744*** (0.0178)(0.0113)(0.0240)(0.0120)(0.0289)(0.0138)(0.0521)(0.0347) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Alcohol abuse and/or dependence in the last 12 months 2SLSOLS Model 3 Model 4 (Direct Effect) OLS Model 1 (Total Effect) Model 2 Female (n=11,779) 2SLSOLS 2SLSOLS 2SLS

PAGE 151

Appendix B-3: Working Full Time Last Week versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Male Sample 137 Variables -0.3232***-0.0380***-0.2157**-0.0228***-0.1672-0.0199**-0.1464-0.0167** (0.0925)(0.0084)(0.0999)(0.0084)(0.1062)(0.0084)(0.1063)(0.0084) Age between 30 and 34 years old 0.0193*0.0297***0.00670.0112 0.00720.0104 0.00930.0123 (0.0100)(0.0089)(0.0094)(0.0089)(0.0093)(0.0089)(0.0093)(0.0089) Age between 35 and 49 years old 0.01060.0329***-0.00910.0020 -0.00740.0008 -0.00180.0060 (0.0109)(0.0078)(0.0100)(0.0080)(0.0100)(0.0080)(0.0102)(0.0080) White 0.0925***0.0910***0.0863***0.0847***0.0854***0.0841***0.0842***0.0828*** (0.0123)(0.0117)(0.0119)(0.0116)(0.0118)(0.0116)(0.0117)(0.0116) Black 0.01150.0065 0.02170.0208 0.0264*0.0271*0.0252*0.0257* (0.0152)(0.0144)(0.0146)(0.0143)(0.0146)(0.0144)(0.0145)(0.0143) Hispanic 0.0977***0.0916***0.0887***0.0837***0.1005***0.0993***0.0993***0.0981*** (0.0145)(0.0137)(0.0142)(0.0136)(0.0141)(0.0139)(0.0140)(0.0138) Living in a MSA with >1 million persons -0.00210.0012 0.00620.0096 0.00130.0025 -0.00040.0004 (0.0079)(0.0074)(0.0077)(0.0074)(0.0077)(0.0075)(0.0076)(0.0075) Living in a MSA with <1 million persons 0.00350.0037 0.00790.0086 0.00400.0037 0.00250.0020 (0.0077)(0.0073)(0.0074)(0.0073)(0.0074)(0.0073)(0.0074)(0.0073) Number of children aged <18 in household 0.00280.0042 0.00420.0053*0.00390.0049* (0.0030)(0.0029)(0.0030)(0.0029)(0.0030)(0.0029) Married 0.0939***0.1094***0.0960***0.1074***0.0960***0.1059*** (0.0112)(0.0077)(0.0113)(0.0077)(0.0112)(0.0077) Widowed 0.03850.0528 0.04440.0561 0.04980.0603 (0.0451)(0.0436)(0.0449)(0.0435)(0.0446)(0.0433) Divorced or separated 0.0687***0.0676***0.0705***0.0703***0.0722***0.0722*** (0.0102)(0.0099)(0.0101)(0.0099)(0.0100)(0.0099) Completed undergraduate/graduate stud y 0.0482***0.0578***0.0358***0.0420*** (0.0121)(0.0098)(0.0113)(0.0100) Had some college educatio n 0.0339***0.0374***0.0242**0.0258*** (0.0102)(0.0098)(0.0101)(0.0099) Completed high school 0.0277***0.0317***0.0191**0.0216** (0.0098)(0.0093)(0.0096)(0.0093) Excellent healt h 0.3456***0.3569*** (0.0402)(0.0388) Very good healt h 0.3436***0.3521*** (0.0396)(0.0386) Good healt h 0.3323***0.3369*** (0.0393)(0.0387) Fair healt h 0.2851***0.2872*** (0.0408)(0.0403) Constant 0.8422***0.7899***0.7716***0.7279***0.7317***0.6940***0.3997***0.3600*** (0.0222)(0.0138)(0.0268)(0.0142)(0.0316)(0.0159)(0.0526)(0.0410) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Alcohol abuse and/or dependence in the last 12 months OLS 2SLSOLS 2SLS Male (n=12,046) OLS Model 1 (Total Effect) Model 2 Model 3 2SLS Model 4 (Direct Effect) 2SLSOLS

PAGE 152

Appendix B-4: Working Full Time Last Week versus Alcohol Abuse and/or Dependence in the Last 12 Months: the Female Sample 138 Variables -0.21910.0205 -0.4763-0.0070 -0.3147-0.0012 -0.27590.0030 (0.2649)(0.0170)(0.2942)(0.0169)(0.3042)(0.0169)(0.3059)(0.0169) Age between 30 and 34 years old -0.0063-0.0029 0.01540.0183 0.01730.0194 0.01890.0210 (0.0138)(0.0131)(0.0137)(0.0131)(0.0134)(0.0131)(0.0134)(0.0131) Age between 35 and 49 years old 0.00880.0161 0.00930.0203*0.01650.0242**0.02160.0291** (0.0138)(0.0112)(0.0137)(0.0115)(0.0138)(0.0115)(0.0142)(0.0115) White -0.0079-0.0113 0.0000-0.0081 -0.0038-0.0093 -0.0123-0.0182 (0.0175)(0.0170)(0.0181)(0.0168)(0.0178)(0.0168)(0.0182)(0.0168) Black 0.0702***0.0730***0.0661***0.0760***0.0759***0.0832***0.0752***0.0814*** (0.0203)(0.0199)(0.0214)(0.0198)(0.0213)(0.0198)(0.0211)(0.0198) Hispanic -0.0154-0.0103 0.00060.0095 0.02710.0355*0.02880.0361* (0.0214)(0.0204)(0.0216)(0.0203)(0.0223)(0.0205)(0.0221)(0.0204) Living in a MSA with >1 million persons -0.0229*-0.0272**-0.0324***-0.0387***-0.0431***-0.0483***-0.0453***-0.0502*** (0.0118)(0.0108)(0.0117)(0.0107)(0.0120)(0.0108)(0.0121)(0.0108) Living in a MSA with <1 million persons -0.0097-0.0107 -0.0150-0.0159 -0.0217**-0.0230**-0.0220**-0.0232** (0.0106)(0.0105)(0.0107)(0.0104)(0.0106)(0.0104)(0.0106)(0.0104) Number of children aged <18 in household -0.0588***-0.0557***-0.0546***-0.0521***-0.0547***-0.0526*** (0.0046)(0.0040)(0.0047)(0.0040)(0.0047)(0.0040) Married -0.0649***-0.0403***-0.0606***-0.0446***-0.0619***-0.0483*** (0.0193)(0.0113)(0.0193)(0.0113)(0.0188)(0.0113) Widowed -0.0981**-0.0903**-0.0879**-0.0818**-0.0852**-0.0796** (0.0404)(0.0388)(0.0398)(0.0388)(0.0396)(0.0387) Divorced or separated 0.0499***0.0444***0.0521***0.0488***0.0518***0.0488*** (0.0144)(0.0135)(0.0141)(0.0135)(0.0140)(0.0135) Completed undergraduate/graduate stud y 0.1076***0.1191***0.0899***0.0975*** (0.0191)(0.0152)(0.0179)(0.0156) Had some college educatio n 0.0875***0.0970***0.0732***0.0800*** (0.0178)(0.0150)(0.0171)(0.0152) Completed high school 0.0714***0.0784***0.0597***0.0647*** (0.0166)(0.0149)(0.0161)(0.0150) Excellent healt h 0.1075*0.1301** (0.0631)(0.0574) Very good healt h 0.1289**0.1461** (0.0608)(0.0572) Good healt h 0.09230.1090* (0.0608)(0.0573) Fair healt h 0.00250.0115 (0.0606)(0.0591) Constant 0.7502***0.7336***0.8584***0.8108***0.7606***0.7202***0.6699***0.6183*** (0.0272)(0.0199)(0.0368)(0.0209)(0.0462)(0.0242)(0.0836)(0.0610) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Alcohol abuse and/or dependence in the last 12 months 2SLSOLS 2SLSOLS 2SLSOLS Female (n=11,779) Model 1 (Total Effect) Model 2 Model 3 2SLSOLS Model 4 (Direct Effect)

PAGE 153

Appendix B-5: Employment (Full Time/Part Time) Last Week versus the Number of Days Drinking Alcohol in the Last 12 Months: the Male Sample 139 Variables -0.00033***-0.00004*-0.00022*-0.00002-0.00011-0.00002-0.00010-0.00001 (0.0001)(0.0000)(0.0001)(0.0000)(0.0001)(0.0000)(0.0001)(0.0000) Age between 30 and 34 years old 0.0170**0.0184***0.00810.0085 0.00750.0077 0.00840.0086 (0.0067)(0.0067)(0.0067)(0.0067)(0.0067)(0.0067)(0.0067)(0.0067) Age between 35 and 49 years old 0.0120**0.0119**-0.0038-0.0046 -0.0055-0.0059 -0.0031-0.0034 (0.0058)(0.0058)(0.0060)(0.0060)(0.0060)(0.0060)(0.0060)(0.0060) White 0.0731***0.0666***0.0674***0.0629***0.0648***0.0626***0.0636***0.0616*** (0.0092)(0.0088)(0.0091)(0.0087)(0.0091)(0.0087)(0.0091)(0.0087) Black 0.00680.0024 0.01340.0108 0.0190*0.0178*0.0181*0.0170 (0.0110)(0.0108)(0.0109)(0.0107)(0.0109)(0.0108)(0.0109)(0.0108) Hispanic 0.0568***0.0574***0.0538***0.0539***0.0695***0.0696***0.0691***0.0692*** (0.0103)(0.0103)(0.0103)(0.0102)(0.0104)(0.0104)(0.0104)(0.0104) Living in a MSA with >1 million persons 0.00560.0032 0.00840.0070 -0.0002-0.0010 -0.0015-0.0022 (0.0057)(0.0056)(0.0056)(0.0055)(0.0057)(0.0056)(0.0057)(0.0056) Living in a MSA with <1 million persons 0.00670.0051 0.00830.0074 0.00240.0018 0.00140.0010 (0.0056)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055) Number of children aged <18 in household 0.00040.0013 0.00200.0025 0.00190.0023 (0.0022)(0.0022)(0.0022)(0.0022)(0.0022)(0.0022) Married 0.0590***0.0612***0.0579***0.0590***0.0572***0.0581*** (0.0060)(0.0058)(0.0059)(0.0058)(0.0059)(0.0058) Widowed 0.02390.0249 0.02740.0279 0.02900.0295 (0.0328)(0.0327)(0.0326)(0.0326)(0.0326)(0.0326) Divorced or separated 0.0265***0.0250***0.0285***0.0278***0.0294***0.0287*** (0.0075)(0.0074)(0.0075)(0.0074)(0.0075)(0.0074) Completed undergraduate/graduate stud y 0.0570***0.0572***0.0480***0.0481*** (0.0073)(0.0073)(0.0075)(0.0075) Had some college educatio n 0.0370***0.0371***0.0309***0.0310*** (0.0073)(0.0073)(0.0074)(0.0074) Completed high school 0.0259***0.0257***0.0208***0.0206*** (0.0070)(0.0069)(0.0070)(0.0070) Excellent healt h 0.1232***0.1243*** (0.0292)(0.0291) Very good healt h 0.1174***0.1182*** (0.0290)(0.0290) Good healt h 0.1047***0.1056*** (0.0291)(0.0291) Fair healt h 0.0851***0.0861*** (0.0303)(0.0303) Constant 0.8942***0.8763***0.8575***0.8435***0.8194***0.8126***0.7112***0.7043*** (0.0125)(0.0103)(0.0134)(0.0107)(0.0143)(0.0119)(0.0322)(0.0308) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% 2SLSOLS 2SLSOLS 2SLSOLS Male (n=12,046) Model 1 (Total Effect) Model 2 Model 3 Number of days drinking alcohol in the last 12 months Model 4 (Direct Effect) 2SLSOLS

PAGE 154

Appendix B-6: Employment (Full Time/Part Time) Last Week versus the Number of Days Drinking Alcohol in the Last Month: the Male Sample 140 Variables -0.00382***-0.00041-0.00246*-0.00016-0.00127-0.00018-0.00113-0.00017 (0.0013)(0.0003)(0.0013)(0.0003)(0.0013)(0.0003)(0.0013)(0.0003) Age between 30 and 34 years old 0.0184***0.0185***0.00900.0085 0.00790.0077 0.00880.0086 (0.0067)(0.0067)(0.0067)(0.0067)(0.0067)(0.0067)(0.0067)(0.0067) Age between 35 and 49 years old 0.0134**0.0121**-0.0030-0.0046-0.005113-0.005838 -0.0027-0.0033 (0.0058)(0.0058)(0.0061)(0.0060)(0.0060)(0.0060)(0.0060)(0.0060) White 0.0734***0.0666***0.0675***0.0628***0.0648***0.0626***0.0636***0.0617*** (0.0092)(0.0088)(0.0092)(0.0087)(0.0091)(0.0087)(0.0091)(0.0087) Black 0.00440.0021 0.01180.0107 0.0183*0.0177*0.01750.0170 (0.0109)(0.0108)(0.0108)(0.0107)(0.0108)(0.0108)(0.0108)(0.0108) Hispanic 0.0547***0.0571***0.0524***0.0538***0.0689***0.0695***0.0686***0.0691*** (0.0104)(0.0103)(0.0103)(0.0102)(0.0104)(0.0104)(0.0104)(0.0104) Living in a MSA with >1 million persons 0.00520.0031 0.00820.0070 -0.0005-0.0010 -0.0017-0.0022 (0.0056)(0.0056)(0.0056)(0.0055)(0.0057)(0.0056)(0.0057)(0.0056) Living in a MSA with <1 million persons 0.00690.0051 0.00840.0074 0.00230.0018 0.00140.0010 (0.0056)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055) Number of children aged <18 in household 0.00050.0013 0.00210.0025 0.00190.0023 (0.0022)(0.0022)(0.0022)(0.0022)(0.0022)(0.0022) Married 0.0593***0.0613***0.0580***0.0590***0.0573***0.0581*** (0.0059)(0.0058)(0.0059)(0.0058)(0.0059)(0.0058) Widowed 0.02390.0250 0.02750.0280 0.02910.0295 (0.0328)(0.0327)(0.0326)(0.0326)(0.0326)(0.0326) Divorced or separated 0.0264***0.0249***0.0285***0.0278***0.0293***0.0287*** (0.0075)(0.0074)(0.0075)(0.0074)(0.0075)(0.0074) Completed undergraduate/graduate study 0.0576***0.0573***0.0485***0.0482*** (0.0073)(0.0073)(0.0075)(0.0075) Had some college education 0.0373***0.0371***0.0311***0.0310*** (0.0073)(0.0073)(0.0074)(0.0074) Completed high school 0.0259***0.0257***0.0208***0.0206*** (0.0070)(0.0069)(0.0070)(0.0070) Excellent health 0.1237***0.1244*** (0.0291)(0.0291) Very good health 0.1178***0.1182*** (0.0290)(0.0290) Good health 0.1050***0.1056*** (0.0291)(0.0291) Fair health 0.0857***0.0861*** (0.0303)(0.0303) Constant 0.8904***0.8757***0.8546***0.8431***0.8177***0.8124***0.7093***0.7041*** (0.0117)(0.0103)(0.0125)(0.0106)(0.0134)(0.0119)(0.0316)(0.0308) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Model 4 (Direct Effect) Number of days drinking alcohol in the last month Male (n=12,046) Model 1 (Total Effect) Model 2 Model 3 2SLSOLS 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 155

Appendix B-7: Employment (Full Time/Part Time) Last Week versus the Number of Days Drank 5+ Al coholic Drinks in the Last Month: the Male Sample 141 Variables -0.00616***-0.00224***-0.00413*-0.00166***-0.00227-0.00127***-0.00204-0.00116*** (0.0021)(0.0004)(0.0022)(0.0004)(0.0023)(0.0004)(0.0023)(0.0004) Age between 30 and 34 years old 0.0171**0.0180***0.00820.0084 0.00760.0077 0.00850.0085 (0.0067)(0.0067)(0.0067)(0.0067)(0.0067)(0.0067)(0.0067)(0.0067) Age between 35 and 49 years old 0.00840.0106*-0.0060-0.0052 -0.0066-0.0063 -0.0041-0.0038 (0.0059)(0.0058)(0.0060)(0.0060)(0.0060)(0.0060)(0.0060)(0.0060) White 0.0682***0.0666***0.0643***0.0632***0.0632***0.0628***0.0622***0.0618*** (0.0088)(0.0088)(0.0088)(0.0087)(0.0088)(0.0087)(0.0087)(0.0087) Black 0.00160.0018 0.00970.0102 0.01660.0170 0.01590.0163 (0.0108)(0.0108)(0.0108)(0.0107)(0.0108)(0.0108)(0.0108)(0.0108) Hispanic 0.0565***0.0571***0.0535***0.0538***0.0684***0.0689***0.0681***0.0686*** (0.0103)(0.0103)(0.0102)(0.0102)(0.0105)(0.0104)(0.0105)(0.0104) Living in a MSA with >1 million persons 0.00090.0021 0.00540.0063 -0.0013-0.0012 -0.0025-0.0024 (0.0056)(0.0055)(0.0056)(0.0055)(0.0056)(0.0056)(0.0056)(0.0056) Living in a MSA with <1 million persons 0.00450.0048 0.00690.0071 0.00200.0019 0.00110.0010 (0.0055)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055)(0.0055) Number of children aged <18 in household 0.00110.0013 0.00230.0024 0.00210.0022 (0.0022)(0.0022)(0.0022)(0.0022)(0.0022)(0.0022) Married 0.0573***0.0598***0.0570***0.0579***0.0564***0.0572*** (0.0062)(0.0058)(0.0062)(0.0058)(0.0062)(0.0058) Widowed 0.02460.0249 0.02760.0278 0.02920.0294 (0.0327)(0.0327)(0.0326)(0.0326)(0.0326)(0.0325) Divorced or separated 0.0260***0.0253***0.0280***0.0279***0.0289***0.0288*** (0.0075)(0.0074)(0.0074)(0.0074)(0.0074)(0.0074) Completed undergraduate/graduate study 0.0536***0.0552***0.0453***0.0465*** (0.0082)(0.0073)(0.0082)(0.0075) Had some college education 0.0353***0.0361***0.0296***0.0302*** (0.0076)(0.0073)(0.0076)(0.0074) Completed high school 0.0250***0.0253***0.0201***0.0203*** (0.0070)(0.0069)(0.0070)(0.0070) Excellent health 0.1221***0.1231*** (0.0292)(0.0291) Very good health 0.1166***0.1173*** (0.0291)(0.0290) Good health 0.1046***0.1051*** (0.0291)(0.0291) Fair health 0.0853***0.0857*** (0.0303)(0.0303) Constant 0.8887***0.8792***0.8543***0.8471***0.8196***0.8161***0.7119***0.7081*** (0.0115)(0.0103)(0.0124)(0.0106)(0.0144)(0.0119)(0.0324)(0.0308) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Model 4 (Direct Effect) 2SLSOLS Number of days drank 5+ alcoholic drinks in the last month Male (n=12,046) Model 1 (Total Effect) Model 2 Model 3 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 156

Appendix B-8: Working Full Time Last Week versus the Number of Days Drinking Alcohol in the Last 12 Months: the Male Sample 142 Variables -0.00056***-0.00002-0.00036**0.00002-0.00026*0.00003-0.000230.00003 (0.0002)(0.0000)(0.0002)(0.0000)(0.0002)(0.0000)(0.0002)(0.0000) Age between 30 and 34 years old 0.0285***0.0310***0.01100.0117 0.01040.0109 0.01220.0127 (0.0091)(0.0090)(0.0090)(0.0089)(0.0090)(0.0089)(0.0089)(0.0089) Age between 35 and 49 years old 0.0360***0.0359***0.00470.0033 0.00290.0019 0.00780.0069 (0.0078)(0.0077)(0.0080)(0.0080)(0.0080)(0.0080)(0.0080)(0.0079) White 0.1032***0.0912***0.0924***0.0840***0.0897***0.0834***0.0877***0.0820*** (0.0124)(0.0118)(0.0122)(0.0116)(0.0122)(0.0116)(0.0121)(0.0116) Black 0.01420.0060 0.0253*0.0204 0.0304**0.0269*0.0285**0.0254* (0.0148)(0.0145)(0.0146)(0.0143)(0.0146)(0.0144)(0.0145)(0.0143) Hispanic 0.0897***0.0908***0.0829***0.0831***0.0988***0.0991***0.0977***0.0980*** (0.0140)(0.0138)(0.0137)(0.0136)(0.0139)(0.0139)(0.0139)(0.0138) Living in a MSA with >1 million persons 0.00630.0018 0.0124*0.0098 0.00460.0024 0.00230.0002 (0.0076)(0.0074)(0.0075)(0.0074)(0.0076)(0.0075)(0.0076)(0.0075) Living in a MSA with <1 million persons 0.00680.0038 0.01030.0086 0.00500.0035 0.00320.0018 (0.0075)(0.0073)(0.0074)(0.0073)(0.0074)(0.0073)(0.0074)(0.0073) Number of children aged <18 in household 0.00270.0044 0.00430.0056*0.00400.0051* (0.0030)(0.0029)(0.0030)(0.0029)(0.0030)(0.0029) Married 0.1073***0.1114***0.1061***0.1092***0.1047***0.1075*** (0.0080)(0.0077)(0.0079)(0.0077)(0.0079)(0.0077) Widowed 0.05260.0547 0.05620.0578 0.06030.0618 (0.0439)(0.0436)(0.0437)(0.0435)(0.0435)(0.0433) Divorced or separated 0.0703***0.0673***0.0722***0.0701***0.0739***0.0720*** (0.0101)(0.0099)(0.0100)(0.0099)(0.0100)(0.0099) Completed undergraduate/graduate stud y 0.0586***0.0591***0.0425***0.0429*** (0.0098)(0.0098)(0.0100)(0.0100) Had some college educatio n 0.0375***0.0379***0.0258***0.0261*** (0.0098)(0.0098)(0.0099)(0.0099) Completed high school 0.0329***0.0322***0.0225**0.0218** (0.0093)(0.0093)(0.0094)(0.0093) Excellent healt h 0.3554***0.3587*** (0.0389)(0.0388) Very good healt h 0.3511***0.3534*** (0.0387)(0.0386) Good healt h 0.3352***0.3378*** (0.0388)(0.0387) Fair healt h 0.2850***0.2877*** (0.0405)(0.0403) Constant 0.8174***0.7838***0.7475***0.7211***0.7069***0.6872***0.3731***0.3527*** (0.0168)(0.0138)(0.0179)(0.0142)(0.0192)(0.0159)(0.0429)(0.0410) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Model 4 (Direct Effect) 2SLSOLS Male (n=12,046) Model 1 (Total Effect) Model 2 Model 3 2SLSOLS 2SLSOLS 2SLSOLS Number of days drinking alcohol in the last 12 months

PAGE 157

Appendix B-9: Working Full Time Last Week versus the Number of Days Drinking Alcohol in the Last Month: the Male Sample 143 Variables -0.00656***-0.00042-0.00414**-0.00001-0.00301*-0.00002-0.002660.00000 (0.0018)(0.0004)(0.0018)(0.0003)(0.0018)(0.0003)(0.0018)(0.0003) Age between 30 and 34 years old 0.0309***0.0311***0.01250.0117 0.01140.0108 0.01310.0126 (0.0091)(0.0090)(0.0090)(0.0089)(0.0090)(0.0089)(0.0089)(0.0089) Age between 35 and 49 years old 0.0385***0.0361***0.00610.0033 0.00390.0020 0.00870.0070 (0.0079)(0.0077)(0.0081)(0.0080)(0.0081)(0.0080)(0.0080)(0.0079) White 0.1040***0.0917***0.0929***0.0845***0.0901***0.0840***0.0880***0.0826*** (0.0124)(0.0118)(0.0122)(0.0116)(0.0122)(0.0116)(0.0121)(0.0116) Black 0.01030.0061 0.02290.0207 0.0288**0.0272*0.0272*0.0257* (0.0147)(0.0144)(0.0144)(0.0143)(0.0145)(0.0144)(0.0144)(0.0143) Hispanic 0.0861***0.0905***0.0805***0.0831***0.0974***0.0991***0.0965***0.0979*** (0.0140)(0.0138)(0.0138)(0.0136)(0.0139)(0.0139)(0.0139)(0.0138) Living in a MSA with >1 million persons 0.00580.0019 0.01210.0100 0.00410.0026 0.00180.0005 (0.0076)(0.0074)(0.0075)(0.0074)(0.0076)(0.0075)(0.0076)(0.0075) Living in a MSA with <1 million persons 0.00720.0040 0.01060.0087 0.00500.0037 0.00320.0020 (0.0075)(0.0073)(0.0074)(0.0073)(0.0074)(0.0073)(0.0074)(0.0073) Number of children aged <18 in household 0.00290.0043 0.00450.0055*0.00410.0050* (0.0030)(0.0029)(0.0030)(0.0029)(0.0029)(0.0029) Married 0.1076***0.1112***0.1063***0.1089***0.1048***0.1072*** (0.0079)(0.0077)(0.0079)(0.0077)(0.0079)(0.0077) Widowed 0.05270.0545 0.05630.0576 0.06050.0616 (0.0438)(0.0436)(0.0437)(0.0435)(0.0434)(0.0433) Divorced or separated 0.0701***0.0675***0.0722***0.0703***0.0739***0.0722*** (0.0100)(0.0099)(0.0100)(0.0099)(0.0100)(0.0099) Completed undergraduate/graduate stud y 0.0598***0.0591***0.0436***0.0428*** (0.0098)(0.0098)(0.0100)(0.0100) Had some college educatio n 0.0381***0.0378***0.0263***0.0260*** (0.0098)(0.0098)(0.0099)(0.0099) Completed high school 0.0329***0.0323***0.0224**0.0219** (0.0093)(0.0093)(0.0093)(0.0093) Excellent healt h 0.3564***0.3584*** (0.0389)(0.0388) Very good healt h 0.3519***0.3532*** (0.0387)(0.0386) Good healt h 0.3357***0.3375*** (0.0388)(0.0387) Fair healt h 0.2861***0.2874*** (0.0404)(0.0403) Constant 0.8113***0.7847***0.7434***0.7227***0.7034***0.6890***0.3693***0.3549*** (0.0158)(0.0138)(0.0167)(0.0142)(0.0180)(0.0159)(0.0422)(0.0410) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% 2SLSOLS 2SLSOLS Male (n=12,046) Model 1 (Total Effect) Model 2 Model 3 2SLSOLS 2SLSOLS Model 4 (Direct Effect) Number of days drinking alcohol in the last month

PAGE 158

Appendix B-10: Working Full Time Last Week versus the Number of Days Drank 5+ Alcoholic Drinks in th e Last Month: the Male Sample 144 Variables -0.01016***-0.00237***-0.00648**-0.00140**-0.00497-0.00102*-0.00435-0.00084 (0.0029)(0.0006)(0.0029)(0.0006)(0.0031)(0.0006)(0.0031)(0.0006) Age between 30 and 34 years old 0.0288***0.0306***0.01120.0116 0.01060.0108 0.01240.0126 (0.0090)(0.0089)(0.0090)(0.0089)(0.0090)(0.0089)(0.0089)(0.0089) Age between 35 and 49 years old 0.0302***0.0346***0.00120.0029 0.00050.0017 0.00550.0067 (0.0079)(0.0077)(0.0080)(0.0080)(0.0080)(0.0080)(0.0080)(0.0079) White 0.0948***0.0918***0.0873***0.0851***0.0860***0.0844***0.0845***0.0830*** (0.0119)(0.0117)(0.0117)(0.0116)(0.0117)(0.0116)(0.0117)(0.0116) Black 0.00550.0057 0.01940.0204 0.0248*0.0267*0.0237*0.0253* (0.0145)(0.0144)(0.0144)(0.0143)(0.0145)(0.0144)(0.0144)(0.0143) Hispanic 0.0893***0.0905***0.0824***0.0829***0.0964***0.0985***0.0956***0.0975*** (0.0139)(0.0138)(0.0137)(0.0136)(0.0140)(0.0139)(0.0139)(0.0138) Living in a MSA with >1 million persons -0.00160.0009 0.00750.0094 0.00220.0025 0.00020.0004 (0.0075)(0.0074)(0.0075)(0.0074)(0.0075)(0.0075)(0.0075)(0.0075) Living in a MSA with <1 million persons 0.00310.0036 0.00810.0086 0.00410.0038 0.00250.0021 (0.0074)(0.0073)(0.0073)(0.0073)(0.0073)(0.0073)(0.0073)(0.0073) Number of children aged <18 in household 0.00380.0042 0.0050*0.0054*0.00460.0049* (0.0029)(0.0029)(0.0029)(0.0029)(0.0029)(0.0029) Married 0.1046***0.1098***0.1043***0.1080***0.1031***0.1064*** (0.0083)(0.0077)(0.0083)(0.0077)(0.0082)(0.0077) Widowed 0.05390.0544 0.05660.0574 0.06070.0615 (0.0437)(0.0436)(0.0436)(0.0435)(0.0434)(0.0433) Divorced or separated 0.0693***0.0679***0.0711***0.0704***0.0729***0.0723*** (0.0100)(0.0099)(0.0100)(0.0099)(0.0099)(0.0099) Completed undergraduate/graduate study 0.0512***0.0575***0.0367***0.0416*** (0.0110)(0.0098)(0.0109)(0.0100) Had some college education 0.0339***0.0370***0.0231**0.0255*** (0.0101)(0.0098)(0.0101)(0.0099) Completed high school 0.0308***0.0320***0.0210**0.0217** (0.0093)(0.0093)(0.0093)(0.0093) Excellent health 0.3533***0.3574*** (0.0390)(0.0388) Very good health 0.3494***0.3524*** (0.0387)(0.0386) Good health 0.3350***0.3370*** (0.0388)(0.0387) Fair health 0.2854***0.2870*** (0.0404)(0.0403) Constant 0.8073***0.7886***0.7416***0.7268***0.70651***0.6925***0.3735***0.3585*** (0.0154)(0.0138)(0.0165)(0.0142)(0.0193)(0.0159)(0.0431)(0.0410) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Number of days drank 5+ alcoholic drinks in the last month 2SLSOLS 2SLSOLS 2SLSOLS Male (n=12,046) Model 1 (Total Effect) Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS

PAGE 159

Appendix B-11: Employment (Full Time/Part Time) Last Week versus the Number of Days Drinking Alcohol in the Last 12 Months: the Female Sample 145 Variables -0.00102***-0.00009***-0.00099***-0.00007**-0.00073***-0.00008**-0.00069***-0.00009*** (0.0002)(0.0000)(0.0002)(0.0000)(0.0002)(0.0000)(0.0002)(0.0000) Age between 30 and 34 years old -0.00210.0006 -0.0041-0.0043 -0.0029-0.0031 -0.0011-0.0014 (0.0077)(0.0074)(0.0078)(0.0075)(0.0076)(0.0075)(0.0075)(0.0074) Age between 35 and 49 years old 0.0111*0.0086 0.00820.0034 0.01100.0075 0.0153**0.0117* (0.0066)(0.0063)(0.0069)(0.0066)(0.0068)(0.0065)(0.0067)(0.0065) White 0.0504***0.0352***0.0481***0.0323***0.0429***0.0316***0.0343***0.0244** (0.0107)(0.0096)(0.0108)(0.0096)(0.0105)(0.0095)(0.0104)(0.0095) Black -0.0056-0.0050 -0.0020-0.0003 0.00530.0059 0.00360.0043 (0.0117)(0.0113)(0.0117)(0.0113)(0.0115)(0.0113)(0.0114)(0.0112) Hispanic -0.0146-0.0072 -0.0146-0.0091 0.01190.0152 0.01160.0148 (0.0121)(0.0116)(0.0121)(0.0116)(0.0119)(0.0116)(0.0118)(0.0116) Living in a MSA with >1 million persons -0.0021-0.0123**-0.0015-0.0100 -0.0134**-0.0188***-0.0156**-0.0205*** (0.0068)(0.0061)(0.0067)(0.0061)(0.0065)(0.0061)(0.0065)(0.0061) Living in a MSA with <1 million persons 0.0023-0.0021 0.0027-0.0010 -0.0052-0.0074 -0.0056-0.0076 (0.0062)(0.0059)(0.0062)(0.0059)(0.0061)(0.0059)(0.0060)(0.0059) Number of children aged <18 in household -0.00160.0023 0.00300.0056**0.00250.0050** (0.0026)(0.0023)(0.0025)(0.0023)(0.0025)(0.0023) Married 0.0192***0.0310***0.0179**0.0262***0.0146**0.0226*** (0.0074)(0.0064)(0.0072)(0.0064)(0.0072)(0.0064) Widowed -0.0541**-0.0438**-0.0425*-0.0358 -0.0409*-0.0348 (0.0231)(0.0222)(0.0226)(0.0220)(0.0224)(0.0220) Divorced or separated 0.00910.0121 0.0139*0.0156**0.0140*0.0155** (0.0080)(0.0077)(0.0079)(0.0077)(0.0078)(0.0077) Completed undergraduate/graduate stud y 0.1116***0.1088***0.0917***0.0903*** (0.0089)(0.0087)(0.0090)(0.0088) Had some college educatio n 0.0992***0.0996***0.0851***0.0863*** (0.0087)(0.0086)(0.0088)(0.0086) Completed high school 0.0638***0.0647***0.0530***0.0544*** (0.0086)(0.0085)(0.0087)(0.0085) Excellent healt h 0.0853**0.0724** (0.0335)(0.0326) Very good healt h 0.0768**0.0640** (0.0333)(0.0325) Good healt h 0.04890.0407 (0.0332)(0.0326) Fair healt h -0.0145-0.0256 (0.0344)(0.0336) Constant 0.9423***0.9131***0.9331***0.8939***0.8376***0.8107***0.7891***0.7743*** (0.0138)(0.0113)(0.0162)(0.0120)(0.0171)(0.0138)(0.0356)(0.0346) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% 2SLSOLS 2SLSOLS 2SLSOLS Female (n=11,779) Model 1 (Total Effect) Model 2 Model 3 Number of days drinking alcohol in the last 12 months Model 4 (Direct Effect) 2SLSOLS

PAGE 160

Appendix B-12: Employment (Full Time/Part Time) Last Week versus the Number of Days Drinking Alcohol in the Last Month: the Female Sample 146 Variables -0.01208***-0.00093**-0.01164***-0.00075*-0.00842***-0.00097**-0.00803***-0.00106*** (0.0028)(0.0004)(0.0029)(0.0004)(0.0027)(0.0004)(0.0027)(0.0004) Age between 30 and 34 years old 0.00030.0008 -0.0024-0.0042 -0.0017-0.0029 0.0001-0.0012 (0.0077)(0.0074)(0.0078)(0.0075)(0.0076)(0.0075)(0.0075)(0.0074) Age between 35 and 49 years old 0.0144**0.0089 0.01050.0036 0.0128*0.0077 0.0170**0.0120* (0.0067)(0.0063)(0.0070)(0.0066)(0.0069)(0.0065)(0.0069)(0.0065) White 0.0523***0.0352***0.0496***0.0323***0.0439***0.0318***0.0354***0.0247*** (0.0108)(0.0096)(0.0109)(0.0096)(0.0106)(0.0096)(0.0105)(0.0095) Black -0.0073-0.0051 -0.0029-0.0003 0.00510.0059 0.00350.0043 (0.0117)(0.0113)(0.0117)(0.0113)(0.0114)(0.0113)(0.0114)(0.0112) Hispanic -0.0135-0.0070 -0.0135-0.0090 0.01360.0153 0.01320.0150 (0.0121)(0.0116)(0.0120)(0.0116)(0.0118)(0.0116)(0.0118)(0.0116) Living in a MSA with >1 million persons -0.0002-0.0123**0.0006-0.0099 -0.0125*-0.0187***-0.0147**-0.0203*** (0.0070)(0.0061)(0.0069)(0.0061)(0.0066)(0.0061)(0.0066)(0.0061) Living in a MSA with <1 million persons 0.0046-0.0020 0.0050-0.0009 -0.0039-0.0073 -0.0043-0.0075 (0.0063)(0.0059)(0.0063)(0.0059)(0.0061)(0.0059)(0.0061)(0.0059) Number of children aged <18 in household -0.00170.0024 0.00310.0056**0.00260.0050** (0.0026)(0.0023)(0.0025)(0.0023)(0.0025)(0.0023) Married 0.0235***0.0313***0.0210***0.0265***0.0176***0.0229*** (0.0070)(0.0064)(0.0068)(0.0064)(0.0068)(0.0064) Widowed -0.0493**-0.0434*-0.0385*-0.0354 -0.0372*-0.0343 (0.0230)(0.0222)(0.0224)(0.0220)(0.0223)(0.0220) Divorced or separated 0.01160.0122 0.0159**0.0158**0.0160**0.0157** (0.0080)(0.0077)(0.0078)(0.0077)(0.0078)(0.0077) Completed undergraduate/graduate study 0.1152***0.1093***0.0951***0.0908*** (0.0091)(0.0087)(0.0091)(0.0089) Had some college education 0.1009***0.0998***0.0868***0.0865*** (0.0087)(0.0086)(0.0087)(0.0086) Completed high school 0.0636***0.0647***0.0528***0.0544*** (0.0086)(0.0085)(0.0087)(0.0085) Excellent health 0.0787**0.0717** (0.0331)(0.0326) Very good health 0.0696**0.0631* (0.0330)(0.0325) Good health 0.04150.0398 (0.0330)(0.0326) Fair health -0.0207-0.0263 (0.0341)(0.0336) Constant 0.9309***0.9119***0.9196***0.8928***0.8257***0.8096***0.7847***0.7739*** (0.0126)(0.0113)(0.0142)(0.0120)(0.0152)(0.0138)(0.0353)(0.0346) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Model 4 (Direct Effect) Number of days drinking alcohol in the last month 2SLSOLS 2SLSOLS 2SLSOLS 2SLSOLS Female (n=11,779) Model 1 (Total Effect) Model 2 Model 3

PAGE 161

Appendix B-13: Employment (Full Time/Part Time) Last Week versus the Number of Days Drank 5+ Al coholic Drinks in the Last Month: the Female Sample 147 Variables -0.02724***-0.00436***-0.02638***-0.00382***-0.02004***-0.00298***-0.01923***-0.00268*** (0.0063)(0.0009)(0.0064)(0.0009)(0.0066)(0.0009)(0.0066)(0.0009) Age between 30 and 34 years old -0.00380.0001 -0.0065-0.0047 -0.0048-0.0034 -0.0032-0.0017 (0.0077)(0.0074)(0.0077)(0.0075)(0.0076)(0.0074)(0.0076)(0.0074) Age between 35 and 49 years old 0.00320.0076 0.00000.0026 0.00420.0066 0.00810.0108* (0.0066)(0.0063)(0.0068)(0.0066)(0.0067)(0.0065)(0.0067)(0.0065) White 0.0367***0.0342***0.0348***0.0316***0.0331***0.0307***0.0263***0.0235** (0.0099)(0.0096)(0.0099)(0.0096)(0.0097)(0.0095)(0.0097)(0.0095) Black -0.0093-0.0056 -0.0065-0.0011 0.00000.0051 -0.00110.0036 (0.0116)(0.0113)(0.0117)(0.0113)(0.0116)(0.0113)(0.0115)(0.0112) Hispanic -0.0120-0.0074 -0.0130-0.0093 0.00950.0147 0.00960.0145 (0.0119)(0.0116)(0.0119)(0.0116)(0.0120)(0.0116)(0.0119)(0.0116) Living in a MSA with >1 million persons -0.0137**-0.0134**-0.0123*-0.0109*-0.0193***-0.0195***-0.0209***-0.0212*** (0.0062)(0.0061)(0.0063)(0.0061)(0.0062)(0.0061)(0.0062)(0.0061) Living in a MSA with <1 million persons -0.0044-0.0028 -0.0036-0.0016 -0.0085-0.0078 -0.0087-0.0080 (0.0061)(0.0059)(0.0061)(0.0059)(0.0060)(0.0059)(0.0060)(0.0059) Number of children aged <18 in household 0.00060.0024 0.00390.0056**0.00350.0051** (0.0024)(0.0023)(0.0024)(0.0023)(0.0024)(0.0023) Married 0.0184**0.0299***0.0174**0.0258***0.0147**0.0224*** (0.0074)(0.0064)(0.0072)(0.0064)(0.0072)(0.0064) Widowed -0.0414*-0.0428*-0.0350-0.0350 -0.0340-0.0339 (0.0228)(0.0222)(0.0224)(0.0220)(0.0223)(0.0220) Divorced or separated 0.01200.0123 0.0148*0.0156**0.0148*0.0156** (0.0079)(0.0077)(0.0078)(0.0077)(0.0078)(0.0077) Completed undergraduate/graduate study 0.0959***0.1066***0.0795***0.0887*** (0.0097)(0.0087)(0.0097)(0.0089) Had some college education 0.0917***0.0984***0.0798***0.0855*** (0.0091)(0.0086)(0.0090)(0.0086) Completed high school 0.0599***0.0641***0.0506***0.0541*** (0.0088)(0.0085)(0.0087)(0.0085) Excellent health 0.0587*0.0689** (0.0333)(0.0326) Very good health 0.05360.0610* (0.0331)(0.0325) Good health 0.03120.0383 (0.0331)(0.0326) Fair health -0.0312-0.0277 (0.0341)(0.0336) Constant 0.9347***0.9142***0.9234***0.8956***0.8402***0.8124***0.8113***0.7777*** (0.0128)(0.0113)(0.0146)(0.0120)(0.0176)(0.0138)(0.0376)(0.0347) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Model 4 (Direct Effect) 2SLSOLS Number of days drank 5+ alcohol drinks in the last month Female (n=11,779) Model 1 (Total Effect) Model 2 Model 3 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 162

Appendix B-14: Working Full Time Last Week versus the Number of Days Drinking Alcohol in the Last 12 Months: the Female Sample 148 Variables -0.000410.00005-0.00072*-0.00005-0.00046-0.00005-0.00040-0.00006 (0.0004)(0.0001)(0.0004)(0.0001)(0.0004)(0.0001)(0.0004)(0.0001) Age between 30 and 34 years old -0.0043-0.0030 0.01860.0184 0.01950.0194 0.02110.0210 (0.0132)(0.0131)(0.0132)(0.0131)(0.0131)(0.0131)(0.0131)(0.0131) Age between 35 and 49 years old 0.01650.0153 0.0242**0.0207*0.0267**0.0245**0.0314***0.0293** (0.0113)(0.0112)(0.0117)(0.0115)(0.0117)(0.0115)(0.0117)(0.0115) White -0.0044-0.0118 0.0040-0.0075 -0.0013-0.0084 -0.0115-0.0171 (0.0183)(0.0170)(0.0184)(0.0168)(0.0182)(0.0168)(0.0181)(0.0168) Black 0.0725***0.0728***0.0748***0.0761***0.0828***0.0832***0.0808***0.0812*** (0.0200)(0.0199)(0.0199)(0.0198)(0.0199)(0.0198)(0.0198)(0.0198) Hispanic -0.0140-0.0103 0.00530.0093 0.03320.0353*0.0339*0.0357* (0.0207)(0.0204)(0.0205)(0.0203)(0.0206)(0.0205)(0.0206)(0.0204) Living in a MSA with >1 million persons -0.0223*-0.0273**-0.0322***-0.0384***-0.0444***-0.0478***-0.0469***-0.0496*** (0.0117)(0.0108)(0.0114)(0.0107)(0.0113)(0.0108)(0.0113)(0.0108) Living in a MSA with <1 million persons -0.0086-0.0108 -0.0131-0.0158 -0.0214**-0.0228**-0.0218**-0.0230** (0.0107)(0.0105)(0.0106)(0.0104)(0.0105)(0.0104)(0.0105)(0.0104) Number of children aged <18 in household -0.0587***-0.0558***-0.0539***-0.0523***-0.0543***-0.0529*** (0.0044)(0.0040)(0.0043)(0.0040)(0.0043)(0.0040) Married -0.0491***-0.0405***-0.0505***-0.0453***-0.0537***-0.0492*** (0.0125)(0.0113)(0.0124)(0.0113)(0.0125)(0.0113) Widowed -0.0982**-0.0907**-0.0866**-0.0824**-0.0837**-0.0803** (0.0393)(0.0388)(0.0391)(0.0388)(0.0389)(0.0387) Divorced or separated 0.0420***0.0442***0.0476***0.0487***0.0479***0.0487*** (0.0137)(0.0135)(0.0136)(0.0135)(0.0136)(0.0135) Completed undergraduate/graduate stud y 0.1211***0.1194***0.0984***0.0976*** (0.0154)(0.0152)(0.0156)(0.0156) Had some college educatio n 0.0968***0.0970***0.0791***0.0798*** (0.0151)(0.0150)(0.0152)(0.0152) Completed high school 0.0778***0.0783***0.0637***0.0645*** (0.0150)(0.0149)(0.0150)(0.0150) Excellent healt h 0.1385**0.1312** (0.0581)(0.0573) Very good healt h 0.1545***0.1473** (0.0579)(0.0572) Good healt h 0.1143**0.1097* (0.0577)(0.0573) Fair healt h 0.01880.0125 (0.0597)(0.0591) Constant 0.7478***0.7335***0.8406***0.8120***0.7390***0.7222***0.6287***0.6204*** (0.0237)(0.0200)(0.0275)(0.0210)(0.0296)(0.0243)(0.0618)(0.0610) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% 2SLSOLS Model 4 (Direct Effect) Female (n=11,779) Model 1 (Total Effect) Model 2 Model 3 2SLSOLS 2SLSOLS 2SLSOLS Number of days drinking alcohol in the last 12 months

PAGE 163

Appendix B-15: Working Full Time Last Week versus the Number of Days Drinking Alcohol in the La st Month: The Female Sample 149 Variables -0.004720.00053-0.00833*-0.00057-0.00529-0.00080-0.00465-0.00090 (0.0048)(0.0007)(0.0049)(0.0007)(0.0048)(0.0007)(0.0047)(0.0007) Age between 30 and 34 years old -0.0034-0.0031 0.01970.0184 0.02030.0195 0.0218*0.0211 (0.0132)(0.0131)(0.0132)(0.0131)(0.0131)(0.0131)(0.0131)(0.0131) Age between 35 and 49 years old 0.01780.0152 0.0257**0.0208*0.0279**0.0248**0.0323***0.0296*** (0.0115)(0.0112)(0.0119)(0.0115)(0.0119)(0.0115)(0.0120)(0.0115) White -0.0037-0.0118 0.0050-0.0073 -0.0007-0.0080 -0.0110-0.0167 (0.0186)(0.0170)(0.0186)(0.0168)(0.0185)(0.0168)(0.0183)(0.0168) Black 0.0718***0.0729***0.0742***0.0760***0.0827***0.0832***0.0808***0.0812*** (0.0200)(0.0199)(0.0199)(0.0198)(0.0199)(0.0198)(0.0198)(0.0198) Hispanic -0.0134-0.0104 0.00610.0094 0.0343*0.0354*0.0348*0.0358* (0.0207)(0.0204)(0.0205)(0.0203)(0.0205)(0.0205)(0.0205)(0.0204) Living in a MSA with >1 million persons -0.0217*-0.0274**-0.0307***-0.0382***-0.0439***-0.0476***-0.0464***-0.0494*** (0.0120)(0.0108)(0.0117)(0.0107)(0.0115)(0.0108)(0.0114)(0.0108) Living in a MSA with <1 million persons -0.0078-0.0109 -0.0115-0.0157 -0.0206*-0.0227**-0.0211**-0.0228** (0.0109)(0.0105)(0.0107)(0.0104)(0.0106)(0.0104)(0.0106)(0.0104) Number of children aged <18 in household -0.0587***-0.0558***-0.0539***-0.0524***-0.0542***-0.0529*** (0.0044)(0.0040)(0.0043)(0.0040)(0.0043)(0.0040) Married -0.0459***-0.0403***-0.0485***-0.0452***-0.0520***-0.0491*** (0.0119)(0.0113)(0.0118)(0.0113)(0.0118)(0.0112) Widowed -0.0947**-0.0905**-0.0840**-0.0822**-0.0815**-0.0800** (0.0391)(0.0388)(0.0389)(0.0388)(0.0388)(0.0387) Divorced or separated 0.0438***0.0443***0.0489***0.0488***0.0490***0.0489*** (0.0136)(0.0135)(0.0136)(0.0135)(0.0135)(0.0135) Completed undergraduate/graduate study 0.1234***0.1198***0.1003***0.0980*** (0.0157)(0.0152)(0.0159)(0.0156) Had some college education 0.0979***0.0972***0.0801***0.0800*** (0.0151)(0.0150)(0.0152)(0.0152) Completed high school 0.0776***0.0783***0.0636***0.0644*** (0.0150)(0.0149)(0.0150)(0.0150) Excellent health 0.1346**0.1308** (0.0576)(0.0573) Very good health 0.1503***0.1468** (0.0574)(0.0572) Good health 0.1100*0.1091* (0.0574)(0.0573) Fair health 0.01510.0121 (0.0593)(0.0591) Constant 0.7430***0.7341***0.8306***0.8115***0.7315***0.7218***0.6261***0.6203*** (0.0216)(0.0199)(0.0242)(0.0209)(0.0263)(0.0242)(0.0615)(0.0610) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Number of days drinking alcohol in the last month Model 4 (Direct Effect) 2SLSOLS 2SLSOLS 2SLSOLS 2SLSOLS Female (n=11,779) Model 1 (Total Effect) Model 2 Model 3

PAGE 164

Appendix B-16: Working Full Time Last Week versus the Number of Days Drank 5+ Alcoholic Drinks in the Last Month: the Female Sample 150 Variables -0.01379-0.00037-0.02044*-0.00263*-0.01374-0.00175-0.01244-0.00145 (0.0109)(0.0016)(0.0111)(0.0016)(0.0114)(0.0016)(0.0114)(0.0016) Age between 30 and 34 years old -0.0055-0.0032 0.01670.0181 0.01820.0192 0.01980.0208 (0.0133)(0.0131)(0.0132)(0.0131)(0.0132)(0.0131)(0.0131)(0.0131) Age between 35 and 49 years old 0.01280.0154 0.01810.0201*0.0223*0.0240**0.0270**0.0287** (0.0114)(0.0112)(0.0116)(0.0115)(0.0116)(0.0115)(0.0116)(0.0115) White -0.0095-0.0109 -0.0054-0.0079 -0.0074-0.0090 -0.0160-0.0179 (0.0171)(0.0170)(0.0169)(0.0168)(0.0169)(0.0168)(0.0169)(0.0168) Black 0.0705***0.0727***0.0713***0.0755***0.0791***0.0827***0.0777***0.0809*** (0.0201)(0.0199)(0.0201)(0.0198)(0.0202)(0.0198)(0.0201)(0.0198) Hispanic -0.0135-0.0108 0.00630.0092 0.03140.0350*0.03230.0356* (0.0206)(0.0204)(0.0204)(0.0202)(0.0208)(0.0205)(0.0207)(0.0204) Living in a MSA with >1 million persons -0.0270**-0.0268**-0.0400***-0.0389***-0.0482***-0.0483***-0.0499***-0.0501*** (0.0108)(0.0108)(0.0108)(0.0107)(0.0108)(0.0108)(0.0108)(0.0108) Living in a MSA with <1 million persons -0.0115-0.0106 -0.0178*-0.0162 -0.0236**-0.0231**-0.0237**-0.0233** (0.0105)(0.0105)(0.0105)(0.0104)(0.0104)(0.0104)(0.0104)(0.0104) Number of children aged <18 in household -0.0572***-0.0558***-0.0535***-0.0523***-0.0538***-0.0528*** (0.0041)(0.0040)(0.0042)(0.0040)(0.0042)(0.0040) Married -0.0503***-0.0412***-0.0513***-0.0454***-0.0543***-0.0491*** (0.0126)(0.0113)(0.0126)(0.0113)(0.0125)(0.0113) Widowed -0.0890**-0.0900**-0.0818**-0.0818**-0.0797**-0.0797** (0.0390)(0.0388)(0.0388)(0.0388)(0.0387)(0.0387) Divorced or separated 0.0441***0.0443***0.0481***0.0487***0.0483***0.0488*** (0.0136)(0.0135)(0.0136)(0.0135)(0.0135)(0.0135) Completed undergraduate/graduate study 0.1105***0.1181***0.0906***0.0967*** (0.0169)(0.0153)(0.0168)(0.0156) Had some college education 0.0916***0.0963***0.0756***0.0794*** (0.0157)(0.0150)(0.0157)(0.0152) Completed high school 0.0750***0.0780***0.0620***0.0643*** (0.0152)(0.0149)(0.0152)(0.0150) Excellent health 0.1222**0.1290** (0.0579)(0.0573) Very good health 0.1404**0.1453** (0.0575)(0.0572) Good health 0.1035*0.1082* (0.0577)(0.0573) Fair health 0.00880.0111 (0.0593)(0.0591) Constant 0.7473***0.7353***0.8352***0.8133***0.7425***0.7229***0.6442***0.6219*** (0.0222)(0.0200)(0.0250)(0.0209)(0.0306)(0.0243)(0.0653)(0.0610) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Number of days drank 5+ alcohol drinks in the last month 2SLSOLS 2SLSOLS 2SLSOLS Female (n=11,779) Model 1 (Total Effect) Model 2 Model 3 Model 4 (Direct Effect) 2SLSOLS

PAGE 165

Appendix C: Complete Second Stage Regression Results Using NELS:88 151

PAGE 166

Appendix C-1: Current Employment (Full Time/Part Time) versus the Number of Occasions Drank Alcohol in the Last 30 Days: the Male Sample 152 Male (n=3,876) Variables -0.01070.0012*-0.01030.0016**-0.01140.0017*** (0.0072)(0.0006)(0.0073)(0.0006)(0.0071)(0.0006) A g e 0.32930.1473 0.31140.1815 0.26470.1234 (0.6135)(0.5779)(0.6075)(0.5772)(0.6125)(0.5772) A g e squared -0.0061-0.0026 -0.0058-0.0033 -0.0049-0.0022 (0.0115)(0.0108)(0.0114)(0.0108)(0.0115)(0.0108) White 0.0296*0.0138 0.02760.0110 0.0318*0.0144 (0.0178)(0.0144)(0.0181)(0.0145)(0.0179)(0.0145) Black -0.0315-0.0315 -0.0297-0.0294 -0.0281-0.0275 (0.0225)(0.0215)(0.0225)(0.0216)(0.0227)(0.0215) Native American or Alaska Native 0.00980.0042 0.0068-0.0102 0.0076-0.0115 (0.0477)(0.0455)(0.0488)(0.0457)(0.0491)(0.0456) Asian or Pacific Islander -0.0911***-0.0917***-0.0886***-0.0854***-0.0838***-0.0785*** (0.0221)(0.0211)(0.0222)(0.0212)(0.0227)(0.0213) Household number of children/stepchildren -0.00500.0014 -0.0072-0.0009 (0.0076)(0.0062)(0.0075)(0.0064) Married 0.02120.0409***0.01760.0391*** (0.0163)(0.0106)(0.0161)(0.0106) Cohabitatin g -0.0682-0.0623 -0.0662-0.0601 (0.0505)(0.0483)(0.0509)(0.0483) Widowed 0.00000.0000 0.00000.0000 (0.0000)(0.0000)(0.0000)(0.0000) Separated -0.0538-0.0519 -0.0545-0.0539 (0.0540)(0.0517)(0.0546)(0.0518) Divorced 0.03010.0282 0.02750.0244 (0.0240)(0.0230)(0.0245)(0.0232) Ph.D. or Professional Doctorate 0.02770.0221 (0.0903)(0.0857) Master de g ree 0.07460.0660 (0.0615)(0.0583) Bachelor de g ree 0.01480.0022 (0.0295)(0.0272) Had some colle g e education 0.0554*0.0539* (0.0294)(0.0279) Completed hi g h school 0.03430.0259 (0.0279)(0.0261) Constant -3.4437-1.1525 -3.2044-1.5869 -2.6007-0.8280 (8.1612)(7.7019)(8.0866)(7.6918)(8.1526)(7.6894) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Number of occasions drank alcohol in the last 30 days Model 2 Model 3 Model 1 (Total Effect) 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 167

Appendix C-2: Current Employment (Full Time/Part Time) versus the Number of Occasions Drank 5+ Alcoholic Drinks in a Row over the Last Two Weeks: the Male Sample 153 Male (n=3,876) Variables -0.03200.0055**-0.03090.0070***-0.0363*0.0069*** (0.0208)(0.0025)(0.0211)(0.0025)(0.0218)(0.0025) A g e 0.28250.1461 0.28340.1800 0.24860.1213 (0.5990)(0.5779)(0.5964)(0.5771)(0.6019)(0.5772) A g e squared -0.0052-0.0026 -0.0052-0.0033 -0.0046-0.0022 (0.0112)(0.0108)(0.0112)(0.0108)(0.0113)(0.0108) White 0.01620.0153 0.01490.0129 0.01960.0161 (0.0148)(0.0144)(0.0149)(0.0144)(0.0151)(0.0145) Black -0.0415*-0.0298 -0.0396*-0.0271 -0.0395*-0.0253 (0.0231)(0.0216)(0.0232)(0.0216)(0.0235)(0.0216) Native American or Alaska Native 0.00930.0040 0.0028-0.0103 0.0034-0.0114 (0.0469)(0.0455)(0.0475)(0.0457)(0.0479)(0.0456) Asian or Pacific Islander -0.1012***-0.0900***-0.0974***-0.0832***-0.0912***-0.0769*** (0.0226)(0.0212)(0.0232)(0.0212)(0.0233)(0.0214) Household number of children/stepchildren -0.00140.0010 -0.0050-0.0011 (0.0066)(0.0062)(0.0069)(0.0064) Married 0.02300.0417***0.01820.0397*** (0.0150)(0.0106)(0.0154)(0.0106) Cohabitatin g -0.0680-0.0620 -0.0674-0.0597 (0.0498)(0.0483)(0.0502)(0.0483) Widowed 0.00000.0000 0.00000.0000 (0.0000)(0.0000)(0.0000)(0.0000) Separated -0.0652-0.0492 -0.0723-0.0505 (0.0540)(0.0517)(0.0549)(0.0518) Divorced 0.02950.0282 0.02320.0252 (0.0237)(0.0230)(0.0241)(0.0232) Ph.D. or Professional Doctorate 0.03680.0202 (0.0892)(0.0857) Master de g ree 0.05770.0689 (0.0607)(0.0583) Bachelor de g ree -0.00260.0050 (0.0285)(0.0272) Had some colle g e education 0.0489*0.0551** (0.0291)(0.0279) Completed hi g h school 0.02930.0265 (0.0271)(0.0261) Constant -2.8841-1.1324 -2.8844-1.5623 -2.4186-0.7986 (7.9778)(7.7008)(7.9442)(7.6905)(8.0144)(7.6892) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% 2SLSOLS Number of occasions drank 5+ alcoholic drinks in a row over the last two weeks Model 1 (Total Effect) Model 2 Model 3 2SLSOLS 2SLSOLS

PAGE 168

Appendix C-3: Currently Working Full Time versus the Number of Occasions Drank Alcohol in th e Last 30 Days: the Male Sample 154 Male (n=3,876) Variables -0.00440.0012 -0.00340.0023***-0.00350.0023*** (0.0092)(0.0008)(0.0092)(0.0008)(0.0090)(0.0008) A g e -0.3876-0.4730 -0.2489-0.3110 -0.2884-0.3514 (0.7840)(0.7670)(0.7709)(0.7599)(0.7719)(0.7612) A g e squared 0.00750.0091 0.00470.0059 0.00550.0067 (0.0147)(0.0144)(0.0145)(0.0143)(0.0145)(0.0143) White 0.0434*0.0360*0.0405*0.0326*0.0407*0.0330* (0.0228)(0.0191)(0.0230)(0.0190)(0.0226)(0.0191) Black -0.0187-0.0187 -0.0111-0.0109 -0.0100-0.0098 (0.0288)(0.0286)(0.0285)(0.0284)(0.0286)(0.0284) Native American or Alaska Native 0.04570.0431 0.01110.0029 0.01130.0028 (0.0609)(0.0604)(0.0619)(0.0601)(0.0619)(0.0602) Asian or Pacific Islander -0.0900***-0.0903***-0.0706**-0.0691**-0.0708**-0.0685** (0.0282)(0.0281)(0.0282)(0.0279)(0.0286)(0.0282) Household number of children/stepchildren 0.01250.0155*0.01300.0159* (0.0096)(0.0082)(0.0095)(0.0084) Married 0.0897***0.0992***0.0889***0.0985*** (0.0207)(0.0139)(0.0203)(0.0140) Cohabitatin g 0.03160.0343 0.03390.0366 (0.0641)(0.0636)(0.0642)(0.0637) Widowed 0.00000.0000 0.00000.0000 (0.0000)(0.0000)(0.0000)(0.0000) Separated -0.0717-0.0708 -0.0683-0.0680 (0.0685)(0.0681)(0.0688)(0.0684) Divorced 0.0866***0.0857***0.0888***0.0874*** (0.0305)(0.0303)(0.0309)(0.0306) Ph.D. or Professional Doctorate 0.10680.1043 (0.1138)(0.1130) Master de g ree 0.04080.0369 (0.0776)(0.0769) Bachelor de g ree 0.01880.0132 (0.0371)(0.0359) Had some colle g e education 0.03420.0335 (0.0371)(0.0369) Completed hi g h school 0.01790.0141 (0.0352)(0.0345) Constant 5.88256.9569 4.10134.8738 4.59365.3844 -10.430-10.221 -10.262-10.127 -10.275-10.142 Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Number of occasions drank alcohol in the last 30 days Model 1 (Total Effect) Model 2 Model 3 2SLSOLS 2SLSOLS 2SLSOLS

PAGE 169

Appendix C-4: Currently Working Full Ti me versus the Number of Occasions Drank 5+ Alcoholic Drinks in a Row over the Last Two Weeks: the Male Sample 155 Male (n=3,876) Variables -0.01360.0091***-0.01070.0134***-0.01180.0135*** (0.0270)(0.0033)(0.0272)(0.0033)(0.0279)(0.0033) A g e -0.4052-0.4879 -0.2569-0.3225 -0.2918-0.3662 (0.7772)(0.7665)(0.7677)(0.7591)(0.7703)(0.7604) A g e squared 0.00780.0094 0.00490.0061 0.00550.0069 (0.0146)(0.0144)(0.0144)(0.0142)(0.0145)(0.0143) White 0.0379**0.0373*0.0364*0.0351*0.0370*0.0349* (0.0192)(0.0191)(0.0192)(0.0190)(0.0193)(0.0191) Black -0.0230-0.0159 -0.0145-0.0066 -0.0137-0.0054 (0.0300)(0.0286)(0.0299)(0.0284)(0.0300)(0.0284) Native American or Alaska Native 0.04560.0424 0.00990.0015 0.01020.0015 (0.0609)(0.0604)(0.0612)(0.0601)(0.0613)(0.0601) Asian or Pacific Islander -0.0943***-0.0875***-0.0737**-0.0646**-0.0733**-0.0650** (0.0294)(0.0281)(0.0299)(0.0279)(0.0298)(0.0281) Household number of children/stepchildren 0.01360.0152*0.01370.0160* (0.0084)(0.0082)(0.0088)(0.0084) Married 0.0901***0.1020***0.0889***0.1015*** (0.0194)(0.0140)(0.0197)(0.0140) Cohabitatin g 0.03150.0353 0.03340.0379 (0.0641)(0.0635)(0.0643)(0.0636) Widowed 0.00000.0000 0.00000.0000 (0.0000)(0.0000)(0.0000)(0.0000) Separated -0.0757-0.0655 -0.0740-0.0613 (0.0695)(0.0681)(0.0702)(0.0683) Divorced 0.0864***0.0856***0.0874***0.0886*** (0.0305)(0.0302)(0.0308)(0.0306) Ph.D. or Professional Doctorate 0.10980.1001 (0.1142)(0.1129) Master de g ree 0.03540.0420 (0.0777)(0.0768) Bachelor de g ree 0.01330.0178 (0.0365)(0.0359) Had some colle g e education 0.03210.0357 (0.0373)(0.0368) Completed hi g h school 0.01640.0147 (0.0347)(0.0344) Constant 6.09157.1532 4.19065.0295 4.63015.5780 (10.352)(10.214)(10.227)(10.115)(10.258)(10.130) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% 2SLSOLS 2SLSOLS Number of occasions drank 5+ alcoholic drinks in a row over the last two weeks Model 2 Model 3 2SLSOLS Model 1 (Total Effect)

PAGE 170

Appendix C-5: Current Employment (Full Time/Part Time) versus the Number of Occasions Drank Alcohol in the Last 30 Days: the Female Sample 156 Female (n=4,638) Variables -0.00770.0065***-0.00130.0013 0.00080.0013 (0.0117)(0.0013)(0.0115)(0.0013)(0.0111)(0.0013) Age 1.8835**1.9751**1.7539**1.7721**1.4486*1.4518* (0.8563)(0.8416)(0.8226)(0.8183)(0.8232)(0.8199) Age squared -0.0366**-0.0381**-0.0335**-0.0338**-0.0276*-0.0277* (0.0161)(0.0158)(0.0155)(0.0154)(0.0155)(0.0154) White 0.0467**0.0285 0.01020.0072 0.00380.0033 (0.0232)(0.0175)(0.0215)(0.0171)(0.0204)(0.0172) Black 0.03290.0361 0.02660.0285 0.02650.0269 (0.0250)(0.0246)(0.0256)(0.0241)(0.0259)(0.0241) Native American or Alaska Native -0.0155-0.0173 0.00980.0093 0.00650.0064 (0.0543)(0.0535)(0.0521)(0.0521)(0.0520)(0.0520) Asian or Pacific Islander -0.0118-0.0248 -0.0775***-0.0780***-0.0844***-0.0844*** (0.0284)(0.0259)(0.0255)(0.0254)(0.0258)(0.0257) Household number of children/stepchildren -0.0897***-0.0880***-0.0817***-0.0815*** (0.0094)(0.0059)(0.0083)(0.0063) Married -0.0416-0.0365***-0.0399-0.0390*** (0.0257)(0.0121)(0.0245)(0.0121) Cohabitating 0.00020.0034 0.00220.0027 (0.0624)(0.0608)(0.0620)(0.0607) Widowed 0.33970.3470 0.34180.3432 (0.3629)(0.3613)(0.3619)(0.3606) Separated 0.01940.0174 0.02270.0222 (0.0465)(0.0456)(0.0467)(0.0456) Divorced 0.02540.0277 0.03140.0318 (0.0282)(0.0263)(0.0274)(0.0264) Ph.D. or Professional Doctorate 0.2187**0.2188** (0.1064)(0.1064) Master degree 0.2108***0.2111*** (0.0658)(0.0654) Bachelor degree 0.1456***0.1453*** (0.0370)(0.0363) Had some college education 0.1614***0.1616*** (0.0366)(0.0362) Completed high school 0.1294***0.1296*** (0.0350)(0.0348) Constant -23.400**-24.800**-22.000**-22.300**-18.200*-18.300* (11.389)(11.182)(10.937)(10.872)(10.940)(10.889) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Model 3 Model 2 OLS 2SLSOLS Model 1 (Total Effect) Number of occasions drank alcohol in the last 30 days 2SLSOLS 2SLS

PAGE 171

Appendix C-6: Current Employment (Full Time/Part Time) versus the Number of Occasions Drank 5+ Alcoholic Drinks in a Row over the Last Two Weeks: the Female Sample 157 Female (n=4,638) Variables -0.02710.0219***0.00650.0059 0.01580.0065 (0.0541)(0.0063)(0.0530)(0.0062)(0.0533)(0.0062) A g e 1.9030**1.9575**1.7699**1.7693**1.4564*1.4485* (0.8505)(0.8429)(0.8201)(0.8183)(0.8213)(0.8199) A g e squared -0.0369**-0.0378**-0.0338**-0.0338**-0.0278*-0.0276* (0.0160)(0.0159)(0.0154)(0.0154)(0.0155)(0.0154) White 0.0390**0.0350**0.00820.0083 0.00340.0042 (0.0181)(0.0174)(0.0175)(0.0170)(0.0176)(0.0171) Black 0.03160.0370 0.02870.0286 0.02850.0269 (0.0255)(0.0246)(0.0259)(0.0241)(0.0259)(0.0241) Native American or Alaska Native -0.0215-0.0124 0.01070.0106 0.00950.0078 (0.0549)(0.0536)(0.0530)(0.0521)(0.0529)(0.0520) Asian or Pacific Islander -0.0215-0.0167 -0.0766***-0.0767***-0.0817***-0.0834*** (0.0266)(0.0260)(0.0273)(0.0255)(0.0274)(0.0258) Household number of children/stepchildren -0.0884***-0.0884***-0.0810***-0.0817*** (0.0068)(0.0058)(0.0075)(0.0063) Married -0.0371*-0.0372***-0.0369*-0.0396*** (0.0197)(0.0120)(0.0198)(0.0120) Cohabitatin g 0.00230.0023 0.00240.0017 (0.0609)(0.0608)(0.0608)(0.0607) Widowed 0.34570.3455 0.34500.3418 (0.3617)(0.3613)(0.3612)(0.3606) Separated 0.01890.0188 0.02480.0240 (0.0458)(0.0456)(0.0458)(0.0456) Divorced 0.02700.0270 0.03210.0314 (0.0265)(0.0263)(0.0267)(0.0264) Ph.D. or Professional Doctorate 0.2173**0.2180** (0.1065)(0.1064) Master de g ree 0.2123***0.2111*** (0.0658)(0.0654) Bachelor de g ree 0.1479***0.1468*** (0.0368)(0.0363) Had some colle g e education 0.1627***0.1617*** (0.0367)(0.0362) Completed hi g h school 0.1301***0.1295*** (0.0349)(0.0348) Constant -23.700**-24.500**-22.300**-22.300**-18.300*-18.200* (11.305)(11.199)(10.899)(10.871)(10.911)(10.889) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Model 2 Model 3 2SLSOLS Number of occasions drank 5+ alcoholic drinks in a row over the last two weeks OLS 2SLSOLS 2SLS Model 1 (Total Effect)

PAGE 172

Appendix C-7: Currently Working Full Time versus the Number of Occasions Drank Alcohol in the Last 30 Days: the Female Sample 158 Female (n=4,638) Variables -0.02010.0076***-0.01400.0014 -0.01140.0012 (0.0145)(0.0015)(0.0140)(0.0016)(0.0135)(0.0016) A g e 2.2789**2.4579**2.1058**2.2135**1.8616*1.9454* (1.0587)(1.0190)(1.0060)(0.9908)(1.0043)(0.9933) A g e squared -0.0441**-0.0472**-0.0401**-0.0420**-0.0353*-0.0369** (0.0199)(0.0192)(0.0189)(0.0186)(0.0189)(0.0187) White 0.0634**0.0278 0.01870.0012 0.0074-0.0052 (0.0287)(0.0211)(0.0263)(0.0207)(0.0249)(0.0208) Black 0.0568*0.0632**0.04380.0551*0.03850.0492* (0.0310)(0.0297)(0.0312)(0.0292)(0.0316)(0.0293) Native American or Alaska Native -0.0451-0.0486 -0.0123-0.0151 -0.0147-0.0170 (0.0671)(0.0648)(0.0638)(0.0631)(0.0635)(0.0630) Asian or Pacific Islander -0.0106-0.0361 -0.0967***-0.0997***-0.1142***-0.1131*** (0.0351)(0.0314)(0.0312)(0.0308)(0.0314)(0.0312) Household number of children/stepchildren -0.1174***-0.1075***-0.1042***-0.0980*** (0.0116)(0.0071)(0.0101)(0.0077) Married -0.0694**-0.0389***-0.0655**-0.0411*** (0.0314)(0.0146)(0.0299)(0.0147) Cohabitatin g -0.0308-0.0121 -0.0271-0.0125 (0.0763)(0.0736)(0.0757)(0.0735) Widowed 0.47120.5145 0.48190.5166 (0.4438)(0.4374)(0.4415)(0.4369) Separated 0.05800.0460 0.06490.0533 (0.0569)(0.0552)(0.0570)(0.0552) Divorced 0.03370.0476 0.04690.0555* (0.0345)(0.0318)(0.0334)(0.0320) Ph.D. or Professional Doctorate 0.3074**0.3107** (0.1298)(0.1289) Master de g ree 0.10160.1096 (0.0802)(0.0792) Bachelor de g ree 0.1377***0.1297*** (0.0451)(0.0440) Had some colle g e education 0.1289***0.1345*** (0.0446)(0.0439) Completed hi g h school 0.0829*0.0878** (0.0427)(0.0421) Constant -28.700**-31.300**-26.800**-28.300**-23.800*-25.000* (14.080)(13.538)(13.377)(13.164)(13.346)(13.192) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Number of occasions drank alcohol in the last 30 days 2SLSOLS 2SLSOLS 2SLSOLS Model 1 (Total Effect) Model 2 Model 3

PAGE 173

Appendix C-8: Currently Working Full Time versus the Number of Occasions Drank 5+ Alcoholic Drinks in a Row over the Last Two Weeks: the Female Sample 159 Female (n=4,638) Variables -0.09900.0338***-0.06230.0153**-0.05250.0160** (0.0671)(0.0076)(0.0648)(0.0075)(0.0651)(0.0075) A g e 2.2986**2.4464**2.1392**2.2197**1.8927*1.9512** (1.0553)(1.0195)(1.0040)(0.9904)(1.0033)(0.9929) A g e squared -0.0444**-0.0470**-0.0407**-0.0421**-0.0359*-0.0370** (0.0198)(0.0192)(0.0189)(0.0186)(0.0189)(0.0187) White 0.0455**0.0348*0.00730.0017 0.0001-0.0053 (0.0224)(0.0211)(0.0214)(0.0206)(0.0215)(0.0207) Black 0.05040.0652**0.04290.0568*0.03910.0510* (0.0316)(0.0298)(0.0317)(0.0292)(0.0316)(0.0292) Native American or Alaska Native -0.0659-0.0413 -0.0262-0.0121 -0.0263-0.0139 (0.0681)(0.0648)(0.0648)(0.0631)(0.0646)(0.0630) Asian or Pacific Islander -0.0388-0.0258 -0.1109***-0.0966***-0.1225***-0.1103*** (0.0331)(0.0314)(0.0334)(0.0308)(0.0335)(0.0312) Household number of children/stepchildren -0.1125***-0.1073***-0.1026***-0.0974*** (0.0083)(0.0071)(0.0091)(0.0076) Married -0.0603**-0.0371**-0.0590**-0.0386*** (0.0242)(0.0145)(0.0241)(0.0145) Cohabitatin g -0.0184-0.0126 -0.0178-0.0127 (0.0746)(0.0736)(0.0743)(0.0735) Widowed 0.48910.5159 0.49500.5189 (0.4428)(0.4373)(0.4411)(0.4367) Separated 0.04210.0483 0.04990.0558 (0.0561)(0.0552)(0.0559)(0.0552) Divorced 0.04210.0474 0.05060.0559* (0.0325)(0.0318)(0.0326)(0.0319) Ph.D. or Professional Doctorate 0.3141**0.3093** (0.1300)(0.1288) Master de g ree 0.10200.1109 (0.0803)(0.0792) Bachelor de g ree 0.1246***0.1322*** (0.0449)(0.0439) Had some colle g e education 0.1284***0.1357*** (0.0448)(0.0439) Completed hi g h school 0.0840**0.0884** (0.0427)(0.0421) Constant -29.000**-31.100**-27.300**-28.400**-24.200*-25.100* (14.027)(13.545)(13.343)(13.159)(13.328)(13.185) Standard errors are in parentheses. ***Statistically significant at 1% **Statistically significant at 5% *Statistically significant at 10% Number of occasions drank 5+ alcoholic drinks in a row over the last two weeks Model 2 Model 3 2SLSOLS 2SLSOLS 2SLSOLS Model 1 (Total Effect)

PAGE 174

About the Author Chanvuth Sangchai received a Bachelors Degree in Economics from Thammasat University, Thailand in 1995 and a Master of Arts in Economics from Cleveland State University, Ohio in 1998. He entered the P h.D. program at the University of South Florida in 2001. While in the Ph.D. program at the University of South Florida, Chanvuth Sangchai taught several classes of Principl es of Microeconomics, Principles of Macroeconomics, and Basic Econometr ics to undergraduate students.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001914392
003 fts
005 20071023143308.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 071023s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001754
040
FHM
c FHM
035
(OCoLC)176152802
049
FHMM
090
HF71 (ONLINE)
1 100
Sangchai, Chanvuth.
4 245
The causal effect of alcohol consumption on employment status
h [electronic resource] /
by Chanvuth Sangchai.
260
[Tampa, Fla] :
b University of South Florida,
2006.
3 520
ABSTRACT: Alcohol consumption may affect labor market outcomes directly through a reduction in productivity and indirectly through human capital accumulation. However, empirical results from previous studies in the economics literature are mixed and inconclusive. While some researchers found negative effects of alcohol use on labor market outcomes, quite a few studies found either positive or insignificant effects. The purpose of this dissertation is to estimate causal effects of alcohol consumption on employment status. It uses three data sets previously unexploited for this purpose and attempts to eliminate any potential estimation problems from previous studies. The results show that previous problematic heavy drinking, i.e. clinically-defined alcohol abuse and/or dependence, has no significant direct effects, but has significant indirect effects on current employment propensity for both genders through human capital components, specifically educational attainment and health status. While general alcohol consumption has only an indirect effect on employment status for females, it has both direct and indirect effects on employment status for males, though the direct effect is very small.
502
Dissertation (Ph.D.)--University of South Florida, 2006.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 159 pages.
Includes vita.
590
Adviser: Donald Bellante, Ph.D.
653
Labor market.
Human capital.
Alcoholism.
Drinking.
Instrumental variable.
690
Dissertations, Academic
z USF
x Business Administration
Doctoral.
773
t USF Electronic Theses and Dissertations.
0 856
u http://digital.lib.usf.edu/?e14.1754