USF Libraries
USF Digital Collections

Friendship networks, perceived reciprocity of support, and depression

MISSING IMAGE

Material Information

Title:
Friendship networks, perceived reciprocity of support, and depression
Physical Description:
Book
Language:
English
Creator:
Huff, Ryan Francis
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Equity Theory
Friendship
Mental Health
Social Network Analysis
Social Support
Dissertations, Academic -- Sociology Social Psychology Mental Health -- Masters -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Using social network analysis as a theoretical framework, the current study examined the associations between self-reported egocentric network characteristics and depression among a sample of United States college students. It is important to understand factors related to depression among this population due to the severity of its potential outcomes (e.g., suicide and interpersonal problems at school). Drawing inspiration from a recent study conducted by Christina Falci and Clea McNeely (2009), the current investigation used OLS regression to test for both linear and curvilinear relationships between egocentric network size and depression. Potential interactions between network size, density, and gender were also explored. As an additional line of inquiry, this project examined whether or not (and to what extent) perceptions of reciprocity mediate the relationships between network characteristics and depression. Data were collected using an online survey, which was proctored to students enrolled in three large undergraduate sociology courses during the fall 2010 semester. In contrast to findings reported by Falci and McNeely (2009), no significant relationships were observed between network characteristics and mental health. However, support reciprocity was found to be a significant predictor of depression at the multivariate level. Additional research will be necessary in order to confirm (or refute) the results of Falci and McNeely (2009) and to further assess the mediating effects of perceived equity.
Thesis:
Thesis (M.A.)--University of South Florida, 2011.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Ryan Francis Huff.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 118 pages.

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:
usfldc doi - E14-SFE0004906
usfldc handle - e14.4906
System ID:
SFS0028158:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Friendship Networks, Perceived Reciprocity of Support, and Depression by Ryan Huff A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Sociology College of Arts and Sciences University of South Florida Major Professor: John Skvoretz, Ph.D. Elizabeth Vaquera, Ph.D. James Cavendish, Ph.D. Date of Approval: April 8, 2011 Keywords: Social Network Analysis, Mental Health, Friendship, Equity Theory, Social Support Copyright 2011, Ryan Huff

PAGE 2

i TABLE OF CONTENTS LIST OF TABLES ............................................................................................................. iii LIST OF FIGURES ........................................................................................................... iv ABSTRACT .........................................................................................................................v CHAPTER 1: INTRODUCTION ........................................................................................1 CHAPTER 2: LITERATURE REVIEW .............................................................................6 Social Network Analysis as a Theoretical Framework ............................................6 So cial Integration and Depression ...........................................................................7 Network Cohesion and Depression ........................................................................11 Perceived Reciprocity of Support ..........................................................................16 CHAPTER 3: SPECIFIC AIMS ........................................................................................21 CHAPTER 4: METHODS .................................................................................................23 Data Collection and Sample...................................................................................23 Measures ................................................................................................................24 Depressive Symptoms ................................................................................24 Network Structure ......................................................................................25 Social Support ............................................................................................28 Reciprocity of Support ...............................................................................28 Demographic Controls and Contextual Questions .....................................29 Analytic Strategy ...................................................................................................31 CHAPTER 5: RESULTS ...................................................................................................33 Descriptive Statistics ..............................................................................................33 Bivariate Correlations ............................................................................................35 Multivariate Models ...............................................................................................37 CHAPTER 6: DISCUSSION .............................................................................................52 Findings Network Structure ................................................................................52 Findings The Importance of Reciprocity ............................................................59 Results When Including Family Members and Partners as Alters ........................62 Conclusion .............................................................................................................64

PAGE 3

ii LIST OF REFERENCES...................................................................................................67 APPENDICES ...................................................................................................................79 Appendix A: Survey Questions .............................................................................80 Appendix B: Supplementary Descriptive Statistics ...............................................91 Appendix C: Supplementary Inferential Statistics ...............................................103

PAGE 4

iii LIST OF TABLES TABLE 1: Descriptive Statistics for College Students at a Large Florida University .........................................................................................................42 TABLE 2: Bivariate Correlations ......................................................................................45 TABLE 3: OLS Regression Models with Depression as the Dependent Variable ............46 TABLE A1: Descriptive Statistics for Individual CES-D Items .......................................91 TABLE A2: Descriptive Statistics for Individual MSPSS Items ......................................95 TABLE A3: Descriptive Statistics for GSS Survey Items.................................................97 TABLE A4: Additional Personal Network Statistics ......................................................100 TABLE A5: Number of Friends (Excluding Family Members and Partners) by Total Number of Alters Reported ...............................................................101 TABLE A6: Network Statistics Including Family Members and Partners as Alters ...........................................................................................................102 TABLE A7: Bivariate Correlations Including Family Members and Partners as Alters ...........................................................................................................103 TABLE A8: Additional OLS Regression Models with Depression as the Dependent Variable ....................................................................................104

PAGE 5

iv LIST OF FIGURES FIGURE 1: Egocentric Networks with Varying Levels of Density ..................................20

PAGE 6

v ABSTRACT Using social network analysis as a theoretical framework, the current study examined the associations between self-reported egocentric network characteristics and depression among a sample of United States college students. It is important to understand factors related to depression among this population due to the severity of its potential outcomes (e.g., suicide and interpersonal problems at school). Drawing inspiration from a recent study conducted by Christina Falci and Clea McNeely (2009), the current investigation used OLS regression to test for both linear and curvilinear relationships between egocentric network size and depression. Potential interactions between network size, density, and gender were also explored. As an additional line of inquiry, this project examined whether or not (and to what extent) perceptions of reciprocity mediate the relationships between network characteristics and depression. Data were collected using an online survey, which was proctored to students enrolled in three large undergraduate sociology courses during the fall 2010 semester. In contrast to findings reported by Falci and McNeely (2009), no significant relationships were observed between network characteristics and mental health. However, support reciprocity was found to be a significant predictor of depression at the multivariate level. Additional research will be necessary in order to confirm (or refute) the results of Falci and McNeely (2009) and to further assess the mediating effects of perceived equity.

PAGE 7

1 CHAPTER 1: INTRODUCTION Approximately 15% of those completing the spring 2008 American College Health Association-National College Health Assessment (ACHA-NCHA), which is a national survey of over 80,000 students located on 106 college campuses, indicated that they had been diagnosed with depression at some point during their respective lifetimes (American College Health Association 2009). Of these individuals, 32% reported having been diagnosed within the past school year, 25% reported currently being in therapy for depression, and 36% reported taking depressionre lated medications (American College Health Association 2009). These findings raise explicit concerns related to the health and well-being of students attending institutions of higher education. At the extreme, depression Association 2000:349), is associated with suicide (Hockenbury and Hockenbury 2003). Far from being limited to the adult population, research has clearly demonstrated that there is a strong relationship between depression and suicidal behavior among adolescents (Spirito et al. 2003; Sadowski and Kelley 1993). Although this outcome may seem to be a marginal possibility, the National Center for Injury Prevention and Control (2006) lists suicide as the third leading cause of death in the United States among those

PAGE 8

2 age 15-24 and the second leading cause of death among those age 25-34. Notably, these age groups account for over 92% of those enrolled at scholarly institutions in America (U.S. Census Bureau 2008). In addition to this extreme outcome, depression has been connected to decreased academic productivity, interpersonal problems at school, and truancy among college students (Heiligenstein and Guenther 1996). Moreover, depression is associated with difficulty concentrating, reduced energy, changes in weight, and changes in the quality or quantity of sleep among those in the general population (Lackey 2008; Hockenbury and Hockenbury 2003). Due of the severity of these potential outcomes, it is important to understand factors which are related to depression among the national collegiate student body. Using social network analysis as a theoretical framework, the current study will examine the associations between self-reported egocentric network characteristics and depression among a sample of United States college students. For the purposes of this investigation, specific focus will be placed on egocentric network size and density, and on the perceived reciprocity of social support exchanges that occur within personal friendship networks. 1 Although not limited to this line of inquiry, social network analysis has been used to study a broad spectrum of mental health outcomes including depression, anxiety, and negative affect (Fiori, Antonucci, and Cortina 2006; Lin, Ye, and Ensel 1999; Lin and Peek 1999). However, several key subject areas within this field remain largely unexplored and/or require further investigation. To elaborate, while numerous benefits (e.g., reduced levels of depression, unhappiness, and suicidal ideation) have been attributed to large egocentric networks (Ueno 2005; Moody 2004; Cannuscio et al. 2004 ; 1 Definitions of network terms are presented in CHAPTER 2.

PAGE 9

3 Field, Diego, and Sanders 2001; Burt 1987; Cohen and Wills 1985; Coates 1985; Fischer and Phillips 1982), relatively few studies have entertained the theoretical notion that over-integration may actually result in greater mental health problems (Pescosolido and Levy 2002) and a sense of obligation that negatively affects the individual (Durkheim 1897/2006). 2 Additionally, while some scholars have reported observing positive associations between network density and mental health (Ueno 2005; Kadushin 1983; Fischer 1982), findings related to this subject have been both inconsistent and inconclusive (Lin and Peek 1999). Providing much of the basis for the current investigation, a recent study conducted by Christina Falci and Clea McNeely (2009) attempted to address several of the gaps and inconsistencies present in the literature. More specifically, using secondary data taken from the National Longitudinal Study of Adolescent Health (Add Health), Falci and McNeely (2009) examined the associations between various egocentric network characteristics and depression among a nationally representative sample of American adolescents. 3 Of particular relevance to this discussion, the authors found that individuals with small or large personal friendship networks both reported experiencing higher levels of depression than those with average-sized networks (Falci and McNeely 2009). Additionally, female adolescents with dense egocentric networks reported lower levels of depressive symptomology than those with fragmented networks ; no significant 2 Although relatively few studies have entertained the possibility that over-integration may actually have a negative impact on mental health, a recent investigation conducted by Kathy Charles is a notable exception. To elaborate, Charles surveyed 200 stude th [. .] friends and 3 The Add Health project consists of survey and interview data collected from a nationally representative sample of American adolescents in grades 7-12 (Add Health 2010).

PAGE 10

4 relationship was found among males of the same age (Falci and McNeely 2009). Finally, it should be noted that females with large, cohesive networks reported experiencing lower levels of depressive symptoms than those with large, fragmented networks; the opposite pattern was found among males (Falci and McNeely 2009). These results are quite unique, as most scholars have focused on delineating independent associations between network variables and depression, rather than searching for interaction effects. 4 In light of the numerous precedents set by Falci and McNeely (2009), the current project will further investigate the relationships between egocentric network size, network density, gender, and depression. As an additional line of inquiry, this study will also examine the relationship between perceived reciprocity of support and mental health. 5 To elaborate, although previous studies have examined egocentric network size and density in relation to depression, and it has been suggested that the observed effects of these characteristics are at least partially related to exchanges of social support, such exchanges have not been investigated directly by network researchers. More specifically, those relatively few network studies which have considered the relationship between social support and depression have focused exclusively on that support which is received by participants, and ignored that support which is given. Using equity theory and social exchange theory as competing frameworks, the current study will expand upon the existing literature by examining the extent to which perceptions of reciprocity mediate 4 A more detailed account of these findings will be presented in the next chapter. 5 Scholars have claimed that those with small egocentric networks may suffer from inadequate levels of and Bjarnason 1998:96). It has also been suggested that the effort which must be exerted to maintain a large network may come to outweigh any benefits or support received from it (Haines et al. 2008; Kessler and McLeod 1984). Moreover, highly cohesive networks are thought to minimize the effort required to maintain individual relationships and to result in the sharing of social burdens (Forrester and Tashchian 2004).

PAGE 11

5 the relationships which have been observed between egocentric network characteristics (i.e., egocentric network size and density) and depression.

PAGE 12

6 CHAPTER 2: LITERATURE REVIEW Social Network Analysis as a Theoretical Framework Although not limited to this line of inquiry, social network analysis has been used to study a broad spectrum of mental health outcomes including depression, anxiety, and negative affect (Fiori, Antonucci, and Cortina 2006; Lin, Ye, and Ensel 1999; Lin and Peek 1999). For the purposes of this discussion, social networks may be described as finite sets of actors who are connected by specific relationships, and social network analysis can be thought of as the study of such networks (Wasserman and Faust 1994). entities, and on the patterns and Faust 1994:3). Rather than treating actors and their actions as independent, autonomous units; researchers guided by this perspective view individuals as interdependent, or reliant upon one another for opportunities and resources (Wasserman and Faust 1994). As a randomly with respect to one another. They form attachments to certain persons, they n 1988:31). Furthermore, it i s argued that these interactions promote the differential flow of information, influence, and social capital (Wasserman and Faust 1994; Coleman

PAGE 13

7 1990). Therefore, it is important to examine the structure, or the patterns of relationships in which people are embed social behavior is a result of 1988:31). Social Integration and Depression The relationship between social integration and depression has been investigated quite extensively since Durkheim (1897/2006) first proposed an association between integration and suicide during the 19 th century. 6 Essentially, social integration refers to (Ueno 2005:485). In order to measure this construct, researchers commonly use network variables, one of which is egocentric network size. To elaborate, an egocentric network may be defined as a network For the purposes of this discussion, direct connections can be thought of as unmediated social ties. In accordance with this description, egocentric network size simply refers to the total number of alters present in an egocentric network (Haines et al. 2008). It should be noted that researchers also typically examine the specific types of ties which are present in a given network. For instance friendship, family, and acquaintanceship ties are commonly distinguished from one another and measured as separate entities or relations (Wasserman and Faust 1994 ; House, Landis, and Umberson 6 According to Durkheim (1897/2006), there are four specific types of suicide: egoistic, anomic, altruistic, and fatalistic. Relevant to the current discussion, egoistic suicide opposite end of the spectrum, altruistic suicide results from over-integration, or an overload of obligations that In addition, it should be noted that anomic suicide is thought to result from a lack of moral regulation, or normlessness, while fatalistic suicide occurs as a result of oppression or over-regulation (Durkheim 1897/2006).

PAGE 14

8 1988). 7 Network studies which examine mental health customarily focus on those ties between kin, friends, neighbors, or co-workers due to the assertion that these types of relations are a medium through which social support is transferred (Haines et al. 2008). Social support which has a negative association with depression, may be defined as the information, emotional relief, material aid, and self-reliance that people retrieve from interpersonal relationships (Bozo, Toksabay, and Kurum 2009). It is believed that friends and family members are especially likely to promote the transfer of social support due to among kin (Wellman and Wortley 1990:559). Furthermore, frequent contact between individuals, as is anticipated with neighbors and co-workers, increases the likelihood that supportive relationships will develop (Wellman and Wortley 1990). For the purposes of this investigation, specific focus will be placed on the significance of friendship ties. Friendships are a key source of social capital which may and cultivation of social relationships Shinew, and Parry 2005:87). The social capital which is developed as a result of increased access to social support (Glover and Parry 2008:211). Reinforcing this point, research has suggested that friendship ties are more likely to transfer emotional aid and companionship than any other relation (Wellman and Wortley 1990). As individuals approach adulthood, it is believed that their friends become increasingly more important 7 It is important to acknowledge that network-study participants are generally responsible for subjectively determining what exactly it is that constitutes a particular type of relation, as specific definitions of friendship and family, for instance, are not always given (Marin and Hampton 2007).

PAGE 15

9 in comparison to family members (Monsour 2002), and studies which have examined is less detrimental than the absence of friends in the context of 2006:25). 8 A noticeable trend in the literature concerning social integration and depression is the relatively consistent association between small egocentric networks, or low levels of integration, and negative emotional arousal. 9 For instance, in a recent study, Ueno (2005) investigated adolescents and found that those with few friendship ties experienced more depressive symptoms than those who were more socially integrated. Additional research, utilizing General Social Survey data, has demonstrated that there is a negative association between the number of people an individual has to discuss important matters with and reported unhappiness (Burt 1987). An association has also been found between social integration and depressive symptomology among college students; those who are wellintegrated with friends tend to report lower levels of depression than less integrated individuals (Fagan 1994). 10 Theoretically, these findings reinforce the Durkheimian notion that social integration provides a form o throwing [an] individual on his own resources, leads him to share in [. .] collective 8 There is reason to suspect that many of the benefits which are associated with friendship might emerge prior to both adolescence and adulthood. For instance, research has indicated that having a large number of friends is associated with good mental health during childhood (Gest, Graham-Bermann, and Hartup 2001). 9 What constitutes a small or a large egocentric network is generally dependent upon the size of the average network in the sample or population under consideration. 10 It should also be noted that Bearman and Moody (2004), using data collected in conjunction with the Add Health project, found that female adolescents without any friendship ties were relatively more likely than their counterparts to think about committing suicide. However, no relationship was found between these two variables (i.e., social isolation and suicidal ideation) among males.

PAGE 16

10 Bjarnason 1998:96). There is also some evidence which suggests that relatively large egocentric networks are associated with high levels of depression, although this relationship has not been as extensively investigated. For example, Falci and McNeely (2009) found that -sized networks (2044). 11 More specifically, a curvilinear relationship was found between egocentric network size and depression, with depressive symptoms declining as network size increased until a specific threshold was reached and this trend reversed (Falci and McNeely 2009). These findings coincide with the theoretical notion that over-integration can result in greater mental health problems (Pescosolido and Levy 2002) and a sense of obligation that negatively affects the individual (Durkheim 1897/2006). In essence, it is believed that the effort which must be exerted to maintain a large network may come to outweigh any benefits or support received from it (Haines et al. 2008). There are few (if any) studies which corroborate McNeely 2009:2032). Notably, methodological issues may account for the lack of clarity concerning this topic in the literature. One important matter which should be addressed is that many previous studies have only tested for, and accordingly found, linear relationships between egocentric network size and depression, with larger network sizes being associated with lower levels of depressive symptoms (Ueno 2005; Cannuscio et al. 2004 ; Burt 1987). In 11 As stated prior, the research of Falci and McNeely (2009) has provided much of the basis for the current investigation.

PAGE 17

11 some instances, these studies have gone so far as to explicitly discount the importance of assessing curvilinearity, referring to it as only relevant in extreme situations (Ueno 2005). However, there seems to be little empirical support for this claim. Also, failure to test for a relationship does not necessarily imply its absence, and due to the nature of what is being investigated, findings of linearity do not refute the possibility that curvilinear relationships exist. In order to expand upon the existing literature, the current study will test for both linear and curvilinear relationships between egocentric network size and depression among United States college students. Network Cohesion and Depression Social cohesion may be defined as the eness, commitment, and harmony characteristic of tightly-knit groups and their members (Schaefer and Kornienko 2009:385). Although conceptually similar, network cohesion refers interconnections original). To elaborate, when examining an egocentric friendship network which is simply an egocentric network composed exclusively of friendship ties, network cohesion 12 Researchers generally measure cohesion by examining another network characteristic, density vels of density are 12 The definitions that are presented for network terms were chosen both due to convention and in order to maintain consistency with the research of Falci and McNeely (2009). However, there are a few minor differences which should be addressed. First, in their own study, Falci and McNeely (2009) used the term current investigation excluded focal persons so that egocentric network size would represent the total Again, this distinction is arbitrary. See Wasserman and Faust (1994 ) for a more comprehensive discussion of network concepts an d terminology

PAGE 18

12 fragmented, with few connections between alters, while those with high levels of density are characterized by many interconnections. This concept is visualized in FIGURE 1. The relationship between network density and mental health has been investigated extensively by researchers, but findings have been inconsistent (Lin and Peek 1999). Moreover, few scholars have focused directly on the relationship of interest: the association between network density and depression. However, there are a few relevant studies present in the literature. For instance, when Falci and McNeely (2009) examined the egocentric friendship networks of adolescents, they found a negative association between network density and depressive symptoms among females, but no significant relationship was found among males of the same age (Falci and McNeely 2009). In a similar study, Ueno (2005) found a general relationship between low network density and high levels of depression among adolescents, but potential gender differences were not examined. Another recent investigation, using an indirect measure of network density, failed to reveal a statistically significant association between this construct and depression among adults of either sex when controlling for other factors (Haines et al. 2008). Although only tangentially related to this discussion, it should be noted that research conducted by Bearman and Moody (2004) revealed an independent, negative association between network cohesion and suicidal ideation among adolescent girls ; no significant relationship was found among those of the opposite sex. So, while there is some evidence which suggests that network density is negatively associated with depression, at least among females, findings have been inconclusive. In previous studies, various egocentric network characteristics (e.g., egocentric network size and density)

PAGE 19

13 This has entailed testing for independent associations between these characteristics and depression (i.e., additive effects), rather than searching for interactions (i.e., multiplicative effects). However, this approach would seem to be counterintuitive, as network constructs do not occur independently in the social world. For instance, it does not matter whether an egocentric network is dense or fragmented; in either case, it must have a specific size or degree. Moreover, regardless separated. Stated more directly, it is certainly possible (for example) that large, dense egocentric networks influence mental health in different ways than large, fragmented networks. The failure to take this into account may partially explain the lack of clarity that has been observed in network studies which have examined depression and its correlates. While most studies have failed to examine potential interactions between network characteristics, there is at least one notable exception present in the literature: Falci and McNeely (2009) and depressive symptoms varies as a function o 2033). The findings of their study indicated that for girls, having a large, fragmented egocentric network was associated with relatively higher levels of depression (Falci and McNeely depressive symptoms for McNeely 2009:2048). There was a different pattern found among boys: large, fragmented networks were associated with low levels of depressive symptoms (Falci and McNeely 2009). However, a curvilinear relationship was found between egocentric

PAGE 20

14 network size and depression among those with cohesive networks large and small networks were both associated with high levels of depression (Falci and McNeely 2009). The above findings have several theoretical implications. To elaborate, it has been suggested by scholars that interactions in dense networks may lead people to Ueno 2005:486). In accordance with this stance, highly cohesive networks are thought to minimize the effort required to maintain individual relationships and to result in the sharing of social burdens (Forrester and Tashchian 2004). However, in conjunction with the results of their study, Falci and McNeely (2009) have speculated that the effects of network density may vary by gender. As Friedkin (2004) purports, identical network structures may differentially influence those networks are qualitatively distinct (413). Providing support for this position, research has demonstrated that females are more likely to engage in mutually supportive interactions with friends than are males, who tend to be more acceptant of negative events and to exhibit relatively independent coping behaviors (Frydenberg and Lewis 1993). Furthermore, males have historically reported friendships which are characterized by impersonal contact and comparatively low levels of emotional involvement; this stands in contrast to females, who are more likely to put the needs of others before their own (Rosenfield, Lennon, and White 2005 ; Umberson et al. 1996; Frydenberg and Lewis 1993). Because of these characteristics, specifically the tendency to seek out and to give social support it is possible that females are more likely to benefit from dense egocentric networks than are males. Also, due to

PAGE 21

15 their aforementioned tendency to deal with problems independently, males may actually be more likely to find large, cohesive networks burdensome. In addition to these factors, research has shown that males generally face more pressure to meet the expectations of their peer groups than females (Zucker et al. 1995). More specifically, adolescents and young adults are einberg and Monahan 2007:1531). Males who fail to meet these expectations are especially susceptible to social rejection (Zucker et al. 1995). In contrast, females are d Monahan 2007:1540). Of further significance, research has suggested that dense egocentric networks tend to exert more normative pressure on individuals than fragmented networks: At least one study has found a positive association between egocentric network density and behavioral accordance among adolescents (Haynie 2001). Stated more directly, research has indicated that adolescents in dense networks are more likely than those in fragmented networks to emulate the (delinquent) behaviors of their peers (Haynie 2001). The above findings, in conjunction with the Durkheimian notion that overregulation can have a negative impact on the individual, further support the claim that males may find large, dense egocentric networks especially burdensome. So, although this topic requires further investigation, there is some empirical and theoretical support for the assertion that large, dense egocentric networks are related to reduced levels of depression among females and elevated levels of depression among males. The current study will expand upon the existing literature by further exploring the relationship

PAGE 22

16 between network density and depression among United States college students. Potential interactions between egocentric network size, density, and gender will also be examined. Perceived Reciprocity of Support Although previous studies have examined egocentric network size and density in relation to depression, and it has been speculated that the observed effects of these characteristics are at least partially related to exchanges of social support, such exchanges have not been investigated directly by network researchers. More specifically, those relatively few network studies which have considered the relationship between social support and depression have focused exclusively on that support which is received by participants, and ignored that support which is given. Despite this shortcoming, it is important to consider relevant findings. For example, Falci and McNeely (2009) found that adolescents who reported, or perceived, receiving high levels of social support from their friends had low levels of depressive symptoms; in addition, this support was found to mediate the aforementioned relationship between small egocentric network size and depression. These results coincide with the existing body of literature related to this subject: Negative associations between social support and depression have commonly been found in other network (Haines et al. 2008) and non-network studies (Symister and Friend 2003). While the mediating effects of support reciprocity have not been investigated directly by network researchers, there is much theoretical and empirical evidence which implies that this line of inquiry is important and should not be overlooked. For example, social exchange theory predicts that in most inst

PAGE 23

17 and Stets 2005:213). In accordance with this perspective, it is has been asserted that positive emotions are few, exchange theory predict ever more [. .] from those who are dependent on them, thus generating [. .] negative exchanges of social support, seem to indicate that those individuals who receive more support than they give should experience lower levels of depression than those individuals who give more than they receive. It may also be inferred from this line of thought that because people with small egocentric networks have relatively fewer options to turn to when attempting to acquire social support, they are more likely to be in nonreciprocal exchange relationships in which they give more than they get. In contrast to this perspective, equity theory seek to maintain symmetry in their relationships with others, and that perceptions of being (Vaananen et al. 2008:1908). In essence, equity theorists propose that giving more than one receives may lead to feelings of resentment, while receiving more than one gives may lead to feelings of guilt or shame (Vaananen et al. 2008). Again, if these principles are applied to exchanges of social support, one would expect individuals who benefit significantly more or less than their alters, or who perceive a lack of equity in their relationships, to experience relatively higher levels of depression.

PAGE 24

18 There is a considerable amount of support for equity theory in the existing literature. For instance, Buunk and Schaufeli (1999) found that individuals who perceived a lack of equity in their relationships with co -workers or superiors were more likely to experience negative affect than those engaged in more mutually supportive interactions. Along the same lines, research by Taniguchi and Ura (2002) indicates that high school students who report inequitable relationships with their closest friends tend to experience higher levels of depression than their counterparts. Additionally, high school teachers who report underbenefiting in comparison to their romantic partners have been found to experience higher levels of depression than teachers in more equitable relationships (Bakker et al. 2000). Studies of this variety have produced more or less consistent results: When considering social support, perceptions of overbenefiting or underbenefiting are both associated with relatively high levels of depression. In light of these findings, the current investigation will further explore the relationship between social support and depression among United States college students. More clearly, when considering the relationship between egocentric network size and depression, specific focus will be placed on the mediating effects of social support. Additionally, using social exchange theory and equity theory as competing frameworks, this study will examine the extent to which perceptions of reciprocity mediate the relationships which have been observed between egocentric network characteristics (i.e., egocentric network size and density) and depression. Since it has been suggested that the effort which must be exerted to maintain a large network may come to outweigh any benefits or support received from it (Haines et al. 2008), there is reason to suspect that perceptions of equity may mediate the relationship between large egocentric network size

PAGE 25

19 and depression. Also, highly cohesive networks are thought to minimize the effort required to maintain individual relationships and to result in the sharing of social burdens (Forrester and Tashchian 2004). Therefore, perceptions of reciprocity may also mediate the relationship between network density and depressive symptomology, especially among females, who are more likely than males to seek out and to give social support (Rosenfield, Lennon, and White 2005 ; Umberson et al. 1996; Frydenberg and Lewis 1993). The current investigation will explore each of these possibilities.

PAGE 26

20 FIGURE 1: Egocentric Networks with Varying Levels of Density

PAGE 27

21 CHAPTER 3: SPECIFIC AIMS The current study will use social network analysis to examine the associations between self-reported egocentric network characteristics and depression among a sample of United States college students. As stated prior, it is important to understand factors related to depression among this population due to the severity of its potential outcomes (e.g., suicide, difficulty concentrating, interpersonal problems, changes in weight, and reduced energy). For the purposes of this investigation, measures of egocentric network size and density, social support, and perceived reciprocity of support will be utilized. In light of the numerous precedents set by Falci and McNeely (2009), this study will test for both linear and curvilinear relationships between egocentric network size and depression. Potential interactions between network size, density, and gender will also be explored. Again, in keeping with the findings of these two scholars (Falci and McNeely 2009), it is anticipated that large and small personal friendship networks will be associated with higher levels of depression than average-sized networks. Moreover, it is predicted that there will be a negative association between network density and depression among females, but no significant relationship is expected to be found among males. It is also hypothesized that females in large, cohesive networks will report lower

PAGE 28

22 levels of depressive symptoms than those in large, fragmented networks. Males are expected to demonstrate the opposite tendency. In addition to focusing on egocentric network size and density, this investigation will further explore the relationship between social support and mental health. Specific attention will be given to whether or not perceptions of reciprocity mediate the relationships which have been observed between egocentric network characteristics (i.e., egocentric network size and density) and depression. While it is acknowledged that the amount of social support received by an individual is important, perceptions of equity, or the lack thereof, may further explain the topic of interest. For instance, if an individual receives little social support from his or her friends, but perceives giving little in return, it is reasonable to speculate that such an exchange may have less of an influence on subjective well-being than if that individual perceives contributing a great deal to his or her friendship network. Stated more clearly, it is anticipated that a negative association will be found between social support and depression and that this support will mediate the relationship between egocentric network size and depressive symptomology. Also, in accordance with the principles of equity theory, it is predicted that there will be a curvilinear relationship between reciprocity of support and depression: Individuals who perceive underbenefiting or overbenefiting in their relationships with friends will have higher levels of depression than those in more equitable networks. It is further hypothesized that perceptions of equity will mediate the relationships that have been observed between egocentric network characteristics (i.e., egocentric network size and density) and depression.

PAGE 29

23 CHAPTER 4: METHODS Data Collection and Sample In order to collect data for this project, an anonymous online survey was created, pilot tested, and then made available to a group of college students at a large, public Florida university during the first eight weeks of the fall 2010 semester. 13 More specifically, participants were recruited from three undergraduate sociology courses : Social Psychology, Introduction to Sociology, and Contemporary Social Problems. 14 Students in each course were told that their participation would allow the primary investigator to gain a better understanding of student friendship networks and their relationship with student attitudes, feelings, and behaviors. In return for taking part in this study, respondents were offered 5 points of extra credit by their respective instructors. As an alternative method of earning these points, individuals were permitted to complete a short, two-page writing assignment. Survey materials were administered using SelectSurvey an online survey interface that was made accessible to students via their respective Blackboard accounts. Blackboard was used to track student participation; this allowed for the allocation of 13 A full, text-based version of this survey is presented in Appendix A. 14 Individuals enrolled in more than one of these courses were only permitted to complete the questionnaire for a single class.

PAGE 30

24 extra credit despite the anonymity of survey responses. 15 It should also be noted that an online consent form was utilized during this investigation. This form preceded survey materials, and it required students to acknowledge that they were over the age of eighteen (minors were not permitted to take part in this study due to their designation as a vulnerable population) and that they were willing participants. A total of 747 students (less an unknown number of individuals enrolled in multiple courses) were given the opportunity to participate in this study. Data were collected from 706 respondents, but 13 minors and 22 individuals who failed to provide any information were excluded from all analyses. This resulted in a final sample size of n = 671 students. While clearly a convenience sample, this method of data collection was deemed appropriate since it taps into the population of interest (i.e., individuals currently enrolled as students at institutions of higher education) and no attempt at generalization will be made. In essence, this may be considered exploratory research it is believed that results will provide meaningful insight and direction for future investigation. Furthermore, the use of convenience sampling is consistent with a long line of mental health research (Hyun et al. 2009; Bailey et al. 2007; Low and Feissner 1998 ). Measures Depressive Symptoms : For the purposes of this investigation, the Center for Epidemiologic Studies Depression Scale (CESD) was used to assess depression. The CES-report scale designed to measure depressive symptomatology in the general po pecifically, it consists of 20 questions that ask individuals to report the frequency with which they have experienced certain 15 More detailed information about Blackboard (http://www.blackboard.com) and SelectSurvey (http:// selectsurvey.net) can be retrieved from their official websites.

PAGE 31

25 thoughts, feelings, and physical conditions during the past week ; these conditions represent symptoms associated with depression (Radloff 1977). Each question has four response categories ranging from (0) (3) 7:387). 16 Overall depression scores are calculated by adding up the values reported for each of the 20 CESD items (Prescott et al. 1998; Radloff 1977 ). Possible scores range from 0 to 60, with larger scores representing higher levels of depressive symptomology. 17 Notably, the CES-D has been shown to have high test-retest reliability among samples of diverse ages and ethnic backgrounds (Prescott et al. 1998). I t has been tested in both general and psychiatric settings, and its internal consistency and construct validity are well established (Prescott et al. 1998; Radloff 1977). Network Structure : name generator has become the standard method to 2007:163; italics in original). To elaborate, name generators are typically administered through surveys or interviews and consist of a prompt which is intended to obtain a list of alters from respondents (Marin and Hampton 2007). This method is especially useful when attempting to measure specific and Hampton 2007). The current investigation utilized a name generator with the Consider who the closest and most important friends in your life are. Put the initials of these people, maximum 5, in the following blanks. Then select the 16 Four CES-D items are scored in reverse about the fu from ( See Appendix A for the exact wording of all survey questions. 17 In practice, it is widely accepted that CESD values represent clinical levels depression ( Cyranowski 2011; SCIRE Project 2010).

PAGE 32

26 Respondents were permitted to list up to 5 alters, who were to be identified by their first and last initials. Initials were reported, rather than full names, in order to ensure the anonymity of those being described. The number of recorded friendships was limited in order to maintain the clarity and manageability of survey documents. This practice is common in network studies, and a cutoff of 5 alters is consistent with guidelines utilized by the General Social Survey (Wellman 2007; Burt 1984 ). Although this method generally underestimates the total number of alters who are present in a given network, there is a high correlation between the number of ties that are reported by participants and the size of their personal networks as determined by more extensive measurement techniques (Marin and Hampton 2007). In essence, name generators and their related follow-up questions may be thought of as providing an adequate, although limited, estimation of egocentric network characteristics. 18 Participants were also asked to indicate whether or not any of their alters could be further categorized as family members or partners. Because friendship is subjective, name generators will not necessarily exclude these relations. However, it is generall y believed that each of these categories (i.e., friends, family members, and partners ) represents a qualitatively distinct set of relationsh ip s (Wellman and Wortley 1990). As a way of accounting for this issue, family members and partners were excluded when calculating network measures. Therefore, the total number of friends reported less the 18 While egocentric network size is most commonly assessed using self-reports, measurements relying upon multiple perspectives (i.e., those of egos and their respective alters) are believed to provide more accurate representations of structural network characteristics (Wellman 1988). Notably, because their data were collected from a series of closed networks (i.e., high schools), Falci and McNeely (2009) were able to count both those friendship nominations that were made and those that were received by focal adolescents when calculating personal network size.

PAGE 33

27 number of alters categorized as family members and partners was used as a measure of egocentric network size; potential values ranged from 0 to 5. 19 For purposes of this investigation, respondents were also asked to describe the relationships between their closest and most important friends. More specifically, participants were given the opportunity to describe each of their friendship pairs using one of the following statements: ther as I When assessing network density, friendship pairs described as b were count ed as 1 friendship tie. Those that were were assigned a value of 0.5, and friendship pairs that were were assigned a value of 0. 20 The total number of reported ties was then divided by the total number of possible ties in order to determine egocentric network density. 21 This method of assessing network density is consistent with guidelines suggested in conjunction with the General Social Survey (Burt 1987 ; Burt and Guilarte 1986). 22 Density scores ranged from 0 (i.e., none of an were friends with each other) to 1 (i.e., all of an were friends with each other). Notably, it is impossible to calculate the density of a personal network that does not have at least two alters. 19 Although family members and partners were not directly excluded by Falci and McNeely (2009), their sample (as stated prior) was comprised of a series of closed networks (i.e., high schools). Therefore, in their own study, family members and partners were unlikely to have constituted a large number of the alters who were reported by participants. 20 Alters categorized as partners or family members were excluded when calculating network density. 21 work [Egocentric Network Size/Possible Number of Ties between Alters: 0/0; 1/0; 2/1; 3/3; 4/6; 5/10]. 22 Ag ain, because Falci and McNeely (2009) used data collected from a series of closed networks, they were respective alters. Therefore, in their own study, friendship ties between alters were either present or absent. In the current investigation, a dichotomous method for calculating network density was also considered assigned a value of 0), but this method was ultimately rejected since it failed to significantly influence results.

PAGE 34

28 Therefore, in order to maintain consistency with the research of Falci and McNeely (2009), all participants with an egocentric network size of 0 or 1 were assigned a density value of 0. Social Support : The Multidimensional Scale of Perceived Social Support -report scale designed to tap perceived social support from family, friends, and significant oth 1995:595). More specifically, this instrument includes three subscales, one for each relation (Cecil et al. 1995). This study utilized the can count on my friends w Zimet, and Walker 1991:757). Answer choices for each item range from 1 (very strongly disagree) to 7 (very strongly agree); in order to determine a total score, values for these items are added together and then divided by 4 (Cecil et al. 1995; Dahlem et al. 1991). Scores on the friendship subscale range from 1 to 7, with larger scores representing higher levels of perceived support (Kazarian and McCabe 1991). Reciprocity of Support : Although social support is a multidimensional construct, research has demonstrated that individuals tend to assess the reciprocity of their relationships holistically (Van Horn, Schaufeli, and Taris 2001). Therefore, it has been -rated Serving as an example, the Hatfield Global Reciprocity Measure is based on a single item and has been used by scholars to assess the perceived reciprocity of individual relationships (Hatfield et al. 1985); a modified version of this instrument has been used to measure the

PAGE 35

29 reciprocity of an entire network (Vaananen et al. 2008; Vaananen et al. 2005). For the purposes of this investigation, a modified version of the Hatfield Global Reciprocity Measure was constructed and then used to assess the perceived equity of each friendship network. Specifically, respondents were asked to consider the following question for each receives more support and help (for example: emotional support, companionship, information, services, or financial help)? How would you describe your Potential answers included the following: (give support and help more than receive support and help more Reported values were added together and then divided by egocentric network size; reciprocity scores ranged from -1 to 1, with negative scores representing perceived underbenefiting and positive scores representing perceived overbenefiting. Because division by 0 is undefined, those participants who failed to report any friendships whatsoever (i.e., those with an egocentric network size of 0) were assigned a reciprocity score of 0. 23 Theoretically, this method was deemed appropriate since individuals without any friends lack the ability to overbenefit or underbenefit in comparison to their alters. Demographic Controls and Contextual Questions : Standard demographic information was also collected from respondents. Specifically, each participant was asked to indicate his or her age, gender, race/ethnicity, and current relationship status. Gender and relationship status were treated as dichotomous variables: Individuals were 23 Again, family members and partners were excluded when calculating network reciprocity scores.

PAGE 36

30 classified as either (0) male or (1) female and as (0) single or (1) partnered. 24 Age was measured in years. Five categories were constructed for race/ethnicity : 25 These categories were treated as polytomous dummy variables when conducting multivariate analyses. The following information was also collected from participants : high school GPA (rounded to two decimal places), current class standing (freshman, sophomore, junior, senior, or other), and distance to campus from current residence (0 miles, 0.1 5 miles, 5.1 10 miles, 10.1 20 miles, 20.1 50 miles, or more than 50 miles). 26 In addition, respondents were asked to indicate how often they had trouble paying for things (never, rarely, sometimes, or always) and the highest level of education completed by either of their parents (highest level of parental education grade school or less, some high school, high school d degree, or some post-graduate education/professional degree). Both of these questions can be thought of as indirect measures of socioeconomic status (Miech and Shanahan 2000; Goodman 1999), which has been found to have an inverse relationship with depression and negative affect (Lorant et al. 2003; Fryers, Melzer, and Jenkins 2003; Hao and Johnson 2000; Link, Lennon, and Dohrenwend 1993). Respective answer choices for 24 25 were ultimately combined since they accounted for less than 6% of all respondents. 26 The specific name of the university where this study was conducted has been omitted in order to maintain the anonymity of respondents.

PAGE 37

31 standing, campus variables when conducting multivariate analyses. 27 Respondents were also asked to describe the gender (male or female) of their alters. Notably, this made it possible to calculate the total number of female friends present in each individual s egocentric network. 28 Because research suggests that females are more likely than males to provide social support to their friends (Haines et al. 2008), it may be important to control for this measure. Potential values ranged from 0 to 5 (female friends). Finally, for contextual reasons, several questions from the General Social Survey (see both Appendix A and TABLE A3) were included in the survey materials presented to respondents. For the purposes of this investigation, none of these items will be considered; however, it should be noted that they were taken from the official GSS website (http://www.norc.org/GSS+Website). Analytic Strategy SPSS Statistics 19 was used for all statistical procedures. No key variables were missing more than 3.7% (n = 25) of their respective values, so listwise deletion of cases was deemed appropriate when conducting multivariate analyses. 29 Specifically, OLS regression models were used to examine the relationships between self-reported egocentric network characteristics (i.e., egocentric network size and density) and depression. Potential interactions between network characteristics and gender were also 27 for distance to campus from current residence 28 Consistent with all other network variables, family members and partners were excluded when calculating 29 On average, key variables were missing approximately 1% (n = 7) of their respective values.

PAGE 38

32 explored. Additional models examined social support and perceived reciprocity of support in relation to depression. 30 Several other measures were considered as potential controls, but they were ultimately excluded since they failed to explain any additional model variance (as determined by F-tests), they had no impact on observed results, and they were not significant predictors of depression. These variables included h 31 No problems with multicollinearity were detected. For all models, VIFs fell well below the acceptable threshold of 10 (Hair et al. 2006). Moreover, in order to avoid potential complications, only those interaction terms that explained additional variance (as determined by F-tests) were kept in subsequent regression models (Kromrey and Foster-Johnson 1998). Skewness (0.92) and kurtosis (0.47) values for the dependent variable (i.e., depression) were also examined and fell within acceptable ranges (Illinois State University 2008). 32 To clarify 1 is considered very good for most psychometric uses, but +/1 to +/(Illinois State University 2008:1). 30 OLS regression was used in order to maintain consistency with the methods of Falci and McNeely (2009). 31 Other potential control variables included recruitment course (Introduction to Sociology, Social Problems, or Social Psychology), total number of family members reported (range = 0 to 5), total number of partners reported (range = 0 to 5), number of same-sex friendships (range = 0 to 5), and having at least one frie nd (0 = no friends; 1 = one or more friend/s). 32 Since the distribution of depression scores was positively skewed, all regression models were rerun using the natural log of depression as the dependent variable. Results did not differ significantly.

PAGE 39

33 CH APTER 5: RESULTS Descriptive Statistics Descriptive statistics are presented in TABLE 1. The sample had a mean age of 20.9, with a standard deviation of 5.0 years. Approximately 97% of all respondents were under the age of 35, and 92% were under the age of 26. These percentages are consistent with national figures: According to the U.S. Census Bureau (2008), over 92% of those enrolled at scholarly institutions in America are between the ages of 15 and 34. Also, the distribution of respondents by race/ethnicity %), wa s similar to the overall distribution of students enrolled at the university where this study was conducted. Official enrollment figures for the fall 2010 semester were as (15.9%), 33 A slightly higher percentage of respondents were single (52.9%), rather than partnered, and a sizeable majority of participants were female (71.5%). 34 Because of the relatively high proportion of females who took part in this investigation, it will be 33 In order to maintain the anonymity of study participants, source material will not be reported for these figures. Additional data are available upon request. 34 During the fall 2010 semester, females made up 57.5% of the student body at the university where this study was conducted.

PAGE 40

34 necessary to interpret findings (especially those related to gender) with caution. 35 Of further significance, there was considerable diversity with regards to the current class standing of respondents. Approximately 33% were freshmen, 24% were sophomores, 19% were juniors, and 22% were seniors. An additional 2% were categorized as Notably, while a small minority of participants (2.6%) indicated that they always had among the remaining three categories for this variable (i.e., never, rarely, and sometimes). Similar findings were observed nly 3.1% of respondents indicated that both of their parents had failed to obtain at least a high school diploma or GED. On average, students reported having 2.2 friends (excluding family members and partners). The largest friendship network that was observed consisted of 5 alters, and the smallest network consisted of 0 individuals. Over 80% of the sample had an egocentri c network size of 1 or greater. The mean value for network density was 0.3; this indicates Roughly 54.8% of respondents had density values that fell below the mean ; 44.9% had values higher than the average score. The mean value for support received (MSPSS friendship subscale) was 5.3; over 78% of all participants had a score that was > 4 (i.e., the neutral midpoint of the friendship subscale). Additionally, it should be noted that the average score for perceived reciprocity (Hatfield Global Reciprocity Measure) was -0.1, which is slightly below the neutral value (0) for this measure and represents perceived underbenefiting. Of further 35 However, it should be made clear that the relatively high proportion of females who took part is this investigation was expected given the courses that students were recruited from.

PAGE 41

35 significance, roughly 10.5% of respondents reported overbenefiting in comparison to their alters, 28% reported underbenefiting, and 61.5% claimed to be in equitable networks. Finally, the mean value for depression (CES-D) was 14.5; approximately 38% of al is commonly used as a threshold for identifying clinical levels depression ( Cyranowski 2011; SCIRE Project 2010). Bivariate Correlations TABLE 2 presents bivariate correlations between key variables (i.e., egocentric network size, network density, social support, reciprocity of support, and depression). Consistent with expectations a weak, negative association was found between social support and depression. Stated more directly, those who reported receiving high levels of support also reported low levels of depressive symptomology. This result coincides with the theoretical notion that insufficient levels of support are associated with feelings of me lancholy and a lack of purpose (Thorlindsson and Bjarnason 1998:96). However, no significant relationships were found between egocentric network size, network density, or reciprocity of support and depression. Notably, the lack of a linear association between egocentric network size and depressive symptomology contradicts previous research which has suggested that there is an inverse relationship between social integration and negative emotional arousal (Ueno 2005; Bearman and Moody 2004; Field et al. 2001; Fagan 1994; Burt 1987). This finding, as well as the possibility that there is a curvilinear relationship between egocentric network size and depression (Falci and McNeely 2009), will be further explored using multivariate techniques. The existing literature also suggests that there is a negative association between network density and depressive symptomology (Ueno 2005; Lin and Peek 1999).

PAGE 42

36 However, recent findings have indicated that this relationship may be stratified by gender, with highly cohesive networks benefiting females exclusively (Falci and McNeely 2009; Bearman and Moody 2004). Therefore, the lack of an observed, linear relationship between these two variables (i.e., network density and depression) is not necessarily surprising. Additionally, it may be necessary to control for egocentric network size when attempting to observe the relationship between network density and depression. To elaborate, for the purposes of this investigation, all respondents with an egocentric network size of 0 or 1 were assigned a density value of 0. This method resulted in a strong correlation between egocentric network size and density, and it limited the extent to which density values were free to vary among those with network sizes less than 2. Stated more directly, it was not possible for there to be an inverse association between network density and depression among those with small egocentric networks. Accordingly, holding network size constant may allow for more accurate results. The relationship between network density and depression will be further explored at the multivariate level. The lack of an observed relationship between perceived reciprocity of support and depression was not unexpected. To elaborate, values for support reciprocity ranged from -1 (perceived underbenefiting) to 1 (perceived overbenefiting). Rather than predicting a linear relationship between this variable and depression, the existing literature suggests that individuals who underbenefit or overbenefit in comparison to the peers are both more likely to experience heightened levels of depressive symptomology (Taniguchi and Ura 2002; Bakker et al. 2000; Buunk and Schaufeli 1999 ). Therefore, one might expect there

PAGE 43

37 to be a curvilinear relationship between reciprocity of support and depression. In the next section, this possibility will be explored using multivariate techniques. Multivariate Models OLS r coef were not found to be significant predictors of depression. Models 1 and 2 were used to examine the relationship between egocentric network size and depressive symptomology. 36 More specifically, Model 1 tested for a linear relationship between these two variables, and Model 2 tested for a curvilinear relationship. Consistent with the approach of Falci and McNeely (2009), the squared term for network size was used to assess curvilinearity. Contrary to expectations, egocentric network size and curvilinear network size both failed to significantly predict depression at the multivariate level. As stated prior, there is a long line of mental health research which suggests that there is an inverse relationship between social integration and depression. However, for the purposes of this investigation, specific focus was placed on replicating the results of Falci and McNeely (2009), who found a curvilinear relationship between egocentric network size and depression among adolescents (i.e., depressive symptoms declined as network size increased until a specific threshold was reached and this trend reversed ). The results of the current investigation failed to support the findings of these two scholars. 36 Depression was the dependent variable in all OLS regression models.

PAGE 44

38 Models 3, 4, 5, and 6 were used to explore the relationship between network density and depression. To elaborate, Model 3 test ed for a linear relationship between network density and depressive symptomology; Model 4 explored the same relationship while controlling for egocentric network size. Because research has suggested that the effects of network cohesion may vary by gender (Falci and McNeely 2009; Bearman and Moody 2004), an interaction term for these two variables (i.e., network density and gender) was constructed. More specifically, Model 5 was used to assess the relationship between network density and depression among females. Again, Model 6 explored the same relationship while controlling for egocentric network size. Contrary to expectations, network density failed to significantly predict depression in all four regression models. So, while Falci and McNeely (2009) found a negative association between network density and depression among female adolescents, the current investigation failed to confirm the presence of such a relationship among United States college students. However, it should be noted that scholars investigating network density and mental health have commonly produced inconsistent results (Haines et al. 2008 ; Ueno 2005; Lin and Peek 1999). Because gender, egocentric network size, and network density were not found to be significant predictors of depression at the multivariate level, further interactions between these three variables were not assessed in subsequent regression models. Models 7 and 8 were used to examine the relationship between social support and depressive symptomology. More specifically, Model 7 tested for a linear relationship between social support and depression. Model 8 assessed the same relationship while controlling for egocentric network size and density. Consistent with expectations (Falci

PAGE 45

39 and McNeely 2009; Haines et al. 2008; Symister and Friend 2003), so cial support was found to be a significant predictor of depression in both regression models. Notably, controlling for egocentric network size and density had little effect on the regression coefficient for social support, which maintained a negative association with depressive symptomology. To elaborate, the unstandardized coefficient for social support was -0.35 in Model 7, and -0.37 in Model 8. Put into context, a 3-point increase on the MSPSS friendship subscale equated to a 1-point decrease on the CES-D. In comparison, antidepressants such as Prozac (fluoxetine), Paxil (paroxetine), Zoloft (sertraline), Effexor (venlafaxine), Serzone (nefazodone), and Celexa (citalopram) have been found to produce improvement scores of approximately 2 points on the 62-point Hamilton Depression Scale, which is roughly equivalent to the depression measure used in this study (Kirsch et al. 2008). Finally, it should be noted that the potential mediating effects of social support were not assessed since egocentric network size was not found to be significant predictor of depression in previous models. Models 9, 10, 11, and 12 were used to examine the relationship between perceived reciprocity of support and depression Model 9 tested for a linear relationship between these two variables, and Model 11 tested for a curvilinear relationship. Again, the squared term for reciprocity of support was used to assess curvilinearity. Models 10 and 12 explored the same relationships as Models 9 and 11, respectively, while controlling for egocentric network size, network density, and social support. Consistent with expectations, the squared term for reciprocity of support was found to be a significant predictor of depressive symptomology; the unstandardized coefficient for this variable was stable across models (Model 11 = 2.47 and Model 12 = 2.46). In general

PAGE 46

40 terms, individuals who perceived overbenefiting or underbenefiting in comparison to their alters experienced higher levels of depression than those in more equitable networks. These findings are consistent with equity theory, which states that individuals seek to maintain symmetry in their relationships with others, and that perceptions of being deprived as well as perceptions of being advantaged are associated w (Vaananen et al. 2008:1908). It should also be noted that the unstandardized coefficient for social support (Model 8 = -0.37 and Model 12 = -0.37) was unaffected by the inclusion of curvilinear reciprocity in regression models. This suggests that social support and perceived reciprocity of support have unique and independent relationships with depression. Again, the potential mediating effects of support reciprocity were not assessed since egocentric network size and density were not found to be significant predictors of depression in previous models. 37 It should also be noted that several demographic control variables were found to have significant relationships with depression. In all twelve OLS regression models, being Black or African American and increased age were associated with relatively low levels of depressive symptomology. Additionally, having trouble paying for things (sometimes or always) Asian were both associated with poor mental health. In Models 1-6 and 8-12, individuals reporting that the highest level of education obtained levels of depression than those answering n all models that did not include social support as a predictor variable, participants who 37 However, the interaction term for curvilinear reciprocity and gender was constructed in order to determine whether or not equity was particularly important for females. This variable failed to significantly predict depression in additional, unreported regression models.

PAGE 47

41 such difficulties. Although not the focus of this investigation, these results were generally consistent with the findings of previous mental health research (Lorant et al. 2003; Fryers et al. 2003; Miech and Shanahan 2000; Hao and Johnson 2000; Goodman 1999; Kelly et al. 1999; Okazaki 1997; Link et al. 1993 ).

PAGE 48

42 TABLE 1: Descriptive Statistics for College Students at a Large Florida University Variable n / Mean % / SD Min Max Independent Variables Ego Demographic s Recruitment Course & Control Variables Introduction to Sociology 233 34.7 Social Problems 220 32.8 Social Psychology 218 32.5 Age (Years) 20.9 5 .0 18 63 Gender Male 191 28.5 Female 479 71.5 Race/Ethnicity White 413 61.7 Black or African American 80 12.0 Hispanic or Latino 97 14.5 Asian 42 6.3 Other 37 5.5 Current Relationship Status Single 355 52.9 Partnered 316 47.1 High School GPA 3.8 0.7 0 .0 6.4

PAGE 49

43 TABLE 1: Descriptive Statistics for College Students at a Large Florida University (Continued) Variable n / Mean % / SD Min Max Independent Variables Ego Demographics Distance to Campus from Current Residence & Control Variables 0 Miles (Campus Housing) 133 19.8 0.1 5 Miles 198 29.5 5.1 10 Miles 59 8.8 10.1 20 Miles 104 15.5 20.1 50 Miles 92 13.7 More than 50 Miles 85 12.7 Current Class Standing Freshman 221 33.0 Sophomore 162 24.2 Junior 127 19.0 Senior 145 21.6 Other 15 2.2 Highest Level of Parental Education Grade School or Less 4 0.6 Some High School 17 2.5 High School or GED 136 20.3 Some College or Associate's Degree 230 34.5 Bachelor's Degree 158 23.6 Some Post Graduate Education 124 18.5 or Professional Degree

PAGE 50

44 TABLE 1: Descriptive Statistics for College Students at a Large Florida University (Continued) Variable n / Mean % / SD Min Max Independent Variables Ego Demographic s Trouble Paying for Things (Frequency) & Control Variables Never 169 25.4 Rarely 261 39.2 Sometimes 218 32.8 Always 17 2.6 Network Structure Egocentric Network Size 2.2 1.6 0 5 (Excluding Family Members and Partners) Network Density 0.3 0.3 0 .0 1 .0 Alter Demographics Number of Female Friends 1.4 1.4 0 5 Social Support Support Received 5.3 1.9 1 .0 7 .0 (MSPSS Friendship Subscale) Perceived Reciprocity of Support 0.1 0.4 1 .0 1 .0 (Hatfield Global Reciprocity Measure) Dependent Variable Depression (CES D) 14.5 9.8 0 54 Notes: (N = 671)

PAGE 51

45

PAGE 52

46

PAGE 53

47

PAGE 54

48

PAGE 55

49

PAGE 56

50

PAGE 57

51

PAGE 58

52 CHAPTER 6: DISCUSSION Using social network analysis as a theoretical framework, the current study examined the associations between self-reported egocentric network characteristics and depression among a sample of United States college students. For the purposes of this investigation, specific focus was placed on egocentric network size and density, and on the perceived reciprocity of social support exchanges that occur within personal friendship networks. To reiterate, it is important to understand factors related to depression among this population due to the severity of its potential outcomes. Specifically, depression has been linked to decreased academic productivity, interpersonal problems at school, and truancy among college students (Heiligenstein and Guenther 1996). Moreover, depression is associated with suicide, difficulty concentrating, reduced energy, changes in weight, and changes in the quality or quantity of sleep among those in the general population (Lackey 2008; Hockenbury and Hockenbury 2003; Spirito et al. 2003; American Psychiatric Association 2000; Sadowski and Kelley 1993). Findings Network Structure As stated prior, a recent study conducted by Christina Falci and Clea McNeely (2009) provided much of the basis for the current investigation. To elaborate, while

PAGE 59

53 numerous scholars (Ueno 2005; Bearman and Moody 2004; Cannuscio et al. 2004; Burt 1987) have reported finding inverse relationships between social integration and depressive symptomology, Falci and McNeely (2009) tested for and found a curvilinear relationship between egocentric network size and depression among a nationally representative sample of American adolescents. Stated more clearly, depressive symptoms were found to decline as network size increased until a specific threshold was reached and this trend reversed (Falci and McNeely 2009). Prior to their investigation, relatively few (if any) network studies had entertained the theoretical notion that overintegration may actually result in greater mental health problems (Pescosolido and Levy 2002) and a sense of obligation that negatively affects the individual (Durkheim 1897/2006). In accordance with the research of Falci and McNeely (2009), the current study tested for both linear and curvilinear relationships between egocentric network size and depression. It was predicted that a curvilinear relationship would be observed between these two variables, with large and small personal friendship networks being associated with higher levels of depression than average-sized networks. Contrary to expectations, egocentric network size and curvilinear network size both failed to significantly predict depression at the multivariate level. More directly, the results of this study not only failed to support the findings of Falci and McNeely (2009), but they also failed to support a long line of mental health research which suggests that there is an inverse relationship between social integration and depression. There are several potential explanations for the lack of an observed relationship between egocentric network size and depressive symptomology. First, it should be noted

PAGE 60

54 that in their own study, Falci and McNeely (2009) found that as egocentric network size increased, depressive symptoms declined until a network size of 12 friends was reached; on average, adolescents with 24 friends reported experiencing roughly equivalent levels of depression as those with no friends. Therefore, it is certainly possible that the failure to observe a curvilinear relationship between egocentric network size and depression was related to the fact that individuals participating in the current study were unable to reach a threshold of 12 reported friends. Clearly, the limited range of observed values for egocentric network size should be considered a methodological issue. Because their data were collected from a series of closed networks (i.e., high schools), Falci and McNeely (2009) were able to count both those friendship nominations that were made and those that were received by focal adolescents when calculating personal network size. 38 This resulted in a range of 0-32 friends (Falci and McNeely 2009), as opposed to the range of 0-5 that was observed in the current investigation. 39 Although research has demonstrated that there is generally a high correlation between the number of ties that are reported by participants and the size of their personal networks as determined by more extensive measurement techniques, the proposition that name generators and their related follow-up questions may provide somewhat limited estimates of egocentric network characteristics (Marin and Hampton 2007) should not be entirely overlooked. 38 A closed network may be described as a closed set of actors. To clarify, the boundary of a set of actors Actors located outside of a closed network are generally not considered when attempting to describe the characteristics of a specific population (Wasserman and Faust 1994). 39 While egocentric network size is most commonly assessed using self-reports, measurements relying up on multiple perspectives (i.e., those of egos and their respective alters) are believed to provide more accurate representations of structural network characteristics (Wellman 1988).

PAGE 61

55 Still, it should be made clear that in their own research, Falci and McNeely (2009) found that hav relationships between egocentric network size and depression have been found in at least two other network studies that restricted the number of alters reported to 5 (Burt 1987) failure to reveal a linear relationship between egocentric network size and depression may simply be an anomaly. Regardless, it is important to treat this result with caution, as data were collected from a single institution of higher education using non-random sampling. Therefore, findings are not necessarily representative of the larger student body. It is also important to consider the possibility that college students represent a unique population, with distinct characteristics that have not been adequately explored by network researchers focusing on either adolescents or adults. 40 To further elaborate, Jeffrey Arnett (2004) has argued that a new and unprecedented period of the life course has taken shape over the past four decades. This period, which Arnett (2004) labels emerging adulthood stretches from the late teens through the midto -late twenties and is characterized by instability, exploration, and opportunity. Arnett (2004) claims that relative fail to achieve their educational goals, get married, become parents, or make long-term career choices until later on in the lifespan. So, while emerging adulthood is marked by freedoms 40 Most studies investigating the relationship between egocentric network size and mental health have focused on either high school students (Falci and McNeely 2009; Ueno 2005; Bearman and Moody 2004) or general (i.e., individuals over the age of 18) adult populations (Haines et al 2008; Cannuscio et al. 2004; Burt 1987).

PAGE 62

56 uncharacteristic of adolescence (e.g., increased independence), it is also devoid of the responsibilities and the relative stability historically associated with adulthood. However, according to t and companionship, given that this is a period when most young people have left their families of origin but have not yet entered As stated prior, over 92% of the individuals who took part in the current investigation were between (2004) age guidelines for emerging adulthood. Notably, in both bivariate and multivariate analyses, social support was found to have a significant, inverse relationship with depressive symptomology. However, egocentric network size was only found to have an extremely weak (0.07) and marginally significant (p < 0.10) bivariate relationship with social support. Clearly, these findings fail to uphold the theoretical notion that those with small egocentric networks are prone to suffering from inadequate levels of support (Haines, Beggs, and Hurlbert 2002 ; Thorlindsson and Bjarnason 1998 ; Walker, Wasserman, and Wellman 1993). If the results of the current investigation are not to be treated as an anomaly, it may be prudent for scholars to further explore the relationships between egocentric network size, social support, and depression among college students in particular, and among emerging adults more generally. Although purely speculation, it is possible that due to their involvement in social activities both at work and at school, college students require relatively little companionship from their

PAGE 63

57 close friends. 41 If this is indeed the case, even small egocentric networks may be able to provide sufficient amounts of social support. Again, however, this is a question for future research. In addition to examining the relationship between egocentric network size and depression, Falci and McNeely (2009) explored potential interactions between network size, density, and gender. This approach wa s quite noteworthy, as most researchers have focused on delineating independent associations between network characteristics and depression, rather than searching for interaction effects. As discussed in further detail above, Falci and McNeely (2009) found a negative association between network density and depression among female adolescents, but no significant relationship was found among males of the same age. Of additional significance, the authors found that females in large, cohesive networks reported lower levels of depressive symptoms than those in large, fragmented networks; the opposite pattern was found among males (Falci and McNeely 2009). Again, in accordance with the research of Falci and McNeely (2009), the current study set out to examine potential interactions between network size, density, and gender. Specifically, it was hypothesized that a negative association would be found between network density and depression among female college students, but no significant relationship was expected among males. Additionally, it was predicted that females in large, cohesive networks would report lower levels of depressive symptoms than those in large, fragmented networks; males were expected to demonstrate the opposite tendency. Contrary to expectations, network density failed to significantly predict depressive 41 Approximately 60% of all college students in America hold jobs while in school, and one fourth of all students work full time (Fitzpatrick and Turner 2006; Arnett 2004).

PAGE 64

58 symptomology at the multivariate level both among females and among the sample as a whole. Because gender, egocentric network size, and network density were not found to be significant predictors of depression at the multivariate level, further interactions between these three variables were not assessed. As was the case for egocentric network size, there are several potential explanations for the lack of an observed relationship between network density and depression. For instance, it should be noted that in their own research, Falci and McNeely (2009) found that the association between network density and depressiv e symptomology was extremely weak among female adolescents with small egocentric networks; the magnitude of this relationship became incrementally larger as network size veal an inverse relationship between network density and depressive symptomology among female college students was related to the fact that there was an artificial limit placed on the number of friends that could be reported by participants. Also, as stated prior, researchers investigating the relationship between network density and mental health have commonly produced inconsistent results (Lin and Peek 1999). Serving as an example, Claude Fischer (1982) examined the social ties of approximately 1,050 individuals residing in 50 different urban localities and found that network density was positively associated with psychological well-being but only among those of low socioeconomic status. Additionally, when Charles Kadushin (1983) studied the interpersonal environments of Vietnam War veterans, an inverse relationship was found between network density and stress, but only among those living in rural areas. Research focusing directly on the mental health of adolescents has been just as erratic.

PAGE 65

59 To elaborate, while Ueno (2005) has reported finding a negative association between network density and depressive symptomology among adolescents, research conducted by Bearman and Moody (2004) suggests that there is only an inverse relationship between network cohesion and suicidal ideation among adolescent females. Complicating matters even further, in a recent study, Haines et al. (2008) failed to find a significant relationship between network density and depressive symptomology among adults of either sex. Clearly, the relationship between network density and mental health requires further investigation, as research in this area has been sporadic, lacked continuity, and provided inconsistent results. Moreover, since Falci and McNeely (2009) were the first (and only) scholars to report finding a three-way interaction between network size, density, and gender additional research will be necessary in order to confirm (or refute) their results. Findings The Importance of Reciprocity In addition to examining the relationship between network structure and depression, the current investigation explored the extent to which social support and perceptions of reciprocity were associated with well-being. As stated prior, inverse relationships between social support and depressive symptomology have commonly been reported by scholars investigating mental health (Falci and McNeely 2009; Haines et al. 2008; Symister and Friend 2003; Laible, Carlo, and Raffaelli 2000). Moreover, in their own research, Falci and McNeely (2009) found that the amount of support received by adolescents mediated the relationship that was observed between small egocentric network size and depression. However, it should be made clear that this support was not

PAGE 66

60 found to mediate the relationship between large egocentric network size and depressive symptomology, or between network density and depression (Falci and McNeely 2009). Accordingly, the present study sought to expand upon the existing literature by exploring the extent to which perceptions of reciprocity mediate the relationships between network characteristics and depression. To reiterate, although the mediating effects of support reciprocity have not been investigated directly by network researchers, there is wealth of empirical evidence which suggests that this line of inquiry is important and should not be overlooked. More specifically, in accordance with the principles of equity theory, numerous studies have demonstrated that perceptions of overbenefiting or u levels of depressive symptomology (Vaananen et al. 2008; Taniguchi and Ura 2002; Bakker et al. 2000; Buunk and Schaufeli 1999). In essence, equity theorists have proposed that giving more than one receives may lead to feelings of resentment, while receiving more than one gives may lead to feelings of guilt or shame (Vaananen et al. 2008). Again, since it has been suggested that the effort which must be exerted to maintain a large network may come to outweigh any benefits or support received from it (Haines et al. 2008), there is reason to suspect that perceptions of equity may mediate the relationship between large egocentric network size and depression. Of further significance, highly cohesive networks are thought to minimize the effort required to ma intain individual relationships and to result in the sharing of social burdens (Forrester and Tashchian 2004). Therefore, perceptions of reciprocity may also mediate the re lationship between network density and depressive symptomology, especially among

PAGE 67

61 females, who are more likely than males to seek out and to give social support (Rosenfield, Lennon, and White 2005 ; Umberson et al. 1996; Frydenberg and Lewis 1993) Ultimately, it was predicted that there would be a negative association between social support and depression among those participating in the current investigation; social support was also expected to mediate the relationship between small egocentric network size and depressive symptomology. Moreover, in accordance with the principles of equity theory, it was predicted that there would be a curvilinear relationship between reciprocity of support and depression: Individuals who perceived underbenefiting or overbenefiting in comparison to their friends were expected to report experiencing higher levels of depression than those in more equitable networks. It was further hypothesized that perceptions of equity would mediate the relationship between large egocentric network size and depressive symptomology, and between network density (among females) and depression. Notably, in the current study, it was not possible to assess the potential mediating effects of either social support or support reciprocity since egocentric network size and density were not found to be significant predictors of depression. However, consistent with expectations, an inverse relationship was found between the amount of support that was received by respondents and depressive symptomology. Also, students who perceived overbenefiting or underbenefiting in comparison to their alters reported experiencing higher levels of depression than those in more equitable networks. Clearly, the most important thing that should be taken from these findings is that support reciprocity has yet to be ruled out as a potential mediator in the relationship between

PAGE 68

62 network structure and depression. Stated more directly, perceptions of equity may still be able to explain the relationships that Falci and McNeely (2009) observed between large egocentric network size and depression and between network density and depressive symptomology. However, as stated prior, additional research will be necessary in order to confirm the findings of these two scholars and to further assess the mediating effects of support reciprocity. Results When Including Family Members and Partners as Alters For the sake of comprehensiveness, network measures (e.g., egocentric network size, network density, and number of female friends) were recalculated to include family members and partners as alters; all analyses were then rerun. 42 As models 13 and 14 illustrate, egocentric network size and curvilinear network size still failed to significantly predict depressive symptomology at the multivariate level. In models 17 and 18, however, a significant inverse relationship was found between network density and depression among female respondents. The magnitude of the association between these two variables was relatively large, and the unstandardized coefficient (Model 17 = -6.95) for network density (among females) remained stable when controlling for egocentric network size (Model 18 = -7.03). Clearly, this result is more in line with findings reported by Falci and McNeely (2009). Standing in contrast to their findings, however, Model 23 indicates that females with large, cohesive networks were especially likely to report experiencing high levels (unstandardized coefficient = 4.90) of depressive symptomology. As stated prior, in their own research, Falci and McNeely (2009) found that there was a reduced risk for 42 Supplementary descriptive and inferential statistics are presented in Appendix B and Appendix C.

PAGE 69

63 depression among female adolescents with large, dense networks. Finally, it should be noted that even when including family members and partners as alters, social support (see models 24 and 25) was found to have a significant, inverse relationship with depressive symptomology at the multivariate level. However, reciprocity of support (see models 2629) was no longer found to be significant predictor of depression. The precise meaning of these results is open to interpretation. On average, respondents nominated 1.8 family members and 0.6 partners as friends. 43 However, it is widely acknowledged that interpersonal relationships with relatives and significant others encompass qualitatively distinct forms of interaction. To elaborate, while research focusing on adults has indicated that that friendship ties are more likely to transfer emotional aid and companionship than any other relation, family members are most commonly relied upon for financial aid and large services such as child care (Wellman and Wortley 1990). In contrast, partners have been found to provide more comprehensive forms of social support (Wellman and Wortley 1990). For the purposes of this discussion, such distinctions may be especially important, as research has demonstrated that many college students rely upon their family members for assistance with food and shelter, college expenses, bills, and other major expenditures (Osgood et al. 2005). Therefore, it is not necessarily surprising, for instance, that support reciprocity loses its ability to predict depressive symptomology when including family members and partners as alters, as it would seem that college students are unlikely to be (or to expect to be) in equitable financial relationships with their families. 43 Again, participants reported an average of 2.2 friends who were not further classified as being family members or partners.

PAGE 70

64 Moreover, although these findings (i.e., those including family members and partners as alters) are certainly interesting, it is somewhat beyond the scope of the current investigation to speculate as to their significance and/or implications. More clearly, the present study was focused on delineating the importance of friendship ties, and there appears to be sufficient theoretical justification for excluding family members and partners from consideration. Although such relations were not directly excluded by Falci and McNeely (2009), their sample was comprised of a series of closed networks (i.e., high schools). Therefore, in their own study, family members and partners were unlikely to have constituted a large number of the alters who were reported by participants. In contrast, these two relations accounted for approximately 52% of all friendship nominations in the current investigation. Specifically, 40% of those nominated were family members, and 12% were partners. This finding is interesting in and of itself. While there is some research which suggests that individuals between the ages of 18 and 30 may come to view their parents as equals (Arnett 2004), studies investigating the extent to which both parents and other family members come to be viewed as friends are notably absent from the literature. Ultimately, this may prove to be a fruitful area for future research. Conclusion The current investigation adds to the limited number of studies which have examined network structure in relation to mental health. This project was somewhat unique, as direct focus was placed on the well-being of United States college students. Although no significant relationships were found between egocentric network characteristics and depressive symptomology, social support and perceptions o f

PAGE 71

65 reciprocity were both found to significantly predict depression at the multivariate level. However, because egocentric network size and density were not found to have significant relationships with depression, it was not possible to assess the potential mediating effects of support reciprocity. Therefore, at least to some extent, perceptions of equity may still explain the relationships that Falci and McNeely (2009) observed between large egocentric network size and depression and between network density and depressive symptomology. As stated prior, additional research will be necessary in order to confirm the findings of these two scholars and to further assess the mediating effects of perceived equity. To reiterate, it is possible that the failure to observe a significant relationship between network structure and depression was related to the fact that an artificial limit was placed on the number of alters who could be reported by individuals participating in the current investigation. In order to address this concern, future studies should consider expanding the number of alters who can be reported by participants; this could be accomplished by assessing friendship using multiple name generators. It is also possible that college students represent a unique population, with distinct characteristics that have not been adequately explored by network researchers focusing on either adolescents or adults. Therefore, it may be prudent for scholars to further explore the relationships between egocentric network characteristics and depression among college students in particular, as relatively few network studies have focused directly on this group. Again, however, it is important to interpret the results of this investigation with caution, as data were collected from a single institution of higher education using non-random sampling. As such, findings are not necessarily representative of the larger student body.

PAGE 72

66 In addition to the limitations that have already been discussed, it should be made clear that because the data used in this investigation are cross-sectional, it is not possible to determine the causality of observed relationships. Stated more clearly, the possibility that depressed college students are highly susceptible to involving themselves in nonreciprocal and non-supportive friendships cannot be ruled out entirely. In order to address this concern, it is suggested that (if at all possible) future studies utilize data collected at multiple points in time. As a final note, the large number of family members and partners who were nominated as friends in the present study should not be overlooked. The precise meaning of friendship among college students has yet to be studied empirically and remains unclear. Ultimately, while the contributions of the current investigation proved to be somewhat limited in scope, it is believed that this project has provided considerable insight and direction for future research.

PAGE 73

67 LIST OF REFERENCES February 17, 2011 (http://www.cpc.unc.edu/projects/addhealth/about). American College Health Associatio National College Health Assessment Spring 2008 Reference Group Data Report Journal of American College Health 57(5):477-488. American Psychiatric Association. 2000. Diagnostic and Statistical Manual of Mental Disorders. Revised 4 th Edition. Washington, DC: American Psychiatric Association. Arnett, Jeffrey J. 2004. Emerging Adulthood: The Winding Road from the Late Teens through the Twenties New York: Oxford University Press. -----American Psychologist 55(5):469-480. Bailey, Mary K., Jaclene A. Zauszniewski, Marjorie M. Heinzer, and Marion HemstromJournal of Child and Adolescent Psychiatric Nursing 20(2):86-95. Bakker, Arnold B., Wilmar B. Schaufeli, Evangelia Demerouti, Peter Janssen, Rennee Van Der Hulst, and Janneke Brouwe

PAGE 74

68 Anxiety, Stress, and Coping 13:247-268. American Journal of Public Hea lth 94(1):89-95. The Journal of Psychology 143(2):193-205. Social Networks 6:293-339. -----Social Networks 9(4):311332. Survey Network Item Response Cat Social Networks 8:387-396. Relationships: An Evolutionary Perspective on Its Importance for Health and WellEuropean Review of Social Psychology 10:259-291. Cannuscio, Carolyn C., Graham A. Colditz, Eric B. Rimm, Lisa F. Berkman, Camara P. Social Science & Medicine 58:1247-1256. Cecil, Heather, Melinda A. Stanley, Patricia G. Carrion, and Alan Swann. 1995. Journal of Clinical Psychology 51(5):593-602.

PAGE 75

69 Coates, Deborah. 1985. -Concept Measures and Social Network Characteristics for Black Adolescents. Journal of Early Adolescence 5:319338. Psychological Bulletin 98:310 357. Coleman James. 1990. Foundations of Social Theory. Cambridge, MA: Harvard University Press. -Body Center; Retrieved March 5, 2011 (http://pmbcii.psy.cmu.edu/core_c/Depression_3_2006.pdf ). Journal of Clinical Psychology 47(6):756-761. Durkheim, Emile. 1897/2006. On Suicide New York: Penguin Books. Edles, Laura D. and Scott Appelrouth. 2005. Sociological Theory in the Classical Era Thousand Oaks, CA: Pine Forge Press. Journal of Youth and Adolescence 23:237-249. Falci, Christina Social Forces 87(4):2031-2062. Field, Tiffany, Miguel Diego, and Christopher Sanders. 2001. and Risk Factors. Adolescence 36:491498.

PAGE 76

70 Journal of Gerontology 61B(1):25-32. Fischer, Claude. 1982. To Dwell among Friends University of Chicago Press, Chicago. Fischer, Claude and Susan Phillips. 1982. People with Small Networks. Pp. 21-39 in Loneliness: A Source Book of Current Theory, Research and Therapy edited by Letitia Peplau and Daniel Perlman. New York: Wiley. to Adulthood, Retrieved February 20, 2011 (http://www.transad.pop.upenn.edu/ downloads/fitzpatrick.pdf). Psychological Reports 95(1):207-214. Annual Review of Sociology 30:409-25. Journal of Adolescence 16:253-266. Social Psychiatry & Psychiatric Epidemiology 38(5):229-237.

PAGE 77

71 Gest, Scott D., Sandra A. Graham -Bermann, and Willard W. Hartup. 2001. Experience: Common and Unique Features of Number of Friendships, Social Network Centrality, and Sociometric Status. Social Development 10(1):23-40. Glover, Troy D. and Diana C Journal of Leisure Research 40(2):208-230. So Leisure Sciences 27:75-92. American Journal of Public Health 89(10):1522-1528. Sex Roles 59:164-175. -----e Structural Contexts of the Support Processes: Social Networks, Social Statuses, 92 in Social Networks and Health edited by Judith A. Levy and Bernice A. Pescosolido. New York: JAI Press. Hair, Joseph F., Ronald L. Tatham, Rolph Anderson, and William C. Black. 2006. Multivariate Data Analysis Revised 6 th Edition. Upper Saddle River, NJ: Prentice Hall.

PAGE 78

72 Emotional WellResearch on Aging 22(6):599-629. Hatfield, Elaine, Jane Traupmann, Susan Sprecher, Mary Utne, and Judy Hay. 1985. -117 in Compatible and Incompatible Relationships edited by William Ickes. New York: SpringerVerlag. American Journal of Sociology 106(4):1013-1057. rment in Journal of American College Health 45(2):59-64. Hockenbury, Don H. and Sandra E. Hockenbury. 2003. Psychology New York: Worth Publishers. Science 241:540-545. Hyun, Myung-Sun, Kyoung Nam, Hee Sung Kang, and William M. Reynolds. 2009. Second Edition: Initial Validation of Journal of Advanced Nursing 65(3):642-651. Illi Psychology; Retrieved March 6, 2011 (http://psychology.illinoisstate.edu/psy138/ resources/spss/spss3.html). Kadushin, Charles, 1983. Mental Health and the Interpersonal Environment: Reexamination of Some Effects of Social Structure and Mental Health. American Sociological Review 48:188198.

PAGE 79

73 MSPSS: Fac Journal of Community Psychology 19:150-160. Kelly, William E., Kathryn E. Kelly, Franklin C. Brown, and Hillary B. Kelly. 1999. i-Cultural College Student Journal 33(1):72-76. Kirsch, Irving, Brett J. Deacon, Tania B. Huedo-Medina, Alan Scoboria, Thomas J. A MetaPLoS Medicine 5(2):260-268. Kromrey, Jeffrey D. and Lynn Fostererated Educational and Psychological Measurement 58(1):42-67. Social Science & Medicine 67:228-237. Laible, Deborah J., Gustavo Carlo, Journal of Youth and Adolescence 29(1):45-59. -258 in A Handbook for the Study of Mental Health: Social Contexts, Theories, and Systems edited by A.V. Horwitz and T.L Scheid. New York: Cambridge University Press.

PAGE 80

74 Lin, Nan, X Mood: Journal of Health and Social Behavior 40:344359. Status and Depression: The Role of Occupations Involving Direction, Control, American Journal of Sociology 98(6):1351-1387. Lorant, V., D. Deliege, W. Eaton, A. Robert, P. Philippot, and M. Ansseau. 2003. ession: A MetaAmerican Journal of Epidemiology 157(2):98-112. Journal of American College Health 47(3):135-138. Marin, A Field Methods 19(2):163-193. Depr Journal of Health and Social Behavior 41(2):162-176. Monsour, Michael. 2002. Women and Men as Friends: Relationships across the Life Span in the 21 st Century. Mahwah, NJ: Lawrence Erlbaum Associates. National Center for Inju

PAGE 81

75 Prevention, Retrieved February 17, 2010 (http://webappa.cdc.gov/cgibin/broker.exe). of Ethnic Differences between Asian American and White Journal of Abnormal Psychology 106(1):52-60. Osgood, D.W., E.M. Foster, Constance Flanagan, and Gretchen R. Ruth. 2005. On Your Own without a Net: The Transition to Adulthood for Vulnerable Populations Chicago, IL: The University of Chicago Press. Health, Illness, Disease, and Healing: The Accepting Present, the Forgotten Past, Advances in Medical Sociology 8:3-25. Prescott, Carol A., John J. McArdle; Earl S. Hishinuma, Ronald C. Johnson, Robin H. Miyamoto, Naleen N. Andrade, Jeanne L. Edman, George K. Makini, Linda B. Depression and Dysthymia from CES-D Scores among Ethnic Minority Journal of the American Academy of Child and Adolescent Psychiatry 37(5):495-503. -D Scale: A Self-Report Depression Scale for Applied Psychological Measurement 1(3):385-401.

PAGE 82

76 Health: Self-Salience and the Emergence of Internalizing and Externalizing Journal of Health and Social Behavior 46:323-340. Retrieved March 5, 2011 (http://mbcalyn.wordpress.com/2011/02/16/too-manyfacebook-friends-causes-stress-webpronews). Adolescen Journal of Consulting and Clinical Psychology 61(1):121-127. Social Psychology Quarterly 72:384-402. emiologic Studies Depression Scale (CESMonkey Hill Health Communications; Retrieved March 5, 2011 (http://www. scireproject.com/outcome-measures/center-epidemiological-studies-depressionscaleces -d). Spirito, Anthony, Sylvia Valeri, Julie Boerger Journal of Clinical Child and Adolescent Psychology 32(2):284-289. in Resistance to Developmental Psychology 43(6):1531-1543. Problematic Support on Optimism and Depression in Chronic Illness: A Prospective Study Evaluating SelfHealth Psychology 22(2):123-129.

PAGE 83

77 among Elementary School and High School Students. Japanese Psychological Research 44(4):247-253. Turner, Jonathan H. and Jan E. Stets. 2005. The Sociology of Emotions New York: Cambridge University Press. American Sociological Review 63:94-110. Social Science Research 34:484-510. Umberson, Debra, Meichu D. Chen, James S. House, Kristine Hopkins, and Ellen Slaten. 19 -Being: Are Men American Sociological Review 61:837-857. --Social and Economic Characteristics of Students: October Statistics Division, Education and Social Stratification Branch; Retrieved February 19, 2010 (http://www.census.gov/population/www/socdemo/school/ Cps2008.html). Vaananen, Ari, Bram P. Bunk, Mika Kivimaki, Jaana Pentti, and Jussi Vahtera. 2005. -Term Health Effects of Journal of Personality and Social Psychology 89(2):176-193.

PAGE 84

78 Vaananen, Ari, Abraham P. Buunk, Mika Kivimaki, Jussi Vahtera, and Markku Social Science and Medicine 67:1907-1916. among Dutch Teachers: Validation of Reciprocity Indices and Their Relation to Stress and WellWork and Stress 15(3):191-213. Walker, John M., Paul M. Wasserman, Sociological Methods and Research 22(1):71-98. Wasserman, Stanley and Katherine Faust. 1994. Social Network Analysis: Methods and Applications New York: Cambridge University Press. -61 in Social Structures: A Network Approach edited by Barry Wellman and S.D. Berkowitz. Cambridge: Cambridge University Press. -----Field Methods 19(2):111-115. American Journal of Sociology 96(3):558588. Zucker, Kenneth J., Debra N. Wilson-Smith, Janice A. Kurita, and Anita Stern. 1995. Sex Roles 33(11/12):703-725.

PAGE 85

79 APPENDICES

PAGE 86

80 Appendix A: Survey Questions Instructions: Please answer the following questions about yourself. 1. Age ____ 2. Gender 3. Race/Ethnicity (Please indicate which of the following categories you MOST identify with) White Asian Native American Indian or Alaska Native Hawaiian or Other Pacific Islander Other 4. Current Relationship Status Involved in a Steady Relationship/Engaged/Married Single/Casually Dating/Divorced/Widowed/Separated 5. What is the highest level of education completed by either of your parents? Grade School or Less Some High School High School Diploma or GED Some Post-Graduate Education or Professional Degree (M.A./PhD/MBA/MD/etc.) 6. How far away do you live from the university? 0 miles (I live on campus) 0.1 5 miles 5.1 10 miles 10.1 20 miles 20.1 50 miles More than 50 miles

PAGE 87

81 Appendix A: Survey Questions (Continued) 7. Current Class Standing freshman sophomore junior senior other 8. High School GPA ____ Instructions: The following items will list some of the things that different people value. Some people say these things are very important to them. Other people say they are not so important. Please explain how important each of these things is to you. 9. Financial security is . one of the most important values you hold very important somewhat important not too important not at all important 10. Being married is . one of the most important values you hold very important somewhat important not too important not at all important 11. Having Children is . one of the most important values you hold very important somewhat important not too important not at all important 12. Having faith in God is . one of the most important values you hold very important somewhat important not too important not at all important

PAGE 88

82 Appendix A: Survey Questions (Continued) 13. Having nice things is . one of the most important values you hold very important somewhat important not too important not at all important 14. Having a fulfilling job is . one of the most important values you hold very important somewhat important not too important not at all important 15. Being cultured is . one of the most important values you hold very important somewhat important not too important not at all important 16. Being self-sufficient and not having to depend on others is . one of the most important values you hold very important somewhat important not too important not at all important Instructions: For each of the following, please indicate how well the description applies to you. 17. I am a kind person. a very good description a good description a fair description not a very good description not a very good description at all 18. I am a dependable person. a very good description a good description a fair description not a very good description not a very good description at all

PAGE 89

83 Appendix A: Survey Questions (Continued) Instructions: Indicate your agreement with the following statements. 19. strongly agree 20. Overall, I expect more good things to happen to me than bad. strongly disagree Instructions: Please answer the following questions to the best of your ability. 21. Would you say that your own health, in general, is excellent, good, fair, or poor? 22. How often do you have problems paying for things that you need (for example: food, clothing, or rent)? Never Rarely Sometimes Always

PAGE 90

84 Appendix A: Survey Questions (Continued) Instructions: Consider who the closest and most important friends in your life are. Put the initials of these people, maximum 5, in the following blanks. Then select the proper alternative suited to these people in the questions that follow. 23. Person 1 (Initials) _____ 24. Person 2 (Initials) _____ 25. Person 3 (Initials) _____ 26. Person 4 (Initials) _____ 27. Person 5 (Initials) _____ 28. Gender (Is this individual male or female?) _____ _____ _____ _____ _____ 29. Is this person currently enrolled as a student at your school? _____ _____ _____ _____ _____ 30. In addition to being a friend, would you characterize this individual as being a . _____ _____ _____ _____ _____ ither of these

PAGE 91

85 Appendix A: Survey Questions (Continued) 31. In your relationship with this person, which of you gives or receives more support and help (for example: emotional support, companionship, information, services, or financial help)? How would you describe your relationship in this respect? _____ _____ and help more than I receive _____ ore than I give _____ _____ s I give

PAGE 92

86 Appendix A: Survey Questions (Continued) 32. Please describe the relationship between each pair of your friends. _____ and _____ other as I am to them _____ and _____ _____ and _____ _____ and _____ _____ and _____ ) _____ and _____ _____ and _____ _____ and _____ _____ and _____ _____ and _____

PAGE 93

87 Appendix A: Survey Questions (Continued) Instructions: We are interested in how you feel about the following statements. Read each statement carefully. Indicate how you feel about each statement. 33. I feel that . My friends really try to help me. I can count on my friends when things go wrong. I have friends with whom I can share my joys and my sorrows. mildly disagree I can talk about my problems with my friends. ly agree

PAGE 94

88 Appendix A: Survey Questions (Continued) Instructions: Below is a list of the ways you might have felt or behaved. Please indicate how often you have felt this way during the past week. 34. During the past week . I was bothered by ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I did not feel like eating; my appetite was poor. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I felt that I could not shake off the blues even with help from my family or friends. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I felt that I was just as good as other people. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I had trouble keeping my mind on what I was doing. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I felt depressed. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days)

PAGE 95

89 Appendix A: Survey Questions (Continued) I felt that everything I did was an effort. of the time (less than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I felt hopeful about the future. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I thought my life had been a failure. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I felt fearful. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) My sleep was restless. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I was happy. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I talked less than usual. rarely or none of the time (less than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days)

PAGE 96

90 Appendix A: Survey Questions (Continued) I felt lonely r none of the time (less than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) People were unfriendly. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I enjoyed life. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I had crying spells. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I felt sad. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) I felt that people dislike me. ess than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days) f the time (less than 1 day) ome or a little of the time (1-2 days) ccasionally or a moderate amount of the time (3-4 days) or all of the time (5-7 days)

PAGE 97

91 Appendix B: Supplementary Descriptive Statistics TABLE A1: Descriptive Statistics for Individual CES-D Items n % I Was Bothered by Things That Usually Don't Bother Me . Rarely or None of the Time (Less than 1 Day) 307 46.7 Some or a Little of the Time (1 2 Days) 225 34.2 Occasionally or a Moderate Amount of the Time (3 4 Days) 103 15.7 Most or All of the Time (5 7 Days) 22 3.4 I Did Not Feel like Eating; My Appetite Was Poor . Rarely or None of the Time (Less than 1 Day) 391 59.5 Some or a Little of the Time (1 2 Days) 152 23.1 Occasionally or a Moderate Amount of the Time (3 4 Days) 86 13.1 Most or All of the Time (5 7 Days) 28 4.3 I Felt That I Could Not Shake Off the Blues . Rarely or None of the Time (Less than 1 Day) 393 59.8 Some or a Little of the Time (1 2 Days) 151 23.0 Occasionally or a Moderate Amount of the Time (3 4 Days) 83 12.6 Most or All of the Time (5 7 Days) 30 4.6 I Felt That I Was Just as Good as Other People . Most or All of the Time (5 7 Days) 251 38.2 Occasionally or a Moderate Amount of the Time (3 4 Days) 234 35.6 Some or a Little of the Time (1 2 Days) 110 16.7 Rarely or None of the Time (Less than 1 Day) 62 9.5 I Had Trouble Keeping My Mind on What I Was Doing . Rarely or None of the Time (Less than 1 Day) 184 28.0 Some or a Little of the Time (1 2 Days) 248 37.7 Occasionally or a Moderate Amount of the Time (3 4 Days) 161 24.5 Most or All of the Time (5 7 Days) 64 9.8 I Felt Depressed . Rarely or None of the Time (Less than 1 Day) 378 57.5 Some or a Little of the Time (1 2 Days) 173 26.3 Occasionally or a Moderate Amount of the Time (3 4 Days) 80 12.2 Most or All of the Time (5 7 Days) 26 4.0

PAGE 98

92 Appendix B: Supplementary Descriptive Statistics (Continued) TABLE A1: Descriptive Statistics for Individual CES-D Items (Continued) n % I Felt That Everything I Did Was an Effort . Rarely or None of the Time (Less than 1 Day) 178 27.1 Some or a Little of the Time (1 2 Days) 206 31.4 Occasionally or a Moderate Amount of the Time (3 4 Days) 181 27.5 Most or All of the Time (5 7 Days) 92 14.0 I felt Hopeful about the Future . Most or All of the Time (5 7 Days) 255 38.9 Occasionally or a Moderate Amount of the Time (3 4 Days) 249 38.0 Some or a Little of the Time (1 2 Days) 116 17.7 Rarely or None of the Time (Less than 1 Day) 36 5.4 I Though My Life Had Been a Failure . Rarely or None of the Time (Less than 1 Day) 538 82.0 Some or a Little of the Time (1 2 Days) 67 10.2 Occasionally or a Moderate Amount of the Time (3 4 Days) 41 6.3 Most or All of the Time (5 7 Days) 10 1.5 I Felt Fearful . Rarely or None of the Time (Less than 1 Day) 368 56.1 Some or a Little of the Time (1 2 Days) 194 29.6 Occasionally or a Moderate Amount of the Time (3 4 Days) 71 10.8 Most or All of the Time (5 7 Days) 23 3.5 My Sleep Was Restless . Rarely or None of the Time (Less than 1 Day) 284 43.3 Some or a Little of the Time (1 2 Days) 196 29.9 Occasionally or a Moderate Amount of the Time (3 4 Days) 103 15.7 Most or All of the Time (5 7 Days) 73 11.1 I Was Happy . Most or All of the Time (5 7 Days) 283 43.1 Occasionally or a Moderate Amount of the Time (3 4 Days) 271 41.3 Some or a Little of the Time (1 2 Days) 87 13.3 Rarely or None of the Time (Less than 1 Day) 15 2.3

PAGE 99

93 Appendix B: Supplementary Descriptive Statistics (Continued) TABLE A1: Descriptive Statistics for Individual CES-D Items (Continued) n % I Talked Less than Usual . Rarely or None of the Time (Less than 1 Da y) 333 50.7 Some or a Little of the Time (1 2 Days) 214 32.6 Occasionally or a Moderate Amount of the Time (3 4 Days) 83 12.7 Most or All of the Time (5 7 Days) 26 4.0 I Felt Lonely . Rarely or None of the Time (Less than 1 Day) 319 48.6 Some or a Little of the Time (1 2 Days) 188 28.7 Occasionally or a Moderate Amount of the Time (3 4 Days) 105 16.0 Most or All of the Time (5 7 Days) 44 6.7 People Were Unfriendly . Rarely or None of the Time (Less than 1 Day) 420 64.0 Some or a Little of the Time (1 2 Days) 177 27.0 Occasionally or a Moderate Amount of the Time (3 4 Days) 45 6.9 Most or All of the Time (5 7 Days) 14 2.1 I Enjoyed Life . Most or All of the Time (5 7 Days) 305 46.5 Occasionally or a Moderate Amount of the Time (3 4 Days) 227 34.6 Some or a Little of the Time (1 2 Days) 104 15.9 Rarely or None of the Time (Less than 1 Day) 20 3.0 I Had Crying Spells . Rarely or None of the Time (Less than 1 Day) 480 73.2 Some or a Little of the Time (1 2 Days) 102 15.5 Occasionally or a Moderate Amount of the Time (3 4 Days) 52 7.9 Most or All of the Time (5 7 Days) 22 3.4 I Felt Sad . Rarely or None of the Time (Less than 1 Day) 318 48.4 Some or a Little of the Time (1 2 Days) 228 34.8 Occasionally or a Moderate Amount of the Time (3 4 Days) 80 12.2 Most or All of the Time (5 7 Days) 30 4.6

PAGE 100

94 Appendix B: Supplementary Descriptive Statistics (Continued) TABLE A1: Descriptive Statistics for Individual CES-D Items (Continued) n % I Felt That People Disliked Me . Rarely or None of the Time (Less than 1 Day) 423 64.6 Some or a Little of the Time (1 2 Days) 159 24.2 Occasionally or a Moderate Amount of the Time (3 4 Days) 58 8.8 Most or All of the Time (5 7 Days) 16 2.4 I Could Not Get "Going" . Rarely or None of the Time (Less than 1 Day) 361 55.1 Some or a Little of the Time (1 2 Days) 191 29.1 Occasionally or a Moderate Amount of the Time (3 4 Days) 77 11.7 Most or All of the Time (5 7 Days) 27 4.1 Notes: (N = 671)

PAGE 101

95 Appendix B: Supplementary Descriptive Statistics (Continued) TABLE A2: Descriptive Statistics for Individual MSPSS Items n % My Friends Really Try to Help Me . Very Strongly Disagree 62 9.5 Strongly Disagree 45 6.9 Mildly Disagree 18 2.7 Neutral 38 5.8 Mildly Agree 82 12.5 Strongly Agree 237 36.2 Very Strongly Agree 173 26.4 I Can Count on My Friends When Things Go Wrong . Very Strongly Disagree 61 9.3 Strongly Disagree 45 6.9 Mildly Disagree 20 3.1 Neutral 31 4.7 Mildly Agree 83 12.7 Strongly Agree 206 31.5 Very Strongly Agree 208 31.8 I Have Friends with Whom I Can Share My Joys and My Sorrows . Very Strongly Disagree 75 11.5 Strongly Disagree 36 5.5 Mildly Disagree 12 1.8 Neutral 22 3.4 Mildly Agree 54 8.3 Strongly Agree 179 27.4 Very Strongly Agree 276 42.1

PAGE 102

96 Appendix B: Supplementary Descriptive Statistics (Continued) TABLE A2: Descriptive Statistics for Individual MSPSS Items (Continued) n % I Can Talk about My Problems with My Friends . Very Strongly Disagree 72 11.0 Strongly Disagree 37 5.7 Mildly Disagree 19 2.9 Neutral 34 5.2 Mildly Agree 58 8.9 Strongly Agree 184 28.1 Very Strongly Agree 250 38.2 Notes: (N = 671)

PAGE 103

97 Appendix B: Supplementary Descriptive Statistics (Continued) TABLE A3: Descriptive Statistics for GSS Survey Items n % Financial Security is . Not at All Important 3 0.5 Not Too Important 6 0.9 Somewhat Important 79 11.9 Very Important 414 62.1 One of the Most Important Values You Hold 164 24.6 Being Married is . Not at All Important 32 4.8 Not Too Important 74 11.1 Somewhat Important 170 25.6 Very Important 227 34.1 One of the Most Important Values You Hold 162 24.4 Having Children is . Not at All Important 37 5.6 Not Too Important 70 10.6 Somewhat Important 161 24.3 Very Important 212 32.0 One of the Most Important Values You Hold 182 27.5 Having Faith in God is . Not at All Important 82 12.4 Not Too Important 81 12.2 Somewhat Important 115 17.3 Very Important 118 17.8 One of the Most Important Values You Hold 267 40.3 Having Nice Things is . Not at All Important 21 3.2 Not Too Important 139 20.9 Somewhat Important 370 55.6 Very Important 116 17.4 One of the Most Important Values You Hold 20 2.9

PAGE 104

98 Appendix B: Supplementary Descriptive Statistics (Continued) TABLE A3: Descriptive Statistics for GSS Survey Items (Continued) n % Having a Fulfilling Job is . Not at All Important 3 0.5 Not Too Important 4 0.6 Somewhat Important 67 10.1 Very Important 336 50.5 One of the Most Important Values You Hold 255 38.3 Being Cultured is . Not at All Important 3 0.5 Not Too Important 49 7.4 Somewhat Important 210 31.6 Very Important 274 41.3 One of the Most Important Values You Hold 128 19.2 Being Self Sufficient and Not Having to Depend on Others is . Not at All Important 1 0.2 Not Too Important 14 2.1 Somewhat Important 62 9.3 Very Important 279 41.9 One of the Most Important Values You Hold 310 46.5 I am a Kind Person . Not a Very Good Description at All 0 0.0 Not a Very Good Description 6 0.9 A Fair Description 51 7.7 A Good Description 283 42.7 A Very Good Description 323 48.7 I am a Dependable Person . Not a Very Good Description at All 3 0.5 Not a Very Good Description 15 2.3 A Fair Description 51 7.7 A Good Description 274 41.3 A Very Good Description 320 48.2

PAGE 105

99 Appendix B: Supplementary Descriptive Statistics (Continued) TABLE A3: Descriptive Statistics for GSS Survey Items (Continued) n % I'm Always Optimistic about My Future . Strongly Disagree 11 1.7 Disagree 95 14.3 Don't Know 31 4.7 Agree 382 57.6 Strongly Agree 144 21.7 I Expect More Good Things to Happen to Me than Bad . Strongly Disagree 13 1.9 Disagree 76 11.4 Don't Know 47 7.1 Agree 341 51.4 Strongly Agree 187 28.2 Would You Say That Your Own Health, in General, is . Poor 12 1.8 Fair 94 14.2 Good 378 57.0 Excellent 179 27.0 Don't Know 0 0.0 Notes: (N = 671)

PAGE 106

100 Appendix B: Supplementary Descriptive Statistics (Continued) TABLE A4: Additional Personal Network Statistics Variable Mean SD Min Max Number of Friends Enrolled at University 0.6 1 .0 0 5 (Excluding Family Members and Partners) Total # of Family Members Reported 1.8 1.6 0 5 Total # of Partners Reported 0.6 0.6 0 5 Notes: (N = 671)

PAGE 107

101 Appendix B: Supplementary Descriptive Statistics (Continued)

PAGE 108

102 Appendix B: Supplementary Descriptive Statistics (Continued) TABLE A6: Network Statistics Including Family Members and Partners as Alters Variable Mean SD Min Max Egocentric Network Size 4.6 0.9 0 5 (Including Family Members and Partners) Network Density 0.6 0.2 0.0 1.0 Perceived Reciprocity of Support 0.0 0.3 1.0 1.0 (Hatfield Global Reciprocity Measure) Notes: (N = 671)

PAGE 109

103 Appendix C: Supplementary Inferential Statistics

PAGE 110

104 Appendix C: Supplementary Inferential Statistics (Continued)

PAGE 111

105 Appendix C: Supplementary Inferential Statistics (Continued)

PAGE 112

106 Appendix C: Supplementary Inferential Statistics (Continued)

PAGE 113

107 Appendix C: Supplementary Inferential Statistics (Continued)

PAGE 114

108 Appendix C: Supplementary Inferential Statistics (Continued)

PAGE 115

109 Appendix C: Supplementary Inferential Statistics (Continued)

PAGE 116

110 Appendix C: Supplementary Inferential Statistics (Continued)

PAGE 117

111 Appendix C: Supplementary Inferential Statistics (Continued)

PAGE 118

112 Appendix C: Supplementary Inferential Statistics (Continued)


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 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2011 flu ob 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004906
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Huff, Ryan Francis.
0 245
Friendship networks, perceived reciprocity of support, and depression
h [electronic resource] /
by Ryan Francis Huff.
260
[Tampa, Fla] :
b University of South Florida,
2011.
500
Title from PDF of title page.
Document formatted into pages; contains 118 pages.
502
Thesis
(M.A.)--University of South Florida, 2011.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
3 520
ABSTRACT: Using social network analysis as a theoretical framework, the current study examined the associations between self-reported egocentric network characteristics and depression among a sample of United States college students. It is important to understand factors related to depression among this population due to the severity of its potential outcomes (e.g., suicide and interpersonal problems at school). Drawing inspiration from a recent study conducted by Christina Falci and Clea McNeely (2009), the current investigation used OLS regression to test for both linear and curvilinear relationships between egocentric network size and depression. Potential interactions between network size, density, and gender were also explored. As an additional line of inquiry, this project examined whether or not (and to what extent) perceptions of reciprocity mediate the relationships between network characteristics and depression. Data were collected using an online survey, which was proctored to students enrolled in three large undergraduate sociology courses during the fall 2010 semester. In contrast to findings reported by Falci and McNeely (2009), no significant relationships were observed between network characteristics and mental health. However, support reciprocity was found to be a significant predictor of depression at the multivariate level. Additional research will be necessary in order to confirm (or refute) the results of Falci and McNeely (2009) and to further assess the mediating effects of perceived equity.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor:
Skvoretz, John .
653
Equity Theory
Friendship
Mental Health
Social Network Analysis
Social Support
690
Dissertations, Academic
z USF
x Sociology Social Psychology Mental Health
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4906