xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001709533
007 cr mnu|||uuuuu
008 060517s2005 flua sbm s000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001367
LeVasseur, Michelle Edington.
Automatic attention to aggression cues and alcohol cues using a dichotic listening task and a parafoveal visual task
h [electronic resource] /
by Michelle Edington LeVasseur.
[Tampa, Fla.] :
b University of South Florida,
Thesis (M.A.)--University of South Florida, 2005.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 148 pages.
ABSTRACT: Ongoing investigations of drunken aggression tend to focus on 1) situational cues, and 2) individual variables such as personality traits. This study investigated the hypothesis that an undergraduates attention would be pulled toward a nonconscious presentation of aggression stimuli, especially in the presence of alcohol cues, and especially if he or she was high on trait anger [as measured using the State Trait Anger Expression Inventory (STAXI); Spielberger, 1988] and had high expectancies for behaving aggressively while drinking alcohol [as measured using the Expectancy Questionnaire for Alcohol and Aggression Lo Dose (EQAAL); Epps, Hunter, LeVasseur, Steinberg, and Hancock, unpublished manuscript]. Seventy-nine of the participants who completed questionnaires also completed one of the two computer tasks (adapted from John Bargh and associates) weeks later in either the Barroom or the Cleanroom.Attention to HiAggression words (as measured by reaction time or error rate difference scores) was significantly higher than attention to NonAggression words using the parafoveal visual task, with observed power at 1. No significant differences were found using the dichotic listening task. Additionally, there was a significant three-way interaction (Word Type X Setting X Angry Temperament) when participants where blocked according to high vs. low angry temperament scores. Follow-up analyses as well as regression analyses for the specific hypothesis provided mixed results. Individuals lower on angry temperament tended to demonstrate higher levels of attentional interference for aggression words, but only in the presence of alcohol cues. Conversely, individuals higher on angry temperament evidenced higher levels of attentional interference, but only in the absence of alcohol cues. It appears that the relationships among these variables are by no means straightforward.Studies that include an opportunity to aggress behaviorally may shed more light on whether ones level of attentional interference and self-reported personality traits can be combined to predict aggression in the presence of alcohol cues. The parafoveal visual task is recommended as the methodology of choice for these future studies.
Adviser: James Epps.
t USF Electronic Theses and Dissertations.
Automatic Attention to Aggre ssion Cues and Alcohol Cues Using a Dichotic Listening Task and a Parafoveal Visual Task by Michelle Edington LeVasseur A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Psychology College of Arts and Sciences University of South Florida Major Professor: James Epps, Ph.D. Cynthia Cimino, Ph.D. Douglas Nelson, Ph.D. Date of Approval: August 26, 2005 Keywords: Automaticity, Trait Anger, Angr y Temperament, Expectancies, Alcohol Myopia Copyright 2005, Michelle Edington LeVasseur
Dedication This thesis is dedicated primarily to my sweetie, James B. LeVasseur. His dedication to walking my path with me was as valuable as his unshakeable faith in my ability to succeed and his keen perspective. These qualities in him made it possible for me to pursue a dream while retaining vital connections with my family and closest friends, while continuing my personal growth, and while keeping my wits about me. This thesis is also dedicated to my children, Aaron Edington and Cristofer Edington, who sometimes had to go without in my pursuit of higher education. I hope that in the final analysis they will be inspired by my tenac ity to also reach beyond the simple or the known. Finally, I dedicate this thesis to my faithful champion, Andrea Edington Shank, who has always understood me, and loved me just the same!
Acknowledgments I wish to thank Dr. James Epps for hi s unequivocal support on numerous projects over the last decade. His flexibility and guidan ce have allowed me to grow as a scientist, clinician, and academician. I would also li ke to thank Dr. Cynthia Cimino and Dr. Douglas Nelson for serving on my committee and helping this project come to fruition. Christine Vaughan, as a fellow graduate stude nt and dear friend, provided her special brand of enthusiasm, humor, and insight th roughout this process. My son, Cristofer Edington, demanded special recognition for help ing me to enter data and organize my materials. Finally, I wish to thank my pare nts, Joseph Tingley and Sylviane Whitehurst, and my brother Joseph Tingley, Jr., for allowing me to broaden my role as a daughter and a sister and for resisting the ur ge to say, Â“ArenÂ’t you done yet?Â”
i Table of Contents List of Tables List of Figures Abstract Introduction Overview Aggression Definitional Issues The Origins of Angry Aggression Frustration-Aggression Hypothesis Social Learning Theory Cognitive Neo-Associationism Social Information Processing Models Knowledge Structures and Network Models Assumptions Generated from Models of Aggression Selective Attention Salience vs. Accessibility Early vs. Late Selection Automatic vs. Controlled Attention Dichotic Listening Tasks for the Measurement of Attention Parafoveal Visual Tasks fo r the Measurement of Attention Limitations of Dichotic Listen ing and Parafoveal Visual Tasks Alcohol and Cognition Physiological and Expectancy Effects Alcohol Myopia Alcohol Cues Alcohol and Aggression Situational Variables Gender Individual Difference Variables Attentional Effects and Automaticity The Current Study Hypotheses iv v vi 1 2 2 2 5 5 6 7 9 12 15 16 17 18 19 23 26 30 30 31 34 38 41 42 43 43 48 50 50
ii Method Design Condition Assignment Power Participants Phase I Materials State-Trait Anger Expression Inventory Expectancy Questionnaire for Alcohol and AggressionÂ—Low Dose Phase II Materials Positive and Negative Affect Schedule Dichotic Listening Task Word Pairs Parafoveal Visual Task Experimental Settings Recognition Task Procedure Results Descriptives Preliminary Analyses Reaction Time for Word Type Error Rate for Word Type Implications of Differenti al Responding Across Word Type Analyses of Alcohol Cue Moderation Hypothesis 1 HiAgg Word Type, Trait Char acteristics, and Alcohol Cue Moderation Difference Scores, Trait Char acteristics, and Alcohol Cue Moderation Hypothesis 2 HiAgg Word Type, Alcohol-A ggression Expectancies, and Alcohol Cue Moderation Difference Scores, Alcohol-Aggression Expectancies, and Alcohol Cue Moderation Hypothesis 3 Alcohol-Aggression Expectancies After Controlling for Trait Anger Supplemental Results PANAS Scores at Time 1 and Time 2 Recognition Task Discussion Comparison of the Dichotic Listenin g Task vs. the Parafoveal Visual Task 52 52 52 52 53 54 54 58 60 60 61 61 62 63 64 64 68 68 70 70 72 73 74 74 75 75 76 76 78 79 79 81 81 82 84 84
iii Error Rate vs. Reaction Time HiAgg Means Vs. HiAggÂ—NonAgg Difference Scores Power for the Regression Analyses Hypothesis 1 Hypothesis 2 Hypothesis 3 Alcohol Cues Limitations of the Current Study and Future Directions References Footnote Appendices Appendix A: Request fo r Further Participation Appendix B: State Trait A nger Expression Inventory Appendix C: Buss Perry Aggression Questionnaire Appendix D: Expectancy Questionnaire for Alcohol and AggressionÂ—Low Dose Appendix E: Alcohol-Aggression Items from Various Measures Appendix F: Demographics Questionnaire Appendix G: Positive a nd Negative Affect Schedule Appendix H: Aggressive and Nonaggressive Word Stimuli Appendix I: Recognition Task Appendix J: Comprehensive Drinker Profile Appendix K: Debrief Statement for Phase I Appendix L: Debriefing Statement for Phase II 84 85 85 86 88 89 90 90 94 108 109 110 111 114 117 119 124 125 126 131 132 136 137 About the Author End Page
iv List of Tables Table 1. Means, Standard Deviations, and Intercorrelations for RT HiAggÂ—NonAgg Difference Scores and Trait Anger Predictor Variables for PVT Completed in the Presence of Alcohol Cues (N = 22) Table 2. Means, Standard Deviations, and Intercorrelations for RT HiAggÂ—NonAgg Difference Scores and EQAAL Predictor Variables for PVT Completed in the Presence of Alcohol Cues (N=22) Table 3. Regression Model Pred icting Attentional Interference (HiAggÂ—NonAgg Difference Scores ) from Aggression Stimuli Presented Parafoveally in the Presence of Alcohol Cues 68 69 81
v List of Figures Figure 1. RT as a function of task type X word type. Figure 2. PVT RT Mean s as a function of setting X word type. Figure 3. Error rate as a function of task type X word type. 71 71 72
vi Automatic Attention to Aggression Cues and Alcohol Cues Using a Dichotic Listening Task a nd a Parafoveal Visual Task Michelle Edington LeVasseur ABSTRACT Ongoing investigations of drunken aggressi on tend to focus on 1) situational cues, and 2) individual variables such as persona lity traits. This study investigated the hypothesis that an undergraduateÂ’s attenti on would be pulled toward a nonconscious presentation of aggression stimuli, especi ally in the presence of alcohol cues, and especially if he or she was high on trait anger [as measured using the State Trait Anger Expression Inventory (STAXI); Spielberge r, 1988] and had high expectancies for behaving aggressively while drinking alc ohol [as measured using the Expectancy Questionnaire for Alcohol and Aggression Â– Lo Dose (EQAAL); Epps, Hunter, LeVasseur, Steinberg, & Hanc ock, unpublished manuscript]. Seventy-nine of the participants w ho completed questionnaires also completed one of the two computer tasks (adapted from John Bargh and associates) weeks later in either the Barroom or the Cleanroom. Attent ion to HiAggression words (as measured by reaction time or error rate difference scores ) was significantly highe r than attention to NonAggression words using the parafoveal visu al task, with observed power at 1. No significant differences were found using the di chotic listening tas k. Additionally, there was a significant three-way interaction (Word Type X Setting X Angry Temperament) when participants where blocked according to high vs. low angry temperament scores. Follow-up analyses as well as regression anal yses for the specific hypothesis provided mixed results. Individuals lower on angry temperament tended to demonstrate higher levels of attentional interference fo r aggression words, but only in the presence of alcohol cues. Conversely, individuals higher on angr y temperament evidenced higher levels of attentional interferen ce, but only in the absence of alcohol cues.
vii It appears that the relationships am ong these variables are by no means straightforward. Studies that include an opportunity to aggress behaviorally may shed more light on whether oneÂ’s level of attenti onal interference and self -reported personality traits can be combined to predict aggre ssion in the presence of alcohol cues. The parafoveal visual task is recommended as the methodology of choice for these future studies.
1 Introduction Aggression represents a global health problem of enormous dimensions and involves behaviors such as homicide, suicide, domestic violence, and sexual violence (World Health Organization, WHO, 2001). WHO identified alcohol abuse as one of the primary individual risk factors for these types of aggression. Various da ta substantiate an alarming association between alcohol use and aggression. The U.S. Department of Justice Bureau of Justice Statistics (1997) reported that the following percentages of state prisoners were under the infl uence of alcohol at the time the offense was committed: murder 45%, negligent manslaughter 52% assault 45%, sexua l assault Â– 40%, and robbery Â– 37%. Additionally, alcohol has been reported to be invol ved in about 70% of fatal automobile accidents, 88% of knifi ngs, and 65% of spouse battering (Steele & Josephs, 1990). Consumption of alcohol appears to be asso ciated with the severity of aggressive behavior. Koss (1988) investig ated this hypothesize using a national college sample of nearly 3,000 men, some of whom were perpetrators of sexua lly aggressive crimes. Male perpetrators reported that alcohol or subs tance use was involved 74% of the time during rape, 67% of the time during attempted rape, 35% of the time during sexual coercion, and 33% of the time during unwanted sexual c ontact (Koss, 1988 as cited in Testa, 2002). Although the majority of us do not become aggressive after consuming alcohol, the regrettable consequences of the interaction exact a heavy toll against our society, rendering the relationship between alcohol a nd aggression worthy of intense scrutiny. Ongoing investigations of this relationship at tempt to identify variables that precipitate, mediate, or moderate drunken aggression, in cluding 1) external socio-cultural or situational cues and 2) indi vidual variables such as c ognitions, mood, or personality traits. Researchers, governmental agencies, and various funding sources continue to invest extensive resources on research th at will increase our unde rstanding of this sometimes deleterious interac tion. It is hoped that a more precise understanding of this interaction will facilit ate the development of more effective intervention programs for those who drink and become aggressive.
2 Overview The current study focused primarily with the cognitive aspects of the alcoholaggression relationshipÂ—attention to salient inte rnal and external cues related to alcohol and aggression. More specifica lly, this study was designed to find out if a personÂ’s attention is more likely to be pulled toward aggression stimuli, especially in the presence of alcohol cues and/or in the presence of high self-reported exp ectations for acting aggressive while consuming alcohol. To this end, a brief summary of the literature regarding aggression, selective attention th eory, alcohol and cognition, and alcohol and aggression is provided below. This is fo llowed by a description of two methods of presenting stimuli that hold promise for enha ncing our understanding of the confluence of aggressive stimuli, alcohol cues and persona lity variables upon a ttention. Finally, the specific hypotheses and methodology for the current study are presented. Aggression Definitional Issues The definitions of the words violence and aggression are similar in their emphasis on the delivery of punishment to another or ganism. Of the two, Â“aggressionÂ” has been operationally defined with greate r precision. Therefore, the term aggression will be used throughout this study. Many definitions of aggression have been offered in the literature. BaronÂ’s (1977) definition of aggression is Â“any form of behavi or directed toward th e goal of harming or injuring another living being who is motivated to avoid such treatmentÂ” (p. 7). This definition excludes cases in whic h 1) hurtful behavior is inte nded to help another person and 2) is acceptable to the target (e.g., the harm received by a surgical or dental patient). It also implies that the essential feature of a ggression is behavior that reflects intention to harm. Renfrew (1997) proposed that Â“aggression is a behavior that is directed by an organism toward a target, resulting in damag eÂ” (p. 6). Renfrew argued that this definition is broad enough to cover a wide range of aggr essive situations such as aggression toward animals or objects, and self-injurious behavi ors. However, the emphasis on Â“resulting in damageÂ” excludes unsuccessful attempts to hur t the target. Also, for many researchers,
3 the impact of aggression toward animals or ob jects is not as interesting or relevant as aggression toward other people. Buss (1961) defined aggressi on as Â“a response that delivers noxious stimuli to another organismÂ” (p. 1). This definition ignores intent or goal to cause damage or injury because Buss insisted that intent is unnecessary in the analysis of aggressive behavior. In his view, Â“the relationship between reinfor cement history of an aggressive response and the immediate situation eliciting the responseÂ” (p. 2) is the critical relationship because it is most likely to predict the occurrence and strength of aggressive responses. BussÂ’ definition fits well within a behavioral approach that circumvents unobservable cognitions such as intent. However, cognitions (e.g., intentions, expectancies, etc.), such as those implied by BaronÂ’s defi nition, are central to an attent ional approach, such as the one taken in this study. Intention to harm another person whether damage is caused or not, appears to be an important aspect of aggression. Therefore, BaronÂ’s definition of aggression will serve as the backdrop for the following disc ussion of the origins of aggression. In addition to various definitional issues concerning aggression, researchers have struggled to distinguish among definitions of aggression anger and hostility One distinction recognizes that these terms are di fferent facets of the same global construct: anger is the affective component; hostility is the complex cognitive, thought, or attitudinal component; and aggression is the behavioral component (e.g., Buss, 1961; Spielberger, Jacobs, Russell, & Crane, 1983; Epps & Kendall, 1995). Unfortunately, some investigators continue to use anger, hostility and aggression interchangeably, contributing to ongoing definitional ambigu ities. One method of minimizing this ambiguity has been to distinguish between angry or hostile aggression on the one hand and instrumental aggression on the other. Th ese distinctions are generally made using BussÂ’s (1961) definitions. Buss characterized all aggressive res ponses as involving an interpersonal context and either being rein forced by the victimÂ’s pain (which is considered angry/hostile aggr ession) or by extrinsic rewards (which is considered instrumental aggression). Angry or hostile aggr ession, then, is reinforced by the victimÂ’s emotional suffering, physiological reaction, or p hysical injury, whereas with instrumental
4 aggression Â“the acquisition of some extrinsi c reinforcer or the cessation of aversive stimuli are the crucial consequences, not the victimÂ’s discomfortÂ” (p. 3). Recently, however, the dichotomy between hostile and instrumental aggression has come under attack. Bushman and Ande rson (2001) recommended Â“pulling the plugÂ” (p. 273) on this dichotomy claiming that it has outlived its usefulness. The authors made a cogent argument that too many acts of aggression serve multiple purposes and include both impulsive anger and a premeditated, instru mental component. For example, when a boy shoves his brother out of a bus seat his inte ntion might be to get the seat for himself, to raise his power status in front of other students, to get revenge because his brother called him a name earlier, or a combination of all three. The relati ve influence of each type of aggression is often incalculable In fact, Bushman and Anderson expressed appreciation for the past utility of the hos tile vs. instrumental a ggression dichotomy, but suggested that psychologist s will realize future advan ces in the study of human aggression by utilizing a know ledge structure (informationprocessing) approach, which will be discussed shortly. Research related to the current study has centered on aggression that is performed concurrently with or secondary to anger ar ousal. While there may be an instrumental component to such aggression, the presence of anger arousal as a common theme makes it appropriate to focus primarily on literatu re related to angry aggression. However, instrumental aggression will be mentioned where appropriate. Salience or salient are concepts encountered frequently within the alcohol, aggression, and selective atten tion literature. Generally, stim uli (e.g., thoughts, attitudes, and environmental objects or events) are rega rded as salient when they stand out and enter conscious thought more readily because their conditions of activation are more easily satisfied (Krech & Cr utchfield, 1948). Higgins ( 1996) has provided convincing arguments that the common view of salien ce is better describe d by the concept of Â“accessibilityÂ” which he defined as the activat ion potential of available knowledge. These distinctions will be elaborated upon later (in the section on Selective Attention). However, in order to be concordant with th e extant literature, the term salience will be used throughout the current study.
5 The Origins of Angry Aggression To understand the variables that may be most fruitful for investigating the alcohol-aggression relationship, it is helpfu l to understand how angry aggression is assumed to develop. Many models have been proposed to explain the development of aggressive behavior including the original frustration-aggression hypothesis (Dollard, Miller, Doob, Mowrer & Sears, 1939), so cial learning theory (Bandura, 1973), a cognitive neo-associationist ic conception (Berkowitz, 1990) two social information processing models (Huesmann, 1988; Crick and Dodge, 1994) and an explication of knowledge structures (Bushman and Anderson, 2001). Each of these models will be discussed briefly followed by a summary of the important themes. Frustration-Aggression Hypothesis. Originally aggression was theorized to result as a direct consequence of frustration brought on by an undesirable in terruption (thwarting) of goal-directed behavior (Dollard, Miller, D oob, Mowrer, & Sears, 1939). The theory specified that once frustration is experienced, the innate drive is to strike ou t at a target. If aggressive behavior is inhibited, the natural response toward aggression is thwarted and more frustration is produced. On the other hand, if aggression is exhibited, relief from the instigation to aggress occurs. This relief ha s been referred to as Â“aggression catharsis.Â” However, several studies have provided evid ence against the cathartic effect. Under conditions in which one would be expected to produce less aggression (e.g., after already having the opportunity to aggress as inve stigated by Geen, Stonner, & Shope, 1975, or over time as investigated by Favata, LeVasse ur, Koenig, Ciarcia, Epps, & Roberts, 2003), participants actually produce more aggression (Lewis and Bucher, 1990). The original frustration-aggression hypothe sis also implied that frustration is always followed by aggression. However, partic ipants who perceive their frustration as resulting from a legitimate reason are less likely to display ag gression (Pastore, 1952; Cohen; 1955) Also, Bandura (1973) argued that awareness of likely punishers may cause a personÂ’s aggressive response to be inhibi ted or even extinguished. Both of these arguments suggest a mediational effect of c ognition regarding the frustrating event, an effect that is not addressed by the original model. Therefore, variables such as prior
6 learning (i.e., expectations) are likely to in fluence oneÂ’s interpreta tion of cues in the environment, mediating whether one considers an event to be frustrating in the first place, and whether a frustrating event even warrants an aggressive response. Both the lack of support for aggression catharsis and resear ch indicating that frustration is not always fo llowed by aggression cast doubt upon the tenability of the frustration-aggression hypothesi s to explain the origins of aggression. Models of learning were instrumental in in creasing our understanding of how aggression develops. Social Learning Theory. Social learning theory, as explicated by Bandura (1973), was the next notable model to describe the origins of aggression. Unlike the frustration-aggression hypothesis, social learning theory views frustration as me rely one example of an emotional state that can lead to aggression. Here aggression is cons idered to result from learningÂ—learning in a social context which feelings to label as Â“angerÂ” and which behaviors are likely to punish another person or lead to reinforcers. According to Bandura (1973), aggressive behavior sequences are learned via direct experience or observation. Through di rect experience, a child may learn by interacting with others behavior s for which he or she is likel y to be punished or rewarded. The frequency of aggressive behavior will be a direct function of how often the behavior was rewarded or punished as it was bei ng learned. Through observation, a child may learn which behaviors exhibite d by influential others (such as role-models) generate reward or punishment. Once an individual uses the modeled behavioral sequence and it is rewarded, this will increase the likelihood that the behavior will be repeated. Conversely, punishment for using the behavioral sequence will result in its extinction. Whether the behavior originated through vicar ious learning or direct expe rience, after more successes than failures in obtaining the desired result s, the behavioral sequence (e.g., aggression) will become part of that individualÂ’s repert oire for controlling his or her environment. It is tenable that frustrati on gives rise to a variety of negative emotions which may instigate the drive to aggress (as in the orig inal frustration-aggressi on hypothesis). It has also been suggested that the experience of negative affect, in general produces emotional arousal (Sandoval, 1997). When this arousal is paired with the right reinforcement
7 contingencies, aggression is produced. Howe ver, Bandura believed emotional arousal is not even necessary in the production of a ggression (Sandoval, 1997) Awareness that an event is aversive may lead directly to aggr ession if the reinforcement contingencies are sufficiently rewarding. Although social learning theory began to specify the role of emotional arousal in the mediation of aggressive behaviors, it di d not address the role of cognition in the mediation of these behaviors (Sandoval, 1997). Bandura (1973) suggested that oneÂ’s cognitive representations of reinforcement contingencies would mediate the interaction between behaviors and the environment. The more specific proce sses of cognition, such as how an individual assesses a situation and se lects an appropriate response, were left to be explicated by information processing theo rists (discussed later). But prior to this, negative affect was elegantly incorporated into a ne w model by Berkowitz (1983). Cognitive Neo-Associationism. In BerkowitzÂ’s modification of the orig inal frustration-a ggression hypothesis, negative affect arising from a range of aversive conditions is considered the basic source of anger and angry aggression (1983, 1989, 1990) Berkowitz continued to expand this model and suggested that aversive conditions produce both flight and fight tendencies (1993). He considered these tendencies to be networks of associatively linked physiological, motoric, and cognitive com ponents (Berkowitz, 1998) and suggested that the associative linkage is relatively primitive, automatic, and can occur in the absence of reportable cognitions. A variety of factorsÂ—genetic, learned, and situationalÂ—influence which of the flight or fight networks are mo st strongly activated. If fight networks are activated, these factor s (e.g., situational) will also in fluence whether the aggressive response is inhibited or exhibited. One situational factor of interest for the current study was attentional focus. Berkowitz found in a series of studies (Berkowitz & Troccoli 1990) that when attention is focused upon oneÂ’s negative affect, emo tional self-regulation is promoted, the link between negative affect and hostility (e.g., negative j udgments about others) is diminished, and aggression is inhibited. Howe ver, not all evidence supports an inhibiting effect of attentional focus. Berkowitz (1998) reported that Â“highly aroused people are apt
8 to focus on the main features of the situati on confronting them to the neglect of matters that are relatively peripheralÂ…. Thus, persons who are emotionally aroused because of an aversive event might well focus their atten tion narrowly on those they blame for the unpleasant occurrenceÂ” (p. 68), disregarding inhi biting cues such as possible punishment. BerkowitzÂ’ speculation parallels the theory of alcohol myopia which asserts that the range of attentional focus may be restricted after the consumption of alcohol, increasing the likelihood of aggression during volat ile situations (Steele & Josephs, 1990). Regardless of whether attentional focus is eventually concluded to inhibit or increase the likelihood of an aggressive response, it is reas onable to assume that attentional focus is an important moderating variable. In some cases, those who are highly ar oused or have overlearned aggressive responses to certain situati ons (to the point of automatic ity), may go from anger (an affect) directly to aggression (a behavior), without any repo rtable intervening cognitions (Berkowitz, 1990). However, once individuals e ngage in a higher order level of cognitive processing, Â“they consider the perceive d causes of their arousal, the possible consequences of any action they might undertak e, the goals they woul d like to attain, and also what sensation they are feeling and what ideas and memories have just occurred to themÂ” (Berkowitz, 1990, p. 497). This indicates that processes (e.g., a ppraisals of rules and consequences, and attributions) not used in its production can mediate aggression. Further, this implies that unde rstanding which types of intern al and external cues render behavioral responses more automatic, and unde rstanding which types of cues facilitate higher order cognitive processing, are worthy goals of aggression research. BerkowitzÂ’s (1990) cognitive-neoassocia tionistic conception of anger and aggression improves upon prior models by offe ring a cognitive bridge between negative affect and aggression. The strength of his model is that it accounts for the original evidence linking frustration to aggression while also linking a variety of aversive events (pain, extreme temperatures, noxious odors, stre ss, provocation, or viewing disgusting or aggressive images) to negative affect, wh ich then leads to aggression, sometimes depending upon the outcome of higher level cognitive processing and sometimes independently of those processes (for research related to aversive events the interested
9 reader is referred to Berkowitz, 1983, 1989; Anderson, 1989; Hearold, 1986; Liebert & Spratkin, 1988; and, Geen 1998). One limitation of BerkowitzÂ’s model is that it does not address what causes the highe r order cognitive processing or how the appraisals and attributions control reactions to the aversive events. Fortuna tely, the task of explaining the origins and consequences of cognitive pr ocessing has been undertaken by information processing theorists. Social Information Processing Models. In information processing theories, a schema is a representation in memory about a general set of facts, and how th ese facts are rela ted (Medin, 2001). A script is a type of schema that contains information about seque nces of ordered actions that occur in a stereotyped situation. Script s help us understand events and make predictions about future events (Medin, 2001). Scripts are l earned and augmented by children (and adults as well) through vicarious and direct social experience and are accessed in order to interpret the social enviro nment and guide behavior. According to HuesmannÂ’s (1988) information processing model of childhood aggression, the conditions most conducive to the learning of aggressive scripts appear to be those in which the child is reinforced for displaying aggression, often observes aggression, and is the object of aggression (Eron, 1994). Salient environmental cues will then activate those aggressive scripts. Salience is affected by oneÂ’s familiarity with those cues as well as oneÂ’s current emotional stat e (Sandoval, 1997). Once the relevant script is activated, the child evaluates 1) the appropr iateness of the script with regards to internalized social norms, and 2) the like lihood that the script will obtain the desired results (Eron, 1994). Aggression is more likely to occur when salient cues activate an aggressive script, when the script has been re peatedly associated w ith the perception of a desired result (e.g., injury to a nother or a reward), and when the child feels confident that he or she can enact the activated script (Huesmann, 1988). Conversely, if salient, activating cues are not present, if there is a perception of negative results such as punishment, or if the child lacks self-confiden ce in enacting the script, aggression is less likely.
10 More recently, Huesmann and his collea gues (Guerra, Nucci, & Huesmann, 1994) proposed that aggressive actions are direct ed by Â“moral judgmentÂ” memory systems or knowledge structures. Guerra, et al. (1994) suggested that which knowledge structures direct aggressive actions depends upon seve ral factors including 1) an individualÂ’s evaluative beliefs (i.e., right or wrong) and informational beliefs (i.e., potential consequences), 2) salience of situational cues (e.g., cues that focus attention on a particular aspect of a situation and activate relevant moral judgments), and 3) interpretive biases that influence whether the cue is perceived at all or distorted upon perception. They offered two sources of interpretive bi ases: interpersonal fact ors (e.g., mood states, personality, and attributional st yle) and sociocultural influe nces (e.g., family and peers, and social contexts such as school and reli gion). They also suggested that judgments become routinized, leading to behaviors that a ppear insensitive to the unique features of a given situation. That is, the behaviors become relatively automatic as a result of the moral judgment knowledge structures. It seem s reasonable to conclude that automatic moral judgments and often-used or well-rehe arsed behavioral scripts may represent an unfavorable combination in the production of aggression. A major strength of information processing th eories is that they can be used to describe how aggressive beha vioral scripts become represented in memory (Sandoval, 1997). Early on, direct and observational learni ng facilitate the creation of knowledge structures (to be discussed in greater detail later) that li nk social cues to aggressive responses. These primitive structures ar e elaborated upon over time. When a child encounters a social problem, attention is direct ed toward the most salient cues (internal cues such as emotional state or external cu es in the environment) and a comparison is made with scripts already in memory. Salient cues that are most similar to those present when the script was first encoded will increase the probability that the previous script will be retrieved. The retrieved behavioral script will then be evaluated for its appropriateness in the current situation. Huesmann (1988) capitalized on an inform ation processing approach to account for how a behavioral sequence is maintained in memory once it is encoded. He postulated that rehearsal is the primar y method by which the behavioral sequence is maintained in
11 memory for later accessibility. Rehearsal includ es such behaviors as recalling the original script, fantasizing, or role-playing. Another social information processing theory (Dodge, Pettit, McClaskey, & Brown, 1986) also postulated aggression as a function of a childÂ’s processing of environmental cues (external and internal ) in social situati ons (Sandoval, 1997). The Dodge et al. paradigm is noteworthy for its elaboration of how scripts are developed and activated. A variety of labels ha ve been applied to what were originally described as the five sequential steps undertaken for skillful social information processing. In general, these five steps consist of 1) cue encoding, 2) cue interpretation, 3) response generation, 4) response decision, and 5) respons e enactment (Dodge & Crick, 1990). More specifically, in the fi rst step of Dodge and CrickÂ’ s model, relevant cues are selected from the environment. Attention to particular cues (e.g., a person, a situation, or an object in the environment) is mediated and moderated by heuristic rules and cognitive schemata that have developed over time. During the second step, cues are mentally represented in long-term memory and given meaning. Meaning is related to oneÂ’s past experience with that particular person, situation, or objec t, as well as oneÂ’s past experience with those general types of stimuli. The third step involves accessing possible behavioral responses to the cues through a ssociative networks of related long-term information. Behavioral responses that have been accessed recently or frequently over time, may be quickly accessed and appear to be relative ly automatic. However, the evaluation of that responseÂ’s probability of achieving a certain outcome or be skillfully enacted (carried out in the fourth step), may l ead to an inhibition of the selected response. Whichever response is chosen, during the final step the selected res ponse is enacted using information from relevant scripts Â“to transf orm the selected res ponse into verbal and motor behaviorsÂ” (p. 14). M onitoring of the responseÂ’s effectiveness for achieving a particular outcome leads to further encoding of social cues, which is hypothesized to alter the response or start the seque nce of social information pro cessing all over again (Dodge & Crick, 1990). Crick and Dodge reformulated their m odel in 1994 and proposed that the processing steps are not executed in a sequent ial fashion and numerous cycles through
12 the steps (possibly in a different order each time) may be performed depending on environmental cues, such as social exchanges with others. Thus, the steps are viewed as more cyclical and transactiona l than linear or sequential as others become involved in the enactment of a script or other cues become more salient. Skillful execution of these cycles is considered necessary for competent behavioral responding, whereas fa ilure at any point may lead to inappropriate behaviors such as aggression. Empirical studies have provided some evidence that childrenÂ’s aggressive behaviors are related to biased or deficient processing during any cycle. Reviews of the research related to the sp ecific Â“stepsÂ” can be found in Sandoval (1997). A review of the evidence for interventions based on the social information processing models discussed above can be f ound in Huesmann and Reynolds (2001). Both of the social information processing models emphasize the impact of information in memory on the perception of new information. That is, a childÂ’s existing memories of situations and po ssible behavioral responses will influence which cues in a new situation are attended to and encoded (Sandoval, 1997). For example, a cue that has preceded punishment in the past may be especially salient for the child, and may inhibit an aggressive response. Both models also assu me that if an aggressi ve behavioral script seems appropriate for the goal of hurting the victim or obtaini ng an extrinsic reward, an aggressive act will be chosen as the suitabl e response. In addition to describing how an aggressive response may be selected, both mo dels imply that inefficient processing of social cues can produce inappropriate responses such as aggression. Knowledge Structures and Network Models. The above theories imply that there is an automatic nature to the cognitive elements of aggression. Bushman and A nderson (2001) relied heavily upon the assumption of automaticity to describe how aggression develops and used the construct of knowledge structures as a framework for appreciating the releva nce of automaticity. First of all, they describe knowledge structur es as organized networ ks of interrelated information (both schemas and scripts) that re sult from frequent activation of bits of related information (p. 276-277). Knowledge structures are comp iled and augmented during childhood in order to guide behavior They are activated by external cues
13 (environmental) and by internal cues (e.g., em otional arousal and goal attainment). In turn, some cues become more salient than others. The most salient cue(s) will prime relevant schemas or activate relevant scripts, which will then influence which behavioral response is finally chosen. Â“The person first se lects a script to repr esent the s ituation and then assumes a role in the scriptÂ” (p. 277). Knowledge structures that are accessed with greater frequency will become overlearned and hence automatic exerting a fast and efficient influence on incoming information and the behavioral response that is selected. That is, once a behavioral sequence becomes ac tivated, it is more likely to be carried out whether the person is aware of a deci sion to act a certain way or not. Salient cues are assumed to activate know ledge structures, which serve to bias some internal and external stimuli. Although salience of stimuli is di fficult to establish a priori, research with children shows some progress. In one study (Dodge & Tomlin, 1987), groups of aggressive and nonaggressive children were to interpret the intentions of a provocateur in a hypothetical situation. Aggr essive children were more likely than nonaggressive children to base their interpretations on schema ta rather than information presented in a story. That is, aggressive children were more likely than their nonaggressive peers to rely upon previous expe rience rather than immediate social cues. In another study (Gouze, 1987), aggressive pa rticipants were more likely than their nonaggressive peers to focus their attention on aggressive social interactions in the environment. They also provided more aggre ssive solutions to hypothetical interpersonal conflict situations. Th is finding suggests that aggressi ve children may pay greater attention to aggressive cues in social situ ationsÂ—presumably because these cues are selfrelevant. Another study of children (Rabiner, Lenhart, and Loch man, 1990) provided some evidence for the role of automaticity. Children were given the task of solving a social problem under conditions that elicited refl ective processing or under conditions that elicited automatic processing. Socially maladjusted childre n had difficulty processing social information adequately only under automatic conditions. Crick and Dodge (1994) suggested that most studies were not designed to evaluate processing deficits under automatic responding conditions. They also suggested that techniques that measure
14 response time and evaluate priming effects w ould be valuable for future research on automatic processes. A major advantage of the knowledge stru cture approach is that it avoids confounding the hostile-instrumental aggressi on dichotomy with the automatic-controlled information processing distinction (Bus hman and Anderson, 2001). As Bushman and Anderson noted, Â“hostile aggression is, by definition, automaticÂ—it is unreasoned, impulsive, uncontrollable, and spontaneous. By contrast, instrumental aggression is, by definition, controlledÂ—it is reasoned, cal culated, and premeditatedÂ” (p. 276). The confound exists because both complex decisions and affect-laden decisions can be made automatically or with careful thought. Bush man and Anderson argued that the knowledge structure approach suggests th at the more frequently a know ledge structure is activated, the more automatic it becomes, regardless of whether it originated from impulsive anger or calculated, conscious inte nt. Additionally, assuming that hostile and instrumental forms of aggression are dichotomous presum es that these knowledge structures are somehow separate in the brain, when it seems more plausible that they interact or are part of the same knowledge structure. The knowle dge structure appro ach facilitates the investigation of conditions that mediate or moderate aggression and avoids reliance upon distinctions that create more ambiguity and questions than clarity and answers. Knowledge structures are conceptually similar to neural network models, especially major network models of sema ntic memory (Bushman and Anderson, 2001). Neural network models are considered to be analogous to the structure and function of neurons in the brain. Over the last six decades neural network models have been revised and extended to a variety of research do mains including computer science, economics and finance, and psychology. Within psychol ogy, AndersonÂ’s latest revision of his adaptive control of thought model (ACT-R) repres ents an attempt to describe a variety of phenomena and data related to memory a nd learning (Medin, 2001). According to the ACT-R model, as information is processe d in working memory, networks of nodes (concepts) are activated. Activation then spr eads to neighboring con cepts through links in the networkÂ—a process called spreading activation Activation of any one node Â“is a function of both the number of activated concep ts it is linked to and also the strength of
15 the links to those nodesÂ” (Medin, 2001, p. 224). Th e strength of a link is directly related to how frequently that link is used. Howe ver, even if the link is strong, activation dissipates from node to node and activation of a particular node may disappear altogether (fall out of working memory) as ne ighboring nodes become less active. The ACT-R model provides a reasonable hypothesis for how information is brought to bear in a situation. First, external and internal cues that are salient for an individual will become activated in working memory. Then, related information (which is most likely to be information that was re lated to it in the pa st, Medin, 2001) will be activated through spread ing activation. Concepts (or behavi oral scripts) with the most activation (i.e., that have been most freque ntly used) will influence which behavioral response is selected. In many ways the knowledge structures appr oach is similar to the ACT-R model. Aggression cues or stimuli that are salient for an individual will activate aggressionrelated knowledge structures or networks. If an individual has often enacted aggressive behavioral scripts in the past he or she is more likely to enact them in the future, compared to other individuals who have been less aggressive. It follows, then, that measuring trait anger of an individual may of fer some predictive utility in determining whether someone is more likely to aggress as compared to others. Although the current study will investigate two methodologies for predicting a ggression cue activation related to trait anger, studies that use one of these methodologies and give participants opportunities to aggress would be need ed to test this last assumption. Assumptions Generated from Models of Aggression Like definitions of aggression, models of aggression have evolved and have successfully accounted for instrumental and angr y aggression, the effect of a variety of aversive events on subsequent aggressi on, and the initiation and maintenance of aggressive behaviors. Several testable assu mptions have emerged from these models. One is that negative emotional reactions (such as frustration or anger) or goal-attainment, or both together, when paired with the right reinforcement contingencies, will produce aggression. Another assumption is that engaging in complex thinking will reduce aggression. Conversely, factors th at limit or interfere with complex thinking are assumed
16 to increase aggression. These factors include 1) automaticity, 2) distraction, which promotes superficial processing of relevant social informati on, 3) individual differences in social reasoning ability, which may be due to poor role models, or inconsistent rewards or punishments for aggressive behavior; 4) individual differences on a variety of personality dimensions, such as impulsivity and trait aggression, 5) intense emotions, which have been shown to interfere with th e processing of incoming information, and 6) anything that would interfere with oneÂ’s cap acity to process a wide range of cues, including alcohol consumption (Smith & Ma ckie, 2000). Some of these factors can be manipulated (e.g., distraction) while others can be measured (e.g., trait aggression). Another assumption generated by aggression models is the reciprocal nature by which salient cues and existing networks of social information (knowledge structures) interact to produce behavior. Cues that ar e salient for an individual will activate knowledge structures, which will in turn gui de behavior. Conversely, existing knowledge structures will bias individuals toward some social cues, especially those that are selfrelevant, while other cues are disregarde d or ignored. That is existing knowledge structures will differentially influence the processing of available information. Determining which cues someone is likely to attend to, whether attent ion is intentionally focused or not, may help us predict behavior a priori. Methods th at help clarify the relationship between knowledge structure activation and a ggressive behavior and can produce automatic effects may also increase our understanding of the alcohol-aggression relationship. Selective Attention In order to investigate assumptions relate d to cognitive mediators of aggression, it is important to consider what is known a bout selective attention and automaticity. There is no universally accepted definition of attent ion in the psychological literature. It has been viewed as mental effort, concentr ation, focalization, or selective processing. Although there is considerable overlap am ong these concepts, a ttention appears to represent a variety of situati ons/processes. Johnston and Dark (1986) in their review of the literature, defined selective attention as Â“the differential processing of simultaneous sources of informationÂ” (p. 44). Although th ey did not define Â“processing,Â” it may be
17 regarded as the series of step s, sequential or simultaneous, taken to detect and analyze a stimulus and to decide upon a course of action based upon that analysis. Johnston and Dark further specified that simultaneous sources of information consist of internal events (memory and knowledge) and external events (environmental objects and situations) and that the information is analyzed perceptual ly, semantically, or both. Selective attention, then, implies that we select some inform ation for further processing (because it is relevant in some way) while ignoring or f iltering out other information (because it is irrelevant). Salience vs. Accessibility Salience is another concept that has been de fined in various ways throughout the literature on selective attenti on and knowledge activation. As discussed earlier, salient stimuli have been viewed as those that st and out and enter thought more readily because their conditions of activation are more eas ily satisfied (Krech & Crutchfield, 1948). Others use the concept of salience to desc ribe anything that commands oneÂ’s attention (see Higgins, 1996). However, this view imp lies that any variable that influences attention is salient. Many othe r factors clearly influence which stimuli we attend to (such as effort, the difficulty of the current task, alertness, mood, etc.). Furthermore, such a broad definition does not allow distinctions to be made between salience and selective attention, salience and accessi bility, or salience and know ledge activation (Higgins, 1996). Higgins (1996) argued that salience is more appropriate ly viewed as Â“something about a stimulus event that o ccurs on exposure, without a prio r set for a particular kind of stimulus, that draws attention selectively to a sp ecific aspect of the eventÂ” (p. 135). Higgins further restricts salience to the properties of the stimulus event only which may include features of the particular stimulus, properties of the immediate context in which the stimulus appears, and the relations among th ese properties. That is, a stimulus can be considered salient because of something a bout its absolute properties or because of something about its properties relative to those of other object s in the immediate situation.
18 HigginsÂ’ (1996) definition of salience does not include properties of the perceiver, such as expectancies or goals. Th ese properties fall unde r the rubric of accessibility which Higgins defined as the activation pot ential of available knowledge. However, salience and accessibility can interact to increa se the likelihood that knowledge structures will be activated (p. 134). That is, the perceptual system is biased toward some internal or external stimuli because of 1) prior experience with that stimuli, 2) chronic or habitual expectancies or attitudes that have devel oped over time, and 3) recent thought or experience. Any of these biases may render one stimulus property as more distinctive when compared to nonbiased stimuli or stimul us properties that are available for further processing. Even for stimulus information that is impoverished, vaguely related, or fits into other knowledge categories better, its act ivation in the perceptual system will be easier and swifter than other cues. Thus, nonbi ased stimuli that w ould normally serve to guide oneÂ’s behavior in a soci al situation may be minimally or not at all processed (Bruner, 1957a, 1957b as cited in Higgins, 1996). Although the above distinctions between salience and accessibility are important for future research intended to parse apar t their relative effect s on knowledge structure activation, the earlier definition of salience is still in common use. It is expected that HigginsÂ’ concise treatise on knowledge activation will facilitate an a ssociated precision in future research endeavors. For the current study, stimuli that are considered salient are best viewed as those stimuli that have highe r potential to activate knowledge structures. However, the term salient will continue to be used instead of accessibility. Early vs. Late Selection The idea that information is different ially attended has been well accepted by researchers. However, agreeing upon the pr ecise point at which salient information becomes selected for further analysis has been the basis of the decades-long early selection versus late selection debate (see Pashler, 1998 for a revi ew). Briefly, Johnston and Dark (1986) summarized the debate as to whether selection of stimuli for further processing takes place after sensory analysis but before semantic anal ysis, or if it always takes place after semantic analysis. In s upport for early selection they concluded, Â“selection based on sensory cues is usually su perior to selection based on semantic cuesÂ”
19 (p. 48). However, they also recognized that early selection of re levant stimuli assumes early rejection of irrelevant stimuli. Th is assumption does not seem tenable since considerable evidence exists th at irrelevant stimuli sometim es undergo semantic analysis (e.g., see Lewis, 1970; Treisman, 1960; M acKay, 1973; Corteen & Wood, 1972; and, Moray, 1959). Thus, the evidence supports bot h early and late selection models. Automatic vs. Controlled Attention Within selective a ttention a distinction was proposed to more fully explicate how environmental stimuli are chosen for further an alysis. One way is via controlled attention, which has been described as a conscious, active, voluntary, effortful, flexible, or intentional cognitive process. The other way is via automatic attention, or automaticity, which has been described as a nonconscious passive, involuntary, effortless, or unintentional cognitive process (see Bargh, 1992 and Johnston & Dark, 1986 for reviews of controlled and automatic attention). Bargh (1992) described controlled attenti on as flexible but resource-limited. That is, controlled attention is a resource that ca n be allocated toward a task. The degree to which a stimulus is processed relates directly to how much attention is directed toward that stimulus, which in turn relates to how demanding the associated task is. BarghÂ’s description mirrors that of cap acity models. According to cap acity theories, simple tasks do not interfere with each other. However, a difficult (or resource-demanding) task interferes with the processing of a simpler ta sk. There are also multiple resource theories (as discussed in Medin, Ross, & Markman, 2001 ), which suggest that there are multiple pools of attentional resources th at can be allocated to various tasks. The degree to which two tasks interfere with each other depends on the degree of overlap between the resource pools. For example, the resource pool for aud itory tasks should not be the same as the resource pool for visual tasks. Therefore, a ttention could be directed toward auditory and visual tasks at the same tim e with minimal interference. Johnston and Dark (1986) revi ewed studies that investig ated spatial attention and likened spatial attention to an attentional s potlight with an adjustable beam. However, adjustment capabilities are limited. The beam can be adjusted to include information directly outside of the fov eal region (the parafoveal re gion), but the processing of
20 information is most efficient for stimuli within the attentional spotlight (i.e., aligned with the center of the fovea and directed toward specific regions of space). The beam of the attentional spotlight can be adjusted in order to complete a task and can even be Â“split.Â” For stimuli outside of the spotlight, proces sing is most likely (albeit minimal) for nonsemantic stimuli but still at a considerable cost to processing speed and accuracy. Regarding semantic information that is pres ented visually, nonattend ed information that is more than about 3 from the visual angle of attended stimuli is un likely to be processed (Rayner, 1978). For information that is pr esented aurally, nonatte nded information is intrusive when it is relevant. For example, in dichotic listening tasks, controlled attention to one channel can be interrupted by informa tion in the other channel if the nonattended information conforms to active sche ma (i.e., is self-relevant). This tendency to be attracted toward in formation that is not being directly attended but is self-relevant is described by the other selection process, the automatic process (Bargh, 1992). This process directs one Â’s attention to e nvironmental stimuli without conscious intent. In the strictest sens e, a process is regard ed as automatic if it requires no cognitive resources to initiate it (i.e., a ttention is not intentionally focused) and if the process runs to completion once it has begun. One possible cause of the automatic process is top-down processing, which refers to the effect that old information (i.e., internal represen tations or expectations about the stimuli) has on the selection of new information. According to Broadbent (a s cited in Johnston & Dark, 1986), as an individual learns associati ons among stimuli, the individua l develops internal biases toward those stimuli. Therefore, even if stimuli are not relevant for a particular task those stimuli may be relevant to the individual These biases direct attention away from the stimuli that should be processed quickly and accurately for su ccessful execution of a given task. Johnston and Dark (1986) presented result s from studies on the intrusiveness of irrelevant (nonattended) stimu li and concluded that selectiv e attention can be guided by active schemata and that this process can be co ntrolled or automatic. Selective attention is more likely to be categorized as controlled when the stimuli are purposefully attended to, but categorized as automatic when the stimuli are attended to because of an internal bias.
21 The internal bias may exist because schemata have been primed (activated) by recent thought or experience, or because those stimu li are self-relevant (chronically accessible) to the individual. Bargh (1992) came to a similar conclusion but assigned this internal bias to one of two types of automaticity: preconscious or postconscious Preconscious automaticity refers to the nonconscious sele ction of stimuli based upon ster eotypical constructs held by the individual (i.e., chronic expectancies a ssembled over years of interacting with the environment). That is, stimuli are pro cessed based upon their mere presence. Postconscious automaticity, on the other hand, is essentially the same except that the nonconscious selection of stimuli is based upon constructs, expectancies, or schemata that have been primed (preactivated ) by recen t conscious thought or experience. That is, the stimuli would not be salient, or proce ssed, based on their mere presence except that they have been recently activated in memo ry. The recent activation results in a lower threshold of accessibility for those stimuli. Priming, in a research context, generall y refers to procedures that activate knowledge structures. Priming can occur at a ny level of stimulus analysis (Rabbitt & Vyas, 1979 as cited in Johnston & Dark, 1986), ranging from low-level sensory analysis (e.g., find something green) to high-level semant ic analysis (e.g., presenting the test word BREAD speeds up the recognition of the test word BUTTER). Considerable evidence for priming effects led Johnston and Dark (1986) to propose that the processing of low-level sensory or high-level semantic information can be primed toward sensory characteristics of stimuli (e.g., auditory vs. visual), toward identity of stimuli (i.e., physical codes in memory), toward semantic representations of stimuli (e.g. word meaning and synonyms), and toward schematic representations of s timuli (e.g., knowledge structures). Thus, Â“all levels of stimulus analysis can be biased simultaneously toward the characteristics of most of the relevant stimuli and some of the irrelevant stimuli. In some instances, these biases can be sufficiently strong that attention to the relevant or irrelevant stimuli appears to be automaticÂ” (p. 65). The interaction between c ontrolled attention and auto matic attention, and the effect of preconscious automaticity were elegantly investigated by Bargh (1982). He
22 demonstrated that even when one is intenti onally focusing on stimuli related to a primary task, stimuli that are unattended (and considered irrelevant to the primary task) can pull oneÂ’s attention away from the relevant stimu li. In a focused-attention dichotic listening task participants were directed to attend to and shadow (repeat) th e stimuli presented to one ear and to ignore the stimuli presented to the other ear. Bargh manipulated the relevance of stimuli by measuring participant levels on particular pe rsonality traits and then presenting traits that the participants were high on to either the attended or unattended channel. For example, a person th at self-reported a high level of independence was presented with words like assertive and nonconformist to either the attended or unattended channel. Bargh hypothesized that se lf-relevant stim uli in the attended channel would facilitate attention (i.e., the stimuli w ould require less attenti onal effort) but that self-relevant information in the ignored channel would inhi bit attention to the attended channel (i.e., the stimuli would require more a ttentional effort for the participants to stay focused). Although the participants demonstr ated no awareness of the words in the unattended ear (as judged late r by a momentary awareness te st) the self-r elevant words facilitated attention if they were in the at tended channel and inhibi ted attention if they were in the unattended channel. Thus, according to Bargh (1982), automatic processes can either facilitate or inhib it the control process, requiring either more or less attentional effort depending on the self relevance of the stimuli that is presented. Bargh (1992) also proposed that primed c onstructs, while they remain active in memory, are equivalent to the effects of preconscious automaticity on the selection of stimuli from the environment. In fact, Bar gh (1996) claimed that Â“preand postconscious automaticity are functionally identical and the processing effects are the same; the only difference is in how the necessa ry level of accessibility is achieved (i.e., via chronic or temporary means)Â” (p. 174). Therefore, it is important for researchers to consider the biases that are created by primed constructs as they design their studi es. In order to rule out the possibility of postconscious automa ticity in his experiment, Bargh (1982) was careful to measure personality traits a month before the dichotic listening task was given. He was then more confident in his conclusi on that attention to the contents of the
23 unattended channel reflected a preconscious automatic process because those items tapped chronic expectancies. After reviewing a plethora of studies relate d to central construc ts and assumptions of selective attention, Johnston and Dark (1986) derived a set of empirical generalizations (eleven to be exact; several of which have been presented here) to explain attentional phenomena. They concluded that theories that view selective attention as the natural priming effects of prior processing on subse quent processing are able to accommodate all eleven empirical generalizations with relativ e ease. In addition, th e view of selective attention as an effect of chronically active schemata precludes any reliance upon an active mental agent (or homunculus) to desc ribe how stimuli are selected from the environment. This view of attention as a by-product of oneÂ’s prior experience with stimuli is the view taken for the purposes of the current study. This view and the associated methods used for this area of re search (notably dicho tic listening tasks and parafoveal vision tasks, to be discussed next) provide a fr amework for investigating the activation of aggression-related knowledge stru ctures and their effect upon the selection of information from the environment. Ultimat ely, one of these methods may provide an index of knowledge structur e strength that is inde pendent of self-report. Dichotic Listening Tasks for the Measurement of Attention Dichotic listening is a procedure that wa s originally developed by Broadbent in 1954 to study the impact of receiving multiple f light bearings, at the same time, on traffic controllersÂ’ attention (Bryden, 1988). Since then the basi c procedure has been used for studying short-term memory, la teralization and ear advantag e. Dichotic listening tasks were also essential in increas ing our understanding of how people are able to focus their attention on one stimulus, but automatically attend to another stimulus that is particularly relevant for them (e.g., the cocktail party phenomenon). Research using the dichotic listening task spawned evidence for both early selection (e.g., Treisman & Geffen 1967) and late selection theories (e.g., MacK ay, 1973; Corteen & Wood, 1972; and Moray, 1959; and, Lewis, 1970) and the procedure contin ues to be used to explore the sequential and simultaneous processing of information.
24 The basic procedure has been adapted in several ways and more recently has proliferated into a variety of psychological contexts such as attention and depression (Ingram, Bernet, & McLaughlin, 1994), atte ntion and schizophrenia (Hugdahl, Rund, Lund, Asbjornsen, Egeland, Landro, Roness, St ordal, & Sundet, 2003) and attention and psychopathy (Hare & Leslie, 1984). In the study of information processing, the procedure has been used to understand when selection of stimuli takes place for further processing and whether selective attention can be guided by active schemata (Johnston & Dark, 1986). More recently, dichotic listening tasks have been used in electrophysiological studies to investigate the influence of atte nded and unattended aud itory stimuli on eventrelated potentials in the brain (e.g., Ben tin, Kutas, & Hillyard, 1995; and Holcomb, & Neville, 1990). The general procedure for di chotic listening tasks is to present verbal stimuli or nonverbal stimuli (e.g., tones, car horns, flushing toilets or music) to the left auditory channel, the right channel, or both at the same time. Instruct ions to attend to one channel while ignoring the other are give n in focused attention tasks. Bryden (1988) reported that the right ear (which ac tivates the left auditory cortex ) processes verbal stimuli more quickly and the left ear (which involves th e right auditory cortex) processes nonverbal stimuli more quickly. However, there are inconsistencies across studies as well as variability among study participants. Thus, when lateralization effects are not under investigation, it is prudent to balance presentati on of stimuli between the left and right ear by balancing the number of participants that are instructed to focus on each channel. Results from dichotic listening tasks have also revealed that stimuli that are presented to the unattended channel are more easily ignored when they are physically different (e.g., the voice in one channel is female while th e other is male; Cherry, 1953). Therefore, the gender generating stimuli (e.g., word pairs or pa ssages of text) is generally the same for both auditory channels when semantic level processing is under investigation. Dependent measures of th e allocation of attention to one channel vs. the other have included intrusions (i.e ., responding with information presented to the unattended channel), and error rates (e.g., incorrect shadowing of stimu li presented to the attended channel). Intrusions and error rates are exp ected to be higher when attention is drawn
25 away from the attended channel. Anothe r method for measuring the allocation of controlled attention was developed by Ba rgh in 1982. He used a secondary taskÂ— reaction time (RT) to a probe stimulusÂ—t o index attention. Study participants were instructed to focus their attention on a prim ary shadowing task and to use any remaining attention to press a button to turn a light stim ulus off as quickly as possible once it came on. BarghÂ’s rationale was that la tency to respond to the second ary task should be directly related to the amount of contro lled attention being given to th e primary task. If response latency were greater, then more effort, and t hus more attention, was being used to stay focused on the primary task. He sought to validate this method of assessing spare processing capacity by including a no probe co ndition. When there were no differences in shadowing errors (the primary task) between the light probe and no probe conditions and no better than chance recognition of target items on a memory test, he concluded that RT to a probe was a valid measure of attenti onal capacity being used by the primary task. Bargh (1982) used this method of dichotic listening, shadowing, and reaction time to a probe to test his main hypothesi s that self-relevant informati on in the attended ear should facilitate faster RTs to the probe and that self-relevant in formation in the unattended ear would interfere with RTs to the probe as the participant struggled to maintain attention on the primary task. As discussed earli er, his data supported his hypotheses. Other researchers have turned to this methodology to help them study cognitive factors that might affect behavior. McCabe and Gotlib (1993) looked at attentional processing in clinically depressed people by having participants complete a focusedattention dichotic listening task and a concurrent light-probe task. They concluded that depressed people had attentional biases (what Bargh would call postconscious automaticity) for negative-content information fed into the unattended ear because their RTs to a stimulus probe were longer on the primary task than t hose of non-depressed people. Interestingly, the resear chers found that when the partic ipants were retested three months later and were no longer depresse d, they no longer demons trated attentional biases. It is reasonable to c onclude that negative-content wo rds were no longer salient for the formerly depressed participants because this information was no longer self-relevant.
26 Although McCabe and GotlibÂ’s (1993) study focused on depression, it does seem clear that self-relevant information must be considered when attentional processes are being studied. When the effects of alcohol are considered in the light of theories of selective attention, it follows th at a drinkerÂ’s attention to stimuli in the internal or external environment is directed by recent t houghts or experiences or by chronic beliefs or expectancies that he or she may hold rega rding the effects of alc ohol on behavior when particular cues are present. BarghÂ’s model shows promise as a method of measuring the salience of internal and external cues. Theref ore, his method will be used to demonstrate an automatic attentional effect of a ggression cues for those who report higher propensities to aggress, especially after the consumption of alcohol. Although the reaction-time methodology has been helpful for understanding the influence of unattended verbal information on attention to a primary task, Bargh has more recently turned his interest toward understa nding the influence of automatic attention on goal-directed behavior (e.g., see Bargh, 1992; Chartrand & Bargh, 1996; Bargh & Ferguson, 2000; and Bargh & Chartrand, 2000). This methodology utilizes a parafoveal vision task to demonstrate that information presented outside of conscious awareness can influence various behaviors. Before this study, a direct comparison of the dichotic listening task and the parafoveal vision task had not been made. Parafoveal Visual Tasks for th e Measurement of Attention In memory and learning, both auditory and vi sual perception play a crucial role in the selection of stimuli, the conversion of stimuli to long-te rm memory, and the selection of responses for behavior. Parafoveal vi sion tasks are a more recent methodological choice for studying the selection of visual s timuli from oneÂ’s environment for encoding and retrieval. Parafoveal visual tasks require the partic ipant to attend to information presented in the center of a computer screen (also the fixation point), while re levant or irrelevant information is presented to the parafoveal (peripheral) region of vision on the computer screen. The foveal region extends from 0 to 2 from the fixation point, whereas the parafoveal region extends from about 2 to 6 from the fixation point. Although parafoveal stimuli theoretically could be presented anywhere between 2 and 6 from the
27 fixation point on any point along the circumfere nce of the circle, us ually the stimuli are presented equidistant from the fixation point to one of the four quadrants encompassed by that circle (e.g., at 45, 135, 225, and 315 as in Chartrand & Bargh, 1996). To determine where the stimuli should be presen ted, Bargh and Chartrand (2000) provided the formula Y = X/tan(2), where X = the distance between the fixation point and the parafoveally presented stimulus, and Y = the distance between the pa rticipantÂ’s eyes and the fixation point at the center of the comput er monitor. Another method is to use Bargh et al.Â’s (1986) existing calculations. That is each word should be placed in one of the quadrants such that the center of the word is 7.6 cm from the fixation point. This will ensure presentation of the stimuli to the pa rafoveal region as long as the participantÂ’s eyes are 99 cm away from (in front of) the monitor. Controlling the placement of the chair and monitor and in structing the participant to sit er ect at all times further ensures that the stimuli are presente d outside of the participantÂ’ s foveal visual field. It is wise to follow parafoveal stimuli with a masking stimulus at the same location for two reasons (Bargh & Chartrand, 2000). The first is th at refresh rates on computer monitors often vary and a stimulus that has not decayed from the monitor is more likely to be perceived by the participan t. The second reason is that a visual iconic memory trace of a stimulus may increase the likelihood of perceiving that stimulus. If a masking stimulus quickly replaces the parafoveal stimulus of interest, the participant is more likely to perceive only the jumble of letters that comprises the masking stimulus (e.g., Bargh generally uses the ma sking string Â“XQFBZRMQWGBXÂ”). Another issue of concern involving paraf oveal presentation of stimuli is the duration that the stimuli are presented. Rayner (1978) found that participants took at least 140 ms to move their eyes from the fixation poi nt to the parafoveal word when they were explicitly instructed to do so. Bargh (2001) recommended using para foveal presentations of 60 ms to 90 ms to avoid even Â“expre ss saccadesÂ” (fast saccad ic jumps of 100 ms; Fischer & Weber, 1993 as cited in Bargh, 2000). Bargh and Chartrand (2000) have also va ried the quadrant that the parafoveal stimulus is presented to as well as the onset of stimulus (between 2 and 7 seconds) so that the participant is unlikely to Â“predictÂ” the lo cation of the next para foveal stimulus. To
28 provide additional reassu rance that the participantÂ’s attent ion is at the fixation point when parafoveal stimuli are presente d, it is prudent to provide a ta sk that involves the fixation point (Bargh & Chartrand, 2000). Since BarghÂ’s (1982) dichotic listening task involves a primary task that is presented to the center of the screen, utilizing the same task for both methodologies will allow a direct test of th e ability of each methodology to provide an index of selective attention to self-relevant stimuli that is presented outside of conscious awareness. Parafoveal visual tasks (or subliminal pr iming tasks as it is referred to by Bargh & Chartrand, 2000) have been used in a variety of research endeavors. In the context of reading, researchers have used these tasks to investigate whether a semantic priming effect occurs for words in sentences that are outside of the fov eal visual region. For example, research by Altarriba, Kambe, Pollatsek, & Rayner ( 2001) did not support a semantic priming effect. However, when wo rds were presented individually, there did appear to be a semantic priming effect as fa r as 3 from central fixation (Rayner, 1978). There is other evidence to suggest that parafoveal words are processed at a semantic level. In a parafoveal priming condition (Di Pace, Longoni, & Zoccolotti, 1991; Experiment 1), participants were presen ted with nonwords at the fixation point concurrently with a lateral parafoveal stim ulus, which consisted of words that were semantically related to the target words and, finally, the target word (e.g., mesod centrally concurrently with flight laterally, followed by the target word eagle ). Participants were expected to show a facilitation effect when parafoveal words were related to target words (i.e., faster reaction times to respond Â“yesÂ” wh en the target word represented an animal) as opposed to the negative priming and base line conditions. The re sults supported their hypothesis. Di Pace, et al. also found that the facilitation eff ect was smaller for parafoveally presented words than for fovea lly presented words and only existed if the inter-stimulus interval between parafoveal word and target word was short (200 ms vs. 2000 ms). They regarded this as evidence fo r the assumption that automatic processing effects decay at a rapid rate. This is consis tent with theories of selective attention. Bargh and his associates have used s ubliminal priming tasks to investigate a variety of automatic effects such as the in creased likelihood for men to sexually aggress
29 when their Â“power sex associationÂ” is activated (Bargh, Raymond, Pryor, & Strack, 1995); the effect of goal activation on impr ession formation (Chartrand & Bargh, 1996); and the effect of subliminal priming on ne gative mood (Chartrand, Ba rgh, & van Baaren, 2003). However, these tasks are generally c onducted in order to prime a particular construct or behavior of inte rest. For example, to investig ate an effect of priming on mood (Chartrand, et al., 2003) part icipants were exposed to st rongly valenced negative or positive stimuli in a subliminal priming task (i.e., a parafoveal vigilance task). The stimuli were presented parafoveally concurren tly with a brief flash to the left or right visual field. Participants indica ted whether they detected a flas h on the left or right of the screen by pressing the appropriate key as quick ly as possible. They found that repeated exposure to positive or negative stimuli produ ced a concurrent mood as self-reported on mood scales. Another study (Chartrand & Bargh, 1996) successfully primed participants to form impressions of a person described in a va riety of behavioral phrases. Participants in the study were subliminally primed with words like impression judgment personality and evaluate (as opposed to a no-goal priming condition) and then read more phrases indicating honest behaviors or more phrases indicating dishonest behaviors. If they had been primed with an impression formation goal, subsequent trait rati ngs of a target person were much more likely to coincide with eith er dishonest or honest ratings (depending on which one they had been exposed to more ofte n within the behavioral phrases). Chartrand & Bargh (1996) concluded that the eff ect of nonconscious activation (subliminal priming) of memory representations was as e ffective as explicit instructions for forming an impression of a target person. As with other studies conducted by Bargh and his associates, reaction times to the vigilance task were not reporte d as being analyzed. It is lik ely that the authors would not see these reaction times as an index of automatic attention to unattended stimuli. However, in a recent review of automatic ity research, Bargh and Chartrand (2000) indicated that using a dual-task parafoveal visual paradigm, such as the one used in BarghÂ’s (1982) dichotic listening task with a concurrent light-probe task, may index how efficiently one is able to process atte nded information under conditions of scarce
30 attentional resources. If a nonatt ended stimulus is self-releva nt, it should interfere with attention to the primary task, which will be reflected in fewer resources available for the secondary task, hence sl ower reaction times. Limitations of Dichotic Listen ing and Parafoveal Visual Tasks The appeal of the dichotic listening task is in its potential to measure the effect of attended versus unattended information on task performance. Ultimately, however, the participantÂ’s attentional focus is not under the experimenterÂ’s direct control. Various techniques have been used to measure atten tional drift to irrelevant stimuli, such as shadowing tasks (e.g., errors and intrusions typi cally indicate attenti onal shift), and posthoc recognition or recal l tasks. One argument against using these techniques is that lack of errors or intrusions or inability to re port unattended stimuli does not equate lack of attention to or nonproce ssing of the stimuli. Parafoveal visual tasks also suffer from lack of experimenter control over intentional shifts to unattended stimuli. In addition to the visual masking procedure described earlier, reassurance that the particip ant did not attend to irrelevant stimuli is available by viewing eye movements (as with a video camera) and, again, recognition or recall tasks. Unfortunately, the use of video cameras to track saccadic eye movement is costly, cumbersome, and intrusive. Additio nally, it is difficult to track saccades concurrently with stimulus presentation. Despite the limitations of using either th e dichotic listening or parafoveal visual dual-task methodologies, either task may be expected to demonstrate aggression-related knowledge structure activation. Th at is, individuals that are higher on measures of trait aggression will demonstrate a form of chronic (preconscious) activation when their attention to a secondary task is made more effortful (response latencies are longer) in the presence of unattended aggre ssion cues. It was hypothesized that this effect would be reliable for both the parafoveal and the dich otic listening methodologi es as long as dual primary and secondary tasks were used. Alcohol and Cognition A great deal of research in the last few decades has been devoted to exploring the nature of alcoholÂ’s cogni tive impairment. Although cogniti ve theory borrows heavily
31 from and is related to various aspects of learning theory a nd social psychology, implicit is the understanding that without clarification of the cogniti ve domain, a comprehensive model of alcoholÂ’s influence on excessive soci al behaviors, such as aggression and risktaking, may never be achieved. Predicting when recently perceived information or information represented in memory (e.g., e xpectancies) will have a disproportionate influence upon behavior (over and above environmental cues, personality variables, or the pharmacological properties of a drug), is a challenge that cognitive theorists face. Predicting when an individual will be the life of the party on one occasion or commit a violent crime the next is a challe nge that alcohol researchers face. Physiological and Expectancy Effects The evidence clearly indicates that alcohol consumption is related to inappropriate and excessive behaviors. Early models attributed these behaviors to the disinhibiting effects of alcohol. That is, alcohol reduces oneÂ’s ability to refrain from acting upon behavioral impulses (e.g., acting upon an impulse to be more sociable or aggressive when one is generally shy or nonviolent). Howeve r, the earlier disinhibition models assumed a more overall effect of alcohol that was not supported by the research. According to Steele and Josephs (1990), alcohol itsel f does not directly cause excessive or inappropriate behavior. They pointed out that an individua lÂ’s specific reactivity to the drug could not account for a personÂ’s behavior because his or her behavior may vary from one occasion to the next. Alcohol consumption has clearly been dem onstrated to cause a slowing of motor responses as it depresses the central nervous system. Howe ver, there is evidence to suggest that even alcoholÂ’s impairment of motor performance is not a pure effect of ethanol on tissue. The degree of impairme nt on motor performance appears to be mediated by the thoughts or expectancies that an individual holds about the drug. Fillmore and Vogel-Sprott (1995), in their e fforts to identify factors that might account for alcoholÂ’s variable effects on behavior, ex amined whether motor response varied with expectations about how alcohol would affect their performance on a specific task. The degree of impairment anticipated on a motor sk ill task after consum ing alcohol accounted for a significant proportion of impairment demonstratedÂ—whether alcohol or placebo
32 was consumed. In fact, expectancies for pa rticipants in the alcohol condition accounted for 12.3% of the variance of a participantÂ’s change in performance. Likewise, expectancies for those in the placebo condition accounted for 17.2% of the variance. These findings are not specific to alcohol. Expectations about task performance after the consumption of caffeine have yielded a simila r pattern of results. That is, increases in actual task performance have been predic ted by the strength of the participantÂ’s expectancies about their perf ormance, regardless of whether they had consumed caffeine or a placebo (Fillmore, Mulv ihill, & Vogel-Sprott, 1994). It is reasonable to hypothes ize that understand ing an individualÂ’s expectancies is an important precursor in predicting behavi or because a drugÂ’s phys iological effects do not occur independently of expectancies. This assump tion prompted an explosion of alcohol expectancy research in the past few decades. However, researchers have had only partial success in using e xpectancy theory to predict the be havior of drinkers, and most of that success applies to the initiation and ma intenance of drinking behaviors (e.g., Carey, 1995; Chassin, & Molina, 1993; Goldman, Brow n, & Christiansen, 1987; Reese, Jones & McMahon, 1994; and, Zucker, Kincaid, Fitz gerald & Bingham, 1996). A potential prevention strategy proposed by Darkes and Goldman (1993) focuses on the arousal and sociability expectations of participants. When these expectancies are challenged in comparison to a no-treatment control, alc ohol consumption and expectancies show reliable decreases at post-treatment and after a booster session six weeks later. Although studies like this are encouraging, researchers have had little success in predicting most other individual behavior, such as when aggression might occur. This lack of predictive power has serious implications for the use of expectancy theory in the assessment and treatment of alcohol-related pr oblems such as aggression. Why has it been so difficult for researcher s to predict an i ndividualÂ’s behavior after consuming alcohol? One reason is that re searchers have struggled to define the term Â“alcohol expectancy.Â” Very generally it refers to an intervening variable of a cognitive nature (Goldman, et al., 1987) and is used in the literature to indicate the belief that one has consumed alcohol and how one thinks it wi ll affect his or her behavior. The use of the word Â“beliefÂ” in association with Â“expectancyÂ” has been criticized for its implication that
33 expectancies are consciously accessible information (Goldman, Del Boca, & Darkes, 1999). Instead, expectancies shou ld be regarded as memo ry templates that organize incoming information; expectancies do not require conscious awareness or focused attention. In relation to alcohol expectancies should be view ed as memory templates that reflect Â“the reinforcement value of alc ohol acquired as a function of biological, psychological, and environmental risk variable sÂ” (Goldman, et al., p. 216). Further, these memory templates or expectancies serve to anticipate which behaviors should be performed under which circumstances depe nding upon what was learned about alcohol and its contexts during previous encounters. Measurement issues may also contribute to the inconsistency of results regarding the relationship between expectancies and alcohol. A major difficulty in separating alcohol expectancy effects from the pharmacol ogical effects is that there appear to be several different types of e xpectancies. For example, e xpectancies about how other people behave after drinking alcohol are diffe rent from expectancies about how alcohol will affect oneÂ’s own behavior. Scales used in the service of predicting an individualÂ’s behavior should consist of items that tap expectancies of oneÂ’s own behavior after consuming alcohol (Leigh & Stacy, 1993). Negative expectancies (e.g., punishment) regarding the outcome of alcohol consumpti on is considered as important as positive expectancies (reinforcement) and attempts have been made to measure both types of outcome expectancies (e .g., Leigh & Stacy, 1993). The expectancy model clearly overlaps with the knowledge structures approach offered by information-processing theory. Bo th approaches imply that networks of memory templates bias the perception of in coming information and that behavioral outputs are selected based upon the similarity of incoming information to the networks of information already represented in memory. The value of expectancy theory is that it may help clarify extant outcome expectancies a nd predict their relevant salience in a given situation. Alcohol myopia is a theory that outlines the mechanisms by which more salient information may have a dispr oportional influence upon behavior.
34 Alcohol Myopia One cognitive theory, developed in the late eighties (Steel e and Southwick, 1985; Steele and Josephs, 1990), suggested that al cohol causes behavioral disinhibition through impaired cognitive processing of relevant cues Thus, alcoholÂ’s variable effects are due to an interaction of the pharmacological and cognitive effects. In this theory, Â“alcohol myopiaÂ” is defined as a state of shortsighted ness in which drinkers process fewer cues less well than non-drinkers. That is, alc ohol intoxication cause s a restriction in information processing that influences the sa lience of both external cues (environmental cues) and internal cues (expectancies, memo ries, and mood), increas ing the likelihood of socially excessive behaviors such as aggression. Even cues that are attended to are assumed to be processed superficially a nd relevant information is not given due consideration before a behavior is initiated and carried out. The theory of alcohol myopia provides some predictive value. An example may demonstrate the utility of alcohol myopia for predicting aggressive behavior. A man in a bar may be drinking, with the expectation that he will become more relaxed. After consuming a moderate amount of alcohol, he notices a large man staring at him in a hostile way. Will there be a bar fight? Accord ing to alcohol myopia theory, the myopic effects of alcohol consumption should increas e with dosage. The more salient a cue is (e.g., the hostilelooking male), and the drunker the person gets, the more likely it is that an aggressive response would prevail over more distal yet more appropriate behavioral responses. The likelihood of an aggressive resp onse may also be mediated by attitudes or beliefs that the inebriated person holds. For example, he may hold a chronic belief that people who stare are rude and de serve to be Â“taught a lesson.Â” If the person holds this belief, becomes intoxicated, and sees a hosti le-looking male staring at him, alcohol myopia theory would predict a bar fight. The above scenario falls under a class of al coholÂ’s social effects that Steele and Josephs (1990) termed drunken excess They posited that whenever there is a conflict between inhibiting and provoking cu es (whether these cues are internal or external), the most salient aspects of th e event will have a dispropor tionate influence upon the behavioral response that is selected. If re levant pressures that would normally inhibit
35 inappropriate responses to salient cues are not allowed normal proces sing, the inebriated personÂ’s behavior will appear impulsive and excessive. As salience of cues change, the strength of the competing responses will change, with the stronger cue saving or wrecking the day. If the knowledge structure that includes information about rudeness (the salient, instigating cue from the previ ous example) is activated frequently and a behavioral sequence that incl udes aggression often follows, st aring may be interpreted as something worth fighting about. Other relevant yet more distal social cues (a bar fight might lead to arrest) may be less likely to overcome the aggressive behavioral sequence once it has been initiated. Thus the drinkerÂ’s perceptions appe ar myopic in that the focus is on stimuli that are nearer in time or proximity. Several research endeavors have utilized the response conflict component of the drunken excess construct to investigate precurs ors to aggression and other socially excessive behaviors such as sexual aggression (Testa 2002); drinking and driving (MacDonald, Zanna, & Fong, 1995); and, high-risk sexual behavior (Kaly, Heesacker, & Frost, 2002; MacDonald, Zanna, & Fong, 1996; Morris & Albery, 2001). Unprotected sex is one example of drunken excess that arguably includes an inherent response conflict. MacDonald, Zanna, and Fong (1996) investigated cognitive precursors to unprotected sex and provided strong eviden ce that alcohol myopia can explain the relationship between alcohol consumption and decreased condom use. Alcohol myopia theory would predict that in a conflicting situation where peop le express intentions to use condoms but condoms are unavailable, intoxicate d people will pay less attention to distal, inhibiting cues (e.g., the risk of pregnancy or getting a sexually transmitted disease), and more attention to immediate, provoking cues (e .g., the attractiveness of the partner, the partnerÂ’s willingness to have sex). As a re sult of this c ognitive process, intoxicated people should endorse higher likelihood of intentions and justifications for having unprotected sex. The relationship between alcohol cons umption and decreased condom use was investigated by having intoxica ted and sober participants watch a video showing a male and female leaving a bar and going to the femaleÂ’s apartment to have consensual sexual intercourse. The couple in the video did not have access to condoms, indicating a risky
36 situation, but the female was a ttractive and will ing to have sexual intercourse. The video was stopped at this point, representing a res ponse conflict for the viewer. The researchers asked students who indicated pos itive attitudes toward usi ng condoms, and reported that they regularly did use them, what they would do next. They found that intoxicated participants were more likely than sober participants to endorse intentions and justifications to have sexual intercourse. Furt her, intoxicated partic ipants indicated more awareness of the potential for condoms to protect them against sexually transmitted diseases and that having interc ourse without them in a situat ion similar to the one in the video could be characterized as Â“extremely f oolishÂ” behavior. The authors interpreted the results as compelling evidence for the alcohol myopia perspective. The central assumptions of the alcoho l myopia model with respect to drunken excess include response conflict, salience of cues, and impaired cognitive processing. Some findings provide evidence for these assu mptions. For example, Mulvihill, Skilling, and Vogel-Sprott (1997) provided evidence that alcohol impairs cognitive processes that govern response inhibition. They demonstrated th is effect using a Â“go-stopÂ” task in which go signals are considered to initiate an activ ating process and stop si gnals are considered to initiate an inhibiting pro cess (Logan, Cowan & Davis, 1984; Mulvihill, et al.). When go and stop signals are presented simulta neously (response activ ation vs. response inhibition), these processes compete. Depe nding on which process finishes first, the response is either executed or inhibited. Mulv ihill et al. showed that participants who were given moderate doses of alcohol were le ss able than participants in a placebo or control group to inhi bit their responses to go signals when they were concurrently provided with a stop signal. Since their m easure of response activation (reaction time) was unaffected for all three groups, they c oncluded that alcohol primarily affects response inhibitionÂ—not response activation. Zeichner, Allen, Petrie, Rasmussen, and Gi ancola (1993) examined the interaction between alcohol drug effects and the salience of cues, specifically information regarding threat. The drug condition included alcohol, placebo, and control. The salience condition included low threat (positive trait) or high threat (negative trait) information that described the participants themselves (salient condition) or describe d others (nonsalient
37 condition). Intoxicated participan ts attended to threat informa tion (negative tr aits) longer than participants in the pl acebo or control group when the traits described themselves (salient condition). Â“Presumably, in the sa lient negative information condition, alcohol limited the subjectÂ’s attention to the most threat ening or salient aspect of their situationÂ” (p. 731). The authors conclude d that these findings were consistent with Steele and JosephÂ’s (1990) attention-allo cation model and that further research should focus on the interactive effects of alcohol intoxicati on and salience of environmental cues in emotionally charged situations su ch as aggressive situations. Another study (Herzog, 1999) is unique in that it investigated the effects of alcohol on the second stage of a two-stage so cial inference (attri butional) process and suggests a link between alcohol and automatic ity. The first stage of social inference involves identifying and classify ing a personÂ’s behavior into dispositional terms (the degree to which the behavior is driven by the enduring personality trai ts of the person) or situational terms (the degree to which the aspects of the personÂ’s environment are influencing his or her behavior). This stage is often regarded as a more automatic process based upon heuristic methods of categorizing the vast array of available information (e.g., stereotypes, schemas, and behavioral scri pts). The second stage involves a corrective stage in which the opposite influences are taken into account. This stage is considered to be a more deliberate, controlled process and requires a higher expenditure of cognitive resources than the first stage. Since alcohol is thought to impair the c ognitive processing of information that is more distal in time or place, Herzog (1999) hypothesized that intoxicated participants would be less likely than sober participants to engage in the more effortful, corrective stage of social inference. When both sober and intoxicated part icipants were asked to rate how influential disposition was in the behavior of actors in a series of videos, intoxicated participants rated dispositional influences as significantly higher than sober participants. Similarly, when both groups were asked to rate how influential situational factors were on the behavior of the actors in the videos intoxicated participants exaggerated the influence of situational factor s. The author suggested that sober participants did not exaggerate the relative influence of either di spositional or situational factors because they
38 were able to consider both influences rega rdless of the condition they were assigned to, and were able to adjust their ratings acco rdingly. The author concluded that these findings were consistent with the alcohol myopia perspective. Although the above studies provide eviden ce for the tenability of the alcohol myopia model, direct evidence for the model is meager. This is partly due to the relatively few studies that have been c onducted and the abunda nce of alternative hypotheses offered by the investigators and ot her authors. Sayette (1999) observed that the alcohol myopia model, among other cognitive models, offers indirect evidence for the alcohol-behavior relationship a nd that, ultimately, validity of models like this will rest on studies that more directly test this relationship. The cu rrent study may provide a method by which salience of a given construct (in this case, aggression and/ or alcohol cues) can be determined before a person is given the opportunity to aggress. Salience, of course, is a central assumption of the alcohol myopia model. Alcohol Cues The current study is intended to demonstrat e that aggression cues are chronically salient to individuals who re port higher levels of aggressi on/trait anger. In a similar fashion, alcohol cues are exp ected to be chronically sali ent to individuals who report higher levels of drinking e xperience. A plethora of evidence exists for this assumption and is exemplified by several modified Stroop color-naming studies. Cox, Blount, and Rozak (2000) used the Stroop paradigm to inves tigate the interference effects of neutral, alcohol-related, and concern-related words on alcohol abusersÂ’ and nonabusersÂ’ attention. Alcohol abusers responded more slowly to na ming the color of stimu li that were related to alcohol (e.g., beer vodka ) than to naming the color of stimuli that were related to personal concerns (e.g., divorce dog ). Nonabusers showed no differential interference. Another modified Stroop color naming study (Sharma, Albery and Cook, 2001) also demonstrated attentional interference fr om alcohol-related words. The investigators found that in-treatment abstinent problem dr inkers were significantly slower to name alcohol-related words than to name neutral words. Interference from alcohol-related stimuli was also found for a Â“high-drinkerÂ” control group. Those in the Â“low drinkerÂ” control group showed no differen tial interference for alcohol-related vs. neutral stimuli.
39 Finally, in a more rigorous investigati on of alcohol cue interference (Stormark, Laberg, Nordby, & Hugdahl, 2000), alcoholics demonstrated longer reaction times to both alcohol-related and emotion-related word s than neutral words on a Stroop task. They also evidenced significantly larger skin conduc tance responses to alcohol words than to any other words. These effects were not dupl icated in the nonalc oholic controls. The researchers concluded that alcohol icsÂ’ attention is biased towa rd alcohol stimuli, and that alcoholics have difficulty disengaging their at tention from those stimuli. They further suggested that alcoholicsÂ’ processi ng of these cues is automated. Another investigation of alcohol-relate d cue salience (Townshend & Duka, 2001) also revealed an attentional bias to alcohol-r elated stimuli. However, in this study a dot probe detection task was used. Participants were instructed to l ook at a fixation cross (presented for 500 ms.) when it appeared in th e center of the comput er monitor. Then two pictures appeared (for 500 ms.), one on each si de of the screen. One picture was related to alcohol and one was related to statione ry (e.g., a hand holding a glass of wine on one side and a hand holding a stapler on the other). After the stimuli were presented, a dot appeared on either the same side as the alc ohol-related picture or on the opposite side. An attentional bias score was calculated for each participant by taking the mean reaction time for when the dot and alcohol -related word were presented in the same location and subtracting it from the mean reaction time for when the dot and alcohol-related words were presented on opposite sides of the scree n. The researchers found that heavy social drinkers responded significantly more quickly than occasional social drinkers to the dot probe when it replaced the alcohol-related pict ure on the same side. They interpreted this as evidence for an attentional bias toward those stimuli. However, the researchers conducted the same task using words inst ead of pictures and found no differences between the two groups. This may be partly e xplained by the nature of the stimuli. The words represented a variety of concepts related to drinking (e .g., withdrawal-related words, craving-related words, and concrete words like beer and wine ). The variety of stimuli but small number of words in each ca tegory may have reduced the sensitivity and power of the task for det ecting attentional biases.
40 Another explanation for the lack of diffe rences regarding alcohol word stimuli in the aforementioned study (Townshend & Duka 2001) is that an initial, automatic orientation toward self-relevant word stimuli has been found only for shorter intervals between onset of the cue word and onset of the target (interstimulus intervals; ISIs; Posner & Snyder, 1975). At longer ISIÂ’s (e.g., 500 ms.), participants have enough time to direct their attention away from the stimuli. Slower RTÂ’s at 100 ms. ISIÂ’s are theorized to represent difficulties in shifting attention, wh ile faster RTÂ’s at 500 ms. ISIÂ’s are assumed to reflect an avoidance of those stimuli and a more conscious effort at shifting attention from those stimuli (Stormark, Field, Hugdahl, and Horowitz, 1997). One study (Stormark, et al., 1997) investigat ed the influence of shorter vs. longer ISIs directly and found that abstinent alcoholics showed longer reaction times when alcohol-related words were presented at a 100 ms ISI (automatic orientation) but faster reaction times (avoidance) when alcohol-rela ted words were invalidly cued at 500 ms. ISI. In the study by Townshend and Duka ( 2001) words were presented at 500 ms ISI. This may have given participants the opportuni ty to avoid some of the word stimuli, especially the ones that were not an integral part of the heavy social drinkersÂ’ knowledge structures (e.g., withdrawal eff ects of drinking). This avoida nce was likely to vary across subjects and may have reduced differences between groups. Although evidence is convergi ng that alcohol-related s timuli interrupt attention for participants with heavier drinking experi ence, it is less clear how the presentation of alcohol cues and aggression cues simultaneou sly would impact attention and how these stimuli would effect a nonalcoholic populat ion. From the above discussion, it seems reasonable to assume for the current study th at college students w ho are presented with aggression cues at a subliminal level of aw areness in a room devoid of alcohol cues, would show longer latencies to respond to those stimuli if they are high on trait aggression. Those who also have a lot of drinking experience may be expected to demonstrate the longest latencies of all. This is related to the assumption that alcohol and aggression knowledge structures will be more tightly interwoven and more frequently activated, resulting in a more automatic eff ect on the processing of aggression-related stimuli.
41 For students tested in a room full of alcohol cues (i.e., a barlab), it is reasonable to assume that alcohol cues would be especi ally salient for those with more drinking experience and higher levels of trait anger. Their levels of interference on attention should be the highest. The current proposal will investigate this hypothesis. Alcohol and Aggression Although the theory of alcohol myopia ma y be useful for predicting alcoholÂ’s various social effects, when attempting to understand the alcohol-a ggression link it is useful to understand the more basic nature of the relationship between alcohol and aggression. Bushman and Cooper (1990) conducte d a meta-analysis of 30 experimental studies with male confederates and male pa rticipants who were social drinkers. They concluded Â“alcohol influences ag gressive behavior as much or more than it influences other social and nonsocial beha viorsÂ” (p. 350). They reported an average effect size for alcohol vs. control to be d (+) = 0.25. The average effect size for alcohol vs. placebo was calculated to be d (+) = 0.61. They speculated that the average effect size for the alcohol vs. placebo condition is larger because th ere are methodological problems with the alcohol vs. control condition. The most seri ous problem is that the control groups generally see through the beverage deception. Sometimes the placebo groups see through the deception as well. However, since the ps ychological and pharmacological effects of alcohol occur together anyw ay, Bushman and Cooper recommended that the alcohol vs. placebo comparison is the best estimate of the effects of alcohol on aggression. Recent research continues to provide evidence for an alcohol-aggression relationship. One study (Lange, 2002) demonstrat ed that participants with higher blood alcohol levels (BACs; .05-.18) who associated alcohol with aggression were more likely to identify ambiguous behavior (via vignettes) as more aggressive than those who associated aggression with amiability. These authors concluded that alcohol affected the perception of aggression. Many studies have investigated the al cohol-aggression rela tionship using the Taylor Aggression Paradigm (TAP; Taylor, 1993) or modifications of it. In these competitive reaction time tasks, participants are generally provided with information (e.g., using feedback lights) about the level of shock the opponent has selected for them if
42 the participant is slower to re spond (i.e., loses the Â“trialÂ”). Pa rticipants are assumed to use this information to set subsequent shock leve ls. Aggression, then, is operationally defined in these tasks as the intensity and/or duration of shocks selected for a fictitious opponent for each competitive trial. Since the amount of wins versus losses is predetermined and distributed evenly in all conditions (Che rmack & Giancola, 1997), direct comparisons can be made between shock levels and durati ons set by participants who are intoxicated and those set by participants who are not. In general, investigators have found that intoxicated participan ts are reliably more aggressive than participants who have r eceived a placebo or a nonalcoholic beverage (Chermack & Taylor, 1995; Laplace, Cherm ack, & Taylor, 1994; Giancola & Zeichner, 1997). This finding may not generalize beyond the college student population or the laboratory. However, as noted by Chermack and Giancola (1997) a few studies have attempted to address external validity and found that aggressive responses within the laboratory correlate positively with peer and counselor rated aggre ssion, with self-report aggression inventories, and with histories of antisocial behavior (see also Anderson & Bushman, 1997; Giancola & Chermac k, 1998; and, Tedeschi & Quigley, 1996). Situational Variables A variety of situational factors have been found to intensify the alcoholaggression relationship. These in clude provocation, frustration, th reat, social pressure (to aggress), and response conflict (as operationa lized in the alcohol myopia perspective of drunken excess). On the other hand, social pr essure (to avoid a ggression) and selffocused attention can also decrease a ggression. (See Chermack & Giancola, 1997; Gustafson, 1993; and Ito, Miller, & Pollock, 1996 for reviews of the literature concerning these variables.) Provocation appe ars to be one of the most important moderators of the alcohol-aggression relationship. In fact, provocation has been claimed to be a more potent elicitor of aggression than either gender or beverage condition (Giancola, Helton, Osborne, Terry, Fuss, & Westerfield, 2002). To date, studies directly investigating the interacti on between aggression and alcohol cues (without the cons umption of alcoholic or nonalcoholic beverages) have not been conducted. Based on the idea of knowle dge structure activation, it may be that
43 participants in a barlab woul d respond more aggressively th an control participants in a room devoid of alcohol cues if they self-report higher levels of aggressi ve states or traits and they have more extensive drinking e xperience. Only one study investigated the impact of drinking experience on alcohol-re lated aggression (Laplace, Chermack, & Taylor, 1994). Surprisingly, of participants categorized with low-, moderate-, or highdrinking experience, only participants w ith low-drinking experience were more aggressive (using the TAP) after consumi ng alcohol. Although the current study would not investigate the interac tion between intoxication and aggression, it may provide a baseline measure of the influence of pers on variables. That is a methodology that can measure the influence of alc ohol and aggression cues to au tomatically interfere with a participantÂ’s attention to a primary task would provide a useful baseline before the participant is given the opport unity to aggress subsequent to provocation or frustration, or before they are given alcohol. Situational factors are clearly essent ial for understanding the alcohol-aggression relationship. Additionally, atten tion (whether automatic or cont rolled) to situational cues (e.g., provocation, threat, social pressure), as specified by th e alcohol myopia perspective, is arguably the most convincing mechanis m by which these variables influence the alcohol-aggression relationship (Gustafson, 1993). However, lack of attention to individual differences (or person variables) limits the explanator y value of cognitive theories. As research explor ing individual differences ha s accumulated, investigators have begun to incorporate these findings into their models (e.g., Chermack & Giancola, 1997). Gender The data regarding the willingness or te ndency for men and women to aggressive at equivalent levels is mixed. In the abse nce of alcohol, there is evidence that both women and men experience anger but that women respond with less physical aggression than men (Frost & Averill, 1982). Another st udy found that men were aggressive toward other men when provoked but not toward wo men, and women were aggressive toward men when provoked but not to ward women (Taylor & Epst ein, 1967). Interestingly, the highest levels of aggression in the study were found for highly provoked women
44 competing with men. However, this finding is in contrast to anot her study (Richardson, Vandenberg, & Humphries, 1986) in which women were less likely than other men to set extreme shock levels toward men. In the presence of alcohol, the data are also mixed. Direct comparisons are often difficult due to variations in the type and am ount of alcohol administ ered, the choice of aggression measures, and the t ype of noxious stimuli inflicte d on or by the participants (Dougherty, Bjork, Bennett and Moeller, 1999) as well as variations in gender of the fictitious opponent, confederate or even e xperimenter. One study found that intoxicated men were more aggressive than sober men toward a fictitious female opponent (Richardson, 1981). Gustafson (1991) did not sh ow an increase in aggression for women as a function of alcohol when a nonaggre ssive response was available toward the fictitious male opponent. Another study (Bond & Lader, 1986) demonstrated an increase in aggression for both intoxicated men a nd women when provocation level was low but only for men when provocation was high. Howe ver, a recent study (Giancola, et al., 2002) demonstrated almost the opposite. Into xicated men were more aggressive then intoxicated women under low provocation, but men and women were equally aggressive under high provocation. These authors (Giancol a, et al. 2002) concluded that alcohol increases aggression for men but that only provocation will lift aggressionÂ–related inhibitions for women. Some studies have focused more on gende r differences related to direct and indirect forms of aggression. In some cases, indirect aggression incr eased for intoxicated women but not men (Rohsenow & Bachorow ski, 1984). Giancola and Zeichner (1995) operationalized direct aggressi on as shock intensity and indi rect aggression as shock duration and found that intoxicated men showed an increase in both forms of aggression, while intoxicated women only showed increas es in indirect aggression. However, the authors recommended interpreting this findi ng with caution since shock duration fits questionably with most definitions of indirect aggression. There appears to be at least one plau sible generalization that can be made regarding gender differences for alcohol-related aggression : men show a consistent increase in aggressive res ponding while drinking. Although th is generalization cannot be
45 applied to women, at least one study found that aggression increased equally for men and women over cumulative doses of alcohol (D ougherty, Bjork, Bennett and Moeller, 1999). In the Dougherty et al. study, participants who showed higher levels of aggressive responding under placebo conditi ons (indicating baseline in dividual differences) also showed the highest increases in aggre ssion under alcohol conditions. This finding suggests that variables beyond gender are critical for understanding the alcoholaggression relationshipÂ—specifically, individual difference variables. Individual Difference Variables The study of individual difference vari ables as moderators in the alcoholaggression relationship has gained momentum over the last two decades. Alcohol expectancies would certainly vary by individual, but, as discussed earlier, alcohol expectancies appear to play a negligible ro le in the alcohol-aggr ession relationship. The role of alcoholaggression expectancies may also be trivia l but the jury is still out. Some studies indicate that alcohol -aggression expectancies do not facilitate aggression beyond dose. One such study (Chermack & Taylor, 1995) used a threequestion scale to determine whether participants had a high or low score on alcohol-aggression expectancies (Effects of Drinking Questi onnaire; EDQ; Dermen & George, 1989) and then randomly assigned participants to a high-dose or placebo-dose condition in which they all performed a competitive reaction time task with a fictitious opponent. Participants in the high-dose condition set hi gher shock intensities for the opponents than those in the placebo condition. The main effect of expectancy was not significant, nor was there a significant interaction of dose X expectancy. On the other hand, participants in the high-dose condition did select more severe shock intensity levels as Â“opponentÂ” shock levels increased. This findi ng appeared to be driven by participants who had scored high on the alcohol-aggression sc ale of the EDQ. Thus, it ap peared that intoxicated participants with high expectancies for alc ohol-related aggression we re the most reactive to increased levels of provoca tion. It is possible that high le vels of alcohol consumption and provocation are both necessary for expe ctancy effects to emerge (Chermack & Taylor, 1995). However, it may also be the ca se that alcohol-aggres sion expectancies are not adequately measured.
46 A few other studies (Dermen & Geor ge, 1989; Leonard & Senchak, 1993; and Quigley & Leonard, 1999) indicated that th e belief that alcohol leads to aggression does moderate the alcohol-aggressi on relationship. These studies ar e correlational and, in one (Quigley & Leonard, 1999), the findings did not hold up over time. A recent study (Leonard, Collins, & Quigley, 2003) investigated a host of variables (e.g., personality and socio-cultural factors) related to alcohol consumption and aggression within bar environments. With respect to alcohol-aggr ession expectancies (measured using the Alcohol Effects Questionnaire; Rohsenow, 1983), the investigators hypothesized that these expectancies would moderate the al cohol-aggression relati onship, and that the belief that alcohol causes aggr ession would be related to both the number of episodes resulting in aggression and a ggression severity. They found that a belief that alcohol causes aggression was not necessary for an association between alcohol consumption and aggression. They also found that participants were more lik ely to behave aggressively during an episode when they held this belief, but that the opponent was less likely to be harmed during the episode. The authors suggest ed that alcohol-aggression expectancies might serve in the initiation of aggression, but not in th e continuation, escalation, or cessation of an aggressive episode. Further, the authors found that once aggression was initiated, forces within the social environmen t (people instigating the participant and his opponent during the aggressive episode, and no one trying to defuse the situation) were predictive of more severe aggression and greater harm to the opponent. Interestingly, angry temperament was not reliably associated with aggression in this study. But, again, the authors suggested that individual differences may in fluence the initiation of aggression, but other factors (i .e., eggers-on) may be more crucial in the escalation of aggression (Leonard, Collins, & Quigley, 2003). Findings from an earlier st udy offer some experimental support for this alternative explanation. Bailey and Taylor (1991) found that when men self-reported moderate to high levels of aggressive tendencies (as m easured by the Assault subscale of the BussDurkee Hostility Inventory, Buss & Durkee, 1957), they were significantly more likely to set higher shock levels at a faster rate toward their pr ovokers in a reaction time task. Although the level of shock intensity se t by men who self-reported nonaggressive
47 tendencies never reached the level set by men who self-reported aggressive tendencies, the former clearly set higher shock levels wh en intoxicated. The authors speculated that when the intentions of the target were ambiguous (during block one), a high dose of alcohol appeared to have an instigative e ffect upon all of the i ndividuals in the study, regardless of whether they se lf-reported high, moderate, or low levels of aggression. When the antagonist was clearly more provoca tive (blocks two and th ree), the effects of alcohol appeared to depend more on dispos ition. That is, thos e who self-reported moderate or high levels of aggression set in creasingly higher shock intensities, whereas those low on aggression were more restrained. A more recent study (Parrott & Zeichner 2002) partially replicated the above results. Participants were categorized as low, moderate, or high tr ait anger ac cording to their responses on the Trait Anger S cale (TAS, Spielberger et al., 1980, 1983). Participants completed a modified TAP in e ither an alcohol or a no-alcohol condition. Shock intensity and duration, as well as the pr oportion of shocks set at the highest level, served as indices of aggression. Regardless of beverage condition, men who were categorized as moderate or high on trait ange r displayed significantly higher aggression on all of the indices of aggr ession. Unexpectedly, only intoxi cated participants rated as moderate trait anger selected higher shock intensities and a greater proportion of shocks at the highest level than their sober counterpa rts. The authors suggest ed that the lack of difference for low trait anger participants like ly reflects an aggressi on-inhibiting effect. For high trait anger participants, the lack of difference may reflect a ceiling effect. They also suggested that a placebo condition and a measure of alcohol-a ggression expectancies might have enhanced the interpre tation of the current findings. As previously mentioned, Giancola et al (2002) have concluded that provocation is the most reliable predictor for alcohol-rela ted aggression across ge nder. Even if this conclusion is accurate, other factors clearly moderate the alcohol-aggression relationship. In an earlier study, Giancola and Zeichner ( 1995) investigated the combined predictive ability of subjective intoxication, BAC level, provoca tion, and aggressive personality traits on physical aggression in men and wome n. They found that aggressive personality traits and BAC level predicted physical aggression under both high and low provocation
48 for men. None of the variables were predic tive of aggression for intoxicated women, and subjective intoxication was not predictive of aggression for men when provocation was high. In a more recent study, Giancola (2002) again found that provocation was the strongest elicitor of aggressi on in a modified TAP. More importantly, alcohol was more likely to increase aggression for men with highe r levels of trait anger as measured by the Spielberger Trait Anger Scale. The author s uggested that research should continue Â“to delineate a multivariable risk profileÂ” in the e ffort to predict when aggression is likely to occur subsequent to alcohol intoxication (G iancola, 2002, p. 1357). In an attempt to specify additional factors with in a Â“risk profileÂ” Giancola used a similar methodology in another study (2003) and measur ed self-reported levels of empathy (empathic concern for others and the ability to see things from a nother personÂ’s point of view as measured by two subscales of the Interpersonal Reactiv ity Index). Alcohol was found to increase aggression for men who self-reported lower empathy on these two subscales. Interestingly, for both of the above studies, alcohol had no effect on female aggression regardless of trait anger or empathic concern (Giancola, 2002; Giancola, 2003). The above studies highlight some variab les that reliably influence the alcoholaggression relationship. Provocation appears to be a crucial situational variable for men and women, and trait anger appears to be a crucial individual difference variableÂ— especially for men. For women, potential Â“riskÂ” vari ables have proven ha rder to identify. One study (as cited in Dougherty, et al., 1999) found that women who self-reported menstrual symptoms were more aggressive than those who did not. The authors summed up the current state of research well when th ey concluded that stud ies like this Â“clearly underscore the need for taking into account in dividual characteristic s that may help us better understand why alcohol increases aggressi on in some persons but not in othersÂ” (p. 329). Attentional Effects and Automaticity Another aspect of the alcohol-aggre ssion relationship i nvolves attention. Although alcohol intoxication do es not appear to change attentional capacity (Lamb & Robertson, 1987), it appears to influence the relative impor tance of the most salient
49 information from the internal and extern al environment. In one study (Jeavons and Taylor, 1985), the attention of half of the intoxicated partic ipants and half of the sober participants were directed to ward a nonaggressive norm intended to reduce participantÂ’s aggression toward a bogus opponent. Mean shock settings by each group clearly indicated that intoxicated participants w hose attention was not directed toward the nonaggressive norm were the most aggressi ve. For intoxicated participants whose attention was directed toward the nonaggressive norm, thei r levels of aggression were comparable to the sober participants, and lowe r than participants who were not provided with a nonaggressive norm. Zeichner, Pihl, Niaura, and Zacchia (1982) al so attempted to eval uate the role of attention in the production of alcohol-mediate d aggression. Some intoxicated participants were forced to attend to the consequences of their behavior (a t one indicating how much pain an opponent felt after receiving shock), so me were distracted from attending to those consequences, and some did not receive any atte ntional instructions at all. Zeichner et al. expected that participants in the forced-att ention condition could not fail to attend to the information about how much pain their opponent was experiencing and that this processing of relevant information woul d lower aggressive responding. They were surprised to find that those in th e forced-attention condition actually increased the duration of time that participan ts pressed the shock button. In contrast, for participants who were distracted from the pain, shock durations were signif icantly shorter. The authors concluded that an information-pr ocessing deficit in terpretation was not applicable. However, they also suggested that alcohol restricted attention to the shock manipulation rather than the behavioral conti ngencies (which are more distal in nature). In this case, their self-focus ed attention combined with alcohol may have been more arousing (e.g., they become more aware of threat of harm). Those that were distracted from the salient threat of harm were able to inhibit their aggr essive responding. Of course, an interpretation such as this awaits furt her investigation. To date, there are no known alcohol studies that have investigat ed the interaction between automaticity and aggression. The current study may validate a methodology that can look at the automatic activation of knowle dge structures relate d to aggression and
50 alcohol cues for those who self -report higher levels of tr ait anger or higher alcoholaggression expectancies. Measurement of a ch ronic, automatic effect is considered necessary for understanding the alcohol-aggression link. The Current Study The current study investigated whether hi gh self-reported levels of aggression, trait anger, or alcohol-aggression expectanci es (which all reflec t chronically accessible knowledge structures) are related to the salien ce of aggression stimuli in the presence or absence of alcohol cues (which both reflect external cues). This study assessed the predictive utility of the parafoveal visual versus dichotic listening methods of presentation to demonstrate the effects of self-relevant aggr ession cues upon two behavioral measures of attenti onÂ—reaction time and error rate. Hypotheses Three main hypotheses were formulated in regard to oneÂ’s performance when aggression cues are presented via a computer task either dichotically or parafoveally: 1. Participants who self-report hi gher levels of trait anger will demonstrate longer latencies and higher error rates (more attent ional interference) when exposed to selfrelevant cues of aggression than those who report lower levels of trait anger. This effect will hold whether participants are test ed in the presence or absence of alcohol cues. 2. Participants who self-report higher levels of alcoho l-aggression expectancies will demonstrate longer latencies and higher erro r rates when exposed to aggression cues than those who report low levels of alc oholaggression expectancies. Setting should moderate this effect. That is, the effect s hould hold only for participants tested in the presence of alcohol cues. 3. Higher alcohol-aggression exp ectancies will predict longe r latencies to respond and higher error rates on the comput er tasks after the effects of trait anger are partialled out. However, this effect will hold only for participants tested in the presence of alcohol cues. Although a specific hypothesis was not formulated in regard to the relative predictive utility of the parafoveal visual computer task versus the dich otic listening computer task,
51 patterns of significance and effect sizes were examined to suggest which methodology may be most helpful for investigating attentional interference, trait characteristics, and alcohol cues.
52 Method Design This study included two dependent measur es (error rate and reaction time) and two independent variables (Word Type and Se tting). Word Type had two within subject (WS) levels (NonAggression and HiAggressi on) and Setting had two between subject (BS) levels (Barroom and Cleanroom). Ther efore, this study was a 3 (WS) X 2 (BS) mixed design. Error rate and reaction time were analyzed separately, as were Dichotic Listening Task (DLT) and Parafov eal Visual Task (PVT) data. Condition Assignment Participants were assigned to one of four conditions: Barroom Parafoveal, Barroom Dichotic, Cleanroom Parafoveal, or Cleanroom Dichotic. The attended Channel (Left vs. Right) was counterbalanced w ithin the DLT condition. Setting and Task assignments were decided by flipping a coin and filling the other cells by default as necessary. Since there were constraints upon depa rtmental availability of room space, if the next Setting condition was not available, the participant was run in the same Setting condition rather than being rescheduled. Power While effect sizes regarding error rate ar e theoretically related to the dependent variables examined in this study, our experi ences suggested that they would be quite small (e.g., Edington, 1996). Alt hough error rate was investig ated in the current study, sample size needed was based upon expect ed effects for reaction time. Similar methodologies investigating reaction time and/ or parafoveal presentation of stimuli (Bargh, 1996; di Pace, Longoni, & Zoccolotti, 1991; Ortells & Tudela, 1996) indicated that 100 participants (25 participants for each computer task and in the presence or absence of alcohol cues) should provide ampl e power for detecting mean differences in reaction times.
53 Power analyses regarding reaction tim e means were conducted for N = 79. Significant differences were found for Wo rd Type on the PVT with power at 1.00. Reaction time means across Word Type for th e DLT were not significant, and power was calculated to be .56. Participan t recruitment was ended at N = 79 (Dichotic X Cleanroom = 20; Dichotic X Barroom = 18; Parafov eal X Cleanroom = 19; and Parafoveal X Barroom = 22). Participants Three hundred eighty five undergraduate students at the University of South Florida completed questionnaire data for the firs t part of this study (P hase I). Eighty-five of the 385 students who completed the Phase I questionnaires participated in Phase II of the study. The data of six participants were not used in any analyses because examination of error rate data revealed that these participants had e rror rates that were unacceptably high. Since the other 79 participants were able to complete either ta sk with error rates no higher than 19%, it is more likely that the six participants with error rates in the range of 47% to 61% did not attend adequately to the instructions. Therefore, data for these six participants was excluded from all reaction tim e (RT) and error rate (ER) data analyses and demographics are reported for N = 791. The mean age for the Dichotic Listening Task (DLT) sample ( N = 38) was 22.95 years (SD = 4.06, range 19 to 35). For th e Parafoveal Visual Task (PVT; N = 41) sample, excluding one 63-year-old female, the mean age was 21.83 years (SD = 2.62, range 19 to 31). Mean ages were not significantly different for the two task types, t (62.75) = 1.44, p = .15. Fourteen of the 79 participants were male (18%) and 65 were female (82%). This overrepresentation of females is expected given the high number of female undergraduates within undergra duate psychology classes at th is university. To evaluate the difference among the proportion of males vs. females completing the PVT vs. the DLT, a contingency table analysis was conduc ted. Gender was not significantly different across tasks, Pearson 2 (1, N = 79) = .03, p = .88. The sample included 14 African American (17.7%), 11 Hispanic (13.9%), 51 Caucasian (64.6%), 2 Asian American (2.5%), and 1 Latino (1.3%) participant. Race/ ethnicity was not significantly different across tasks, Pearson 2 (4, N = 79) = 5.72, p = .22.
54 To evaluate whether students who never came in for Phase II were significantly different on demographic charac teristics or on the measures of interest for Phase I, a comparison group of Phase II noncompleters was randomly selected from the 300 remaining participants. This resulted in a comparison of 79 Phase II completers and 79 Phase II noncompleters. No differences em erged for age, gender, race/ethnicity, household income, the trait anger variables, or the alcohol-ag gression expectancy variables. Phase I Materials Participants were given a nu mber of measures to complete during Phase I. These measures were the State Trait Anger Expr ession Inventory (STAXI; Appendix B), the Aggression Questionnaire (BPAQ; Appendix C), the Expectancy Questionnaire for Alcohol and AggressionÂ—Low Dose (EQAAL; Appendix D), a questionnaire made up of other alcohol-aggression expectancy subscales (Appendix E), a demographics questionnaire (Appendix F), and a request fo r further participati on (Appendix A). Only the psychometric properties of the STAXI a nd the EQAAL measures will be considered below. The BPAQ and the questionnaire incl uding other expectancy subscales were included in Phase I for later study. The de mographics questionnaire and request for further participation will not be discussed further. State-Trait Anger Expression Inventory. The State-Trait Anger Expression Inve ntory (STAXI; Spielberger, 1988) was used to measure participantsÂ’ levels of anger proneness (tra it anger) as well as the manner in which they typically expre ss their anger. The STAXI evolved from earlier measures of the experience and expression of anxiety and anger as important factors in the etiology of hypertension and coronary heart disease in the late 1960Â’s (e.g., see Spielberger, Gorsuch, & Lushene, 1970; Spielberger, 1980) Spielberger has spent decades refining the measurement of state anger (operationali zed as a relatively short-lived emotional state) and trait anger (operationalized as a longer-standing personality characteristic) to assess individual differences in the experien ce of anger (State-Trait Anger Scale or STAS; Spielberger, 1980; Spielberger, et al., 1983). Spielberger maintained that individuals high in trait anger would more frequently perceive a wider range of situations
55 as anger provoking than those low in trait anger and that they would experience more frequent and more intense eleva tions in state anger over time. Although the measurement of st ate and trait anger proved us eful in some contexts, Spielberger, Sydeman, Owen, and Marsh (1999) realized that understanding the experience of anger is not enough to develop strategies and treatments for maladaptive anger. They claimed that it is essential Â“not only to distinguish, both conceptually and empirically, between the experience of a nger as an emotional state (S-Anger) and individual differences in anger proneness as a personality trait (T-Anger), but also to identify and measure the characteristic ways in which people express their angerÂ” (p. 1006). This led to the development of the Anger Expression scale or AX Scale, which provided a distinction between anger-in (suppressed anger; AX/In) and anger-out (verbally or physically expre ssed anger; AX/Out). Research with the AX scale indicated an anger control factor, which was develope d into the third subs cale of the AX, the anger-control (AX/Con) subscale. The STAS a nd the AX scales were combined to create the State-Trait Anger Expression Inve ntory (STAXI; Spielberger, 1988). The STAXI has continued th e role of the STAS and AX in research on the relationship between anger and health-related factors such as hypert ension, as well as a variety of other constructs (for examples of the use of the STAS, AX, and STAXI in research the interested reader is referred to Spielber ger, Reheiser, & Sydeman, 1995). The STAXI has also been proposed and eval uated as a screening and outcome measure with mixed but promising results (e.g., Fo ley, Hartman, Dunn, Smith, & Goldberg, 2002; Mahon, Yarcheski, & Yarcheski, 2000; Co rnell, Peterson, & Richards, 1999). The STAXIÂ’s ability to measure anger-related states, traits, and its expression (i.e., aggression) suggests that it can be useful for providing eviden ce that these constructs are highly correlated with aggressive behavior in laboratories, and, more importantly, in naturalistic settings. The STAXI is comprised of: a 10-item tr ait anger (T-Anger) scale that measures individual differences in a nger proneness; a 10-item stat e anger (S-Anger) scale that measures oneÂ’s current subjective feelings of anger; and a 24-item Anger Expression (AX) scale that measures internalized, s eething anger (AX/InÂ—8 ite ms), externalized
56 aggressive behavioral tendencies (AX/OutÂ—8 items), and effort expended controlling the expression of anger (AX/ConÂ—8 items). The AX /EX measure indexes the frequency that anger is experienced and expressed and is calculated by combining items from other STAXI scales. A factor stru cture analysis of the TAnger and S-Anger scales (Spielberger, 1988) indicated that the T-Anger Scale should be broken into two subscales. Angry Temperament (T-Anger/T) is intended to measure oneÂ’s tendency to experience and express anger without provocation. Angr y Reaction (T-Anger/R) is intended to measure oneÂ’s tendency to express anger when criticized or treated unfairly. The STAXI has been found to have good psychometric properties (Fuqua, et al., 1991; Moses, 1992; and, Spielberger, 1988). Fact or analysis of the S-Anger scale yielded high item-remainder correlations and alpha coef ficients of .93 for both sexes. Internal consistency of the T-Anger subscales was ev aluated separately for males and females using college and Navy samples (Spielberger 1988, p. 8). Item-remainder correlations were acceptably high for both subscales and al pha coefficients ranged from .84 to .89 for the T-Anger/T subscale and from .70 to .75 for the T-Anger/R subscale. In the current study, internal consistency estimates of reliab ility were calculated with both males and females and yielded a somewhat lower, al though still acceptable, result for S-Anger (alpha = .85). For the T-Anger subscales, co efficients for males and females together were comparable to previously obtained results with alpha = .85 for T-Ang/T but somewhat lower for T-Ang/R with alpha = .68. Item-remainder correlations for the AX/In and AX/Out subscales of the AX scale in SpielbergerÂ’s (1988) study were much lower but still satisfactory and alpha coefficients for all thr ee of the AX subscales ranged from .73 to .85. Coefficients were highest for the AX/Con subscale (.84 for females and .85 for males) and lowest for the AX/Out s ubscale (.75 for females and .73 for males) with AX/In coefficients falling in between (.81 for females and .84 for males). In the current study, AX/In coefficient alpha wa s .78, AX/Con alpha was .82, and AX/Out alpha was .78. Correlations among the STAXI scales in the Spielberger (1988) study were as expected (e.g., essentially zero between AX/In and AX/Out). Test-retest stability coefficients for the state and trait anger scal es of the State-Trait Personality Inventory
57 (STPI; Spielberger, Jacobs, Crane, Russell, Westberry, Barker, et al., 1979) and for the AX subscales have been reported for males and females separately (Jacobs, Latham, & Brown, 1988). As would be expected, coefficien ts for the state anger scale (.27 for males and .21 for females) were much lower than fo r the trait anger scale (.70 for males and .77 for females). Additionally, coefficient values for the AX subscales (a range of .64 to .70 for males and .73 to .81 for females) were comparable to the trait scale values. The T-Ang/T and T-Ang/R subscales were of most interest for the current study. The mean T-Ang/T score reported by Spielb erger (1991) for college students was 6.56 for males ( SD = 2.67) and 6.71 for females ( SD = 2.73). The mean T-Ang/R score reported by Spielberger for colle ge students was 9.84 for males ( SD = 2.55) and 10.18 for females ( SD = 2.60). In the current sample, mean T-Ang/T scores for the Parafoveal Visual Task (PVT) were 5.57 for males ( SD = 1.81; N = 7) and 6.00 for females ( SD = 1.71; N = 34). For the Dichotic Listening Task (DLT), mean T-Ang/T scores were 5.71 for males ( SD = 1.89; N = 7) and 5.90 for females ( SD = 2.34; N = 31). For the T-Ang/R subscale, mean scores for the PVT were 8.71 for males ( SD = 2.75) and 8.44 for females ( SD = 1.74). On the DLT, mean scores were 8.14 for males ( SD = 2.12) and 8.19 for females ( SD = 2.56). The maximum obtainable score for each subscale is 16 and the lowest is 4. This indicates that most res pondents indicated having an angry temperament somewhere between Â“almost neverÂ” and Â“som ewhat.Â” For the T-Ang/R subscale, sample means were somewhat higher than the means for T-Ang/T but they were still considerably lower than the means that Spielberger reported. Concurrent validity of the STAXI was provided using 270 naval recruits and 280 college undergraduates (Spielberger, 1988). T-A nger scale scores were compared with scores from the Buss-Durkee Hostility Inve ntory (BDHI) and the Hostility and Overt Hostility scales of the Minnesota Multiphasic Personality Inventory (MMPI). All correlations were significant at <.01 for both males and female s. Spielberger (1988) also studied the correlations between the STAXI T-Anger and S-Anger scales and the Eysenck Personality Questionnaire (EPQ) subs cales and the Trait and State Anxiety and Curiosity Scales of the State-Trait Personal ity Inventory (STPI). Moderate correlations between the EPQ Neuroticism scales and the T-Anger scale as well as between the State
58 and Trait Anxiety scales from the STPI and the T-Anger scale were significant and interpreted by Spielberger (1988) as consistent with the theory that individuals high in neuroticism and trait anxiety frequently experi ence angry feelings th at they suppress (p. 12). Correlations of the T-Anger scale with the EPQ Extraversion scale and the STPI Sand T-Curiosity scales were es sentially zero and suggested that T-Anger is not related to those personality constructs. Convergent and divergent validity for th e AX/EX scale has also been provided (Spielberger, Johnson, Russell, Crane, Jac obs, & Worden, 1985). Correlations of the AX subscales with the STPI state and trait curios ity subscales were rela tively nonexistent, but significant correlations were found with trait anxiety for both males and females (ranging from .24 to .34). Correlations between the AX/ EX total anger expression scores and the STPI anger measures were lower although still significant with the exception of the STPI T-Anger/T and AX/EX correlation for fema les, which was essentially zero. In an analysis of the 44-item STAXI (Fuqua, Leonard, Masters, Smith, Campbell & Fischer, 1991), as well as a replication analysis (Forgays, Forgays, & Spielberger, 1997) the structure of the measure was examin ed to determine whether the use of the different subscales is justified. The researcher s concluded that the st ructural validity of the STAXI was better than expected and that th e scale structure they found was similar to that claimed for this instrument. Expectancy Questionnaire for Alcohol and AggressionÂ—Low Dose. The current study views aggres sive scripts as knowledge structures in memory that represent information about when and w hy it might be appropriate to use aggression in a given situation, and what will happen as a result. Alc ohol scripts are viewed in a similar fashion. It is reasona ble to expect that knowledge structur es about aggression secondary to alcohol use will contain info rmation about the circumstances under which someone drinking alcohol would be aggres sive and what the outcomes might be. Therefore the Expectancy Questionnaire fo r Alcohol and AggressionÂ—Lo Dose version (EQAAL; Epps, Hunter, LeVasseur, Steinber g, & Hancock, unpublished manuscript) was used in the current study to measure expectan cies that participants hold about behaving aggressively following the consumption of lo w but behaviorally significant doses of
59 alcohol. Questions from other measures that tap alcohol a nd aggression related expectancies were also included for la ter study (i.e., the Alcohol Expectancy Questionnaire, Brown, Goldman, Inn & A nderson, 1980; the Drinking Expectancy Questionnaire, Young & Knight, 1989; the Comprehensive Effects of Alcohol Questionnaire, Fromme, Stroot, & Kaplan, 1993; the Effects of Drinking Alcohol scale, Leigh & Stacy, 1993; and, the Alcohol Exp ectancy Questionnaire-3, George, Frone, Cooper, Russell, Skinner, & Windle, 1995). Th e items taken from these measures are provided in Appendix E but the ps ychometric properties of their respective scales will not be discussed. The EQAAL is a 23-item scale representing a cognitive-behavioral taxonomy of alcohol-aggression expectancies divided into f our factors. Factor an alysis indicated two affective factors labeled Unprovoked Ange r Expectancies (UnpAng; 8 items) and Reactive Anger Expectancies (AngReac; 7 items), one cognitive factor labeled Expectancies of Hostile Cognitions (HostCog; 3 items), and one behavioral factor labeled Expectancies to Maintain Control (ExpCon; 5 items). Confirmatory factor analysis revealed the following factor loadings: UnpAng = .70; AngReac = .68; HostCog = .71; and, ExpCon = .78. High internal consistency was demonstrated by computing ChronbachÂ’s alpha (UnpAng = .88, AngReac = .81, HostCog = .76, and ExpCon = .82). Intercorrelations between the various s cales ranged from -.014 to .53. The higher intercorrelations were among the two anger f actors and the hostile cognitions factor. Sixweek test-retest reliability with a separa te group of students revealed the following Pearson product-moment correlations : UnpAng = .80, AngReac = .57, HostCog = .56, and ExpCon = .79. No significant differences were found on any of the subscales according to gender or ethnici ty. Although the EQAAL shows pr omise as a more precise measure of alcohol-aggression expectancies studies that provide evidence of discriminant and convergent validity are lacking. The current study provides additional evidence for internal consistency. Coefficient alphas for three of the four fact ors were quite high and were as follows: Unprovoked Anger Expectancies = .93, Exp ectancies of Hostile Cognitions = .77, Reactive Anger Expectancies = .88, and Exp ectancies to Maintain Control = .90.
60 Phase II Materials During Phase II, participants comple ted the Positive and Negative Affect Schedule (PANAS; Appendix G). Then they co mpleted either the Dichotic Listening Task (DLT) or the Parafoveal Visual Task (PVT) in the presence or absence of alcoholrelated stimuli. During either task they we re exposed to aggressive and nonaggressive stimuli (word stimuli are included in Appe ndix H) while they responded to a reaction time task. Upon completion of the computer task, they completed a recognition task (Appendix I) followed by the PANAS again, an d the Brief Drinker Profile (Appendix J; for later study). Finally, part icipants were debriefed about the study (Appendix K) and awarded their extra credit points. Positive and Negative Affect Schedule. It was possible that exposure to aggr ession stimuli would induce or increase negative affect. To investigate this possibility, the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988) was gi ven to participants before and after they completed the computer task in Phase II. Both times the participants were instructed to read the adjectives and Â“indi cate to what extent you feel th is way right now, that is, at this very moment.Â” The PANAS consists of 10 positive adjectives and 10 negative adjectives that are assumed to represent two orthogonal dimensi ons of moodÂ—positive affect and negative affect. In order to develop these scales, a ra nge of descriptors (60 adjectives taken from Zevon and Tellegen, 1982) that loaded .40 or gr eater on the relevant positive or negative factor and that did not load |.25| on the other factor, were selected. Of 12 positive descriptors, two more items were dropped th at had relatively high secondary loadings on the negative affect factor for a final pool of 10 items. Of th e 25 negative descriptors, two content categories were dropped altogether l eaving two items from each of five content categories (distressed, angry, fearful, guilty, a nd jittery). For example, the items that represent Â“angryÂ” are hostile and irritable The authors obtained PANAS ratings using six time frames (moment, today, past few days, past few weeks, year, and in general) and found acceptably high alpha reliabilities (rangi ng from .86 to .90 for positive affect and from .84 to .87 for negative affect) and low correlations between the scales (ranging from
61 -.12 to -.23, indicating about 15% shared variance). Test-r etest reliabil ity indicated stability over a two-month time period, with greater stability over longer time frames. Validity for the PANAS was also accept able, based upon good factorial validity and expected correlations with measures of re lated constructs (Watson, Clark, & Tellegen, 1988). In the current study, coefficient alphas for the positive affect scale were .87 (Time 1) and .88 (Time 2). Coefficient alphas for the negative affect scale were .64 (Time 1) and .72 (Time 2). When comparing the positive and negative affect scales at Time 1, alpha was .22. At Time 2 alpha was .04. Overa ll, the scales appear to measure orthogonal constructs as maintained by Watson, et al. (1988). Dichotic Listening Task (DLT) The participantsÂ’ primary task was to attend to one channel of a DLT while they simulta neously performed a computer task. The computer program was created using Superl ab Pro, Version 2, which presented word pairs dichotically (through h eadphones) and number strings ( on the monitor) using a Dell PC. During the computer task, participants i ndicated by the press of a button whether the center number in a five-digit string of nu mbers was odd or even. Between number strings, a capital X was placed in the center of the screen and served as the fixation point. Two seconds later, the .wav file played and the number string appeared. After two seconds, or as soon as the odd or even button was pressed, the X reappeared. The computer program recorded, in milliseconds, the time between onset of the number string to the pressing of the red (Â“oddÂ”) key or the blue (Â“evenÂ”) key. If no key was pressed within 2 seconds of stimulus presentati on, RT was recorded as .00 seconds and the response was marked as incorrect. All particip ants were presented with 10 Blank trials, 10 Practice Trials, 80 Word Type Trials mixed with 10 Blank Trials, and 10 final Practice Trials. Within a type of trial, a differe nt randomly generated order of stimuli was presented to each participant to control for order effects. Word Pairs. The words presented to the unatt ended channel consisted of words rated as high aggressive, ambiguously aggres sive, low aggressive or nonaggressive (as used in Edington, 1997; see Appendix H). Tw o types of aggressive words (high and
62 nonaggressive) were selected in order to examine whether these levels would be predictive of differential cognitive inte rference. Low aggressive and ambiguous aggressive words were also included. However, these two wo rd types were included for later study. All of the words pr esented simultaneously to the attended channel consisted of additional nonaggressive words. Twenty word pairs comprised each of the Word Type conditionsÂ—called NonAgg, LoAgg, AmbAgg, and HiAggÂ—for a total of 80 word pairs. All four types of words were selected from a normed databa se of over 5000 words and their associated links compiled by Nelson, McEvoy and Schr eiber (1999). Words in the HiAgg category were associated with aggression-related c oncepts from 60-95% of the time. Examples include brawl, stab and fight Words in the AmbAgg category were related to aggression concepts 35-50% of the time. Examples include tank strike and punch LoAgg words were associated with aggression-related concepts 10-20% of the time and included mask fray and hide The NonAgg words had no evident a ssociation with aggression-related concepts. Examples included store few and desk Additionally, both words in the word pairs began with the same consonant sound. For example, stab was paired with store Ten words, without a paired word, were presented before the practice words (Blanks). Ten NonAgg word pairs were presen ted at the beginning of the computer task to investigate practice effects (Practice words) and another 10 Blanks were mixed in with the 80 pairs of experimental words. This yi elded 100 word pairs and 20 Blanks that were presented at a rate of one pair approximately every four seconds. Word pairs were recorded and saved as sound .wav files using the Adobe Audition program. Detailed information about the word stimuli is provided in Appendix H. All of the words were recorded using the same male voice. The Superlab program accessed these files and used them for th e Dichotic Listening Task. The audio presentation typically lasted from .50 to .80 seconds. Parafoveal Visual Task. The Parafoveal Visual Task (PVT) computer program was also written with Superlab Pro, Ve rsion 2.14 and presented on a Dell PC. Participants completed the same odd vs. ev en number decision-making task as in the Dichotic Listening Task. However, the expe rimental words (Word Type) were presented
63 as written words within the parafoveal vi sual field of the pa rticipant. Single nonaggressive words played in stereo on the headphones. The PVT methodology resembled that of Chartrand and Bargh ( 1996), Experiment 2. On a computer screen, these researchers presented a word and a subseq uent masking stimulus at an angle of 45, 135, 225, or 315 from asterisks in the center of the screen (the fixation point). These four quadrants coincided with an area appr oximately 7.6 cm from the fixation point. Thus, stimuli were presented to the parafoveal visual field, which has been shown to be from 2 to 6 of visual angle. In order for the experimental word stimu li to appear within the parafoveal range, participants in the current study were situated in front of the screen such that their eyes were 65 cm from the fixation point. The cen ter of the word stimulus appeared 2.6 cm above, below, to the left, or to the right of the fixation point. Detailed information about the parafoveal stimuli is provided in A ppendix H. The distance of 65 cm from the monitor insured that if the participant lean ed forward 5 cm (approximately 2 inches) or back 5 cm, the stimulus would still be pres ented within 2 to 6 of visual angle. Chartrand and Bargh (1996) used brief prime word duration, immediate masking, and parafoveal presentation of words to pr event conscious awaren ess of subliminally primed stimuli. These precautionary measures were observed in the current study as well. Experimental words were presented for a dur ation of 100 milliseconds, which is halfway between the duration Bargh et al. (1995) used and the minimum that Rayner (1978) suggested. The masking stimulus, Â“WQXQW,Â” replaced the experimental word for the next 60 milliseconds. Response time to pr ess the odd or even key after stimulus presentation was recorded by the Cedrus RB-610 Response Pad. Cedrus has reported that the response pad provides reac tion time data that is accura te to within 1 millisecond. Experimental Settings. The barroom is a room at the University of South Florida created as an analogue drinking environment filled with typical drinking paraphernalia. The Â“cleanÂ” room is another room in the sa me building that is devoid of any alcohol related cues. Participants were assigned to one of the two settings in order to examine whether alcohol cues would increase or decr ease overall reaction times or error rates to aggression stimuli.
64 Recognition Task. Recognition tasks are often used to determine whether awareness of stimuli has occurred (B argh, 1982; Edington, 1996; Epps, Hunter, LeVasseur, Steinberg, & Hanc ock, 1997). Usually, some words that were presented during the relevant task are listed on a sheet of paper along with some words that the participant had not been exposed to, and the pa rticipant is asked to check off words he or she recognizes. Better than chance recognition of words to which participants were exposed may indicate awareness of those st imuli. In the current study, minimal to no awareness of the stimuli presented via th e Dichotic Listening Task (DLT) or the Parafoveal Visual Task (PVT ) was expected to occur. For this study, one recognition task was cr eated to determine wh ether participants were momentarily aware that alcohol or aggr ession related stimuli were presented to the unattended channel in the DLT or parafoveally in the PVT. The recognition task that was used is provided in Appendix I. The rec ognition task contained 14 Â“ControlÂ” words, which were words that participants were instru cted to listen to in the attended channel of the DLT or in stereo for the PVT. The recogn ition task also contained eight Experimental words that had been presented to the unatte nded channel or parafoveally. Eighteen New words were selected that pa rticipants had not been expos ed to during any task. These words were similar in length and general mean ing to words that they had been exposed (e.g., hate and maim were selected as equiva lent aggression words for shout and gun ). New words were selected from Nelson, et al. (1999) database. Participants were instructed to check off words that they recognized seeing or hearing during their respective task. The additional questions on the recogn ition task were inte nded to provide information about whether the participants ha d guessed the purpose of the computer task, or if they were aware of a connection between completing th e computer task and filling out a variety of aggression and alcoholaggression related que stionnaires during a previous, supposedly unrelated expe riment (i.e., during Phase I). Procedure Phase I was conducted over Fall 2004 through Summer 2005. Students were recruited via an online research data mana gement program, Experimentrak, used by the
65 University of South FloridaÂ’s Psychology Department. Participants completed the questionnaires either online or in groups of 10 to 20 in sm all classrooms. Minor changes were made to the questionnaires, Informed C onsent, and the Debrief form in order to accommodate an online format. Participants were asked to complete the AQ, the EQAAL, the STAXI, and a demographic ques tionnaire (see Appendix F) for extra credit toward their course grade. After completing th e questionnaires, partic ipants were asked to indicate their willingness to be contacted at a later date for particip ation in a larger study (see Appendix A) and were given a debrie f form (see Appendix K). For the online participants, the request for participa tion in another study was omitted. Whether participants completed the Phase I questionnaires in classrooms or online, they were contacted at least two later and asked to participate in Phase II. The time delay was included to reduce the possibili ty that alcoholor aggressi on-related concepts would be active, or recently primed, during Phase II of the study. Participants who had indicated interest in the larger study on the questionnaires; who did not indicate motor, hearing, or vi sual impairments; and who indicated that English was their native/first language were contacted at least two weeks later by phone or e-mail and invited to participate in a com puter task study (Phase II). Those completing the questionnaires online had al ready passed a screener (which included similar questions about impairments and first language) and were contacted at least two weeks later by email and invited to participate in Phase II. Whether recruited by phone, e-mail, or online, some deception was necessary to increase the likelihood that participants were unaware that the two phases of the st udy were related. All participan ts of Phase II were fully debriefed about the connection between th e two phases before leaving the lab. Recruitment of participants to return for Phase II of the study was inadequate. A few weeks after Phase II was initiated, a lottery was instituted. When participants were contacted about particip ation in Phase II, they were told that when they arrived for the study their participation identification number (IDN) would be entered into a lottery that included three $20 drawings. They were told that if they then decided to participate, their IDN would be entered into the drawing ag ain. When recruitment did not appear to improve, Informed Consent was changed to specify monetary remuneration and
66 participants were paid $5 for their particip ation in addition to being included in the lottery. Before a participant arrived, he or she was assigned to one of the two tasks and one of the two settings. Upon arri val, participants were asked if they had any impairment that would prevent them from completing th e computer tasks and if English had been their primary language since birth. Those that were appropriate for the study were shown to the designated area of the lab and seated in front of the computer. They were asked to complete the PANAS before instructions were given regarding the computer task. For the Dichotic Listening Task, participan ts were told that they would be doing two tasks at one time. Their most important task was to listen to the word presented in the ear they were instructed to attend and count the number of words that started with the letter Â“L.Â” They were also told to ignore a ny words they might hear in the other channel. Participants were told that if they could re port the correct number of L words within plus or minus two, their IDN would be entered into the lottery again. At the same time that word pairs were presented, participants co mpleted a decision-making task. Participants were instructed to decide wh ether the center number of a five -digit stimulus appearing in the center of the computer screen represen ted an even number or an odd number (e.g., 04863). Participants were asked to respond as quickly and as accurately as possible by pressing a key to indicate that the number wa s odd, or a different ke y indicating that the number was even. For example, for the word pair kill-key participants would have ignored the word kill and would not have counted the word key and, at the exact same time, they would have pressed the EVEN butt on to indicate that the center number in the string 04863 was even. For the Parafoveal Visual Task, participan ts completed the same two tasks. There were two fundamental differences: 1) all of the words presented on the headphones were single nonaggressive control words presented in stereo, and 2) the experimental word of the word pair was presented on the computer sc reen to the participan tÂ’s parafoveal region (i.e., at 2 to 6 angle from the fixation point). For the word pair scream-lake participants would have heard the word lake in stereo on their headphones, they would have counted it as an L word, they would not have been expected to notice the word scream flashed to
67 their parafoveal region, and, at the same ex act time, they would have pressed the ODD button because the center number in the string 23501 is odd. To insure that the experimental word stimuli were always presented to the parafoveal region, a ruler was used to meas ure 65 centimeters betw een the participantÂ’s eyes and the center of the monitor. The ch air was stationary and participants were instructed to maintain this position during th e computer task and told that the distance between their eyes and the computer would be measured again after completion of the task. For either computer task, all 100 word-p airs (and 20 Blanks) took approximately 11 minutes to complete. Once instructions had been given, the participant donned the headphones, the experimenter left the room and the participant began the task. Upon reentering the room, the experimenter asked a ll participants to report the number of L words they heard and to complete the Recogni tion Task (Appendix I). Participants were then given the PANAS to complete for a sec ond time. Next, they were asked a number of questions about their patterns of drinking in order to obtain an estim ate of their standard ethanol consumptions units ove r the last three months. This estimate was obtained using a brief version of the Comprehe nsive Drinker Profile (Marlatt & Miller, 1986; Appendix J). Finally, all participants were told about the purp ose of the study and the minor deception involving the connection between Phase I and Ph ase II. They were then given a debrief form (Appendix K) to take with them.
68 Results Descriptives Visual inspection of histograms indicated that the Reaction Time (RT) sample means for the two word types were norma lly distributed. Measures of kurtosis and skewness were within acceptable ranges (all within 1). For Error Rate, sample means were not normally distributed for the Dichotic Listening Task (DLT), with measures of kurtosis and skewness as high as 5.31 and 2.27, respectively. Overall, ER means for the DLT tended to bunch up around .00 (no errors) with relatively few means over .10 for word type HiAgg. Therefore, any results usi ng Error Rate means from the DLT should be interpreted with appropriate caution. Table 1 Means, Standard Deviations, and Intercorre lations for RT HiAggÂ—NonAgg Difference Scores and Trait Anger Predictor Variables for PVT Completed in the Presence of Alcohol Cues (N = 22) ________________________________________________________________________ Variable M SD 1 2 ________________________________________________________________________ RT HiAggÂ—NonAgg DS 52.76 64.67 -.68* .01 Predictor Variable 1. Trait Anger/Angry Temperament 6.36 1.71 -.30 2. Trait Anger/Angry Reaction 8.82 1.26 -________________________________________________________________________ Note. RT = Reaction Time; DS = Difference Scores; HiAgg = High Aggression Word Type; NonAgg = Non Aggression Word T ype; PVT = Parafoveal Visual Task p < .001
69 The possible bias that could exist in the data due to outliers for the RT data was considered. Response latencies higher than 2000 milliseconds (2 seconds) were recorded as missing data for RT and as an error. By design, this cutoff limited the potential for extreme outliers to exist in the data. When considering the two Word Types of interest (NonAgg and HiAgg) and a cutoff of three st andard deviations, there were roughly 45 outliers out of 3,160 data points (approximately 1.4%). However, the data were fairly evenly spread across the data and we re not deleted from the analyses. Table 2 Means, Standard Deviations, and Intercorre lations for RT HiAggÂ—NonAgg Difference Scores and EQAAL Predictor Variables for PVT Completed in the Presence of Alcohol Cues (N=22) ________________________________________________________________________ Variable M SD 1 2 3 4 ________________________________________________________________________ RT HiAggÂ—NonAgg DS 52.76 64.67 .01 -.29 -.47* .06 Predictor Variable 1. EQAAL Â– AngReac 21.18 6.81 -.22 .48 .62** 2. EQAAL Â– ExpCon 18.96 6.11 -.57** -.19 3. EQAAL Â– HostCog 6.86 2.27 -.27 4. EQAAL Â– UnpAng 16.59 6.40 -________________________________________________________________________ Note. RT = Reaction Time; HiAgg = High A ggression Word Type; NonAgg = Non Aggression Word Type; EQAAL = Expect ancy Questionnaire for Alcohol and Aggression Â– Low Dose; DS = Difference Score; AngReac = Angry Reaction Expectancies; ExpCon = Expectancies to Main tain Control; HostCog = Expectancies for Hostile Cognitions; UnpAng = Unprovoked Anger Expectancies; PVT = Parafoveal Visual Task p < .05, ** p < .01
70 Correlations among the trait anger measur es and HiAggÂ—NonAgg difference score means used in the regression analyses are presented in Table 1. Correlations among the EQAAL subscales and HiAggÂ—NonAgg differen ce score means are presented in Table 2. Preliminary Analyses Reaction Time for Word Type. The first step in evaluating each of the hypotheses for this study was to demonstrate differe ntial responding across Word Type. Mean RTs were calculated for each type of word and a Repeated Measures Analysis of Variance (ANOVA) was used with two levels (NonA gg and HiAgg Word Type) of the withinsubject factor and two levels (Setting: Ba rroom vs. Cleanroom) of the between subject factor. ANOVA was used in an attempt to c ontrol familywise error. DLT and PVT data were analyzed separately. Statistical analyses failed to indicate a significant interaction between Word Type and Setting for the DLT [ F (1, 36) = 3.05, p > .05] or the PVT [ F (1, 39) = .46, p > .05]. Therefore, main effects were inspected. The main effect of Word Type for the DLT was not significant [ F (1, 36) = .76, p > .05] with low power (.14). The main effect of Word Type for the PVT was significant [ F (1, 39) = 25.87, p < .001] with power at 1.00. RT means in milliseconds for the DLT were 696.32 ( SD = 172.53; NonAgg), and 684.82 ( SD = 154.97; HiAgg). RT means for the PVT for the two word types were 651.15 ( SD = 211.42, NonAgg) and 698.17 ( SD = 219.62; HiAgg). RT means as a function of Task X Word Type are shown in Figure 1. RT means in milliseconds for the D LT in the barroom setting were 665.24 ( SD = 152.69; HiAgg) and 703.79 ( SD = 176.95; NonAgg). In the cleanroom setting, RT means were 702.44 ( SD = 158.79; HiAgg) and 689.59 ( SD = 172.77; NonAgg). RT means in milliseconds for the PV T in the barroom setting were 730.15 ( SD = 239.29; HiAgg) and 677.39 ( SD = 219.48; NonAgg). For the PVT in the cleanroom setting, RT means were 661.14 ( SD = 194.12; HiAgg) and 620.76 ( SD = 203.26; NonAgg). RT means for the PVT as a functi on of Setting X Word Type are shown in Figure 2.
71 Figure 1. RT as a function of task type X word type. Figure 2 PVT RT Means as a function of setting X word type. RT as a Function of Task Type X Word Type620 640 660 680 700 720 DichoticVisual Task TypeReaction Time (Ms) NonAgg HiAgg PVT RT Means as a Function of Setting X Word Type550 600 650 700 750 CleanroomBarroom SettingReaction Time (Ms) NonAgg HiAgg
72 Error Rate for Word Type. The number of errors (incorrectly selecting Odd vs Even) made by each participant was summed and converted to a mean error rate (percentage) for both Word Types (NonAgg and HiAgg). The analyses used for ER mirrored those used for RT. Statistical analyses indicated that the interaction between Word Type and Setting for ER was not significant for the DLT [ F (1, 36) = .31, p > .05] or the PVT [ F (1, 39) = .67, p > .05]. The main effect of Word Type for the DLT was not significant [ F (1, 36) = .31, p > .05] with low power (.08). The main effect of Word Type for the PVT was significant [ F (1, 39) = 28.94, p < .001] with power at 1.00. ER mean percentages for the DLT for the two word types were 3% ( SD = 4.7%; NonAgg) and 2.6% ( SD = 4.6%; HiAgg). ER means for the PVT were 4.5% ( SD = 5%, NonAgg) and 10.6% ( SD = 6.1%; HiAgg). ER means are shown in Figure 3 as a function of Task Type X Word Type. Figure 3 Error rate as a function of task type X word type. Error Rate as a Function of Task Type X Word Type0 5 10 15 DichoticVisual Task TypeError Rate (Percentages) NonAgg HiAgg
73 Main effects for Setting were not signif icant for the DLT when considering ER means [ F (1, 36) = .39, p > .05; power = .09]. Main eff ects for Setting also were not significant for PVT ER means [ F (1,39) = .63, p > .05; power = .12]. ER mean percentages for the DLT in th e cleanroom setting were 3% ( SD = 4%; NonAgg) and 3% ( SD = 3%; HiAgg). In the barroom setti ng, ER mean percentages were 3% ( SD = 6%; NonAgg), 3% ( SD = 6%; HiAgg). ER mean percenta ges for the PVT in the cleanroom setting were 4% ( SD = 5%; NonAgg) and 11% ( SD = 4%; HiAgg) For the PVT in the barroom setting, ER mean percentages were 10% ( SD = 8%; HiAgg). It is apparent when looking across task that the pres ence of alcohol cues did not appear to make a difference in the distribution of erro r rates or reaction times. Implications of Differential Responding Across Word Type. The significant main effect of Word Type on Error Rate was unexp ected. As previously noted, error rate was not found useful for detecting mean differences in word type on a dichotic listening task (Edington, 1996). The lack of mean differences for the DLT was replicated in the current study. But, for the PVT, ER means differed si gnificantly by Word Type. According to the theory driving the current methodology, highe r error rates for words that are more aggressive in nature should reflect more at tentional interference fr om (or the pulling of attention toward) those words. The current results for the PVT supported this assumption for ER and RT. Both ER and RT appeared to measure th e ability of aggression words presented parafoveally to pull attention away from a decision ma king task. Although this finding is important because it suggests that the PVT may be a useful methodology for investigating automatic atte ntion, the predictiv e value of the overall methodology was central to this thesis. Therefore, regression analyses were necessary to better understand the relationship between participantÂ’s self-re ported trait aggressi on or alcohol-aggression expectancies and the attentional interference that could arise from aggression words. Overall, ER means were higher and late ncies to respond (RT means) were longer when participants were exposed to high aggr ession stimuli presented parafoveally. This finding made it possible to use the magnitude of the difference to reflect each individualÂ’s sensitivity to HiAgg words vs. NonAgg words. That is, the difference scores in the
74 current study were regarded as a meaningf ul index of the magnitude of attentional interference caused by HiAgg words for each pa rticipant. To investigate the potential usefulness of such an index, a difference sc ore was calculated by simply subtracting the NonAgg mean from the HiAgg mean for both ER and RT data. This method of calculating a difference score was comparable to that used by Townshend and Duka (2001). To investigate Hypotheses 1 through 3, all of the regression analyses were conducted using ER and RT HiAgg means as the criterion variables first and then the analyses were repeated using the ER and RT HiAggÂ—NonAgg mean difference scores as the criterion variables. The results of both approaches are provided. Analyses of Alcohol Cue Moderation For each hypothesis, Setting (and hence al cohol cues) was tested as a moderator of the specified relationship. In the current study, moderation was considered to be the combined influence of two variables after controlling for the eff ects of each variable alone. If the interaction of the two variables successfully predicts the criterion variable, moderation could be said to have occurred. If moderation occurs, it is reasonable to disambiguate the combined effects by repea ting the analyses at each level (e.g., Barroom vs. Cleanroom) of the moderating variable. Th is method follows the approach suggested by Pedhazur (1997). Therefore, in the current study, the moderating effect of Setting was explored by entering a trait anger subscale or EQAAL subscale and Setting (coded as 1 for Cleanroom and 2 for Barroom) into the re gression equation first, and the product of those two variables second. Si gnificant interaction terms were then parsed apart by entering the relevant subscale in one step with either Barroom or Cleanroom cases selected, and vice versa. Since the main eff ects for Word Type were not significant for the Dichotic Listening Task (DLT) using RT or ER data, DLT data were not examined. Hypothesis 1 Participants who self-report hi gher levels of trait anger will demonstrate longer latencies and higher error rates (more attent ional interference) when exposed to selfrelevant cues of aggression than those who repor t lower levels of trait anger. This effect will hold whether participants are tested in the presence or absence of alcohol cues.
75 HiAgg Word Type, Trait Characteri stics, and Alcohol Cue Moderation. To evaluate Hypothesis 1, multiple regres sion analyses using Parafoveal Visual Task (PVT) data were conducte d. Each trait anger subscale (representing the predictor variable) from the STAXI was regressed on HiAgg word RT means and ER means (representing the dependent variable). The subscales included th e Trait-Anger/Angry Temperament (T-Ang/T) subscale and th e Trait-Anger/Angry Reaction (T-Ang/R) subscale. Neither T-Ang/T nor T-Ang/R with Setting as a moderator predicted responses to HiAgg words presented parafoveal ly when using either ER data or RT data. Therefore, difference scores were examined. Difference Scores, Trait Characteri stics, and Alcohol Cue Moderation. Using HiAggÂ—NonAgg ER difference scor es, the regression equation including T-Ang/T, Setting, and the intera ction term was significant [ R2 = .22, adjusted R2 = .16, F (1, 37) = 4.35, p < .05]. Examination of the bivari ate correlations between T-Ang/T and ER difference scores for participants te sted in the Barroom revealed that T-Ang/T was positively associated with ER difference scores [ r (22) = .50, p < .01; t (22) = 2.55, p < .05]. That is, angry temperament tended to in crease as attentional interference (mean ER differences) increased. The bivariate correla tion in the Cleanroom was not significant [ r (19) = -.02, p > .05] indicating that alcohol cues (S etting) moderated the effect of trait angry temperament on the magnitude of the difference for HiAgg vs NonAgg error rates. Using HiAggÂ—NonAgg RT difference scor es, a different pattern emerged. The regression equation including TAng/T, Setting, and the in teraction term was again significant [ R2 = .37, adjusted R2 = .31, F (1, 37) = 16.86, p < .001. However, examination of the bivariate correlations be tween T-Ang/T and RT difference scores for participants tested in the Barr oom revealed that T-Ang/T was negatively associated with RT difference scores [ r (22) = -68, p < .001, t (22) = -4.09, p < .01]. That is, angry temperament tended to increase as the magnit ude of attentional interference (mean RT differences) decreased. This time, the bivari ate correlation in the Cleanroom was also significant [ r (19) = .41, p < .05]. However, the model was not [ R2 = .17, adjusted R2 = .12, F (1, 18) = 3.44, p > .05]. This may be due to the instability of the regression
76 coefficient [Beta = .41, t (19) = 1.86, p > .05]. For the Trait A nger/Angry Reaction (TAng/R) subscale of the STAXI, no main eff ects or interaction e ffects were observed. In order to more fully investigate the relationship between trait angry temperament and reaction time for participants in general (that is, between subjects), PVT participants were blocked into high or lo w trait angry temperament. This yielded 20 participants with a score of 5 or lower (the Low T-Ang/T group) and 17 participants with a score of 7 or higher (the High T-Ang/T group). Therefore, a 2 (Word Type: NonAgg vs. HiAgg) X 2 (Setting: Barroom vs. Cleanr oom) X 2 (Angry Temperament: High vs. Low) repeated measures ANOVA was conducted with RT as the dependent variable. As expected, the main effect for Word Type wa s significant. However, there was also a significant three-way interaction between Word Type, Setting, and Angry Temperament, F (1, 34) = 11.86, p < .01 with power observed at .92. Next, four follow-up paired t-tests were conducted to examine simple effects for Word Type. For participants completing the PVT in the barroom, t hose who self-reported lower trait anger had significantly lo nger RT means for HiAgg words ( M = 704.52, SD = 297.18) than for NonAgg words [ M = 621.35, SD = 248.31; t (8) = -3.37, p = .01]. For those higher on T-Ang/T, no differe nces emerged in the barroom [ t (10) = -1.20, p > .05] for HiAgg words ( M = 616.43, SD = 195.18) versus NonAgg words ( M = 599.93, SD = 227.66). For participants completing the task in the cleanroom, the pattern of results was reversed. That is, for pa rticipants who reported higher levels of angry temperament, RTs to HiAgg words ( M = 739.90, SD = 191.78) were significantly longer than RTs to NonAgg words [ M = 670.54, SD = 186.63; t (5) = -5.11, p = .004]. The RT means for HiAgg words ( M = 732.50, SD = 215.94) were not significan tly different from NonAgg words ( M = 713.20, SD = 216.45) for participants lower on angry temperament when tested in the cleanroom [t (10) = -1.48, p > .05)]. It is apparent that the A NOVA results and t-tests replicated the results produced by examining the data using regression and interaction terms. Overall, support for Hypothesis 1 using either method of analysis was mixed. Using ER difference scores, the positive relationship between angry temperament and the magnitude of attentional
77 interference in the Barroom was predicted. The negative association that resulted in the Barroom when using RT differen ce scores was not predicted. Hypothesis 2 Participants who self-report higher levels of alcoho l-aggression expectancies will demonstrate longer latencies and higher error rates when exposed to aggression cues than those who report low levels of alcohol-aggr ession expectancies. Setting should moderate this effect. That is, the effect should hold only for participants tested in the presence of alcohol cues. HiAgg Word Type, Alcohol-Aggressi on Expectancies, and Alcohol Cue Moderation. To investigate Hypothesis 2, each alc ohol-aggression subscale (representing the predictor variable) from the EQAAL wa s regressed individually on HiAgg word means (representing the dependent variable). The four subscales of the EQAAL include Unprovoked Anger Expectancies (UnpAng) Angry Reaction (AngReac), Hostile Cognitions (HostCog), and Expectancies fo r Maintaining Contro l (ExpCon). It is important to keep in mind that partic ipants reported likely alcohol-aggression expectancies for when they were drinking a low dose of alcohol. Using ER data, none of the regressions fo r the main effect of an EQAAL subscale or Setting or the interaction term approached significance. Using RT data, the regression equation including AngReac, Setting, and the in teraction term approached significance in relation to HiAgg RT means [ R2 = .19, adjusted R2 = .12, F (1, 37) = 4.03, p = .05]. Examination of the bivariate correlations for participants in the Barroom revealed that AngReac was negatively associated with HiAgg RT means [ r (22) = -.51, p < .05, t (22) = -2.62, p < .05]. Overall, when participants repo rted that they were less likely to react with anger after consuming a low dose of alcohol, they were more likely to show attentional interference from aggression cues when in the presence of alcohol cues. This effect did not hold in the absence of alcohol cues [ r (19) = -.03, p > .05]. The regression equation including HostCog, Setting, and the interaction term was significantly related to HiAgg RT means [ R2 = .28, adjusted R2 = .22, F (1, 37) = 7.97, p < .01]. Examination of the bivariate correlations for participants in the Barroom revealed that HostCog was negatively asso ciated with HiAgg RT means [ r (22) = -.64, p < .01,
78 t (22) = -3.75, p < .01]. When participants reported that they were less likely to be suspicious of others after consuming a low dose of alcohol, they were more likely to show attentional interference fr om aggression cues when in th e presence of alcohol cues. Again, the relationship was not significan t in the absence of alcohol cues [ r (19) = .03, p > .05]. Although the regression equation using the interaction term representing Expectancies to Maintain Control (ExpCon) and Setting to predict HiAgg RT means was not significant [ R2 = .16, adjusted R2 = .09, F (1, 37) = 2.33, p > .05], the moderate bivariate correlation between HiAgg RT means and ExpCon was significant [ r (41) = -.31, p < .05]. Thus, the effect of Setting wa s examined. In the Barroom, ExpCon was negatively associated wi th HiAgg RT means [ R2 = .22, adjusted R2 = .18, F (1, 37) = 5.47, p < .05]. In addition, the regression coefficient was significant [ t (22) = -2.34, p < .05]. Since lower scores on this scale reflect d ecreased expectancy to maintain behavioral control while drinking a low dose of alcohol the negative association suggested that participants who reported fewe r control expectancies were also likely to show greater attentional interference to HiAgg stimuli. Additionally, the lack of an association between these two variables for participants tested in the Cleanroom provides additional evidence for the moderating effects of alcohol cues [ r (19) = -.02, p > .05]. Regression equations using Unprovoked Anger Expectancies and Setting were not significant. Difference Scores, Alcohol-Aggression Expectancies, and Alcohol Cue Moderation. As with trait variables, scores re presenting the difference between HiAggÂ— NonAgg RT means and HiAggÂ—Non Agg ER means were used as the criterion variable with EQAAL subscales as the predictor variables in multiple regression analyses. The potential moderation effect of Setting was inve stigated as well. For ER difference scores none of the regression equations were signi ficant. When considering RT difference scores, the regression equation including Ho stCog, Setting, and the interaction term was significantly related to the magnitude of the difference between HiAgg vs. NonAgg RT means [ R2 = .19, adjusted R2 = .12, F (1, 37) = 7.52, p < .01]. Examination of the bivariate correlations for participants tested in the Barroom revealed that HostCog was negatively associated with RT difference scores [ r (22) = -.47, p = .05, t (22) = -2.39, p <
79 .05]. This relationship suggested that when part icipants reported that they were less likely to be suspicious of others after consum ing a low dose of alcohol, the magnitude of attentional interference was greater when in the presence of alcohol cues. An inverse relationship approached but did not r each significance in the Cleanroom [ r (19) = .31, p = .10] probably due to the unsta ble regression coefficient [ t (19) = 1.36, p > .05].There were no significant regression equations for the ot her three EQAAL subscale predictors when using ER or RT difference scores. Although significant results were obtained in relation to the Angry Reaction and Expectancies for Hostile Cognitions subscal es of the EQAAL, the relationships with HiAgg words were negative. Expectancies to Maintain Control were also negatively associated with HiAgg RT means but lowe r scores on this subscale reflect fewer expectancies to maintain cont rol while drinking alcohol. Thus, in the presence of alcohol cues, greater interference to aggression cues was also associated with higher expectancies for losing control while drinking. Converse ly, greater attentio nal interference to aggression cues (using HiAgg means or RT difference scores) was associated with a lower level of expectancies to react with a nger or view othersÂ’ intentions suspiciously when drinking a low dose of alcohol. Hen ce, support for this hypothesis was mixed. Hypothesis 3 Higher alcohol-aggression expectancies wi ll predict longer latencies to respond and higher error rates on the co mputer tasks after the effects of trait anger are partialled out. However, this effect will hold only for pa rticipants tested in the presence of alcohol cues. Alcohol-Aggression Expectancies Af ter Controlling for Trait Anger. In order to investigate whether the E QAAL measure could predict HiAgg means beyond trait anger for participan ts tested in the Barroom, TAng/T was entered on step 1 and then the EQAAL subscale that demonstrat ed the highest correla tion with HiAgg RT means (HostCog) was entered on Step 2. The analysis was then repeated using RT difference scores and then again using ExpC on since this variable had successfully predicted HiAgg RT means in the expected direction. It wa s decided to use predictor variables sparingly because the ratio of the number of predictors to the sample size was
80 too small and inclusion of all four EQAAL s ubscale scores was likely to result in an overestimation of R2 in addition to unstable regressi on coefficients (Pedhazur, 1997). For participants tested in the Barroom the regression equation was significant when regressing HostCog onto Hi Agg RT means after T-Ang/T [ R2 = .41, adjusted R2 = .35, F (1, 19) = 13.00, p < .01]. The regression coeffi cient for HostCog was also significant [Beta = -.65, t (22) = -3.31, p < .01]. When considering HiAgg RT means as the criterion variable it appears that participants showed more attentional interference from high aggression words when they re ported fewer expectancies for thinking suspiciously about othersÂ’ intentions if they had consumed a low dose of alcohol, regardless of their standing on th e trait anger construct. This finding held after trait anger was partialled out but only in the presence of alcohol cues. That is, the bivariate correlation between HostCog and HiAgg mean s was not significant in the Cleanroom [ r (19) = -.03, p > 05]. When predicting RT difference scores fo r participants in the Barroom, HostCog was significantly predictive beyond trait anger [ R2 = .57, adjusted R2 = .53, R2 = .11, F (1, 19) = 5.04, p < .05]. Additionally, regression coeffi cients were significant for both TAng/T [Beta = -.60, t (22) = -9.92, p < .01] and HostCog [Beta = -.35, t (22) = -2.25, p < .05]. In the Cleanroom, the bivariate correla tion between T-Ang/T and the RT difference score was significant and in a positive direction [ r (19) = .41, p < .05]. However, the bivariate correlation using Ho stCog was not significant [ r (19) = .31, p = .10]. Additionally, neither regression equation was significant in the Cleanroom, probably due to the instability of the regression coefficients [Beta of T-Ang/T = .36, t (19) = 1.21, p > .05; Beta of HostCog = .09, t (19) = .29, p > .05]. When using ER difference scores instead of RT difference scores, regression equations were not significant. Also, when considering ExpCon and HiAgg RT means, ExpCon was not predictive after partia lling out the effects of T-Ang/T [ R2 = .03, F (1, 19) = 1.25, p > .05]. For participants in a barroom enviro nment, a consistent and significant relationship was uncovered between a meas ure of alcohol-aggression expectancies (HostCog) and attentional interference from aggression stimuli even after controlling for
81 trait anger (see Table 3). Unfort unately, the direction of the associations between alcoholaggression expectancies and attentional in terference were contrary to expectation. Although the finding is intriguing, this hypothesis was not supported. Table 3 Regression Model Predicting Attentional Interference (HiAggÂ—NonAgg Difference Scores) from Aggression Stimuli Presented Pa rafoveally in the Presence of Alcohol Cues ________________________________________________________________________ Predictor Variable B t p R2 Adj R2 R2 F ________________________________________________________________________ Step 1 STAXI T-Ang/T -25.60 -.68 -4.09 .001 .46 .43 .46 16.75* Step 2 STAXI T-Ang/T -22.85 -.60 -3.92 .001 EQAAL HostCog -9.83 -.35 -2.25 .037 .57 .53 .11 5.04* ________________________________________________________________________ Note. RT = HiAgg = High Aggression Word Type; NonAgg = Non Aggression Word Type; STAXI T-Ang/T = State-Trait Anger Expression Inventory Trait Anger/Angry Temperament subscale; EQAAL = Expect ancy Questionnaire for Alcohol and Aggression Â– Low Dose; HostCog = E xpectancies for Hostile Cognitions p < .05. Supplemental Results PANAS scores at Time 1 and Time 2. For the Dichotic Listening Task (DLT), the overall mean for the Positive Affect scale at Time 1 ( M = 2.90, SD = .77) was significantly higher than the mean at Time 2 [ M = 2.69, SD = .84; t (36) = 2.76, p < .01]. For the Parafoveal Visual Task (PVT), Positi ve Affect means were equivalent from Time 1 ( M = 2.85, SD = .62) to Time 2 [ M = 2.75, SD = .66; t (38) = 1.71, p > .05. For the Negative Affect scale, overall means were not significantly different between Time 1 and
82 Time 2 for either computer task. DLT means for Negative Affect were 1.28 ( SD = .30) at Time 1 and 1.21 ( SD = .32) at Time 2. PVT means for Negative Affect were 1.27 ( SD = .24) at Time 1 and 1.26 ( SD = .27) at Time 2. Overall, par ticipants experienced a slight decrease in Positive Affect when completing only the DLT and did not experience a change in Negative Affect when completing either task. It is possible that feelings related to the construct of anger could change when a person is exposed to words of an aggressive na ture even if overall ne gative affect did not change. Two negative affect adjectives were of special interest in the current studyÂ— hostile and irritable Â—since they are considered to represent the construct of angry When these two adjectives were examined acro ss computer task and time, no differences emerged. Given the overall pattern of results for the PANAS, it was unlikely that, at least subjectively, participants were experiencing a meaningful cha nge in negative affect as a result of their exposure to aggression related words. Recognition Task. Participants were expected to check off a higher percentage of Control Words (words that they were instructed to attend to) than Experimental words (unattended and parafoveal) or New words (wor ds they had never been exposed to). On average, participants checked off 41% of th e 14 Control words. On average, they also checked off 20% of the eight Experimental wo rds, and 19% of the New words. Thus, the rate of endorsement for words participants ha d never seen or heard was equivalent to the rate of endorsement for unattended words. Regarding the words on the Recognition Task that had been chosen to represent HiAgg (e.g., shout and gun ), comparable distractor words we re selected to represent this category (e.g., hate and maim ). Comparison of the averag e number of HiAgg vs. New (HiAgg equivalent) words checked off revealed no difference ( t < 1). In fact, a majority of participants checked off none of these wo rds. Additionally, there appeared to be no discernible difference in the overall patte rns of checked words for participants completing either task. Although it cannot be unequivocally concl uded that conscious awareness of the experimental aggressi on-related words did not occur for a few participants, overall the pattern suggests that participants were una ble to meaningfully
83 discriminate between unattended stimuli and ne w stimuli. So, results were generally not a product of conscious awareness of unattended stimuli. Participants were also as ked to write down any othe r words they thought they might have seen or heard that were no t listed on the recognition task. It was not surprising that most of the wo rds participants recalled started with the letter Â“L.Â” Other experimental words listed that participan ts had been exposed to (in the unattended channel or parafoveally) included rape kill and shoot All of these words were reported by participants in the DLT condition. Other words that participants in the PVT condition reported included wow quick yellow and window It is interesting to note that the masking stimulus for the PVT was represented by the string Â“WQXQW.Â” It is reasonable to consider that wow and quick are examples of words that someone might report when a stimulus is only presented for 100 milliseconds outside of his or her foveal region.
84 Discussion Comparison of the Dichotic Listening Ta sk vs. the Parafoveal Visual Task A specific hypothesis was not formulated regarding whether the Parafoveal Visual Task (PVT) methodology or Dichotic Listen ing Task (DLT) methodology would be most helpful for investigating the confluence of trait characteristics, alcohol-aggression expectancies, and aggression stimuli on atte ntion. However, a comparison of the two methodologies was the overarching purpose of the current investigation. Power estimates, as discussed earlier, provided strong evidence that with an equal num ber of participants, the PVT (observed power = 1) was a more se nsitive and useful task for investigating qualitative differences among Word Type React ion Time and Error Rate means than the DLT (observed power ranged from .10 to .56). It is possible that, with more particip ants, differences would have emerged across Word Type for the DLT. However, this stil l suggests that the PVT is a ore sensitive procedure. Further, it would be difficult to explain a trend for which the data did not conform to a relatively linear relationship (see Figures 1 and 2). The quantitative difference between NonAgg and HiAgg words on the PVT made it possible to calculate difference scores that may more meaningfully reflect within subject differences to such an elusive Â“black-boxÂ” phenomenon as attentional interference. Error Rate vs. Reaction Time Error Rate (ER) means showed an equiva lent trend to Reaction Time (RT) means for PVT data. Overall, ERs were low indica ting that most participants had no trouble completing the primary Â“L-wordÂ” counting task as well as the secondary odd vs. even decision-making task. ER data proved extrem ely useful for detecting a small group of participants who were eith er unable to, or, more likely, did not choose to follow directions to Â“respond as quick ly and accurately as possible. Â” Poor compliance with the directions was indicated by ERs no better than chance.
85 ER means when used as th e criterion variable in regres sion analyses were not as useful as RT means. Given the frequency with which participants made no or relatively few errors, it is reasonable to conclude that ER was not sensitive enough to detect minute changes in attention. HiAgg Means Vs. HiAggÂ—NonAgg Difference Scores Although ER means differed significantly across Word Type in the expected direction, individual ER means were not predicted by any of the trait anger or alcoholexpectancy measures. HiAgg RT means were predicted by three of the four alcoholexpectancy measures. Using a difference score that should reflect an individualÂ’s level of interference from HiAgg words as opposed to NonAgg words also uncovered significant relationships among predictor a nd criterion variables. The na ture of the relationships were contrary to predictions with the ex ception of the positive association between T/Ang-T and ER difference scores in the presence of alcohol cues. Overall, it appears that ER means, RT means, and difference scores all had something important to add in the investigation of differen tial responding to Word Type for the Visual Task. That is, error rate mean s pointed out participants who may not have attended to task directio ns, RT means provided information about how attentional interference varied across participants, and difference sc ores provided information about how attentional interference varied within participants. Power for the Regression Analyses In this study, ER and RT means reflec ted meaningful differences across Word Type. However, for ER, only the regression equa tion used to predict ER difference scores (HiAgg mean Â– NonAgg ER m ean) in relation to Trait A nger/Angry Temperament was significant. For HiAgg RT means and RT diffe rence scores, several regression equations were significant. A larger sample may clar ify the relative useful ness of using an RT difference score instead of an RT mean as the criterion variable in regression analyses. Overall, effect sizes for the significant regression equations were medium (~ .30) to large ~.50). Estimates of adjusted R2 ranged from .12 to .38. Although effect sizes were encouraging and moderate to strong bi variate correlations were observed, some regression coefficients were unstable. This is likely due to the small ratio between
86 predictors and sample size. For example, if all of the EQAAL subscales had been used to predict attentional interference above and beyond trait angry temperament, the ratio would have been approximately 1: 4. Th is is far below the most liberal of recommendations (i.e., 1:15; Pedhazur, 1997). Overall, the data did not support the prediction hypotheses or the support was mixed. Given the concerns about adequate sa mple size for the regr essions, the following conclusions are speculative. Hypothesis 1 Support for the hypothesis that higher self -reported levels of trait anger would predict higher levels of attentional in terference from aggression stimuli was only provided when using ER mean difference scores (mean HiAgg ERÂ—mean NonAgg ER) for participants tested in the presence of alcohol cues. The association between these two variables was positive, as specified in the hypothesis, and the effect did not hold in the absence of alcohol cues. Since this is the only hypothesis for which ER mean difference scores were successfully predicted, and the resu lt does not fit with the patter n of the rest of the results, it may be that this finding is spurious, especi ally given the sample size for the regressions and the overall inability of the current me thodology to predict error rate differences. Reaction time difference scores (HiAgg RT mean Â– NonAgg RT mean), on the other hand, appeared to be more meaningf ul and useful for uncovering relationships among trait anger, alcohol-aggression exp ectancies, and atten tional interference. However, for Hypothesis 1, the association between trait angry temperament and the magnitude of attentional inte rference was predicted to be positive and was, in fact, negative. One possible explanation for this finding is that students who generally experience fewer angry feelings may be more attentive to stimuli in their environment that represents a potential threat. Since th e negative association between trait angry temperament was found only in the Barroom, it is likely that alcohol cues and aggression cues combined to put lower angry temperam ent participants on th e alert. Conversely, those with relatively higher levels of tr ait angry temperament were able to ignore aggression stimuli while in the presence of alcohol cues.
87 It may be, for the current study, that lower trait anger participan ts in the Barroom (which might be considered low stress envi ronment when alcohol cues interact with aggression cues), experienced a similar attenti onal bias while participants higher in trait anger were able to ignore aggression stimuli in the Barroom. Literature on the interference of threat words wa s reviewed in order to shed some light on this unexpected finding. One study (MacLeod & Rutherford, 1992) found an interaction between high vs. low anxious participants under high vs. low stress and naming latencies to threatening Stroop words. The authors used a difference sc ore (color naming latencies to threat words minus color naming latencies to nonthreatening words) to index individual susceptibility to attentional interference (p. 486). Low trait anxious individuals under high stress displayed a lower magnitude of interferen ce from threatening Stroop words than high trait anxious individuals. Th is effect was reversed when participants were under low stress. That is, low trait anxi ous individuals under low stress showed greater interference to threat words than high trait anxious indi viduals under low stress. However, it appears that the findings of MacLeod and Rutherford do not support this al ternative explanation. Another explanation for the lack of findings in the predicted direction involves the conditional compensatory responses (CCR) theo ry (Siegel, Baptista, Kim, McDonald, & Weise-Kelly, 2000). CCRs are initiated by the ce ntral nervous system to counteract the effects of a drug such as alcohol. Further, CCRs have been found to occur in the presence of cues that had earlier been paired with the consumption of a particular drug. Examples of these cues are drug-related paraphernali a and the context (environment) under which the drug is consumed. It is possible that a cl assically conditioned re sponse to alcohol cues in the barroom prompted a comp ensatory emotional reaction. Although the interaction between Word Type X Setting was found to be nonsignificant, and follow-up analyses for Sett ing at each level of Word Type were not significant, Barroom clearly showed a moderati ng effect in the regression analyses. Also, the pattern of Reaction Time (RT) means, w ithout exception, indicate d that RT means in the barroom for a given word type were alwa ys higher than RT means in the cleanroom for the same word type. This does not sugge st an overall compensatory effect. If a compensatory effect were occurring, it woul d most likely occur for participants with
88 heavier drinking experience since CCRs are pr esumed to represent the effects of drug tolerance. Although drinker status could be calculated from the Comprehensive Drinker Profile data collected during th is study, the sample size was t oo small to investigate such a hypothesis. The tenability of a conditional compensatory response must await further investigation. Hypothesis 2 It was predicted that alcohol-aggression expectancies would predict attentional interference from aggression stimuli but that this effect would be moderated by the presence of alcohol cues. When AngR eac, HostCog, and ExpCon, were used as predictors of HiAgg RT means, each regres sion equation was significant for participants tested in the Barroom. Howeve r, the direction of the rela tionships for the AngReac and HostCog variables were nega tive, and, thus, did not conf orm to our predictions. The relationship between ExpCon a nd HiAgg RT means was also negative, but the scale is scored such that lower scores reflect higher expectancies for losing control while drinking a low dose of alcohol. This finding supported our predictions. Higher scores on the AngReac scale would suggest higher expectancies to react with anger while drinking a low dose of alcohol in response to a particular event. It is possible that participants w ith higher AngReac expectancies more easily dismissed the aggression cues since there was nothing obviously threatening about the Barroom environment (i.e., nothing to which to reac t with anger). The same may be said for HostCog scores. Higher scores on this scal e would indicate a higher tendency for an individual to have suspicious thoughts about the intentions of othe rs while drinking a low dose of alcohol. Again, those with higher sc ores may have more readily dismissed aggression stimuli in a clearly nonhostile situation. That is, the activation of aggression scripts would undoubtedly include information about the types of circumstances under which internal tendencies to feel angry w ould manifest. Those with lower scores, on the other hand, may have fewer or less strongly interconnected scripts for what would happen in the presence of alcohol cues, and, consequen tly, their attentional sy stem was more alert to the relatively Â“novelÂ” aggr ession stimuli. As with hypothesis 1, it could be illuminating
89 to measure drinking experience with a larger sample to explore the relationship between higher vs. lower drinking experience and atten tional interference from aggression stimuli. Regarding the capacity of Expectancies to Maintain Control (ExpCon) to predict automatic attentional interference, the signi ficant association was in the expected direction. That is, participants who reported that they would be more likely to lose control while drinking a low dose of alcohol were also more likely to show greater interference to aggression stimuli. This result is encouraging in that expectancies to behave aggressively should more clearly relate to act ually behaving aggressively than to merely having angry feelings that are not ne cessarily expressed behaviorally. In fact, numerous studies are emerging in the literature in which distinctions are being made about the qualitative differences among behavioral, affective, and cognitive elements of aggression. For example, Gianco la, Saucier, and Gussler-Burkhardt (2003) found that self-reported affec tive aggression was not related to behavioral aggression whether or not a low dose of alcohol was cons umed. They suggested that the experience of anger is causally unrelated to behavioral manifestations of anger (p. 1951). Although the present study did not provide an opportunity for individuals to exhibit aggression and no alcohol was consumed, it is important to understand the extent to which individuals are biased toward aggression stimuli in th e presence of alcohol cues. The tendency of participants with a lower rate of expectan cies to maintain control to demonstrate attentional interference suggested a behavioral component that may not be tapped by trait angry temperament alone. Unfortunately, the re sults in respect to Hypothesis 3 indicated that expectancies to maintain control did not predict attentional interference beyond trait angry temperament. Future research may ev entually disambiguate the components of aggression and determine which ones are causa lly related to attentional interference or behavioral aggression. Hypothesis 3 Overall, the use of T-Ang/T and Ho stCog provided the greatest explanatory evidence for the relationship between trait anger, alcohol-expectancies and attentional interference. Not surprisingly, given the prior results, the di rection of the relationships
90 did not conform to our predictions and ExpC on did not predict atte ntional interference beyond trait angry temperament. The finding, that in the Barroom, T-Ang/ T and HostCog contributed uniquely to HiAgg RT means and HiAggÂ—NonAgg differen ce scores, and that HostCog contributed beyond T-Ang/T, is indeed intriguing. The resu lts suggested that afte r holding trait anger constant, lower scores on negative hostile c ognitions explained 53% of the variance in the magnitude of attentional in terference from aggression st imuli. This supports the continued need to evaluate alcohol-aggression expectancy networks in addition to trait characteristics. Alcohol Cues Repeated Measures ANOVA failed to reve al a main effect for Setting or an interaction effect for Word T ype X Setting. However, it was apparent from the regression analyses that alcohol cues moderated the re lationships among the variables. This was as predicted. It is likely that al cohol cues served to activate aggression networks in some fashion. Whether that knowledge structur e activation will ultimately be found to represent a suppression of attent ion toward aggression cues or not remains to be explored. Depending upon the outcome, a next step could be to explore whether actual behavioral aggression increases or decreases. Predic ting the salience of internal cues (e.g., knowledge structures that incl ude alcohol-aggression-related information) and external cues (alcohol stimuli and aggression stimuli) is a goal implied by alcohol myopia theory. Although the current methodology did not directly investigate assumptions of the alcohol myopia model, the mixed results of this st udy support the contention that the salience of cues is a highly complex matter. Studies such as this may eventually help identify and predict relative cue salience. Limitations of the Current St udy and Future Directions The most problematic concern for interpre ting the results of this study is that the sample size was not optimal for the regression analyses. It might be possible to increase sample size for both the questionnaire and comput er task data by collecting all of the data over one session. Such a procedure would need to avoid a priming effect if nonconscious processes are being investigated.
91 The possibility was explored that the differences in reaction time between NonAgg and HiAgg words could be due to the increase in reaction times caused by responding to Â“LÂ” words. Seven Â“LÂ” words we re presented during th e DLT and eight Â“LÂ” words were presented during the Parafoveal Visual Task (PVT). For the DLT, none of the Â“LÂ” words were presented within the two word types of interest. For the PVT, two of the Â“LÂ” words were presented within word types of interest. Within the NonAgg category, note-lamp was presented resulting in a mean reaction time (RT) of 698.18 ( SD = 321.92). The mean RT for the other 19 NonAgg words was 631.71 ( SD = 200.50). The NonAgg Â“LÂ” word RT mean was not significantly different from the mean RT of the other nineteen NonAgg words, t (38 ) = 1.70, p = .10. Leaving the RTs for this word pair in the anal yses accounted for an increase in total mean RT by approximately 4 ms. Within the HiAgg category, scream-lake was presented yielding a mean reaction time of 853.49 ( SD = 411.82). The mean RT for the other 19 HiAgg words was 689.88 ( SD = 216.27). These two RT means were significantly different, t (40) = 3.32, p < .01. Leaving the RTs for the word pair in the analyses resulted in an increase in total mean RT by about 9 ms. When the data were reanalyzed, the same pattern of results emerged. Therefore, it is highly unlikely that the re sults of the current study were due to the inclusion of RTs for two Â“LÂ” words. Although it is unclear why the mean for the HiAgg Â“LÂ” word was significantly higher than the m ean for the other 19 HiAgg words, in similar future studies it would be prudent to dele te words such as the Â“LÂ” word that may contribute to measurement error. Another concern for the current study is that the sample was over represented by females. It may be that females expect aggressiveness to increase for males who are drinking, but not for themselves. The range restriction caused by endorsing items in a manner consistent with this assumption (as noted in the s ection regarding the psychometric properties of the STAXI) may have dampened the predictive ability of the measures of interest. Since participants were aware that the questi onnaires would include questions about alcohol and aggr ession, perhaps this sample of college students wanted to present themselves in a more positive light th an samples that provide responses outside of
92 an alcohol-aggression context. In the future, adding a measure of social desirability (e.g., the Marlowe-Crowne Desirability Scale; Reyno lds, 1982) may shed lig ht on this potential threat to valid responses. Using the current methodology, an indicati on of how often a pa rticipant may have actually aggressed (with or wit hout the consumption of alcohol) was not available. Future research will need to index the occurrence of behavioral aggression in order to more fully understand obtained differences in attentional interf erence for aggression stimuli. That is, perhaps an individual that had self-reported higher levels of trait anger and had reported prior instances of aggression w ould be biased toward aggressi on stimuli. Or perhaps he or she would be more likely to ignore aggre ssion stimuli. Further it may be possible to predict behavioral aggression from attentiona l interference when the participant is then given the opportunity to aggress. Vali dating a methodology to investigate these hypotheses was the ultimate goal of this projec t. In the final estim ation, the Parafoveal Visual Task provide s such a methodology. The results of the current study suggested that the relationship between attentional interference and trait anger or alcohol-aggression expectanci es is not straightforward, especially when research into these relati onships relies upon self-report. One of the hopes of this study was to develop a methodology th at may someday augment or even replace self-report. That is, if attentional interfer ence (and therefore automatic attention to external cues) can be indexed and then shown to relate to aggressive behavior, the need for self-report would be obviate d. This is a worthy goal of research since people are apt to give reports about feelings, thoughts, and behaviors that include biases in how that information was stored in memory, biases re garding how they want to be viewed, and other biases that may serve any number of unknown or indeterminate goals. It is concluded that the current methodology holds promise for further exploring attentional interference to aggression stimuli be yond what an individual may report. Another limitation of the current study is that the obtained results may not generalize to other aggression stimuli or beyond this analog drinking environment. Attentional interference is ar guably a complex phenomenon th at is challenging to study. This study represents a small step toward a methodology that may help us understand the
93 link between attentional interference from (or bias toward) aggression stimuli in the presence of alcohol cues and how that inte rference relates to tra it characteristics and alcohol-aggression expectancies. Understanding these relationships may eventually guide us toward new strategies and interventions for those who drink and become aggressive.
94 References Altarriba, J., Kambe, G., Pollatsek, A., Rayner, K. (2001). Semantic codes are not used in integrating information across eye fixations in reading: Evidence from fluent Spanish-English bilinguals. Perception & Psychophysics, 63 (5), 875-890. Anderson, C. A. (1989). Temperature and aggr ession: Ubiquitous effects of heat on occurrence of human violence. Psychological Bulletin, 106 74-96. Anderson, C. A., & Bushman, B. J. (1997). Exte rnal validity of "trivial" experiments: The case of laboratory aggression. Review of General Psychology, 1 (1), 19-41. Bailey, D. S., & Taylor, S. P. (1991). Effect s of alcohol and aggressive disposition on human physical aggression. Journal of Research in Personality, 25 (3), 334-342. Bandura, A. (1973). Aggression: A social learning analysis Englewood Cliffs, NJ: Prentice Hall. Bargh, J. A. (1992). The ecology of automa ticity: Toward establishing the conditions needed to produce automatic processing effects. American Journal of Psychology, 105 (2), 181-199. Bargh, J. A. (1996). Automatic ity in social psychology. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 169183). New York: Guilford Press. Bargh, J. A., Bond, R. N., Lombardi, W. J., & Tota, M. E. (1986). The additive nature of chronic and temporary sources of construct accessibility. Journal of Personality & Social Psychology, 50 (5), 869-878. Bargh, J. A., & Chartrand, T. L. (2000). The mind in the middle: A practical guide to priming and automaticity research. In H. T. Reis & C. M. Judd (Eds.), Handbook of research methods in so cial and personality psychology. (pp. 253-285). New York City: Cambridge University Press. Bargh, J. A., & Pietromonaco, P. (1982). Au tomatic information processing and social perception: The influence of trait info rmation presented outside of conscious
95 awareness on impression formation. Journal of Personality & Social Psychology, 43 (3), 437-449. Bargh, J. A., Raymond, P., Pryor, J. B., and Strack, F. (1995). Attractiveness of the underling: An automatic power --> sex association and its consequences for sexual harassment and aggression. Journal of Personality & Social Psychology, 68 (5), 768-781. Bargh, J. A. (1996). Automaticit y in social psychology. In E. T. Higgins, & Kruglanski, A. W. (Eds.). Social psychology: Handbook of basic principles (pp. 169-183). New York: Guilford Press. Baron, R. A. (1977). Human aggression. New York: Plenum. Bentin, S., Kutas, M., & Hillyard, S. A. (1995). Semantic processing and memory for attended and unattended words in di chotic listening: Behavioral and electrophysiological evidence. Journal of Experiment al Psychology: Human Perception & Performance, 21 (1), 54-67. Berkowitz, L. (1983). Aversively stimulated ag gression: Some parallels and differences in research with animals and humans. American Psychologist, 38 1135-1144. Berkowitz, L. (1989). Frustrat ion-aggression hypothesis: Ex amination and reformulation. Psychological Bulletin, 106 59-73. Berkowitz, L. (1990). On the formation a nd regulation of anger and aggression: A cognitive-neoassociationistic analysis. American Psychologist, 45 (4), 494-503. Berkowitz, L. (1998). Affective aggression: The role of stre ss, pain, and negative affect. In R. G. Geen & E. Donnerstein (Eds.), Human aggression: Theories, research, and implications for social policy (pp. 49-72). San Diego: Academic Press. Berkowitz, L., & Troccoli, B. T. (1990). Feelin gs, direction of attention, and expressed evaluations of others. Cognition and Emotion, 4 (4), 305-325. Bond, A., & Lader, M. (1986). The relationshi p between induced behavioural aggression and mood after the consumption of two doses of alcohol. British Journal of Addiction, 81 (1), 65-75.
96 Brown, S. A., Goldman, M. S., Inn, A., & Anderson, L. R. (1980). Expectations of reinforcement from alcohol: Their doma in and relation to drinking patters. Journal of Consulting and Clinical Psychology, 43 419-426. Bryden, M. P. (1988). An overview of the dichot ic listening procedur e and its relation to cerebral organization. In K. Hugdahl (Ed.), Handbook of dichotic listening: Theory, methods and research. Chichester, NY: Wiley. Bushman, B. J. (1995). Moderating role of tra it aggressiveness in the effects of violent media on aggression. Journal of Personality & Social Psychology, 69 (5), 950960. Bushman, B. J., & Anderson, C. A. (2001). Is it time to pull the plug on the hostile versus instrumental aggression dichotomy? Psychological Review, 108 (1), 273-279. Bushman, B. J., & Cooper, H. M. (1990). E ffects of alcohol on hu man aggression: An integrative research review. Psychological Bulletin, 107 (3), 341-354. Buss, A. H. (1961). The psychology of aggression. New York: Wiley. Buss, A. H., & Durkee, A. (1957). An inventor y for assessing different kinds of hostility. Journal of Consulting Psychology, 21 (4), 343-349. Buss, A. H., & Perry, M. (1992). The Aggression Questionnaire. Journal of Personality & Social Psychology, 63 (3), 452-459. Chartrand, T. L.,& Bargh, J. A. (1996). Auto matic activation of impression formation and memorization goals: Nonconscious goal pr iming reproduces e ffects of explicit task instructions. Journal of Personality & Social Psychology, 71 (3), 464-478. Chartrand, T. L., Bargh, J. A., & van Baaren, R. B. (2003). Linking automatic evaluation to mood and information processing style: Consequences for experienced affect, impression formation, and stereotyping. Manuscript submitted for publication, Ohio State University. Chermack, S. T. & Giancola., P. R. (1997). The relation between al cohol and aggression: An integrated biopsychoso cial conceptualization. Clinical Psychology Review, 17 (6), 621-649.
97 Chermack, S. T. &Taylor, S. P. (1995) Alcohol and human physical aggression: Pharmacological versus expectancy effects. Journal of Studies on Alcohol, 56 (4), 449-456. Cherry, E. C. (1953). Some experiments on the recognition of speech, with one and with two ears. Journal of the Acoustical Society of America, 25 975-979. Cohen, A. R. (1955). Social norm s, arbitrariness of frustratio n, and status of the agent of frustration in the frustration-aggression hypothesis. Journal of Abnormal and Social Psychology, 511 222-226. Cornell, D. G., Peterson, C. S., & Richards, H. (1999). Anger as a pr edictor of aggression among incarcerated adolescents. Journal of Consulting & Clinical Psychology, 67 (1), 108-115. Corteen, R. S., & Wood, B. ( 1972). Autonomic responses to shock-associated words in an unattended channel. Journal of Experime ntal Psychology, 94 (3), 308-313. Cox, W. M., Blount, J. P. and Rozak, A. M. (2000). Alcohol abusers' and nonabusers' distraction by alcohol and concern-related stimuli. American Journal of Drug & Alcohol Abuse, 26 (3), 489-495. Crick, N. R., & Dodge, K. A. (1994). A review and reformulation of social informationprocessing mechanisms in children's social adjustment. Psychological Bulletin, 115 (1), 74-101. Darkes, J., and Goldman, Mark S. (1993). E xpectancy challenge and drinking reduction: Experimental evidence for a mediational process. Journal of Consulting & Clinical Psychology, 61 (2), 344-353. Dermen, K. H., & George, W. H. (1989). Alcohol expectancy and the relationship between drinking and physical aggression. Journal of Psychology, 123 (2), 153161. di Pace, E., Longoni, A. M., & Zoccolotti, P. (1991). Semantic processing of unattended parafoveal words. Acta Psychologica, 77 (1), 21-34. Dodge, K. A., & Crick, N. ( 1990). Social information-proc essing bases of aggressive behavior in children. Personality & Social Psychology Bulletin, 16 (1), 8-22.
98 Dodge, K. A., & Tomlin, A. M. (1987). Utiliz ation of self-schemas as a mechanism of interpretational bias in aggressive children. Social Cognition, 5 (3), 280-300. Dodge, K. A., Pettit, G. S., McClaskey, C. L., & Brown, M. M. (1986). Social competence in children. Monographs of the Society for Research in Child Development, 51 (2), 1-85. Dollard, J., Miller, N. E., Doob, L. W., Mowrer, O. H., & Sears, R. R. (1939). Frustration and aggression New Haven: Yale University Press. Dougherty, D. M., Bjork, J. M., Bennett, R. H., Moeller, & F. G. (1999). The effects of a cumulative alcohol dosing procedure on laboratory aggression in women and men. Journal of Studies on Alcohol, 60 (3), 322-329. Edington, M. A. (1996). Alcohol myopia, selective atten tion and expectancy interference. Unpublished Honors Thesis, University of South Florida, Tampa. Epps, J., & Kendall, P. C. (1995). Hos tile attributional bias in adults. Cognitive Therapy & Research, 19 (2), 159-178. Epps, J., Hunter, B., LeVasseur, M. E ., Steinberg, H., & Hancock, D. (2002). Alcohol and aggression expectancies: Development and partial validation of the Expectancy Questionnaire for Alcohol and Aggression--Low Dose Version (EQAAL). Unpublished manuscript, University of South Florida, Tampa. Eron, L. D. (1994). Theories of aggression: From drives to cognitions. In L. R. Huesmann (Ed.), Aggressive behavior: Current perspectives. (pp. 3-11). New York, NY: Plenum Press. Favata, J., LeVasseur, M. E., Koenig, D., Ci arcia, K., Epps, J. & Roberts, A.M. (2003, March). Multimodal behavioral aggression in men and women: The role of alcohol expectancy network activation. Poster session presented at the Southeastern Psychological A ssociation, New Orleans, LA. Fillmore, M. T. & Vogel-Sprott, M. (1995). Expectancies about alcohol-induced motor impairment predict individual differences in responses to alcohol and placebo. Journal of Studies on Alcohol, 56 (1), 90-98.
99 Fillmore, M. T., Mulvihill, L. E., & VogelSprott, M. (1994). The expected drug and its expected effect interact to determine placebo responses to alcohol and caffeine. Psychopharmacology, 115 (3), 383-388. Foley, P. F., Hartman, B. W., Dunn, A. B., Sm ith, J. E., & Goldberg, D. M. (2002). The utility of the State-Trait Anger E xpression Inventory with offenders. International Journal of Offender Therapy & Comparative Criminology, 46 (3), 364-378. Forgays, D. G., Forgays, D. K., & Spielberger, C. D. (1997). Factor structure of the StateTrait Anger Expression Inventory. Journal of Personality Assessment, 69 (3), 497507. Frost, W. D. & Averill, T. R. (1982). Difference between men and women in the everyday experience of anger. In J. R. Averill (Ed.), Anger and aggression: An essay on emotion. (pp. 281-316). New York: Springer-Verlag. Fuqua, D. R., Leonard, E., Masters, M. A., Smit h, R. J., Campbell, J. L., & Fischer, P. C. (1991). A structural analysis of the State-Trait Anger Expression Inventory. Educational & Psychological Measurement, 51 (2), 439-446. Geen, R. G., & Donnerstein, E. (Eds.). (1998). Human aggression: Theories, research, and implications for social policy. San Diego: Academic Press. Geen, R. G., Stonner, D., & Shope, G. L. (1975). The facilitation of aggression by aggression: Evidence agains t the catharsis hypothesis. Journal of Personality & Social Psychology, 31 (4), 721-726. George, W. H., Frone, M. R., Cooper, M. L., Russell, M., Skinner, J. B., & Windle, M. (1995). A revised Alcohol Expectancy Questionnaire: Factor structure confirmation and invariance in a general population sample. Journal of Studies on Alcohol, 56 (2), 177-185. Giancola (2003Â… need to reference a bunch of these from about page54).** Giancola, P. R., & Chermack, S. T. (1998). C onstruct validity of laboratory aggression paradigms: A response to Tedeschi and Quigley (1996). Aggression & Violent Behavior, 3 (3), 237-253. Giancola, P. R., and Zeichner, A. (1995). An investigation of gender differences in alcohol-related aggression. Journal of Studies on Alcohol, 56 (5), 573-579.
100 Giancola, P. R., Helton, E. L., Osborne, A. B ., Terry, M. K., Fuss, A. M., & Westerfield, J. A. (2002). The effects of alcohol a nd provocation on aggressive behavior in men and women. Journal of Studies on Alcohol, 63 (1), 64-73. Giancola, P. R., Saucier, D. A., & Gussi er-Burkhardt, N L. (2003). The effects of affective, behavioral, an d cognitive components of trait anger on the alcoholaggression relation. Alcoholism: Clinical and Ex perimental Research 27 (12), 1944-1954. Goldman, M. S., Brown, S. A., & Christia nsen, B. A. (1987). Expectancy theory: Thinking about drinking. In H. T. Blane & K. E. Leonard (Eds.), Psychological theories of drinking and alcoholism (pp. 181-226). New York: Guilford Press. Goldman, M. S., Del Boca, F. K., & Darkes, J. (1999). Alcohol expectancy theory: The application of cognitive neuroscience. In H. T. Blane & K. E. Leonard (Eds.), Psychological theories of drinking and alcoholism (2nd ed., pp. 203-246). New York: Guilford Press. Gouze, K. R. (1987). Attention and social pr oblem solving as correlates of aggression in preschool males. Journal of Abnormal Child Psychology, 15 (2), 181-197. Guerra, N. G., Nucci, L., & Huesmann, L. R. (1994). Moral cognition and childhood aggression. In L. R. Huesmann (Ed.), Aggressive behavior: Current perspectives (pp. 13-33). New York: Plenum Press. Gustafson, R. (1991). Aggressive and nonaggres sive behavior as a function of alcohol intoxication and frustration in women. Alcoholism: Clinical & Experimental Research, 15 (5), 886-892. Gustafson, R. (1993). What do experimental paradigms tell us about alcohol-related aggressive responding? Journal of Studies on Alcohol, Suppl 11 20-29. Hare, R. D. M., Leslie M. (1984). Psychopat hy and perceptual asymmetry during verbal dichotic listening. Journal of Abnormal Psychology, 93 (2), 141-149. Hearold, S. (1986). A synthesis of 1043 effects of television on social behavior. In G. Comstock (Ed.), Public communications and behavior (Vol. 1, pp. 65-133). New York: Academic Press.
101 Herzog, T. (1999). Effects of alcohol intoxication on social inferences. Experimental & Clinical Psychopharmacology, 7 (4), 448-453. Higgins, E. T. (1996). Knowledge activation: Ac cessibility, applicabil ity, and salience. In E. T. Higgins, & Kruglanski, A. W. (Eds.), Social psychology: Handbook of basic principles. (pp. 133-168). New York: Guilford Press. Holcomb, P. J., & Neville, H. J. (1990). Audito ry and visual semantic priming in lexical decision: A comparison using ev ent-related brain potentials. Language & Cognitive Processes, 5 (4), 281-312. Huesmann, L. R. (1988). An information processing model for the development of aggression. Aggressive Behavior, 14 (1), 13-24. Huesmann, L. R. R., Meredith A. (2001). C ognitive processes and the development of aggression. In A. C. Bohart & D. J. Stipek (Eds.), Constructive & destructive behavior: Implications fo r family, school, & society (pp. 249-269). Washington, DC: American Psychological Association. Hugdahl, K., Rund, B. R., Lund, A., Asbjornse n, A., Egeland, J., Landro, N. I., Roness, A., Stordal, K. I., & Sundet, K. (2003). Attentional and execu tive dysfunctions in schizophrenia and depression: Evidence fr om dichotic listening performance. Biological Psychiatry, 53 (7), 609-616. Ingram, R. E., Bernet, C. Z., & McLaugh lin, S. C. (1994). Attentional allocation processes in individuals at risk for depression. Cognitive Therapy & Research, 18 (4), 317-332. Ito, T. A., Miller, N., & Pollock, V. E. ( 1996). Alcohol and aggre ssion: A meta-analysis on the moderating effects of inhibitory cu es, triggering events, and self-focused attention. Psychological Bulletin, 120 (1), 60-82. Jacobs, G. A., Latham. L. E., & Brown, M. S. (1988). Test-retest reli ability of the StateTrait Personality Inventory a nd the Anger Expression scale. Anxiety Research, 1 263-265. Jeavons, C. M., & Taylor, S. P. (1985). Th e control of alcohol -related aggression: Redirecting the inebriate's attenti on to socially ap propriate conduct. Aggressive Behavior, 11 (2), 93-101.
102 Johnston, W. A., and Dark, Veroni ca J. (1986). Selective attention. Annual Review of Psychology, 37 43-75. Kaly, P. W., Heesacker, M., & Frost, H. M. (2002). Collegiate alcohol use and high-risk sexual behavior: A literature review. Journal of College Student Development, 43 (6), 838-850. Krech, D., & Crutchfield, R. S. (1948). Theory and problems of social psychology New York: McGraw-Hill. Lamb, M. R., & Robertson, L. C. (1987). Eff ect of acute alcohol on attention and the processing of hierarchical patterns. Alcoholism: Clinical & Experimental Research, 11 (3), 243-248. Lange, J. E. (2002). Alcohol's effect on aggres sion identification: A two-channel theory. Psychology of Addictive Behaviors, 16 (1), 47-55. Laplace, A. C., Chermack, S. T., & Taylor, S. P. (1994). Effects of alcohol and drinking experience on human physical aggression. Personality and Social Psychology Bulletin, 20 439-444. Leigh, B. C., and Stacy, A. W. (1993). Alc ohol outcome expectancies: Scale construction and predictive utility in higher order confirmatory models. Psychological Assessment, 5 (2), 216-229. Leonard, K. E., Collins., R. L., & Quigley, B. M. (2003). Alcohol consumption and the occurrence and severity of aggression: An event-based analysis of male to male barroom violence. Aggressive Behavior, 29 (4), 346-365. Leonard, K. E., & Senchak, M. (1993). Al cohol and premarital aggression among newlywed couples. Journal of Studies on Alcohol, Suppl. 11 96-108. Lewis, J. L. (1970). Semantic processing of unattended messages using dichotic listening. Journal of Experime ntal Psychology, 85 (2), 225-228. Lewis, W. A., & Bucher, A. M. (1992). Anger, catharsis, the reformulated frustrationaggression hypothesis, a nd health consequences. Psychotherapy: Theory, Research, Practice, Training, 29 (3), 385-392. Liebert, R. M., & Spratkin, J. (1988). The early window: Effects of television on children and youth (3rd edition). New York: Pergamon Press.
103 Logan, G. D., Cowan, W. B., & Davis, K. A. (1984) On the ability to inhibit simple and choice reaction time response: A model and a method. Journal of Experimental Psychology: Human Perception, 10 276-291. McCabe, S. B., and Gotlib, Ian H. (1993). Atte ntional processing in clinically depressed subjects: A longitudinal investigation. Cognitive Therapy & Research, 17 (4), 359377. MacDonald, T. K., Zanna, M. P., & Fong, G. T. (1995). Decision making in altered states: Effects of alcohol on att itudes toward drinking and driving. Journal of Personality & Social Psychology, 68 (6), 973-985. MacDonald, T. K., Zanna, M. P., & Fong, G. T. (1996). Why common sense goes out the window: Effects of alcohol on intentions to use condoms. Personality & Social Psychology, 22 (8), 763-775. Mackay, D. G. (1973). Aspects of the theory of comprehension, me mory and attention. Quarterly Journal of Expe rimental Psychology A, 25 (1), 22-40. MacLeod, C. & Rutherford, E. M. (1992). Anxiety and the sel ective processing of emotional information: Mediating roles of awareness, trait and state variables, and personal relevance of stimulus materials. Behavioral Research and Therapy, 30 (5), 479-491. Mahon, N. E., Yarcheski, A., & Yarcheski, T. J. (2000). Positive and negative outcome of anger in early adolescents. Research in Nursing & Health, 23 (1), 17-24. Marlatt, G. A., & Mi ller, W. R. (1986). CDP: Comprehensive Drinker Profile. Odessa, FL: Psychological Assessment Resources. Medin, D. L., Ross, B. H., & Markman, A. B. (2001). Cognitive Psychology (3rd ed.). Fort Worth: Harcourt. Moray, N. (1959). Attention in dichotic listening: Affective cues and the influence of instructions. Quarterly Journal of Ex perimental Psychology, 11 56-60. Morris, A. B., & Albery, I. P. (2001). Alc ohol consumption and HIV risk behaviours: Integrating the theories of alcohol myopia and outcome-expectancies. Addiction Research & Theory, 9 (1), 73-86.
104 Mulvihill, L. E., Skilling, T. A., & Vogel-Sp rott, M. (1997). Alcohol and the ability to inhibit behavior in men and women. Journal of Studies on Alcohol, 58 (6), 600605. Nelson, D. L., McEvoy, C. L. & Schreiber, T. A. (1999). The University of South Florida Word Association, Rhyme, and Fragment Norms. http://luna.cas.usf.edu/~nelson/ Ortells, J. J., & Tudela, P. (1996). Positive and negative sematic pr iming of attended and unattended parafoveal words in a lexical decision task. Acta Psychologica, 94 (2), 209-226. Parrott, D. J., & Zeichner, A. (2002). Effe cts of alcohol and trait anger on physical aggression in men. Journal of Studies on Alcohol., 63 (2), 196-204. Pashler, H. E. (1998). The psychology of attention Cambridge, MA: MIT Press. Pastore, N. (1952). The role of arbitrarin ess in the frustratio n-aggression hypothesis. Journal of Abnormal and Social Psychology, 47 728-731. Pedhazur, E. J. (1997). Multiple regression in behavioral research: Explanation and prediction (3rd ed.). Fort Worth: Harcourt. Posner, M. I., & Snyder, C. R. (1975). Attent ion and cognitive control. In R. L. Solso (Ed.), Information processing and cognition: The Loyola symposium (pp. 55-85). Hillsdale, NJ: Erlbaum. Quigley, B. M., & Leonard, K. E. (1999). Husband alcohol exp ectancies, drinking, marital conflict styles as predictors of severe marital violence among newlywed couples. Psychology of Addictive Behaviors, 13 49-59. Rabiner, D. L., Lenhart, L., & Lochman, J. E. (1990). Automatic vers us reflective social problem solving in relation to children's sociometric status. Developmental Psychology, 26 (6), 1010-1016. Rayner, K. (1978). Foveal and parafoveal cues in reading. In J. Requin (Ed.) Attention and Performance VII: Proceedings of the Seventh International Symposium on Attention and Performance. Hillsdale, NJ: Lawrence Erlbaum. Renfrew, J. A. (1997). Aggression and its causes. New York: Oxford University Press. Reynolds, W. M. (1982). Development of reliable and valid short forms of the MarloweCrowne social desirability scale. Journal of Clinical Psychology, 38 119-25.
105 Richardson, D. (1981). The effect of alcohol on male aggression toward female targets. Motivation & Emotion, 5 (4), 333-344. Richardson, D. R., Vandenberg, R. J., & Humphries, S. A. (1986). Effect of power to harm on retaliative aggression among males and females. Journal of Research in Personality, 20 (4), 402-419. Rohsenow, D. J. (1983). Drinking habits and ex pectancies about alcoho l's effects for self versus others. Journal of Consulting & Clinical Psychology, 51 (5), 752-756. Rohsenow, D. J., Bachorowski, J. (1984). Eff ects of alcohol and expectancies on verbal aggression in men and women. Journal of Abnormal Psychology, 93 (4), 418-433. Sayette, M. A. (1999). Cognitive theory and re search. In K. E. Leonard & H. T. Blane (Eds.), Psychological theories of drinking and alcoholism (2nd ed, pp-247-291). New York: Guilford Press. Sandoval, A. R. (1997). The relationship between mode of aggression and sex differences. Unpublished master's thesis, Univer sity of South Florida, Tampa. Scheier, M. F., Buss, A. H., Buss, D. M. (1978). Self-consciousness, self-report of aggressiveness, and aggression. Journal of Research Personality, 12 133-140. Sharma, D., Albery, I. P. and Cook, C. (2001) Selective attentiona l bias to alcohol related stimuli in problem dri nkers and non-problem drinkers. Addiction, 96 (2), 285-295. Smith, E. R., & Mackie, D. M. (2000). Social Psychology (2nd ed.). Philadelphia, PA: Psychology Press. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the State-Trait Anxiety Inventory Palo Alto, CA: Consulting Psychologists. Spielberger, C. D., Jacobs. G., Russell, S., & Crane, R. (1983). Assessment of anger: The State-Trait Anger Scale (STAS). In J. N. Butcher & C. D. Spielberger (Eds.), Advances in Personality Assessment (Vol. 2, pp. 159-187). Hillsdale, NJ: Lawrence Erlbaum. Spielberger, C. D., Jacobs, G. A., Crane, R. S., Russell, S. F., We stberry, L., Barker, L., Johnson, E. H., Knight, J., and Marks, E. (1979). State-Trait Personality Inventory Tampa: Human Resources Institute University of South Florida.
106 Spielberger, C. D., Johnson, E. H., Russell, S. F., Crane, R. J., Jacobs, G. A., & Worden, T. I. (1985). The experience and expressi on of anger: Construction and validation of an anger expression scale. In M. A. Chesney & R. H. Rosenman (Eds.), Anger and hostility in cardiovascul ar and behavioral disorders (pp. 5-30). New York: Hemisphere/McGraw-Hill. Spielberger, C. D., Sydeman, S. J., Owen, A. E., & Marsh, B. J. (1999). Measuring anxiety and anger with the State-Trait Anxiety Inventory (STAI) and the StateTrait Anger Expression Inventory (ST AXI). In M. E. Maruish, Ed. (Ed.), The use of psychological testing for trea tment planning and outcomes assessment. (2nd edition ed., pp. 993-1021). Mahwah, NJ: Lawrence Erlbaum. Spielberger, C. D., Reheiser, E. C., & Syde man, S. J. (1995). Measuring the experience, expression, and control of a nger. In H. Kassinove (Ed.), Anger disorders: Definition, diagnosis, and treatment. Series in clinical and community psychology. (pp. 49-67). Washington DC : Taylor & Francis. Spielberger, C. D. (1988). State-Trait Anger Expression Inventory Odessa, FL: Psychological Assessment Resources. Steele, C. M. & Josephs, R. A. (1990). Alcohol myopia: Its prized a nd dangerous effects. American Psychologist, 45 (8), 921-933. Steele, C. M., and Southwick, L. (1985). Alc ohol and social behavi or I: The psychology of drunken excess. Journal of Personality & Social Psychology, 48 (1), 18-34. Stormark, K. M., Field, N. P., Hugdahl, K., and Horowitz, M. (1997). Selective processing of visual alcohol cues in abs tinent alcoholics: An approach-avoidance conflict? Addictive Behaviors, 22 (4), 509-519. Stormark, K. M., Laberg, J. C., Nordby, H., and Hugdahl, K. (2000). Alcoholics' selective attention to alcohol stimuli: Automated processing? Journal of Studies on Alcohol, 61 (1), 18-23. Taylor, S. P. (1993). Experimental inves tigation of alcohol-induced aggression in humans. Alcohol Health & Research World, 17 (2), 108-112. Taylor, S. P., & Epstein, S. ( 1967). Aggression as a function of the interaction of the sex of the aggressor and th e sex of the victim. Journal of Personality, 35 (3), 474-485.
107 Tedeschi, J., & Quigley, B. (1996). Limita tions of laboratory paradigms for studying aggression. Aggression and Violent Behavio r: A Review Journal, 1 163-177. Testa, M. (2002). The impact of men's alc ohol consumption on perpetration of sexual aggression. Clinical Psychology Review, 22 (8), 1239-1263. Townshend, J. M., & Duka, T. (2001). Attenti onal bias associated with alcohol cues: Differences between heavy and occasional social drinkers. Psychopharmacology., 157 (1), 67-74. Treisman, A. M. (1960). Contextual cues in selective listening. Quarterly Journal of Experimental Psychology, 12 242-248. Treisman, A. M., & Geffen, G. (1967). Sele ctive attention: Perception or response? Quarterly Journal of Ex perimental Psychology, 19 (1), 1-17. U.S. Department of Justice Bureau of Jus tice Statistics. (n.d.). Substance abuse and treatment of state and federal pris oners, 1997. Retrieved January 17, 2004, from http://www.ojp.usdoj.gov/bj s/abstract/satsfp97.htm Watson, D., Clark, L. A., & Tellegen, A. ( 1988). Development and validation of brief measures of positive and negative affect: The PANAS Scales. Journal of Personality and Social Psychology, 54 181-202. World Health Organization. (2001, May 21). Vi olence: An enormous, but preventable global health problem. Retrieved Janua ry 17, 2004, from http://www.who.int/infpr-2001/ en/note2001-WHA6.html Zeichner, A., Allen, J. D., Petrie, C. D., Rasmussen, P. R., & Giancola, P. R. (1993). Attention allocation: Effects of alcoho l and information salience on attentional processes in male social drinkers. Al coholism: Clinical and Experimental Research, 17(4), p. 727-732. Zeichner, A., Pihl, R. O., Niaura, R., & Z acchia, C. (1982). Attentional processes in alcohol-mediated aggression. Journal of Studies on Alcohol, 43 (7), 714-724. Zevon, M. A., & Tellegen, A. (1982). Th e structure of mood change: An idiographic/nomothetic analysis. Journal of Personality and Social Psychology, 43 111-122.
108 Footnote 1 Since significant mean error rate diff erences were not found in an earlier study (Edington, 1996), the potential effect of unreliable data on mean reaction time was considered to be of primary importance in th e current study. Analyses revealed that when reaction time data from participants with chance-level error rates ( N = 6) was compared with data from those that ma de relatively few mistakes ( N = 79), standard errors calculated using Repeated Measures ANOVA we re quite different. For low error rate participants completing the Dichotic Listeni ng Task the standard error was estimated at 24.77. For the Parafoveal Visual Task, standa rd error was 33.55. For high error rate participants completing the Dichotic Listen ing Task, standard error was 153.58 and, for the Parafoveal Visual Task, standard error was 162.10. Standard errors were roughly five times higher for participants with an error ra te at a level no better than chance, further supporting the notion that the reaction time data from these six participants was unreliable.
110 Appendix A Request for Further Participation Our lab is conducting several other experi ments at this time. These include the completion of various computer tasks. Most of these tasks require from 45 minutes to 1 hour and 15 minutes. Individuals may receive up to 3 experimental points for participating in any one of these experiments. ARE YOU INTERESTED IN BEING CO NTACTED FOR PARTICIPATION IN ANY OF THESE STUDIES? ______ NO ______ YES (If YES PLEASE COMPLETE THE FO LLOWING QUESTIONS.) Do you have any hearing, visual, or motor im pairments, or any other impairment, that would prevent you from completing various computer tasks? _______ NO _______ YES _______ DONÂ’T KNOW/UNSURE (Please descr ibe_______________ ________________.) Is English your first/native language? _______ NO _______ YES NAME (please print) _____________________________________________________ PHONE NUMBER(S) (where you can be reached): (_____)_____________________ (_____)_____________________ E-MAIL ADDRESS : ____________________________________________________
111 Appendix B State Trait Anger Expression Inventory (STAXI) Part 1 Directions: A number of statements that peopl e use to describe themselves are given below. Read each statement and then ci rcle the appropriate number to indicate how you feel right now There are no right or wrong answers. Do not spend too much time on any one statement, but give the answer that seems to best describe your present feelings Not At AllSomewhat Moderately So Very Much So 1. I am furious 1 2 3 4 2. I feel irritated 1 2 3 4 3. I feel angry 1 2 3 4 4. I feel like yelling at somebody 1 2 3 4 5. I feel like breaking things 1 2 3 4 6. I am mad 1 2 3 4 7. I feel like banging on the table 1 2 3 4 8. I feel like hitting someone 1 2 3 4 9. I am burned up 1 2 3 4 10. I feel like swearing 1 2 3 4 Part 2 Directions: A number of statements that peopl e use to describe themselves are given below. Read each statement and then circle the appropriate number on the answer sheet to indicate how you generally feel. There are no right or wrong answers. Do not spend too much time on any one statement, but give the answer that seems to best describe how you generally feel. Almost Never Sometimes Often Almost Always 11. I am quick tempered 1 2 3 4 12. I have a fiery temper 1 2 3 4
112 Appendix B (Continued) Almost Never Sometimes Often Almost Always 13. I am a hotheaded person 1 2 3 4 14. I get angry when IÂ’m slowed down by others/ mistakes. 1 2 3 4 15. I feel annoyed when I am not given recognition for doing good work 1 2 3 4 16. I fly off the handle 1 2 3 4 17. When I get mad, I say nasty things 1 2 3 4 18. It makes me furious when I am criticized in front of others 1 2 3 4 19. When I get frustrated, I feel like hitting someone 1 2 3 4 20. I feel infuriated when I do a good job and get a poor evaluation 1 2 3 4 Part 3 Directions: Everyone feels angry or furious from time to time, but people differ in the ways that they react when they are a ngry. A number of statements are listed below which people use to describe th eir reactions when they feel angry or furious Read each statement and the circle the appropriate number on the answer sheet to indicate how often you generally react or behave in the manner desc ribed when you are feeling angry or furious. There are no right or wrong answers. Do not spend too much time on any one statement. WHEN ANGRY OR FURIOUSÂ… Almost Never Sometimes Often Almost Always 21. I control my temper 1 2 3 4 22. I express my anger 1 2 3 4 23. I keep things in 1 2 3 4 24. I am patient with others 1 2 3 4 25. I pout or sulk 1 2 3 4
113 Appendix B (Continued) WHEN ANGRY OR FURIOUSÂ… Almost Never Sometimes Often Almost Always 26. I withdraw from people 1 2 3 4 27. I make sarcastic remarks to others 1 2 3 4 28. I keep my cool 1 2 3 4 29. I do things like slam doors 1 2 3 4 30. I boil inside, but I donÂ’t show it 1 2 3 4 31. I control my behavior 1 2 3 4 32. I argue with others 1 2 3 4 33. I tend to harbor grudges that I donÂ’t tell anyone about 1 2 3 4 34. I strike out at whatever infuriates me 1 2 3 4 35. I can stop myself from losing my temper 1 2 3 4 36. I am secretly quite critical of others 1 2 3 4 37. I am angrier than I am willing to admit 1 2 3 4 38. I calm down faster than most other people 1 2 3 4 39. I say nasty things 1 2 3 4 40. I try to be tolerant and understanding1 2 3 4 41. IÂ’m irritated a great deal more than people are aware of 1 2 3 4 42. I lose my temper 1 2 3 4 43. If someone annoys me, IÂ’m apt to tell him or her how I feel 1 2 3 4 44. I control my angry feelings 1 2 3 4
114 Appendix C Buss Perry Aggression Questionnaire Instructions: Following are some statemen ts which may or may not describe YOU. Beside each statement, circle the number re presenting the rating which best describes YOU. Extremely Unlike Me Mostly Unlike Me Somewhat Like Me Mostly Like Me Extremely Like Me 1 2 3 4 5 Extremely Unlike Me Mostly Unlike Me Somewhat Like Me Mostly Like Me Extremely Like Me 1. Once in a while I canÂ’t control the urge to strike another person. 1 2 3 4 5 2. I tell my friends openly when I disagree with them. 1 2 3 4 5 3. I flare up quickly but get over it quickly. 1 2 3 4 5 4. I am sometimes eaten up with jealousy. 1 2 3 4 5 5. At times I feel I have gotten a raw deal out of life. 1 2 3 4 5 6. I often find myself disagreeing with people. 1 2 3 4 5 7. When frustrated, I let my irritation show. 1 2 3 4 5 8. Given enough provocation, I may hit another person. 1 2 3 4 5 9. Other people always seem to get the breaks. 1 2 3 4 5 10. I sometimes feel like a powder keg ready to explode. 1 2 3 4 5
115 Appendix C (Continued) Extremely Unlike Me Mostly Unlike Me Somewhat Like Me Mostly Like Me Extremely Like Me 11. If somebody hits me, I hit back. 1 2 3 4 5 12. I wonder why sometimes I feel so bitter about things. 1 2 3 4 5 13. I am an even-tempered person. 1 2 3 4 5 14. Some of my friends think IÂ’m a hothead. 1 2 3 4 5 15. I get into fights a little more than the average person. 1 2 3 4 5 16. If I have to resort to violence to protect my rights, I will. 1 2 3 4 5 17. When people annoy me, I may tell them what I think of them. 1 2 3 4 5 18. I know that Â“friendsÂ” talk about me behind my back. 1 2 3 4 5 19. Sometimes I fly off the handle for no good reason. 1 2 3 4 5 20. There are people who pushed me so far that we came to blows. 1 2 3 4 5 21. I am suspicious of overly friendly strangers. 1 2 3 4 5 22. I canÂ’t help getting into arguments when people disagree with me. 1 2 3 4 5 23. I can think of no good reason for ever hitting a person. 1 2 3 4 5
116 Appendix C (Continued) Extremely Unlike Me Mostly Unlike Me Somewhat Like Me Mostly Like Me Extremely Like Me 24. I have become so mad that I have broken things. 1 2 3 4 5 25. I sometimes feel that people are laughing at me behind my back. 1 2 3 4 5 26. My friends say that IÂ’m somewhat argumentative. 1 2 3 4 5 27. When people are especially nice, I wonder what they want. 1 2 3 4 5 28. I have threatened people I know. 1 2 3 4 5 29. I have trouble controlling my temper. 1 2 3 4 5
117 Appendix D Expectancy Questionnaire for Alcoho l and AggressionÂ—Low Dose (EQAAL) version Instructions: Many people believe that drinki ng alcohol can influence how angry they feel and how aggressive they ac t. We would like to know how you think having a few drinks of alcohol (enough to make you buzzed) affects you Please circle the number that best describes to what extent you agree or disagree with each statement below. (If you do not drink at all, you can still fill this out. Just answer the questions according to what you think you would feel like if you did drink.) When I have had a few drinks of alcohol I am more likely to: _____________________ Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 1. get furious. 1 2 3 4 5 6 2. get angry when I am in line to get something and someone cuts in front of me. 1 2 3 4 5 6 3. think that people who act like theyÂ’re being honest really have something to hide 1 2 3 4 5 6 4. keep my cool. 1 2 3 4 5 6 5. feel angry. 1 2 3 4 5 6 6. get angry if I am trying to concentrate, but someone keeps making noise. 1 2 3 4 5 6 7. get frustrated and feel like hitting someone 1 2 3 4 5 6 8. get angry when I need to get somewhere in a hurry, but I get stuck in traffic. 1 2 3 4 5 6 9. wonder about the hidden reasons if someone does something nice for me. 1 2 3 4 5 6
118 Appendix D (Continued) Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 10. control my behavior. 1 2 3 4 5 6 11. fly off the handle. 1 2 3 4 5 6 12. get angry when I am singled out for correction, while someone else who is doing the same thing is ignored. 1 2 3 4 5 6 13. stop myself from losing my temper. 1 2 3 4 5 6 14. feel like yelling at somebody. 1 2 3 4 5 6 15. get angry with someone who looks through my things without permission. 1 2 3 4 5 6 16. feel that other people always seem to get the breaks. 1 2 3 4 5 6 17. get mad. 1 2 3 4 5 6 18. get angry when I am accused of something I didnÂ’t do. 1 2 3 4 5 6 19. try to be tolerant and understanding. 1 2 3 4 5 6 20. have a fiery temper. 1 2 3 4 5 6 21. get angry with someone who is always contradicting me. 1 2 3 4 5 6 22. control my angry feelings. 1 2 3 4 5 6 23. get burned up. 1 2 3 4 5 6
119 Appendix E Alcohol-Aggression Items from Various Measures Power and Aggression Subscale ( from the Alcohol Effects Questionnaire (G eorge, Frone, Cooper, Russell, Skinner, & Windle, 1995; Rohsenow, 1983). Instructions: Please respond to the following statem ents according to your own personal thoughts, feelings and beliefs about alcohol. We are interest ed in what you think about alcohol, regardless of what other people might think. Agree Strongly Agree Moderately Agree Slightly Disagree Slightly Disagree Moderately Disagree Strongly 1. Drinking makes me feel warm and flushed. 1 2 3 4 5 6 5. I feel powerful when I drink, as if I can really make other people do as I want. 1 2 3 4 5 6 9. If I have had a couple of drinks, it is easier for me to tell someone off. 1 2 3 4 5 6 16. Drinking makes me more aggressive. 1 2 3 4 5 6 32. IÂ’m more likely to get into an argument if IÂ’ve had some alcohol. 1 2 3 4 5 6 37. After a few drinks it is easier for me to pick a fight. 1 2 3 4 5 6
120 Appendix E (Continued) Risk and Aggression Subscale (from the Comprehensive Effects of Alcohol scale; Fromme, Stroot, & Kaplan, 1993) Instructions: Please respond to the following statemen ts by circling the number that best completes the following sentence. Disagree Slightly Disagree Slightly Agree Agree 1 2 3 4 If I were under the influence from drinking alcoholÂ…. DisagreeSlightly Disagree Slightly Agree Agree 1. I would take risks. 1 2 3 4 2. I would act aggressively. 1 2 3 4 3. I would be loud, boisterous, or noisy. 1 2 3 4 4. I would act tough. 1 2 3 4 5. I would feel dominant. 1 2 3 4
121 Appendix E (Continued) Arousal/Aggression Subscale (from the Alcohol Expectancy Questionna ire; Brown, Goldman, Inn, & Anderson, 1980) Instructions: Read each statement carefully and re spond according to your own personal thoughts, feelings and beliefs about alcohol now. We are interested in what you think about alcohol, regardless of what other peopl e might think. If you think that the statement is true, or mostly true, or true some of the time, then circle "Agree" on the answer sheet. If you think the statement is fals e, or mostly false, then circle "Disagree" on the answer sheet. When the statements refer to drinki ng alcohol, you may think in terms of drinking any alcoholic beverage, such as beer, wine whiskey, liquor, rum, scotch, vodka, gin, or various alcoholic mixed drinks. Whether or not you have had actual drinking experiences yourself, you are to answer in te rms of your beliefs about alcohol 1. Agree Disagree Drinking makes me feel flushed 2. Agree Disagree After a few drinks, it is easier to pick a fight 3. Agree Disagree I feel powerful when I drink, as if I can really influence others to do as I want 4. Agree Di sagree Drinking increases male aggressiveness 5. Agree Disagr ee At times, drinking is like permission to forget problems
122 Appendix E (Continued) Aggression Subscale (from the Drinking Expectancy Qu estionnaire; Young, & Knight, 1988) Instructions: The following questions ask about the effects that dr inking alcohol has on you. There are no right or wrong answers to these items. We would like to know how you feel about them. All that is re quired is that you circle the appropriate number beside each statement, using the following key: Strongly Disagree Disagree Neither Agree nor Disagree Agree Strongly Agree 1 2 3 4 5 Strongly Disagree Disagree Neither Agree nor Disagree Agree Strongly Agree 1. I control my temper more easily when drinking alcohol. 1 2 3 4 5 2. Little things annoy me less when IÂ’m drinking. 1 2 3 4 5 3. Drinking increases my aggressiveness. 1 2 3 4 5
123 Appendix E (Continued) Social Subscale (from the Effects of Drinking Alc ohol scale; Leigh & Stacy, 1993) Instructions: Here are some effects or cons equences that some people experience after drinking alcohol. How likely is it that these things happen to you when you drink alcohol? Please circle the number that best describes how drinkingalcohol would affect you. (If you do not drink at all, you can still fill this out; just answer it according to what you think would happen to you if you did drink.) When I drink alcohol: ________________________________ No Chance Very Unlikely Unlikely Likely Very Likely Certain to Happen 1. I become aggressive. 1 2 3 4 5 6 2. I get into fights. 1 2 3 4 5 6 3. I get mean 1 2 3 4 5 6
124 Appendix F Demographics Questionnaire Please provide the following background information: Date of Birth: _____________ Gender: ____ Male ____ Female Race/Ethnicity: ____ African American ____ Asian American ____ Caucasian ____ Hispanic ____ Latino ____ Native American ____ Other (Please specify:__________________________________) Household yearly income (home that you were raised in): ____ Less than $10,000 ____ $40,000 $79,000 ____ $10,000 Â– $24,999 ____ More than $80,000 ____ $25,000 -$39,999 ____ I do not know our yearly income.
125 Appendix G Positive and Negative Affect Schedule (PANAS) This scale consists of a number of word s that describe diffe rent feelings and emotions. Read each item and then mark th e appropriate answer in the space next to that word. Indicate to wh at extent you feel this way RIGHT NOW, THAT IS, AT THIS VERY MOMENT. Use the following scale to record your answers: 1 2 3 4 5 very slightly a little bit mode rately quite a bit extremely or not at all _____ guilty _____ determined _____ scared _____ attentive _____ hostile _____ jittery _____ enthusiastic _____ active _____ interested _____ irritable _____ distressed _____ alert _____ excited _____ ashamed _____ upset _____ inspired _____ strong _____ proud _____ nervous _____ afraid
126 Appendix H PARAFOVEAL VISUAL TASK WORD PAIRS Parafoveal (Unattended; On the Computer) Type Digits Auditory (Attended; On the Headphones) Word (1st of Pair) Pixel Position (X/Y axis : range in cm) Word (2nd of Pair) Db Word Begin Word End Length of .wav 10 Blank 10 Blank Control (0 Â“LÂ” words) Odd 27935 1. Bird -1 1.01 1.80 2.002 Even 12476 2. Cone -1 1.07 1.78 2.000 Odd 71392 3. Dream -1 1.02 1.83 1.999 Even 49231 4. Math -1 1.05 1.65 2.000 Odd 92758 5. Mint -1 1.04 1.67 2.000 Even 25879 6. Porch -1 1.09 1.67 2.000 Odd 84167 7. Rhyme -1 0.93 1.83 2.000 Even 51683 8. Sheep -1 1.05 1.74 2.002 Odd 01589 9. Shirt -1 1.03 1.70 2.002 Even 39615 10. Well -1 1.09 1.70 2.000 10 Practice Words 10 Practice Control (1 Â“LÂ” word) 1. Barn Up (Y87 : 2.6 Â– 3.4) Odd 52146 1. Calm -1 1.00 1.82 2.001 2. Dull Dn (Y-90 : 2.6 Â– 3.4) Even 82651 2. Prune -1 1.07 1.76 2.001 3. Grace Lf (X-124 : 6.3 Â– 2.6) Odd 32768 3. Spice -1 0.84 1.82 2.000 4. Grin Rt (X110 : 2.6 Â– 5.1) Even 60492 4. Roof -1 1.03 1.72 2.001 5. Hall Up (Y87 : 2.6 Â– 3.4) Even 46280 5. Lack -1 1.04 1.75 2.001 6. Lend Dn (Y-90 : 2.6 Â– 3.4) Odd 74325 6. Cough -1 1.10 1.77 2.000 7. More Lf (X-115 : 5.6 Â– 2.6) Even 91847 7. Pink -1 1.08 1.81 2.000 8. Ranch Rt (X127 : 2.6 Â– 6.3) Odd 08573 8. Shade -1 0.92 1.83 2.002 9. Truth Up (Y87 : 2.6 Â– 3.4) Odd 15904 9. View -1 1.08 1.78 1.999 10. Yacht Dn (Y-90 : 2.6 Â– 3.4) Even 28469 10. Then -1 1.04 1.68 2.001 20 NonAgg Words 20 NonAgg Control (1 Â“LÂ” word) 1. Air Lf (X-97 : 4.4 Â– 2.6) Odd 62793 1. Tent -1 1.12 1.64 2.002 2. Bake Rt (X115 : 2.6 Â– 5.3) Even 17428 2. Tube -1 1.02 1.77 2.001 3. Blank Up (Y87: 2.6 Â– 3.4) Even 90837 3. Shine -1 0.99 1.80 2.000 4. Chance Dn (Y-90 : 2.6 Â– 3.4) Odd 26180 4. Fruit -1 1.03 1.65 2.001 5. Chill Lf (X-113 : 5.5 Â– 2.6) Ev en 83219 5. Truck -1 1.07 1.74 2.002 6. Crust Rt (X127 : 2.6 Â– 6.3) Odd 50374 6. Scarf -1 1.00 1.79 2.000 7. Find Up (Y87 : 2.6 Â– 3.4) Odd 38516 7. Pump -1 1.04 1.62 2.001 8. Five Dn (Y-90 : 2.6 Â– 3.4) Even 47632 8. Hip -1 1.07 1.65 2.001 9. Flag Lf (X-108 : 5.3 Â– 2.6) Odd 71905 9. Mop -1 1.06 1.75 2.000 10. Groom Rt (X132 : 2.6 Â– 6.2) Even 08459 10. Up -1 1.13 1.56 2.001 11. Note Up (Y87 : 2.6 Â– 3.4) Even 81625 11. Lamp -1 1.13 1.77 2.000 12. Plot Dn (Y-90 : 2.6 Â– 3.4) Odd 24730 12. Frog -1 1.17 1.71 2.000 13. Rain Lf (X106 : 5 Â– 2.6) Even 90238 13. Spoon -1 0.98 1.86 2.001 14. Scoop Rt (X128 : 2.6 Â–6.4) Odd 17584 14. Guide -1 1.03 1.76 2.000 15. Scrap Up (Y87 : 2.6 Â– 3.4) Odd 62103 15. Teach -1 1.04 1.76 2.001 16. Shop Dn (Y-90 : 2.6 Â– 3.4) Even 59472 16. Grape -1 1.03 1.76 2.000 17. Shrimp Lf (X-130 : 6.6 Â– 2.6) Odd 32941 17. Stamp -1 0.99 1.83 2.000 18. Space Rt (X124 : 2.6 Â– 6.1) Ev en 43628 18. Desk -1 1.05 1.78 2.001 19. Stew Up (Y87 : 2.6 Â– 3.4) Even 71254 19. Fresh -1 0.99 1.80 2.000 20.Year Dn (Y-90 : 2.6 Â– 3.4) Odd 08396 20. Mom -1 1.05 1.77 2.001 20 LoAgg Words 20 LoAgg Control (1 Â“LÂ” word) 21. Cage Lf (X-112 : 5.5 Â– 2.6) Even 91426 21. Reach -1 1.06 1.77 2.000 22. Dare Rt (X116 : 2.6 Â– 5.5) Odd 68304 22. Cup -1 1.11 1.69 2.001 23. Fray Up (Y87 : 2.6 Â– 3.4) Odd 59712 23. Blush -1 1.02 1.78 2.001 24. Fraud Dn (Y-90 : 2.6 Â– 3.4) Even 27830 24. Laugh -1 1.00 1.81 2.000 25. Free Lf (X-111 : 5.4 Â– 2.6) Odd 14589 25. Plum -1 1.21 1.73 2.001 26.Friend Rt (X129 : 2.6 Â– 6.5) Even 73625 26. Youth -1 1.06 1.80 2.001 27. Ghoul Up (Y87 : 2.6 Â– 3.4) Even 30257 27. Scale -1 0.97 1.80 2.001 28. Grief Dn (Y-90 : 2.6 Â– 3.4) Odd 47152 28. Flunk -1 1.07 1.80 2.000 29. Hide Lf (X-106 : 5 Â– 2.6) Even 04863 29. Fool -1 1.19 1.75 2.000 30. Lie Rt (X96 : 2.6 4) Odd 83921 30. Shrub -1 1.10 1.81 2.000
127 Appendix H (Continued) 31. Like Up (Y87 : 2.6 Â– 3.4) Odd 34785 31. Need -1 1.00 1.82 2.001 32. Make Dn (Y-90 : 2.6 Â– 3.4) Even 71452 32. Crave -1 1.19 1.80 2.000 33. March Lf (X-125 : 6.3 Â– 2.6) Odd 23501 33. Shed -1 1.00 1.80 2.001 34. Mask Rt (X116 : 2.6 Â– 5.5) Even 19238 34. Chew -1 1.04 1.76 2.001 35. Pat Up (Y87 : 2.6 Â– 3.4) Even 58614 35. Few -1 1.09 1.80 2.000 36. Take Dn (Y-90 : 2.6 Â– 3.4) Odd 48327 36. Age -1 1.13 1.68 2.000 37. Tight Lf (X-117 : 5.8 Â– 2.6) Even 60439 37. Scroll -1 0.92 1.85 2.000 38. Toil Rt (X108 : 2.6 Â– 4.9) O dd 89106 38. Beach -1 1.02 1.65 2.000 39. Trap Up (Y87 : 2.6 Â– 3.4) Odd 05983 39. Cheese -1 1.02 1.76 2.000 40. Urge Dn (Y-90 : 2.6 Â– 3.4) Even 92840 40. Great -1 1.15 1.66 2.000 20 AmbAgg 20 AmbAgg Control (0 Â“LÂ” words) 41. Bang Lf (X-114 : 5.4 Â– 2.6) Odd 62793 41. Key -1 1.11 1.67 2.000 42. Beat Rt (X114 : 2.6 Â– 5.1) Even 17428 42. Frost -1 0.93 1.86 2.000 43. Bruise Up (Y87 : 2.6 Â– 3.4) Even 90837 43. Have -1 1.03 1.81 2.001 44. Chop Dn (Y-90 : 2.6 Â– 3.4) Odd 26180 44. Film -1 1.04 1.70 2.000 45. Curse Lf (X-125 : 6.2 Â– 2.6) Even 83219 45. Food -1 1.01 1.83 1.999 46. Cut Rt (X108 : 2.6 Â– 4.8) Odd 50374 46. Feel -1 1.04 1.77 2.000 47. Grab Up (Y87 : 2.6 Â– 3.4) Odd 38516 47. Boat -1 1.10 1.78 2.000 48. Guard Dn (Y-90 : 2.6 Â– 3.4) Even 47632 48. Yawn -1 0.99 1.83 2.000 49. Hit Lf (X-98 : 4.3 Â– 2.6) Odd 71905 49. Spell -1 1.02 1.80 2.001 50. Lash Rt (X114 : 2.6 Â– 5.3) Even 08459 50. Scrub -1 1.01 1.80 2.000 51. Mad Up (Y87 : 2.6 Â– 3.4) Even 81625 51. Hint -1 1.11 1.67 2.001 52. Mob Dn (Y-90 : 2.6 Â– 3.4) Odd 24730 52. Guess -1 1.05 1.66 2.001 53. Punch Lf (X-126 : 6.3 Â– 2.6) Even 90238 53. Store -1 1.10 1.71 2.001 54. Push Rt (X116 : 2.6 Â– 5.5) Odd 17584 54. Room -1 1.11 1.79 2.003 55. Rude Up (Y87 : 2.6 Â– 3.4) Odd 62103 55. Mist -1 1.06 1.67 2.000 56. Stern Dn (Y-90 : 2.6 Â– 3.4) Even 59472 56. Brave -1 1.18 1.79 2.001 57. Strike Lf (X-126 : 6.3 Â– 2.6) Odd 32941 57. Broom -1 1.14 1.76 2.001 58. Tank Rt (X116 : 2.6 Â– 5.5) Even 43628 58. Share -1 1.09 1.81 2.000 59. Tough Up (Y87 : 2.6 Â– 3.4) Even 71254 59. Feed -1 1.03 1.72 2.001 60. Whip Dn (Y-90 : 2.6 Â– 3.4) Odd 08396 60. Rose -1 1.09 1.74 2.001 20 HiAgg Words 20 HiAgg Control Words (1 Â“LÂ” word) 61. Brawl Lf (X-127 : 6.3 Â– 2.6) Even 91426 61. Stream -1 1.00 1.80 2.001 62. Feud Rt (X114 : 2.6 Â– 5.4) Odd 68304 62. Stock -1 1.00 1.80 2.000 63. Fight Up (Y87 : 2.6 Â– 3.4) Odd 59712 63. Tape -1 1.10 1.63 2.001 64. Fist Dn (Y-90 : 2.6 Â– 3.4) Even 27830 64. Care -1 1.22 1.67 2.001 65. Force Lf (X-124 : 6.2 Â– 2.6) Odd 14589 65. Chair -1 1.13 1.72 2.001 66. Gun Rt (X110 : 2.6 Â– 4.9) Even 73625 66. Mud -1 1.15 1.75 2.000 67. Harm Up (Y87 : 2.6 Â– 3.4) Even 30257 67. Moon -1 1.07 1.65 2.000 68. Hurt Dn (Y-90 : 2.6 Â– 3.4) Odd 47152 68. Shell -1 1.08 1.78 2.001 69. Kill Lf (X-104 : 4.7 Â– 2.6) Even 83921 69. Shape -1 0.97 1.86 2.001 70. Rage Rt (X116 : 2.6 Â– 5.5) O dd 04863 70. Pouch -1 1.22 1.75 2.001 71. Rape Up (Y87 : 2.6 Â– 3.4) Odd 34785 71. Wheat -1 1.18 1.68 2.001 72. Scold Dn (Y-90 : 2.6 Â– 3.4) Even 71452 72. Race -1 1.11 1.74 2.000 73. Scream Lf (X-136 : 7 Â– 2.6) Odd 23501 73. Lake -1 1.14 1.75 2.000 74. Shoot Rt (X128 : 2.6 Â– 6.3) Even 19238 74. Book -1 1.05 1.75 2.001 75. Shot Up (Y87 : 2.6 Â– 3.4) Even 58614 75. Touch -1 1.17 1.66 2.001 76. Shout Dn (Y-90 : 2.6 Â– 3.4) Odd 48327 76. Cute -1 1.21 1.71 2.001 77.Shove Lf (X-124 : 6.2 Â– 2.6) Even 60439 77. Green -1 1.16 1.77 2.000 78. Spank Rt (X125 : 2.6 Â– 6.1) Odd 89106 78. Horse -1 1.12 1.80 2.000 79. Stab Up (Y87 : 2.6 Â– 3.4) Odd 05983 79. Park -1 1.14 1.72 2.000 80. Yell Dn (Y-90 : 2.6 Â– 3.4) Even 92840 80. Gum -1 1.22 1.70 2.001 10 Random Blanks 10 Control Words (1 Â“LÂ” word) Odd 27935 1. Cake -1 1.21 1.69 2.000 Even 12476 2. Dusk -1 1.12 1.72 2.001 Odd 71392 3. List -1 1.11 1.80 2.001 Even 49231 4. Play -1 1.13 1.80 2.001 Odd 92758 5. Short -1 1.10 1.74 2.000 Even 25879 6. Show -1 1.14 1.81 2.001 Odd 84167 7. Spin -1 0.95 1.84 2.001 Even 51683 8. Train -1 1.14 1.85 2.000
128 Appendix H (Continued) Odd 01589 9. Turn -2 1.19 1.78 2.000 Even 39615 10. Your -1 1.07 1.83 2.000 10 Final Practice Words 10 Final Practice Control Words (3 Â“LÂ” words) 1. Card Lf (X-115 : 5.5 Â– 2.6) Odd 52146 1. Heart -1 1.18 1.79 2.001 2. Dig Rt (X100 : 2.6 Â– 4.3) Even 82651 2. Lean -1 1.16 1.77 2.000 3. Dough Up (Y87 : 2.6 Â– 3.4) Odd 32768 3. Gloss -1 1.14 1.82 2.000 4. Droop Dn (Y-90 : 2.6 Â– 3.4) Even 60492 4. Last -1 1.08 1.81 2.001 5. Kind Lf (X-108 : 4 Â– 2.6) Even 46280 5. Lunch -1 1.13 1.82 1.999 6. News Rt (X119 : 2.6 Â– 5.7) Odd 74325 6. Mile -1 1.13 1.73 2.001 7. North Up (Y87 : 2.6 Â– 3.4) Even 91847 7. Wait -1 1.12 1.85 2.001 8. Please Dn (Y-90 : 2.6 Â– 3.4) Odd 08573 8. Street -1 1.15 1.84 2.000 9. Proud Lf (X-127 : 6.3 Â– 2.6) Odd 15904 9. Vast -1 1.04 1.82 2.001 10. Taste Rt (X124 : 2.6 Â– 5.9) Even 28469 10. Shoe -1 1.15 1.82 2.000 Total Parafoveal Words = 100 Total Auditory Words = 120; Total Â“LÂ” words = 8 DICHOTIC LISTENING TASK WORD PAIRS Unattended Channel of Headphones Attended Channel of Headphones Length of .wav Type Digits Word (1st of Pair) Db Word Begin Word End Word (2nd of Pair) Db Word Begin Word End 10 Blank 10 Blank Control (1 Â“LÂ” words) 1. Cough -1 1.18 1.69 2.000 Odd 27935 2. Less -1 1.20 1.75 2.000 Even 12476 3. Good -1 1.10 1.70 2.001 Odd 71392 4. Math -1 1.18 1.77 2.002 Even 49231 5. News -1 1.06 1.78 2.001 Odd 92758 6. Please -1 1.06 1.77 1.999 Even 25879 7. Rhyme -1 1.06 1.75 2.002 Odd 84167 8. Soft -1 1.11 1.80 2.002 Even 51683 9. Stray -1 1.08 1.80 2.000 Odd 01589 10. Taste -1 1.19 1.72 2.002 Even 39615 10 Practice 10 Practice Control (1 Â“LÂ” words) 1. Barn -19 1.03 1.84 1. Bird -1 1.03 1.71 2.000 Odd 52146 2. Kind -19 1.06 1.88 2. Calm -1 1.06 1.84 2.000 Even 82651 3. Dull -19 1.19 1.78 3. Dig -1 1.20 1.78 2.000 Odd 32768 4. Grin -19 1.14 1.66 4. Grace -1 1.14 1.74 2.000 Even 60492 5. Heart -19 1.12 1.77 5. Hall -1 1.14 1.84 1.999 Even 46280 6. Last -19 1.10 1.84 6. Lunch -1 1.10 1.73 2.000 Odd 74325 7. Spin -19 1.00 1.88 7. Spice -1 1.04 1.89 2.001 Even 91847 8. Shade -19 1.05 1.87 8. Short -1 1.07 1.74 2.000 Odd 08573 9. Train -19 1.08 1.83 9. Truth -1 1.08 1.71 2.001 Odd 15904 10. View -19 1.06 1.80 10. Vast -1 1.03 1.81 2.001 Even 28469 20 NonAgg Words 20 NonAgg Control (0 Â“LÂ” words) 1. Air -19 1.12 1.77 1. Age -1 1.12 1.79 2.000 Odd 62793 2. Bake -19 1.16 1. 66 2. Beach -1 1.16 1.73 1.999 Even 17428 3. Blank -19 1.20 1.78 3. Blush -1 1.20 1.79 2.001 Even 90837 4. Chance -19 1.16 1.80 4. Chew -1 1.19 1.73 2.000 Odd 26180 5. Chill -19 1.04 1.77 5. Chees e -1 1.04 1.81 2.001 Even 83219 6. Crust -19 1.13 1.76 6. Crave -1 1.13 1.80 2.002 Odd 50374 7. Find -19 1.00 1.85 7. Few -1 1.00 1.63 2.000 Odd 38516 8. Five -19 1.03 1.83 8. Fool -1 1.03 1.70 2.001 Even 47632 9. Flag -19 1.00 1.87 9. Flunk -1 1.00 1.87 1.999 Odd 71905 10. Groom -19 1.11 1.66 10. Great -1 1.12 1.82 2.000 Even 08459 11. Note -19 1.15 1.76 11. Need -1 1.15 1.84 2.000 Even 81625 12. Plot -19 1.19 1.76 12. Plum -1 1.19 1.77 2.001 Odd 24730 13. Rain -19 1.16 1. 83 13. Reach -1 1.19 1.80 2.001 Even 90238 14. Scoop -19 1.03 1.71 14. Scale -1 1.03 1.85 2.001 Odd 17584 15. Scrap -19 1.03 1.80 15. Scroll -1 1.03 1.80 2.000 Odd 62103 16. Shop -19 1.10 1.85 16. Shed -1 1.13 1.80 1.999 Even 59472
129 Appendix H (Continued) 17. Shrimp -19 1.05 1.85 17. Shrub -1 1.07 1.85 2.000 Odd 32941 18. Space -19 1.01 1.85 18. S poon -1 1.01 1.76 2.000 Even 43628 19. Stew -19 1.19 1.82 19. Stamp -1 1.18 1.79 2.001 Even 71254 20. Year -19 1.12 1.81 20. Youth -1 1.18 1.73 2.002 Odd 08396 20 LoAgg Words 20 LoAgg Control (2 Â“LÂ” words) 21. Cage -19 1.08 1.84 21. Cup -1 1.14 1.53 2.000 Even 91426 22. Dare -19 1.14 1.85 22. Desk -1 1.14 1.81 2.000 Odd 68304 23. Fraud -19 1.02 1.88 23. Fresh -1 1.05 1.83 2.000 Odd 59712 24. Fray -19 1.10 1.81 24. Frost -1 1.10 1.88 2.000 Even 27830 25. Free -19 0.99 1.78 25. Frog -1 1.02 1.88 1.999 Odd 14589 26. Friend -19 1.05 1.85 26. Fruit -1 1.08 1.68 2.002 Even 73625 27. Ghoul -19 1.00 1.89 27. Guide -1 1.00 1.89 2.000 Even 30257 28. Grief -19 1.26 1.75 28. Grape -1 1.27 1.71 1.999 Odd 47152 29. Hide -19 0.97 1.89 29. Hip -1 1.09 1.64 1.999 Even 04863 30. Lie -19 1.16 1.80 30. Laugh -1 1.16 1.80 2.002 Odd 83921 31. Like -19 1.10 1.77 31. Lamp -1 1.12 1.79 1.998 Odd 34785 32. Make -19 1.16 1.73 32. Mist -1 1.17 1.75 1.999 Even 71452 33. March -19 1.10 1.78 33. Mop -1 1.10 1.79 2.002 Odd 23501 34. Mask -19 1.16 1.78 34. Mom -1 1.08 1.69 1.999 Even 19238 35. Pat -19 1.17 1.73 35. Pump -1 1.20 1.73 2.001 Even 58614 36. Take -19 1.20 1.65 36. Tent -1 1.20 1.68 1.999 Odd 48327 37. Tight -19 1.20 1. 67 37. Teach -1 1. 20 1.72 2.002 Even 60439 38. Toil -19 1.20 1.74 38. Tube -1 1.20 1.79 2.000 Odd 89106 39. Trap -19 1.22 1.79 39. Truck -1 1.22 1.71 2.000 Odd 05983 40. Urge -19 1.19 1.78 40. Up -1 1.20 1.61 1.999 Even 92840 20 AmbAgg Words 20 AmbAgg Control (1 Â“LÂ” word) 41. Bang -19 1.10 1.85 41. Boat -1 1.10 1.68 2.002 Odd 62793 42. Beat -19 1.16 1.76 42. Book -1 1.17 1.72 2.002 Even 17428 43. Bruise -19 1.16 1.77 43. Brave -1 1.15 1.80 2.000 Even 90837 44. Chop -19 1.10 1.79 44. Chair -1 1.13 1.80 2.001 Odd 26180 45. Curse -19 1.20 1.79 45. Cute -1 1.20 1.68 2.000 Even 83219 46. Cut -19 1.19 1.64 46. Care -1 1.19 1.75 2.001 Odd 50374 47. Grab -19 1.12 1.80 47. Green -1 1.15 1.78 2.002 Odd 38516 48. Guard -19 1.19 1.86 48. Gum -1 1.19 1.74 2.002 Even 47632 49. Hit -19 1.20 1.69 49. Horse -1 1.20 1.78 1.999 Odd 71905 50. Lash -19 1.07 1.85 50. Lake -1 1.13 1.61 2.000 Even 08459 51. Mad -19 1.01 1.87 51. Moon -1 1.00 1.81 2.001 Even 81625 52. Mob -19 1.08 1.81 52. Mud -1 1.10 1.70 2.001 Odd 24730 53. Punch -19 1.13 1.81 53. Park -1 1.13 1.70 2.000 Even 90238 54. Push -19 1.11 1.64 54. Pouch -1 1.10 1.81 2.001 Odd 17584 55. Rude -19 1.12 1.78 55. Race -1 1. 12 1.80 2.000 Odd 62103 56. Stern -19 1.01 1.88 56. Stock -1 1.05 1.72 2.000 Even 59472 57. Strike -19 1.12 1.79 57. Stream -1 1.15 1.83 2.002 Odd 32941 58. Tank -19 1.20 1.74 58. Touch -1 1.21 1.71 2.000 Even 43628 59. Tough -19 1.23 1.73 59. Tape -1 1.23 1.70 2.000 Even 71254 60. Whip -19 1.28 1.68 60. Wheat -1 1.29 1.72 1.999 Odd 08396 20 HiAgg Words 20 HiAgg Control (0 Â“LÂ” words) 61. Brawl -19 1.20 1.80 61. Broom -1 1.20 1.77 1.999 Even 91426 62. Feud -19 1.07 1.79 62. Film -1 1.08 1.80 1.999 Odd 68304 63. Fight -19 1.15 1.70 63. Feed -1 1.13 1.80 2.000 Odd 59712 64. Fist -19 1.20 1.79 64. Feel -1 1.20 1.76 2.000 Even 27830 65. Force -19 1.08 1.70 65. Food -1 1.08 1.84 1.999 Odd 14589 66. Gun -19 1.11 1.81 66. Guess -1 1.12 1.75 2.000 Even 73625 67. Harm -19 1.02 1.88 67. Hint -1 1.02 1.64 2.001 Even 30257 68. Hurt -19 1.18 1.74 68. Have -1 1.18 1.80 2.001 Odd 47152 69. Kill -19 1.11 1.77 69. Ke y -1 1.11 1.84 2.002 Even 04863 70. Rage -19 1.12 1.79 70. Room -1 1.12 1.68 2.001 Odd 83921 71. Rape -19 1.12 1.63 71. Rose -1 1.11 1.81 2.002 Odd 34785 72. Scold -19 1.00 1.81 72. Scarf -1 1.00 1.74 1.999 Even 71452 73. Scream -19 1.03 1.80 73. Scrub -1 1.03 1.83 2.000 Odd 23501
130 Appendix H (Continued) 74. Shoot -19 1.06 1.60 74. Shine -1 1.06 1.80 2.002 Even 19238 75. Shot -19 1.11 1.74 75. Share -1 1.10 1.81 1.999 Even 58614 76. Shout -19 1.00 1.75 76. Shell -1 1.01 1.81 2.001 Odd 48327 77. Shove -19 1.10 1.78 77. Shape -1 1.05 1.80 2.000 Even 60439 78. Spank -19 1.05 1.85 78. Spell -1 1.06 1.81 2.000 Odd 89106 79. Stab -19 1.01 1.88 79. Store -1 1.02 1.79 2.000 Odd 05983 80. Yell -19 1.13 1.79 80. Yawn -1 1.12 1.80 1.999 Even 92840 10 Blank 10 Blank Control (0 Â“LÂ” words) 1. Cake -1 1.20 1.70 1.999 Odd 27935 2. Dusk -1 1.12 1.72 2.001 Even 12476 3. Mint -1 1.17 1.74 2.001 Odd 71392 4. Play -1 1.19 1.77 2.000 Even 49231 5. Porch -1 1.11 1.80 2.000 Odd 92758 6. Read -1 1.14 1.76 1.999 Even 25879 7. Slurp -1 1.15 1.81 2.001 Odd 84167 8. Snore -1 1.05 1.79 2.000 Even 51683 9. Street -1 1.13 1.78 2.003 Odd 01589 10. Thumb -2 1.18 1.77 2.000 Even 39615 10 Practice Words 10 Practice Control (2 Â“LÂ” words) 1. Card -19 1.10 1.85 1. Cone -1 1.10 1.70 2.000 Odd 52146 2. Dream -19 1.22 1.78 2. Droop -1 1.22 1.70 1.999 Even 82651 3. Lack -19 1.16 1.71 3. Lean -1 1.16 1.76 2.001 Odd 32768 4. Lend -19 1.20 1.77 4. List -1 1.20 1.77 2.001 Even 60492 5. More -19 1.19 1.75 5. Mile -1 1.19 1.75 2.000 Even 46280 6. Proud -19 1.17 1.80 6. Prune -1 1.18 1.80 2.002 Odd 74325 7. Shoe -19 1.06 1.80 7. Shirt -1 1.06 1.78 2.000 Even 91847 8. Show -19 1.15 1.76 8. Sheep -1 1.15 1.75 2.001 Odd 08573 9. Wait -19 1.20 1.78 9. Well -1 1.20 1.77 2.001 Odd 15904 10. Yacht -19 1.18 1.80 10. Your -1 1.19 1.81 1.999 Even 28469 Total Unattended Words = 100 Total Attended Words = 120; Total Â“LÂ” Words = 7 Note. NonAgg = NonAggressive Words; LoAgg = Low Aggressive Words; AmbAgg = Ambiguously Aggressive Words; HiAgg = High Aggressive Words. Words obtained from Nelson, McEvoy, & SchreiberÂ’s (1999) normed database of words and their associated links.
131 Appendix I Recognition Task Instructions: Some of the following words were presented to you on the computer task that you just completed. Please check off any words that you think you heard. _____ mop [LoA Control] _____ shine [HiA Control] _____ shop [NonA] _____ bench [New] _____ film [AmbA Control] _____ grab [AmbA] _____ arm [New] _____ bend [New] _____ beach [NonA Control] _____ cute [AmbA Control] _____ shout [HiA] _____ hate [New] _____ play [BL Control] _____ kind [New] _____ pump [NonA Control] _____ shave [New] _____ touch [AmbA Control] _____ up [LoA Control] _____ give [New] _____ trick [New] _____ patch [New] _____ harsh [New] _____ crave [NonA Control] _____ broom [HiA Control] _____ march [LoA] _____ last [New] _____ fun [New] _____ gun [HiA] _____ take [LoA] _____ screen [New] _____ find [NonA] _____ rude [AmbA] _____ ram [New] _____ please [New] _____ mean [New] _____ laugh [LoA Control] _____ hard HiA Control] _____ maim [New] _____ feast [New] _____ gum [AmbA Control] Also, please list any words you think you h eard that are not on the above list. __________ __________ __________ __________ __________ __________ __________ __________ __________ __________ What do you think the purpose of this experiment was? ________________________________________________________________________ ________________________________________________________________________ Did this experiment remind you of any other ex periments that you have participated in? ______ No, it did not remind me of any other experiments. ______ Yes, and the experiment was about ____________________________________ and it reminded me because _________________________________________
132 Appendix J Calculation of Standard Ethanol Consum ption (SEC) units using part of the Comprehensive Drinker Profile (Marlatt & Miller, 1986) Steady Pattern Chart If the client drinks at l east once per week complete th e Steady Pattern Chart, then complete Q/F data summary. (If client does no t drink at least once per week, proceed to the Episodic Pattern Chart.) For each time period enter the type of beverage % alcohol amount consumed and approximate time span during which it was consumed. Morning Afternoon Evening Total for Day Mon ___________ Total SECs Monday Tues ___________ Total SECs Tuesday Wed ___________ Total SECs Wednesday Thur ___________ Total SECs Thursday Fri ___________ Total SECs Friday
133 Appendix J (Continued) Sat ___________ Total SECs Saturday Sun ___________ Total SECs Sunday Formula for calculating SECs: # oz. X % alcohol x 2 = SECs A. TOTAL SECs per week ___________ B. TOTAL drinking (nonabs tinent days) reported: ___________ C. AVERAGE SECs per drinking day (A/B): ___________ D. ESTIMATED Peak BAC for week: ___________ Quantity/Frequency Summary Data (Steady Drinking Pattern Only ) Total SECs per week from table: [ ] SECs per week Multiply by 13 weeks X 13 -----------Total SECÂ’s in past 3 months: = [ ] SECs (From Steady Pattern Chart Only )
134 Appendix J (Continued) Episodic Pattern Chart (Periodic and Combination Patte rns Only; For Steady Drinkers, skip to Pattern History.) Type and Amount of Beverage Consumed: Number of Episodes in past 3 months: Multiply Quantity (SECs per episode) by Frequency (episodes per 3 months) for each episode type: *Total SECs per episode: __________ *Hours: ______ *Peak BAC: ______ mg % X ___________ episodes per 3 mo = ________________ SECs per 3 months *Total SECs per episode: __________ *Hours: ______ *Peak BAC: ______ mg % X ___________ episodes per 3 mo = ________________ SECs per 3 months *Total SECs per episode: __________ *Hours: ______ *Peak BAC: ______ mg % X ___________ episodes per 3 mo = ________________ SECs per 3 months For COMBINATION PATTERN DRI NKERS, subtract from this total the number of SECs already accounted for in the Steady Pattern Chart and record here only SECs in excess of the steady drinking pattern. No drink should be counted on both charts. For PERIODIC DRINKERS, however, record all drinks here (since for these drinkers the Steady Pattern Chart is left blank). _______________ Grand Total SECs 3 mo. from all episodic drinking
135 Appendix J (Continued) Total Q/F Add the Total SECs from the Quantity/Frequency Summary Data section ____________ to the Grand Total SECs from the Episodic Pattern Chart + ____________ for Total Q/F SECs for past 3 mos = ____________ Pattern History What is the largest amount of alcohol that you have ever drunk in one day? Beverage Amount ___________________ _____________________ ___________________ _____________________ ___________________ _____________________ ___________________ _____________________ ___________________ _____________________ Over ____________ Hours Total SECs: ______________ Estimated Peak BAC: ____________ mg%
136 Appendix K Debriefing Statement for Phase I You were asked to respond to statements regarding personality characteristics and expectations you may hold about a variety of behaviors. Resear ch indicates that some of our personality characteristics (such as being more or less ag gressive) may be related to the expectations we hold about certain beha viors (such as drinking). Today you filled out some measures that may support or refute this research. Thank you very much for participating. If you ha ve questions or conc erns please contact Dr. James Epps at (813) 974-0388 or Mi chelle LeVasseur at (813) 974-1520.
137 Appendix L Debriefing Statement for Phase II Today you completed a computer task designed to measure the interf erence of aggression stimuli (and for some, alcohol stimuli) on atte ntion. Research suggest s that those who are higher on certain personality ch aracteristics may show more attentional interference to certain types of information (in this case, some aggression-related stimuli) that are presented on headphones (auditorily) or on a co mputer monitor (visually). During Phase I (several weeks ago), we measured your expe ctancies (with a variety of questionnaires) about how you would behave after drinking al cohol or how aggressive you consider yourself typically to be. We then asked fo r volunteers to participate in another study (Phase II) that was supposed to be unrelate d to Phase I. The time delay and the deception were both necessary so that you would not be more reactive to al cohol and aggression stimuli simply because you had recently been asked many questions about those constructs. The decisions you made on the computer task (odd vs. even numbers) provided us with a measure of reaction time. Reaction times gave us an indication of whether aggressive words interfer ed with your ability to attend to the computer task more than nonaggressive words. Thank you very much for participating. Please do not discuss this experiment with other students until they have completed the expe riment. If you have que stions or concerns, contact Dr. James Epps at (813) 974-0388 or Michelle LeVasseur at (813) 974-1520. If you would like to learn more about these topi cs please refer to the following references: 1. Bargh, J. A., Bond, R. N., Lombardi, W. J., & Tota, M. E. (1986). The additive nature of chronic and temporary so urces of construct accessibility. Journal of Personality & Social Psychology, 50 (5), 869-878. 2. Bargh, J. A., & Chartrand, T. L. (2000). The mind in the middle: A practical guide to priming and automaticity researc h. In H. T. Reis & C. M. Judd (Eds.), Handbook of research methods in social and personality psychology. (pp. 253285). New York City: Cambridge University Press. 3. Chermack, S. T. & Taylor, S. P. (1995) Alcohol and human physical aggression: Pharmacological versus expectancy effects. Journal of Studies on Alcohol, 56 (4), 449-456.
138 About the Author Michelle Edington LeVasseur received her Asso ciate in Arts at Pasco Hernando Community College in 1993. She was designated Honor Graduate with Highest Honors and was awarded a transfer scholarship. She attended the University of Sout h Florida, participated in the Department of Psychology Honors Program, and r eceived her BachelorÂ’s Degree in 1996. After working in the mental health field for three year s, she was accepted into the University of South FloridaÂ’s doctoral progra m in Clinical Psychology. Since entering the doctoral program in Augus t of 2000, Ms. LeVasseur has co-authored a published journal article and developed a manuali zed treatment intervention for teenaged girls with histories of emotional prob lems, substance abuse, and violen ce. She authored a chapter in a similar manual for women, and co-a uthored two poster presentations. In her spare time, Ms. LeVasseur likes to r econnect with significant others, including her two children, sail her 34Â’ catamaran, drink martinis, garden, and read.