Influence of Beverage on Aggression - Research Paper Example

Running Head: INFLUENCE OF BEVERAGE ON AGGRESSION Introduction Alcohol and violent crime seem to be all-too-common companions. While the literature concerning this correlation is beset with methodological problems, the pervasiveness of the phenomenon argues against it being a spurious finding [Pemanen, 1976; Pihl, 1983]. Controlled laboratory research has also demonstrated that intoxicated subjects respond more aggressively than sober subjects [see Taylor, 1983 and Pihl and Ross, 1987 for reviews]. As yet though, no clear model for the manner in which alcohol increases aggressive responding has emerged. Debate continues over the relative contribution of personal versus pharmacological factors. Two variables that have also been shown to have mediating effects are beverage type and dose. Distilled beverages produce more verbal [Takala et al., 1957; Boyatzis, 1974] and physical [Pihl et al., 1984a] aggression than do brewed beverages. A dose effect has been demonstrated for verbal [Boyatzis, 1974] and physical [Taylor and Gammon, 1975] aggression although failures to find a dose effect have also been reported [Bennetetal., 1969]. One type of research in this area that is conspicuously absent is field research. With the exception of the quasi-experimental study of Boyatzis [1974] and one observational study [Graham et al., 1980], the research to date on the alcohol-aggression link has been either retrospective crime investigations or controlled laboratory research. Although informative, the crime studies lack controls and the experimental studies lack ecological validity. In crime research, it is quite possible that since both victims and perpetrators have high blood alcohol levels, the correlation may simply define where crime takes place. Furthermore, Pemanen [1976] has suggested that intoxicated individuals are less likely to respond appropriately to a medical emergency, which would result in more homicidal fights, and they are more likely to be caught. External validity is a concern with laboratory studies, as they usually employ atypical subjects, drinking in isolation in laboratory waiting rooms despite the finding that subjects drinking alone experience an increase in aggressive mood unlike those drinking in groups [Pliner and Cappell, 1974]. Further, and typically, subjects are assessed with dependent variables involving unnatural responses. Several authors have expressed the need for more naturalistic research of this phenomenon and have warned of the danger of generalizing from laboratory results to the natural setting [Boyatzis, 1974; Taylor and Gammon, 1975]. The viability of such research in the alcohol field has been demonstrated [Cutler and Storm, 1975; Kessler and Gomberg, 1974; Plant et al., 1977; Somer, 1965]; however, only one study to date [Graham et al., 1980] has applied these observational techniques to the problem of alcohol and aggression. That study focussed on the characteristics of bars rather than individuals and demonstrated that more aggression occurred in bars with a higher percentage of intoxicated patrons but did not state whether or not it was the intoxicated patrons who were being aggressive. The present study represents an attempt to test in a natural setting the hypotheses that subjects with higher blood alcohol levels (BALs) will be more aggressive than subjects with low BALs and that subjects drinking distilled beverages will be more aggressive than subjects drinking brewed beverages. The venue for the experiment will be bars. Bars were chosen because one third of all assaults involving alcohol and resulting in police action occur in barroom settings [Gersen, 1978]. Furthermore, the barroom is one of the most accessible drinking locations and one that involves a minimal invasion of privacy since all the patrons are by choice drinking in a public place open to observation. Patrons can be observed without their awareness and are not solicited subjects. Hence the problems of volunteer subjects are avoided. Methodology Subjects The subjects were the male patrons of predominantly English Montreal bars. The sample consisted of 38 males with a mean age of 24.4 and a range of 19 to 34. The mean BAL, estimated using the formula developed by Watson et al., [1981], was .061. The mean SES of the subjects and their fathers were near the mid-point of the Blishen and McRoberts [1976] scale. All subjects were drinking in male groups of an average size of 2.63 people. Materials Surveyors carried clipboards with the McGill insignia and "Alcohol Research Group" printed on the cover. The questionnaire asked what beverage the subject was drinking at the time, what his usual beverage was, how much beer, wine, and liquor he had consumed that evening, how intoxicated he felt, the time of his first drink, and questions concerning height, weight, age, and the subject's and his father's occupation. Confederates were equipped with Aiwa TP-MIO micro cassette recorders and Aiwa tie-clip microphones. The female confederates wore their microphones as jewellery and the males wore theirs clipped under sweaters. They were dressed appropriately for the bars visited. The conversations were rerecorded in a random order on a reel to reel tape recorder using a graphic equalizer to increase the voice to noise ratio. Raters listened on stereo headphones to the tapes played back over the same tape recorder, amplified by an integrated amplifier. The interactions were also transcribed and typed for the raters to read. Procedure Bars were randomly selected from a list of currently licensed premises provided by La Regie des Permis d'Alcohol du Quebec. A bar was excluded if it had any of the following characteristics: only male patrons served only non-distilled beverages, nude dancers, a gay bar, primarily a restaurant, table service only, or a clientele predominately over 35. Bars were visited on a Thursday, Friday, or Saturday evening between 9:00 and 2:00, as these are the times when a disproportionately high level of alcohol related aggression occurs [Gersen and Preston, 1979; Wolfgang, 1958]. The research team divided into two randomly determined pairs: a pair of surveyors and a pair of confederates with a male and a female in each pair. The surveyors entered the bar first and approached the owner or shift manager for permission to conduct a survey on drinking preferences. Permission was not sought in advance in order to avoid the possibility of a change in the population due to forewarning. If permission was obtained, the surveyors began to interview patrons. If permission was not granted, they left the bar and tried the next one on the list. Of the 28 bars visited, permission was refused in 6 bars, most of which could be termed "exclusive." After 5 minutes, if the surveyors did not exit the building, the two confederates entered. They did not acknowledge the surveyors but took seats, ordered drinks, and began a conversation as though they were just another couple. The surveyors selected subjects in the following manner. They started with a random patron in the middle of the bar and counted four to the right. If this person was drinking beer or a distilled beverage (including mix), and was drinking in a same sex group, the surveyors approached and stated they were with the "McGill Alcohol Research Group." Lone drinkers were excluded as they appear to represent a distinctive group from social drinkers [Pliner and Cappell, 1974]. Surveyors then asked if the individual would be willing to answer a short questionnaire on drinking preferences. If the potential subject did not meet this criterion, they again counted four to the right until they found a customer meeting the criterion. Only 7 patrons of the 123 approached refused permission. A respondent was considered a subject if he had been drinking only beer or only distilled liquor and was under the age of 35. These criteria resulted in the rejection from the study of 78 of those interviewed. The surveyor then indicated by eye contact that the randomly selected confederate could approach the subject. The confederate waited 5 minutes, got up, and walked past the subject. Just as he/she walked past the subject, he/she looked at his/her watch, stopped and turned to the subject and asked: "Do you have the correct time?" The confederate waited for the answer and asked the subject to repeat the answer if it did not seem loud enough to be recorded. The confederate turned away after the answer and started to look for change in his/her pocket/purse. The confederate then turned back to the subject and said: "Would you have change for a dollar?" When the answer was given, the confederate then requested: "Would you have a piece of paper I could write a message on?" After the answer to this question, the confederate queried: "Maybe you could save me a phone call. Do you know where Girgaldi Street is?" (There is no such street in the Montreal area.) The presentation of the four questions followed a natural progression, requiring progressively more effort on the part of the subject. After the reply to this last question, the confederate thanked the subject for the help, went to the telephone, and then rejoined the other confederate. The confederate wrote down relevant information as soon as possible after the interaction such as any nonverbal responses that replaced a verbal answer. If the bar was large enough, the procedure was repeated in another comer of the bar out of sight of the first encounter. The team then left the bar and proceeded to the next establishment. Transcripts of the encounters were made the following day. The confederates were not blind to the experimental hypotheses. However, a check during nightly debriefing showed that they could recall the beverage type at no better than a chance level. Further, blind raters scored how provoking the confederates' questions were and found no notable differences in questioning style. Ethical Considerations Concerns for privacy and anonymity are relevant to this study, although the methodology is not without precedent [Feldman, 1968]. The following precautions were taken: All research participants were thoroughly briefed on the rules of confidentiality concerning research, names were never solicited nor recorded by the surveyor when offered by the subject, and any reference to last names was deleted from the transcripts and the tapes. Concern also exists about eliciting aggression. For this reason the study used verbal aggression as the dependent measure. Furthermore, we did not use any direct provocation in the form of insults or challenges as is commonly used in laboratory experiments. The irritant in this experiment is the number of requests the confederate makes and the increasing effort required of the respondent. These questions are innocuous in themselves and are the type of questions regularly asked in bars. In addition, the subject was never touched, insulted, challenged or territorially violated, and the only conversation recorded was that between the confederate and the subject. Dependent Measures Standardized verbal aggression scales for the kind of structured question-answer interaction used in this study are lacking. Cook and Campbell [1979] recommend rank ordering responses in this type of situation. As it would be a considerably arduous task for raters to rank 152 (38 x 4) responses, while remembering each subject's tone of voice, the responses of the subjects were formalized into an exhaustive list that comprised 18 categories, which are listed in Table I. These formalized responses were typed on 3 X 5 index cards. Ten psychologists rank-ordered these cards according to degree of hostility-aggression from least to most. The resulting rankings were analyzed using a Kendall test of concordance [Kendall, 1955]. The results revealed a high degree of concordance (w = .8444, X2 = 151.993, d.f = 18, P < .001). In addition, the rankings were analyzed using dual scaling [Nishisato, 1980]. Dual scaling is a technique that locates points relative to one another in space such that more similar points are located closer together. This involves two variables: a set of categories and a set of respondents. This resulted in a three-dimensional solution with the first dimension, an aggression dimension, accounting for 86% of the variance and the other two dimensions less than 10% combined. The results of these two tests indicate that the responses could be reliably ordered along an aggression dimension and that most raters used similar criterion in deciding that order. The final ranking based on the average ranks was used as the scoring measure. Two blind female raters were then trained in the use of this scale. The raters characterized each of the responses of the sample by listening to the tapes and reading the descriptions of the interactions. In this way, raters could rate responses in terms of content and also evaluate nonverbal cues such as voice, inflection, volume, speed, and nuance. Female raters were chosen since research shows they are more accurate in assessing nonverbal cues [Hall, 1978; Gitter et al., 1972]. It has been demonstrated that male and female raters show no bias in rating same or opposite sex subjects [Horn and Hays, 1981]. The inter-rater agreement was 80.6% and the test-retest reliability based on 15 randomly selected subjects (60 responses) was 98.3% for rater one and 90.6% for rater two. When the raters agreed on a category, that score was used as the dependent measure. When a disagreement occurred, one of the two scores was randomly selected in order to avoid any systematic biases. Control Variables The survey asked the respondents their occupation, their father's occupation, the time of their first drink, how intoxicated they felt on a 10-point scale, height, weight, age, and the number of drinks they had consumed. These last four variables were also used to calculate an approximate BAL using a revised form of the Widmark equation [Watson et al., 1981]. This equation has been found to provide a more precise estimate of BAL. In the Watson et al., [1981] study comparing actual measured BALs to estimates using the revised equation it was found that for men, 44% of the estimated BALs fell within ±5% of the measured value; 62% of the estimates fell within ±10% and .87% fell within ±15% of the measured value. The BAL score was used as a control variable in the analysis of the beverage effect. In addition, the surveyor noted the time of the survey and the size of the subject's drinking group. The time of the subject's first drink was subtracted from the time of the survey to calculate the amount of time the subject had been drinking. The surveyors and the confederates also rated the bars using a five-point scale on four physical variables: decor [Graham et al., 1980], crowdedness [Baum and Koman, 1976], noise level [Sherrod et al., 1979], and temperature [Palamarek and Rule, 1979]. The interrater reliabilities of these variables was r = .64 for decor, r = .88 for crowdedness, r = .87 for noise, and r = .71 for temperature. The occupations of subjects were converted into social economic status scores using the Blishen and McRoberts Scale [1976] for Canadian males. The usual problems with self report exist for some of the above questions and most critically, for the amount of drink the subjects had consumed. There is, however, no reason to assume that one group would systematically over or under report their consumption. Results For purposes of analyzing control variables, the subjects were divided into two groups. The high BAL group were those with a BAL over .0499 (N = 17, ten beer and seven liquor drinkers) and the low BAL group were those with a BAL under .0499 (N = 21, 9 beer and 12 liquor drinkers). The mean BAL of the high group was .097 and the low group had a mean BAL of .032. This difference is significant (F = 24.26; df = 1, 36; p < .0001). The data was then analyzed in two stages: the possible influence of control variables on the dependent measure was evaluated and aggression scores were than analyzed with an ANOVA using the BMDP program (Dixon and Brown, 1969). The control variables were analyzed three ways. First, they were correlated with the dependent variable for each of the four questions. Occupation, father's occupation, time of first drink, subject's feeling of intoxication, height, weight, age, and number of drinks were not correlated with aggression scores for each question. Only time of first drink correlated with any of the responses to the questions (question 1, r = .31; question 2. r = .41; both P < .05). However, these correlations are not strong enough to warrant an analysis of covariance. The questions themselves were not correlated with each other. Second, the control variables were analyzed using the BAL groups and sex of confederate to test for possible confounds. There were no significant confounds on the following variables: noise levels, temperature, decor, size and crowdedness of the bars, size of the drinking group, subject's and subject's father's occupations, and the time of the survey. There was an age by BAL interaction (F = 4.68; df = 1,34; P < .038) with low BAL subjects interacting with male confederates being older than high BAL subjects interacting with male confederates (P < .05). Further, the low BAL group was heavier (F = 9.1; df = 1,34; P < .005) and started drinking earlier in the evening (F = 10.41; df = 1,34; P < .0003) than the high BAL group. Third, the control variables were tested for possible confounds with the sex of confederate and the beverage type factors using a series of 2 x 2 ANOVAs. There was no difference between the two beverage groups on any of the intoxication measures. Only the two way interaction (sex of confederate by type of beverage) on the crowdedness factor approached significance (F = 4.02; df = 1,34; P < .053). The beer drinkers interacting with the male confederate were in less crowded bars than the beer drinkers interacting with female confederates. The same sex confederates were compared to see if one or the other was receiving consistently more aggressive responses. There were no significant differences. Aggression scores were analyzed using a four-way ANOVA (BAL group X sex of confederate X beverage X question) with question being the repeated measure. The effect of question was significant (F = 2.56; df = 3; P < .0597) and showed a rise in aggression scores across the four questions. There were no significant effects for BAL groups, or the sex of confederate, or the interaction. There was a significant beverage x question interaction (F = 4.59; df = 3; P < .0049). This is shown graphically in Figure 1. This figure illustrates that the distilled beverage drinkers became steadily more aggressive across questions (question 1 vs. question 4) while beer drinkers did not become more aggressive across questions. Post-hoc analysis [Scheffe, 1953] showed that beer drinkers were more aggressive on question one and distilled beverage drinkers were more aggressive on question 4 (all differences: P < .05). It should be noted that the difference between beer drinkers and distilled beverage drinkers on question 1 was entirely due to two subjects who made passes at the female confederate. If these two scores are dropped, there is no difference in the means of the two beverage groups on question 1. There was no difference between male and female confederates nor was there any interaction between beverage type and the sex of the confederate. There were no homogeneity of variance problems [Greenhouse-Geiser Epsilon = .9484; Winer, 1971]. Discussion The hypothesis that men drinking distilled beverages would be more aggressive than men drinking brewed beverages is confirmed. While the increase in aggression scores across questions could be the result of the increasing amount of effort required of the subject this possibility cannot account for the difference in aggression scores between the two beverage groups. The finding that the drinking of distilled beverages is associated with increased aggression in males appears to be robust having been demonstrated in experimental group drinking [Takala et al., 1957], in an experimental party situation [Boyatzis, 1974], under controlled laboratory conditions using a modified Buss aggression paradigm [Pihl et al., 1984a] and now in the natural setting. This study adds the ecological validity that was missing from previous work. Each of these studies was done in a different setting with different dependent measures and three of the four studies were done by different research groups. As this was a field experiment lacking the rigorous controls of the laboratory, any explanation of the results is speculative. In particular the absence of the placebo and non-drinking control is limiting. Nevertheless, a number of salient considerations are evident. First, distilled beverage drinkers were not more intoxicated than beer drinkers. There were no differences between these groups on either objective measures, the estimated BALs nor the actual number of drinks consumed, nor subjective measures, their own estimate of their inebriation. Further, there were also no differences in the amount of time subjects in the two groups had been drinking, nor the time of night they were approached. Second, social class factors also do not appear relevant to the results. Since distilled beverage drinks in general costs twice as much as beer in the bars studied, one might expect that these drinkers would have the higher social economic status. However, when the SES scores were analyzed, there were no significant differences on either the subject’s current SES or the SES of his father. Further, the physical characteristics of the subjects did not vary with beverage choice as there were no significant differences for height, weight, or age. Third, the possibility that different beverages are consumed in different types of bars, some of which are more conducive to aggression appears inappropriate to the present study. The bars in this study were rated on five characteristics: temperature, noise, decor, size and, crowdedness, factors that did not appear to affect the results. Another possible social-psychological explanation, the size of the drinking group, also failed to show any confounding effects. An alternate explanation not controlled for in this experiment is that of personality. Since subjects selected the beverage, a common personality trait might lead to both the beverage choice and aggressive responding. However, in previous studies with random assignments [Boyatzis, 1974; Pihl et al., 1984a; Takala et al., 1957], the effect is found thus arguing against a personality explanation. While personality factors can be predictive of who will become more aggressive when drinking [Boyatzis, 1975; Pihl et al., 1982, 1984b] these variables do not explain the beer versus liquor difference. The Pihl et al. [1984b] study does show that a previous negative experience with alcohol is predictive of aggression for distilled beverage drinkers but not beer drinkers. Another explanation for the beverage effect that has been suggested is that congeners, the by-products of making alcohol, are responsible for the increases in aggression [Boyatzis, 1974; Pemanen, 1976; Takala et al., 1957]. However, in terms of volume consumed, beer has a much higher congener content that distilled spirits [Greizerstein, 1981]. Differences have also been demonstrated for distilled beverages with differing amounts of congener content [Taylor and Gammon, 1975]. In all cases the beverage with the higher congener content is responsible for the lower aggression scores. This would suggest that one or more congeners might have the pharmacological property of moderating alcohol's aggression-producing property. However, such an explanation would require that a direct cause existed between alcohol and aggression. This is an untenable position as this beverage effect has been shown with placebo beverages [Pihl et al., 1984a]. Suggested is an alternative hypothesis, e.g., it is the beliefs about the beverages that produce the effect, i.e., "a real man drinks whiskey" phenomenon. This notion is further supported by a survey conducted by Lang et al. [1983]. They found that beer has more positive associations than distilled beverages and that people believe that it requires fewer drinks of straight liquor than beer to become either mildly intoxicated or drunk despite the fact that a 12 ounce beer has as much absolute alcohol as 1.25 ounce of liquor. The failure to find a dose effect in the present study was unexpected but not surprising. A dose effect has not always been found [Bennett et al., 1969] nor does alcohol inevitably lead to aggression [Graham, 1980; Pernanen, 1976; Pihl, 1983]. Taylor et al., [1976] have suggested that both alcohol and a threatening opponent must be present before alcohol shows any differential effect in aggressive responding. Taylor and Gammon [1975] further showed that only when subjects are both intoxicated and threatened do they use potentially hazardous levels of shock. Mayfield [1975] and Virkkunen [1974] reported that in one half of the alcohol-related homicides in their respective studies, the victim was the instigator or had acted in a manner the perpetrator considered threatening. In this study, for ethical reasons, the confederates were purposefully not threatening. Furthermore, the lack of a control group does not permit a conclusion on whether there was an alcohol effect per se. The formation of legitimate control groups is logistically impossible in a bar situation. Patrons could not be given placebo drinks to control for a placebo effect and since people generally frequent bars to drink it is difficult to find non-drinking controls. It is possible that the two groups are too close in BAL to see reliable differences. As well, BAL estimates may not be entirely accurate for several reasons. First, the equation used is based on the assumption that the subject has been fasting, a condition that cannot be controlled for in the naturalistic setting. Second, the equation does not take into account the time span over which alcoholic beverages were consumed and the affect this may have on BAL. Finally, this study demonstrates along with the work of Cutler and Storm [1975], Graham et al. [1980], Kessler and Gomberg [1974], Plant et al. [1977], and Somer' [1965] that naturalistic research on drinking and related behaviours is both viable and valuable. An advantage of the present procedure is the incorporation of a more experimental type of naturalistic research. This type of combination of laboratory and field research could add much to the knowledge of the alcohol-aggression link. References Baum A, Koman S (1976): Differential response to anticipated crowding: The effects of social and spatial density. Journal of Personality and Social Psychology, 34:525-536. Bennett RM, Buss AH, Carpenter JA (1969): Alcohol and human physical aggression. Quarterly Journal of Studies on Alcohol 30:870-876. Blishen BR, McRoberts HA (1976): A revised socioeconomic index for occupations in Canada. Canadian Review of Sociology and Anthropology 13:71-79. Boyatzis RE (1974): The effects of alcohol consumption on the aggressive behaviour of men. Quarterly Journal of Studies on Alcohol 35:959-972. Boyatzis RE (1975): The predisposition toward alcohol-related interpersonal aggression in men. Journal of Studies on Alcohol 36:1196-1207. Cook TD, Campbell DT (1979): "Quasi Experimentation: Design and Analysis Issues for Field Settings." Chicago: Rand McNaily. Cutler RE, Storm T (1975): Observational study of alcohol consumption in the natural setting. Journal of Studies in Alcohol 36:1173-1183. Dixon WJ, Brown MB (eds) (1969): "BMDP:Biomedical Computer Programs: (pseries). Los Angeles: University of California Press. Feldman RE (1968): Responses to compatriot and foreigner who seek assistance. Journal of Personality and Social Psychology 10:202-214. Gerson LW (1978): Alcohol related acts of violence: Who was driving and where the acts occurred. Journal of Studies on Alcohol 39(7): 1294-1296. Gersen LW, Preston L (1979): Alcohol consumption and the incidence of violent crime. Journal of Studies on Alcohol 40:307-311. Gitter AG, Kozel NJ, Mostofsky DJ (1972): Perception of emotion: The role of race, sex and presentation mode. The Journal of Social Psychology 88:213-222. Graham K, LaRocque L, Yetman R, Ross TJ, Guistra E (1980): Aggression and barroom environments. Journal of Studies on Alcohol 41:277-292. Greizerstein HB (1981): Congener contents of alcoholic beverages. Journal of Studies on Alcohol 42:1030-1037. Hall JA (1978): Gender effects in decoding verbal cues. Psychological Bulletin 85:845-847. Hom WF, Hayes SN (1981): An investsigation of sex bias in behavioural observations and ratings. Behavioural Assessment 3:173-183. Kendall MG (1955): "Rank Correlation Methods" (2 ED). London: Charles Griffin and Company. Kessler M, Gomberg C (1974): Observations of barroom drinking: methodology and preliminary results. Quarterly Journal of Studies on Alcohol 35:1392-1396. Lang, AR, Kaas L, Bames P (1983): The beverage type stereotype: An unexplored determinant of the effects on alcohol consumption. Bulletin of the Society of Psychologists in Additive Behaviors 2:46-49. Mayfield D (1975): Alcoholism, alcohol, intoxication & assault behaviour. Diseases of the Nervous System 37:288-291. Nishisato S (1980): "Analysis of Categorical Data: Dual Scaling and its Applications." Toronto: University of Toronto Press. Palamarek DL, Rule BG (1979): The effects of ambient temperature and insult on the motivation to retaliate or escape. Motivation and Emotion 3:83-92. Pemanen K (1976): Alcohol and crimes of violence. In B. Kissin and H. Begleiter (eds): "The Biology of Alcoholism." New York: Plenum Press, Vol. 4. Pihl RO (1983): Alcohol and aggression: A psychological perspective. In E. Gottheil (ed): "Alcohol, Drug Abuse and Aggression" Springfield, HI: C. Thomas, pp 292-313. Pihl RO, Zacchia C, Zeichner A (1982): Predicting levels of aggression after alcohol intake in male social drinkers: A preliminary investigation. Journal of Studies on Alcohol 43:599-602. Pihl RO, Smith M, Farreil B (1984a): Alcohol and aggression: A comparison of brewed and distilled beverages. Journal of Studies on Alcohol 45:278-282. Pihl RO, Smith M, Farrell B (1984b): Individual characteristics of aggressive beer and distilled beverage drinkers. The International Journal of Addictions 19:689-696. Pihl RO, Ross D (1987): Research on alcohol related aggression: A review and implications for understanding aggression. Drugs and Society 1:105-126. Plant MA, Kreitman N, Miller, TI, Duffy I (1977): Observing public drinking. Journal of Studies on Alcohol 38:867-880. Pliner P, Cappell H (1974): Modification of affective consequences alcohol: A comparison of social and solitary drinking. Journal of Abnormal Psychology 83:418-425. Scheffe H (1953):' A method for judging all contrasts in the analysis of variance. Biometrika 40:87-104. Sherrod DR, Moore BS, Underwood B (1979): Environmental noise, perceived control and aggression. Journal of Social Psychology 109:245-252. Somer R (1965): The isolated drinker in the Edmonton beer parlor. Quarterly Journal of Studies on Alcohol 26:95-110. Takala M, Pihkanen TA, Markkanen T (1957): The effects of distilled and brewed beverages: A physiological, neurological and psychological study. Helsinki: The Finnish Foundation for Alcohol Studies. Taylor SP, Gammon CB (1975): Effects of type and dose of alcohol on human physical aggression. Journal of Personality and Social Psychology 32:169-175. Taylor SP, Gammon CB, Capasso DR (1976): Aggression as a function of alcohol and threat. Journal of Personality and Social Psychology 34:938-941. Taylor SP (1983): Alcohol and human physical aggression. In E. Gottheil (ed): "Alcohol Drug Abuse and Aggression." C. Thomas: Springfield, III. pp 280-291. Virkkunen M (1974): Alcohol as a factor precipitating aggression and conflict behaviour leading to homicide. British Journal of Addiction 67:149-154. Watson PE, Watson I, Batt RD (1981): Prediction of blood alcohol concentrations in human subjects. Journal of Studies on Alcohol 42:547-556. Winer BJ (1971): Statistical principles in experimental design (2 ed.). New York, McGraw Hill. Wolfgang ME: (1958): "Patterns in criminal homicide." London: Oxford University Press. Read More