PATTERNS OF ALCOHOL DRINKING IN A POPULATION OF YOUNG SOCIAL DRINKERS: A COMPARISON OF QUESTIONNAIRE AND DIARY MEASURES

J. M. Townshend and T. Duka,*

Laboratory of Experimental Psychology, University of Sussex, Falmer, Brighton BN1 9QG, UK

Received 31 May 2001; in revised form 31 July 2001; accepted 27 August 2001


    ABSTRACT
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ACKNOWLEDGEMENTS
 REFERENCES
 
Studies of alcohol use often depend on self-reported alcohol intake measured by quantity/frequency questionnaires. Previous research has shown that alcohol consumption may be underestimated by this type of retrospective questionnaire. The primary aim of this study was to compare the accuracy of an Alcohol Use Questionnaire (AUQ) with a 4-week diary account. A further aim was to explore patterns of drinking in young social drinkers, with particular attention to binge drinking, which has been suggested as a factor in increasing the risk of alcohol dependency. University students completed the AUQ in the laboratory. They were then asked to keep a record of their alcohol, nicotine and caffeine consumption over a 4-week period (diary). The questionnaire and the diaries were compared on factors of alcohol intake (units per week) and patterns of drinking behaviour (speed of drinking, number of times being drunk and percentage of times getting drunk when drinking). The two measures (AUQ and diary) were highly correlated on alcohol consumption and the other questions relating to drinking behaviour. However, differences were found between the two measures on alcohol intake, speed of drinking (drinks per hour) and number of times being drunk. Alcohol consumption was underestimated by ~12% on the questionnaire, and, when the accuracy of estimation of drinking habits was examined, it was found that high drinkers tended to underestimate their drinking behaviour, whereas lower drinkers tended to overestimate. The results suggest that the AUQ can be used with a reasonable degree of confidence, bearing in mind the tendency for high drinkers to underestimate consumption and drinking behaviour. Relationships between ‘binge scores’, beverage specificity and alcohol consumption support the idea that the criteria for binge drinkers should be based on patterns of drinking rather than alcohol consumption.


    INTRODUCTION
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ACKNOWLEDGEMENTS
 REFERENCES
 
Studies of alcohol use often depend on self-reported alcohol intake, measured by quantity/frequency questionnaires. This method of measurement consists of a question to assess average frequency of drinking combined with a question to assess average quantity. Additionally, this type of questionnaire may ask for beverage-specific information (i.e. quantities of beer, wine and spirits separately), or for total number of alcoholic drinks. In a review paper based on twelve studies using different methods of measuring alcohol consumption, it was reported that quantity/frequency questionnaires and prospective diary studies completed on a daily basis give intake estimates 20% higher than retrospective diaries which incorporate a ‘time-line follow back method’ (Feunekes et al., 1999Go). Beverage specificity was also the only significant predictor of alcohol intake and resulted in a 19% higher reported intake. A quantity/frequency questionnaire which also incorporates beverage specificity is the Alcohol Use Questionnaire (AUQ; Mehrabian and Russell, 1978Go). The widespread use of this questionnaire in alcohol research leads directly to the question of accuracy in self-reported accounts of alcohol consumption and drinking patterns. The primary aim of this study was to examine the validity of a commonly used, modified version of the AUQ in terms of estimation of alcohol consumption based on retrospective report, in comparison with a 4-week diary account.

Patterns of drinking among young healthy volunteers could provide important insights into the development of alcohol dependence. In particular, Stephens (1995) has suggested that repeated withdrawal from alcohol could increase the degree of alcohol dependence. Consequently, periods of abstinence from alcohol interspersed with bouts of heavy drinking (e.g. ‘bingeing’) in social drinkers could increase the risk for alcohol dependence. Binge drinking in a student population has been shown to predict the frequency with which alcohol-related problems were experienced (Wechsler et al., 1994Go), and additionally Hunt (1993) has suggested that binge drinkers may be more at risk of developing brain damage. It has been proposed that the number of drinks in a row differentiates bingers from non-bingers (Wechsler and Austin, 1998Go), and, although this may be the case, it also means that bingers and non-bingers will almost certainly consume different quantities of alcohol. Consequently any differences between bingers and non-bingers could be due to absolute alcohol intake, rather than patterns of drinking. In a recent study (R. Weissenborn and T. Duka, unpublished data), a group of social drinkers was divided into bingers and non-bingers on the basis of drinking behaviour. The two groups in the study consumed the same amount of alcohol per week but their drinking patterns were very different. The results of this study provided evidence in support of Hunt's (1993) proposal that binge drinkers may be more at risk of developing brain damage when bingers were found to perform worse on two frontal lobe tasks, compared to non-bingers. Studies such as these have highlighted the need to explore patterns of drinking as well as quantities of alcohol consumed.

The AUQ asks specific questions about drinking behaviour. Speed of drinking or number of drinks per hour is one such question and it has not previously been clear whether subjects' responses to this question refer to their maximum number of drinks in 1 h, their usual number of drinks, or a working average. The two other questions that refer to drinking patterns ask the frequency of getting drunk and the percentage of times that drinking leads to drunkenness. A secondary aim of this diary study was to examine the drinking patterns of young social drinkers and to compare the validity of retrospective report of drinking behaviour with a daily, diary measure.


    MATERIALS AND METHODS
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ACKNOWLEDGEMENTS
 REFERENCES
 
Subjects
Sixty-one university students replied to the original advertisement for social drinkers to complete a 4-week diary study. From the 61 who collected the diary sheets, 56 (42 female and 14 male; mean age 20.54 years, range 18–28) completed the diary sheets for the full 4-week period. One subject was discarded, due to the very large discrepancy between her AUQ score (197) and her diary score (67). This discrepancy was not due to the reported alcohol intake that was similar for both measures, but for the AUQ drinking behaviour questions 11 and 12 described below. Although, of course, the purpose of the study was to compare the two measures, it was felt that this subject may have misunderstood or misinterpreted the questions on the AUQ. Of the remaining 55, 14 were smokers and 41 non-smokers.

Materials
Diary sheets (Appendix IGo) were created to provide the same information as the questions on the AUQ (Appendix IIGo): average weekly beverage-specific alcohol consumption (wine, AUQ questions 1–3; beer, AUQ questions 4–6; spirits, AUQ questions 7–9); speed of drinking (number of drinks per hour, AUQ question 10); number of times intoxicated by alcohol (AUQ question 11: in the past 6 months; diary: during the 4-week period); and percentage of times getting drunk when going out drinking (question 12). The sheets were divided into hourly slots, from 12 noon to 12 midnight, plus a section for before 12 noon and for after midnight. Subjects were asked to enter the type of drink and the quantity consumed in the space provided for each hour. They were instructed to record all alcoholic drinks, giving the type and the brand. Within the hourly slots, there were also sections to record nicotine and caffeine intake in order to explore relationships between alcohol use and other legal drug use. One sheet covered a 7-day period.



View larger version (73K):
[in this window]
[in a new window]
 
APPENDIX A

 


View larger version (39K):
[in this window]
[in a new window]
 
APPENDIX B

 
Procedure
The time period chosen for the study was mid-spring term, an exam-free term falling between the festival periods of Christmas and Easter. It was felt that this time would represent a fairly typical or average drinking period that might correspond quite closely with what the subjects thought of as their usual drinking behaviour. It should be noted that the AUQ asks for habitual use of alcohol, and suggests that participants use the previous 6-month period as a guide, rather than trying to remember the precise quantities. Birthday celebrations that fell within the diary period were noted and later checked for excessive, atypical alcohol intake, in which case the day was excluded from the diary analysis.

Subjects were first asked to come to the laboratory, where they completed the AUQ. They were then given four weekly diary sheets and both verbal and written instructions for completing the sheets correctly. They were asked to record all alcohol, caffeine and nicotine use on a daily basis and to return the sheets each week. It was stressed that the diaries should be completed at the end of each day, or, if this was not possible, then directly the following morning. Additionally, they were asked to record at the end of each day whether they had considered themselves to be drunk. Being drunk was defined as the inability to speak clearly, loss of co-ordination or nausea. They were paid a small sum for each returned sheet plus a bonus for returning all four sheets. When the final sheet was returned the subjects were asked whether they felt that monitoring their alcohol intake had had an effect on their consumption.

Scoring
Intake. The AUQ gives a score based on number of drinks per week (drinks being pints of beer, glasses of wine or single measures of spirit), speed of drinking (number of drinks per hour), number of times being drunk in the previous 6 months, and percentage of times getting drunk when going out drinking. The equation for the AUQ score is as follows:

The diaries were scored in the same way, number of beverage-specific drinks per week averaged over the 4 weeks. Weekly alcohol intake was calculated by the number of standard drinks, in order to reflect the scoring of the AUQ, and additionally by the number of units of alcohol per week calculated using a ‘unit calculator’ provided by the Portman group, and based on information found in The Sensible Drinking Report provided by the Department of Health (1995). The AUQ responses to weekly beverage-specific alcohol intake questions were also converted into alcohol units using the same information.

Patterns of drinking. Drinking speed was also calculated in several different ways, as it was not known exactly what measure subjects were referring to on this particular question of the AUQ. One measure was the maximum number of drinks in any 1 h between 12 noon and 12 midnight. The second measure was the modal number of drinks for each hour of drinking; and thirdly, the average of the maximum and modal measures over all 4 weeks. Additionally, in accordance with the five/four measure of binge drinking behaviour (Wechsler and Austin, 1998Go), ‘drinks in a row’ were calculated by counting the maximum number of drinks in a row with a gap of not more than 2 h between drinks. Number of times drunk in the diary was calculated by counting the number of times that subjects claimed to be drunk during the 4 weeks and multiplying this by 6 to estimate a number for the 6 months. This measure therefore may not be equivalent to the measure in the AUQ which represents an average number of times being drunk during the last 6 months.

‘Binge score’. To further assess the relationship between drinking patterns and alcohol intake, a ‘binge score’ was calculated for all subjects on the basis of the information given in items 10, 11 and 12 of the AUQ. The score is calculated in the same way as the AUQ score (Mehrabian and Russell, 1978Go) but without the items 1–9 that refer to quantity and type of alcohol intake. This score gives a picture of the drinking patterns of the participants, rather than just a measure of alcohol intake. Subjects who have a high ‘binge score’ and drink frequently but irregularly may have a similar intake of alcohol to those with a lower ‘binge score’ who drink on a regular basis.

Caffeine and nicotine intake
Scoring of caffeine intake was based on the amounts given in James (1991) and the total amount over the 4-week period averaged for an individual weekly caffeine intake. Nicotine intake was scored in a similar way, but not analysed further as there were relatively few smokers in the sample (13/55).

Statistical methods
Differences between the AUQ and diaries were examined with paired-sample t-tests and similarities of ranking with Pearson correlation coefficients. In addition, rank order Spearman analysis of the relationship was performed to ensure appropriate analysis in case the data were not normally distributed. Data from the AUQ and diary for the number of units drunk was log-transformed for Fig. 1Go to reduce the variance. In addition, relationships between ‘binge scores’, alcohol consumption, the five/four measure (Wechsler and Austin, 1998Go), beverage specificity (wine, beer and spirits) and caffeine intake, were also explored with Pearson correlation coefficients.



View larger version (17K):
[in this window]
[in a new window]
 
Fig. 1. Estimated alcohol intake from the Alcohol Use Questionnaire (AUQ) correlated with diary recorded alcohol intake.

r = 0.975, P < 0.01.

 

    RESULTS
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ACKNOWLEDGEMENTS
 REFERENCES
 
Comparisons between AUQ and diary (alcohol consumption and drinking patterns)
Table 1Go shows the means (±SD) for the AUQ and diary scores on units of alcohol, the three questions concerning patterns of drinking and the overall score from each method. The best correlation for speed of drinking (question 10) from information given on the AUQ and information from the diary was attained by the average measure (r = 0.592, P < 0.01). Consequently, this measure was used in calculating the total diary score. Using a paired-sample t-test, differences were found between the two measures on alcohol units, with alcohol intake higher from the diary information than from the AUQ, and on the number of times the subjects had been drunk in the previous 6 months at the significance level of P < 0.01. Differences with P < 0.05 were found on speed of drinking and the total scores.


View this table:
[in this window]
[in a new window]
 
Table 1. Means ± SD of Alcohol Use Questionnaire (AUQ) and diary scores
 
A Pearson correlation between estimated alcohol intake and diary recorded alcohol intake measured in alcohol units was highly significant, showing a strong correlation (Fig. 1Go).

Significant differences were found between the calculated total scores on the two measures when all the questions, including those regarding drinking behaviour, were taken into account, based on the formula from the AUQ. In order to indicate the accuracy of subject's retrospective estimation across all aspects of the questionnaire, the differences between the two scores were plotted (Fig. 2Go). It was found that higher drinkers tended to underestimate their drinking behaviour, whereas lower drinkers tended to overestimate.



View larger version (18K):
[in this window]
[in a new window]
 
Fig. 2. Differences between the calculated total scores on the two measures [Alcohol Use Questionnaire (AUQ) and diary] illustrating the accuracy of subject's retrospective estimation.

 
Binge scores
To establish the relationship between the amount of alcohol consumed per week with the ‘binge scores’ and ‘drinks in a row’ (five/four) measure, Pearson correlation coefficients were calculated. A highly positive correlation was found between number of drinks in a row and amount of alcohol consumed, but no significant correlation between ‘binge scores’ and alcohol intake (Table 2Go).


View this table:
[in this window]
[in a new window]
 
Table 2. Pearson correlation coefficients between the ‘binge score’ and drinks in a row with units of alcohol consumed
 
In addition, the beverage of choice for binge drinking was examined in a further Pearson correlation on the two factors ‘binge scores’ and ‘drinks in a row’, with information about specific beverages taken from the diary accounts (Table 3Go). It can be seen that consumption of wine was significantly positively correlated with both the ‘binge score’ and the ‘drinks in a row’. Additionally, spirits were also significantly correlated with both measures but more highly with ‘drinks in a row’, whereas the amount of beer drunk did not show any relationship with either measure of binge drinking.


View this table:
[in this window]
[in a new window]
 
Table 3. Pearson correlation coefficients between the ‘binge score’ and drinks in a row with specific beverages (wine, beer and spirits)
 
Caffeine
Caffeine intake was not significantly correlated with alcohol consumption (r = 0.124, ns) or with either the ‘binge score’ (r = 0.034, ns) or drinks in a row (r = –0.096, ns). A post hoc analysis of the relationship of caffeine with specific beverages produced a significantly positive correlation of caffeine and wine (r = 0.286, P < 0.05). However, a Bonferroni correction for multiple comparisons increased the P value to above the significance level, making the result less reliable. The means (±SD) and ranges for all the items used in the correlations, are reported in Table 4Go.


View this table:
[in this window]
[in a new window]
 
Table 4. Means ± SD and ranges of ‘binge scores’, maximum number of drinks in a row and specific beverages
 
Finally, the subjects had been asked whether they felt that monitoring their alcohol intake had had an effect on their consumption. Some were surprised at their quantity of alcohol consumption but none of them felt that they had changed their drinking habits as a result. In order to investigate this possibility further, a within-subjects, repeated measures analysis was carried out on the number of units per week for all subjects, to compare drinking across the 4 weeks of the diary period. This proved to be non-significant [F(3) = 0.097; P = 0.961] indicating that there had been no systematic change in drinking behaviour.

Rank-order Spearman correlations for all the relationships examined showed the same results as the Pearson correlation, therefore they are not reported.


    DISCUSSION
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ACKNOWLEDGEMENTS
 REFERENCES
 
The present study set out to compare the AUQ with a diary record and to examine patterns of drinking in social drinkers. The social drinkers in this study were drawn from a student population with a higher number of females than males represented in the sample. Therefore a generalization of these findings to the general population may have its limitations. The high correlation between the AUQ and the diary record on measures of alcohol intake suggest that the ranking of individuals by alcohol intake can be reliably measured using the AUQ. However, the measure of absolute intake of alcohol appears to be less reliable, and was generally underestimated by the AUQ. This is in accordance with previous studies (e.g. Poikolainen and Karkkainen, 1983Go; Lemmens et al., 1988Go), which also found that questionnaire measures were underestimated in comparison with diary accounts.

When other drinking habits are also considered (speed of drinking, number of times getting drunk and percentage of time drinking to getting drunk) as in the total score, these data also suggest that there may be an element of ‘wishful thinking’ in young social drinkers at either end of the scale, in as much as the low drinkers tend to overestimate, and the high drinkers tend to underestimate, their drinking habits. These results confirm the findings of a previous study using shorter time periods, in which it was shown that at the lowest intake levels, there was a tendency to over-report consumption, and that under-reporting increased with consumption (Lemmens et al., 1988Go). This could be an important factor to consider when using the AUQ as a measure of alcohol use, as it is the high drinkers who may be at particular risk from alcohol-related diseases.

A key problem with the design of the present study is that the period that the AUQ covers precedes the diary period. However, previous studies have used a similar design (e.g. Poikolainen and Karkkainen, 1983Go; Lemmens et al., 1988Go; O'Callaghan and Callan, 1992Go). Another problem to consider is that the AUQ asks for information on the quantity of alcohol intake on a weekly basis, whereas, for number of times being drunk (question 11), information is requested from the previous 6 months. For this reason, the number of times being drunk during the 4-week diary period needs to be extrapolated in order to match information from the AUQ. Table 1Go shows a large difference on question 11 between the estimated number of times drunk and the diary-recorded number, which may more accurately reflect a problem with the data analysis, rather than a genuine difference.

When patterns of binge drinking were examined, a highly positive correlation was found between number of drinks in a row and amount of alcohol consumed. This suggests, not unexpectedly, that, as the number of drinks in a row increases, the total amount of alcohol consumed also increases. However, the total amount of alcohol drunk may not provide an accurate description of binge drinking, as it is proposed that patterns of binge drinking behaviour, rather than quantity of alcohol consumed, increases the degree of alcohol dependence. This idea is derived from a withdrawal sensitization theory of addiction (Stephens, 1995Go). Thus, the ‘binge score’ as presented here, found to be unrelated to the amount of alcohol drunk, may be a better predictor of future dependency problems. R. Weissenborn and T. Duka (unpublished data), using the same ‘binge score’, have shown differences between bingers and non-bingers in ‘novelty seeking’ and ‘impulsivity’ on the temperament and character inventory (TCI; Cloninger et al., 1994Go) and on the CANTAB (Cambridge Neuropsychological Test Automated Battery) tasks of spatial working memory and pattern recognition.

As the correlations revealed, wine would appear to be the beverage of choice for binge drinkers among students, whether this is measured by drinks in a row or a ‘binge score’. This may have been due to the higher proportion of females in the population sample, although, overall, there was less wine drunk than beer and spirits. Spirits made up the majority of alcohol consumed, and the significant correlation between spirits and drinks in a row provides additional evidence that the five/four measure refers mainly to quantity of alcohol consumed, rather than to drinking patterns. Wine drinking among binge drinkers would seem to be an obvious choice, as it enables a high alcohol intake without the volume of beer and is relatively inexpensive in comparison to spirits.

The close relationship of estimated alcohol intake as measured by a quantity/frequency, beverage-specific questionnaire compared to a 4-week diary study, suggests that the AUQ can be used with a certain degree of confidence. The additional information from the AUQ regarding drinking behaviour was also closely related to the diary information, although not to the same extent as quantity consumed. The efficacy of using retrospective questionnaires compared with longer-term diary methods, means that abandoning retrospectively collected data entirely is not feasible. However, human memory is a selective, constructive process and it is important that researchers who use self-reported questionnaires are aware of the limitations in the data produced. ‘Because of memory distortions, self-reports are biased narratives, rather than incomplete, if otherwise accurate, evocations of past events’ (Hammersley, 1994Go). To illustrate this point, it is interesting to compare questionnaire versus diary reports of sexual activity. Unlike alcohol consumption, sexual activity was found to be consistently over-reported on questionnaires compared with diaries, in a group of homosexual men (Coxon, 1999Go) and in adolescents (Leigh et al., 1998Go). However, it must be pointed out that even diary records cannot be relied on as precise accounts of behaviour and may also be, to a lesser extent, subject to biases and inaccuracies in memory.

This current study has also added to the knowledge that quantity of alcohol consumed appears to be consistently underestimated by a retrospective questionnaire method and that this may be the case particularly for high drinkers. In addition, the study has shown that binge drinking, as measured by the five/four measure (Wechsler and Austin, 1998Go), is more closely related to quantity of alcohol consumed than to patterns of drinking. The use of ‘binge scores’ to identify bingers and non-bingers is important, as bingeing behaviour may be a key factor for increased dependence and cognitive impairment (Hunt, 1993Go; Stephens, 1995Go; R. Weissenborn and T. Duka, unpublished data). Consequently, a method of calculating a binger score based on patterns of drinking as measured by the AUQ would appear to have more validity. Further research will be needed to investigate whether this method of distinguishing bingers and non-bingers can differentiate social drinkers on dependence measures and cognitive tasks.


    ACKNOWLEDGEMENTS
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ACKNOWLEDGEMENTS
 REFERENCES
 
This work was supported by MRC Grant No. G9806260.


    FOOTNOTES
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ACKNOWLEDGEMENTS
 REFERENCES
 
* Author to whom correspondence should be addressed. Back


    REFERENCES
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ACKNOWLEDGEMENTS
 REFERENCES
 
Cloninger, C. R., Przybeck, T. R., Svrakic, D. M. and Wetzel, R. D. (1994) The Temperament and Character Inventory (TCI): A Guide to its Development and Use. Center for Psychobiology of Personality, St Louis, MO.

Coxon, A. P. M. (1999) Parallel accounts? Discrepancies between self-report (diary) and recall (questionnaire) measures of the same sexual behaviour. AIDS Care — Psychological and Socio-Medical Aspects of AIDS/HIV 11, 221–234.

Department of Health (1995) The Sensible Drinking Report. The Results of an Interdepartmental Working Group. HMSO, London.

Feunekes, I. J., van't Veer, P., van Staveren, W. A. and Kok, F. J. (1999) Alcohol intake assessment: the sober facts. American Journal of Epidemiology 150, 105–112.[Abstract]

Hammersley, R. (1994) A digest of memory phenomena for addiction research. Addiction 89, 283–293.[ISI][Medline]

Hunt, W. A. (1993) Are binge drinkers more at risk of developing brain damage? Alcohol 10, 559–561.[ISI][Medline]

James, J. E. (1991) Caffeine and Health. Academic Press, London.

Leigh, B. C., Gillmore, M. R. and Morrison, D. M. (1998) Comparison of diary and retrospective measures for recording alcohol consumption and sexual activity. Journal of Clinical Epidemiology 51, 119–127.[ISI][Medline]

Lemmens, P., Knibbe, R. A. and Tan, F. (1988) Weekly recall and diary estimates of alcohol consumption in a general population survey. Journal of Studies on Alcohol 49, 131–135.[ISI][Medline]

Mehrabian, A. and Russell, J. A. (1978) A questionnaire measure of habitual alcohol use. Psychological Reports 43, 803–806.[ISI][Medline]

O'Callaghan, F. V. and Callan, V. J. (1992) Young adult drinking behaviour: a comparison of diary and quantity–frequency measures. British Journal of Addiction 87, 723–732.[ISI][Medline]

Poikolainen, K. and Karkkainen, P. (1983) Diary gives more accurate information about alcohol consumption than questionnaire. Drug and Alcohol Dependence 11, 209–216.[ISI][Medline]

Stephens, D. N. (1995) A glutamatergic hypothesis of drug dependence — extrapolations from benzodiazepine receptor ligands. Behavioural Pharmacology 6, 425–446.[ISI][Medline]

Wechsler, H. and Austin, B. (1998) Binge drinking: the five/four measure. Journal of Studies on Alcohol 59, 123–125.[ISI]

Wechsler, H., Davenport, A., Dowdall, G., Moeykens, B. and Castillo, S. (1994) Health and behavioural consequences of binge drinking in college: a national survey of students at 140 campuses. Journal of the American Medical Association 272, 1672–1677.[Abstract]