- Email: [email protected]
Excessive drinking situations in German alcoholics: Replication of a three-factor model used for North Americans
DEPENDENCE ELSEVIER
Drug and Alcohol Dependence 41 (1996) 75-79
Excessive drinking situations in German alcoholics: replication of a three-factor model used for North Americans A. Victorio-Estradaa,b,*, R.F. Mucha”,“, E.R. Stephand “Institute of Psychology, University of’ Cologne, Herbert-Lewin-Strasse 2, D-50931 K&z, Germany bCenter of Psychological Service, National Autonomous University of Mexico, Mexico City, Mexico ‘Institute for Medical Psychology, University qf Tiibingen, D- 72074 Tiibingen, Germany ‘Institute qf Businessand Social Psychology, University of Cologne, Herbert-Lewin-Strasse, D-50931 Kiiln, Germany Received 8 June 1995; accepted 7 March 1996
Abstract German alcoholics were examined for a profile of drinking situations which was found recently in North American alcoholics. A mixed sample of alcoholics (N = 336) receiving in-patient rehabilitation treatment for alcoholism was administered a validated translation of the English version of Annis’s Inventory ofDrinking
Situations (IDS). Analysis of the 100 items in the questionnaire,
using principal component methods and varimax rotation, revealed the same three factorial scales seen in the previous study in North American
alcoholics (Cannon et al. (1990) Addictive Behaviour 15, 265-269).
Drinking
was most likely to take place in
situations (1) of negative states, (2) of positive states, and (3) where personal control was tested. Although there are large differences in the level and social acceptability of drinking between Central Europe and English-speaking data confirm the cross-cultural validity of basic determinants of drinking.
countries, the present
Keywords: Drinking situations; Cross-cultural; Alcoholics; Questionnaire; Motivation
1. Introduction
in which alcoholics show excessive drinking may be important for understanding the maintenance of drinking and relapse in abstinent individuals (Annis et al., 1987). However, there is still no clear idea whether the drinking situations reported by alcoholics are representative of identifiable, fundamental processes of addiction or simply reflect social/cultural organisation in the environment of the alcoholic. Recently, a population of American alcoholics were asked about the situations in which they drank excessively using an inventory of drinking situations (Annis et al., 1987). Using principle component analysis with varimax rotation of the data, it was found that there were three situations that give rise to drinking behaviour: negative states, positive states and testing of personal control (Cannon et al., 1990). The situations
* Corresponding author, Pacific0 159-201, Los Reyes, Coyoacan, 04330 Mexico D.F., Mexico.
However, comparison of the findings of Cannon et al. (1990) to other data suggests that drinking situations
may be actually specific to the population tested. In a study of proximal determinants of relapse in Germans, testing personal control, unpleasant feelings, work problems and conflict with a partner were identified as reasons (Pfeiffer et al., 1988); there had been, however, no indication that Germans drink excessively in positive states. Other studies failed to find any proximal relapse determinants in German drinkers (Rist et al., 1989). Prominent differences exist between American and other English-speaking and German cultures, regarding frequency and quantity of alcohol drinking, levels of abstinence, acceptability and availability of alcohol (Cockerham et al., 1989); for example, the national per
capita consumption of alcohol in Germany is at least 50% higher than in countries like Canada, England, Australia and the United States (Junge, 1993). Therefore, looking at the data reported in the literature, it may not seem surprising that the drinking situations reported by German alcoholics should be different from those of North Americans. This would suggest socio-
0376-8716/96/$15.00 0 1996 Elsevier Science Ireland Ltd. All rights reserved
PII SO376-8716(96)01241-O
16
A. Vicrorio-Estrada
et al. II Drug und Alcohol Dependence 41 (1996) 75 79
cultural influences in the fundamental processesof alcoholism. A major problem with comparisons across cultural groups is that parallel tests are not usually carried out in the different populations. Indeed, the findings of Cannon et al. (1990) have not been tested in a wide range of alcoholics so at the present time it is not possible to draw any firm conclusions concerning differences in the drinking situations reported by different populations of alcoholics. It is, for example, not clear how drinking situations examined by Cannon et al. (1990) correspond to the proximal determinants of relapse studied by Pfeiffer et al. (1988) and Rist et al. (1989). Moreover, it was only recently recognised in the English-speaking literature that alcohol and drugs may be consumed for their positive reinforcing effects rather than due to their negative reinforcing effects (Stewart et al., 1984; Robinson and Berridge, 1993). Therefore, failure to recognise positive states as important for drinking during preparation of the German studies may have lead to a failure to include such drinking determinants in the test instruments. Finally, only females had been examined in a study that failed to find any proximal relapse determinants in German drinkers (Rist et al., 1989); in the study by Cannon et al. (1990) only males were used. It is known that females tend not to drink in social situations where alcohol would be expected to augment pleasure (Olenick and Chalmers, 1991). Therefore, the purpose of the present study was to replicate the study carried out by Cannon et al. (1990) in a population of German alcoholics. This was possible since the same instrument used in the American study was recently translated and tested for its applicability to the study of drinking situations in German subjects (Victoria-Estrada, 1993). To the extent that the results of Cannon et al. (1990) reflect fundamental processesof alcoholism, it was predicted that the use of the same test instrument should yield the same significant profile of drinking determinants. If the cultural context plays a role in determining drinking situations in alcoholics, then one would expect to confirm differences in the data of the German alcoholics as opposed to North Americans. The present report represents data described more fully elsewhere (Victoria-Estrada, 1993). 2. Methods 2.1. Subjects A total of 336 alcoholics was tested in in-patient rehabilitation treatment. Alcoholics in treatment from 11 different centres were asked to participate in the study, and participation was entirely voluntary. The
data collection was completed over a 16-month period. There were 38 questionnaires not correctly completed. The study comprised 20% women; the average age of the sample was 42.7 years (SD. = 8.7; average male age = 42.9, average female age = 41.8 years), with an average education level of 9.1 years of schooling (S.D. = 2.1). In addition, 47% of the subjects were married, and 26% were unemployed. The average self-reported duration of problem drinking was 6.3 years (SD. = 5.9), and the average weekly intake was estimated to be 1.4 litres of pure alcohol (S.D. = 1.1). It was also found that 34% reported an episode of in-patient treatment within the previous 12 months. 2.2. Instruments A validated German version of the IDS was used along with a short questionnaire about demographic information and drinking patterns (Victoria-Estrada, 1993). The IDS is a questionnaire with 100 items contributing to eight subscales designed to assessthe frequency of heavy drinking in the last year in situations described by the item. The items are scored from 0 to 3 (Annis et al., 1987; Victoria-Estrada, 1993). The scales and the number of items contributing to each scale (in parentheses) are as follows: Unpleasant Emotions (20), Physical Discomfort (lo), Pleasant Emotions (lo), Testing Personal Control (lo), Urges and Temptations (lo), Conflict with Others (20), Social Pressure to Drink (lo), and Pleasant Time with Others (10). 2.3. Procedure In accordance with the procedures of Cannon et al. (1990), questionnaires with more than five omissions, with lack of variability in their answers, and those answered for the second time were excluded from the study. All 100 items from IDS were included in the factor analysis using (SPSS).Missing items were substituted with the mean and only loadings greater than 0.5 were considered. Then, the correlation matrix was analysed with principal components. The scree plot revealed three factors as the largest ones; they explained 43% of the variance. The three largest factors were extracted and varimax rotated. Also in accordance with recommendations by Cannon et al. (1990), any derived factors were to be used as a basis for the construction of new factorial scales; Cannon et al. (1990) criteria were also used. These scales were then evaluated for their reliability using tests of internal consistency (Anastasi, 1990) and for correlation with appropriate data from the original loo-item test. The new factor scales were then used to examine the determinants of drinking in different subgroups of subjects in the test population, using tests for correlation and repeated measures ANOVAs and t-test (with Bonferroni adjust-
A. Victoria-Estrada
et al. /Drug
and Alcohol Dependence 41 (1996) 75-79
77
Table 1 Psychometrical characteristics of the three scales for drinking situations Scales
Items
Item-difficulty
Item-total correlations
Cronbach’s alpha
Negative States Positive States Testing Personal Control
15 15 10
0.39-0.69 0.43-0.58 0.24-0.53
10.62-0.75 1X66-0.89 0.55-0.80
0.94 0.96 0.92
Minimum and maximum values of item difficulty and item-total correlations are shown.
ments for the alpha levels of 0.016 for the first comparison and 0.025 for the second).
3. Results Analysis of the data indicated a factor consisting of 37 items from the subscales ‘Negative Emotional States’, ‘Conflict with Others’ and two items from ‘Physical Discomfort’. This factor explained 27.6% of the variance and was named Negative States (NS). The second factor consisted of 30 items from the subscales ‘Pressure to Drink’, ‘Pleasant Times with Others’ and ‘Positive Emotional States’; it explained 14.4% of the variance and was named Positive States (PS). The last factor was made up of the 10 items from the subscale ‘Testing Personal Control’ and explained 5.2% of the variance; it was also called Testing Personal Control (‘W. The three factors were used as a basis for the construction of a new factorial scale. The first 1.5 items with the largest loadings from the first and second factors of the factor analysis were used to construct two new factorial scales; for the third scale the 10 items from the last factor of the factor analysis were used. The new scales had a total of 40 items as follows: Negative States (NS) using 21, 24, 3 1, 40,41,47, 51, 54, 64, 71, 74, 81, 87, 94 and 97; Positive States (PS) using 8, 13, 19, 28, 39, 48, 49, 63, 68, 69, 73, 79, 88, 89 and 99; and Testing Personal Control (TC) using all items. The results indicate acceptable internal consistency, variety of item difficulty and differentiation accuracy. The correlation between the 100 items from the whole questionnaire and the 40 items from the scaleswas high (r = 0.96) (Table 1). The scores from the 40 items were then analysed to assessthe relative importance of the three factors as determinants for drinking. To make the data from the three scalescomparable they were first weight adjusted: the scores from the first two scaleswere divided by 1.5. Results of a repeated measures ANOVA revealed an effect for the scales (F = 28.6, P < 0.001) and an interaction of sex by scale (F = 10.2, P < 0.001). Test of the correlation between the three scales were all significant: NS and PS r = 0.27, NS and TC r = 0.19, and PS and TC r = 0.37 (P < 0.001).
Further tests revealed that the scores on the three scales differed significantly from one another (Table 2) with the largest effect seen between NS and TC (t = 10.05, P < 0.001) and the smallest for NS versus PS (t = 3.48, P < 0.001). In addition, it was found that the relation of the scores on the different scales were different for men and women (Table 2). For men the scores on the NS and PS scaleswere significantly higher than for TC (t = 9.43 and 8.18, both P < 0.001). For women, however, the scores for NS were significantly higher than PS (1 = 4.38, P < 0.001) and TC (t = 3.59, P < 0.001). It was confirmed that the men and women differed significantly on the scale PS (t = 3.25, P < 0.01).
4. Discussion The present study found that on a standard test for drinking situations (Annis et al., 1987) the factorial structure of a German sample of alcoholics conformed very well with the structure reported for a sample of North American male alcoholics (Cannon et al., 1990). Scales were found which corresponded to drinking in situations (1) of negative states, (2) of positive states, and (3) where personal control was tested. The factors, their structure and their explanation of variance are almost identical in the present and previous study. There are a number of other similarities between the
Table 2 Means (S.D.) of the three scales for drinking situations for the group as a whole and for men and women Scales
Total (N = 336)
Men (n = 269)
Women (n = 67)
Negative States Positive States Testing Personal Control
17.02 (6.38)
17.03 (6.43)
17.01 (6.19)
15.34 (8.11)
16.10 (7.76)
12.27 (8.83)
12.21 (7.32)
11.95 (7.30)
13.27 (7.37)
Minimum and maximum values of item difficulty and item-total correlations are shown.
78
A. Victoria-Estraa’a
et al. / Drug and Alcohol Dependence 41 (1996) 75-79
present study population and that of the previous report (Cannon et al., 1990, p. 266), such as age (42.7 vs. 44.3 years), and marital status (47% vs. 43% married). Although only 32 items of the German scales were the same as the 40 items of the Cannon study, this should be considered high since the German sample also included women. Moreover, the subjects in the American study were recruited from a more uniform population (Veterans). Unfortunately, Cannon et al. (1990) did not report data for their sample regarding percentage of unemployed subjects, duration of problem drinking, weekly intake of ethanol, and percentage of subjects with inpatient treatment in the previous year. Thus, we are not able to compare both samples on these variables. In addition, the data in the present study were further evaluated to determine the test characteristics of the three derived scales. It was noted that the obtained factorial scales showed high internal consistency and were highly correlated with the original queitionnaire. These results confirmed the reliability and validity of the new factorial scales. Comparison of the scales to one another revealed differences in response to each scale. NS were the most important determinant for drinking, followed by PS and then TC. It was further indicated that there were sex differences in the responses of different scales. In men both NS and PS were associated with drinking and both were more important than TC as a determinant of drinking. These findings resembled the pattern obtained by Cannon et al. (1990) with their male-only sample. In women, on the other hand, NS were more important for drinking than PS and TC. Consistent with this, Olenick and Chalmers (1991) reported that female alcoholics tend to drink alone rather than in social situations, as is seen in female normal drinkers and male alcoholics. It is likely that the social situations would be associated with pleasure-enhancing effects of alcohol. Unfortunately, the relatively small group of female alcoholics in the present study made it difficult to test whether PS were indeed a significant factor in the women drinkers alone. Accordingly, our results replicate and extend the findings from Cannon et al. ,(1990). The data provide a cross-cultural validation of a pattern of empirical determinants to drink. It is seen that drinking situations involve (1) negative emotions and social conflict, (2) positive emotions, positive social interactions and social pressure to drink, and (3) situations regarding testing personal control. This pattern of determinants to drink is comparable to that previously thought by Litman et al. (1977) to represent high risk situations for relapse in a sample of British alcoholics as well as to the pattern of high-risk situations found by Isenhard (1991) with a sample of American male veterans alcoholics tested with the IDS and using a different factorial procedure.
Alcoholics appear, therefore, to show a similar pattern of risk situations for drinking irrespective of prevailing cultural differences in drinking patterns in the general population. Moreover, the results of the present study are consistent with the North American study despite many differences in the level of alcohol consumption on both sides of the Atlantic. In addition, in-patient rehabilitation treatment programs in Germany involve durations of many months compared to only a few weeks in North America (Feuerlein, 1990). Accordingly to the similar profile of situations where excessivedrinking takes place seenin alcoholics in both Germany and North America, despite major cultural and attitudinal differences with regard to alcohol consumption in the society at large, cultural factors may not have an obvious influence in those alcoholics presenting for treatment. However, the size of our sample, and the limitations of the instrument we tested, i.e. limited item content, prevent us from extending our interpretations to other populations of drinkers. Further, the present study does not in any way question the role of sociocultural factors as an important determinant of drinking in other sectors of the drinking population at large. These factors may be involved in the initiation of alcohol and substance consumption, in the consumption of levels that do not lead to dependence, or in other behaviours associated with alcohol (e.g., sexuality, aggression).
Acknowledgements We wish to thank Dr Helen Annis for allowing us to use the IDS. This research is supported by the Deutscher Akademischer Austauschdienst (DAAD). Preparation of the manuscript was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG Mu 1136/2-l).
References Anastasi, A. (1990) Psychological Testing, Macmillan, New York. Annis, H.M., Graham, J.M. and Davis, C.S. (1987) Inventory of Drinking Situations (IDS) User’s Guide, Addiction Research Foundation, Toronto. Cannon, D.S., Leeka, J.K., Patterson, E.T. and Baker, T.B. (1990) Principal components analysis of the Inventory of Drinking Situations: empirical categories of drinking by alcoholics. Addict. Behav. 15, 265-269. Cockerham, W.C., Kunz, G. and Lueschen, L. (1989) Alcohol use and psychological distress: a comparison of Americans and West Germans. Int. J. Addict. 24, 951-961. Feuerlein, W. (1990) Treatment for alcoholics in German speaking countries. Br. J. Addict. 85, 353-356. Isenhard, C.E. (1991) Factor structure of the Inventory of Drinking Situations. J. Subst. Abuse, 3, 59-71.
A. Victoria-Estrada
et al. / Drug and Alcohol Dependence 41 (1996) 75-79
Junge, B. (1993) Alkohol. In: Jahrbuch Sucht ‘94 [Adiction Yearbook ‘941(Htlllinghorst, R., Wtinschmann, B. and Bermeister, J., eds.), pp. 81-99. Neuland, Geesthacht. Litman, G.K., Eiser, J.R., Rawson, N.S.B. and Oppenheim, A.N. (1977) Towards a typology of relapse: a preliminary report. Drug Alcohol Depend. 2, 157-162. Olenick, N.L. and Chalmers, D.K. (1991) Gender-specific drinking styles in alcoholics and nonalcoholics. J. Stud. Alcohol, 52, 325330. Pfeiffer, W., Fahrner, E-M. and Feuerlein, W. (1988) Soziale Anpassung und Riickfallanalyse bei ambulant behandelten Alkoholabhangigen [Social adjustment and relapse analysis in dependent alcoholics in out-patient treatment]. Suchtgefahren 34, 357-368. Rist, F., Watzl, H. and Cohen, R. (1989) Versuche zur Erfassung von
19
Rtickfallbedingungen bei Alkoholkranken [Experiences on the characterization of conditions of relapse in alcoholics]. In: Riickfall und Riickfall-prophylaxe [Relapse and Relapse Prevention]. (Watzl, H. and Cohen, R., eds.), pp. 126-138. Springer, Berlin. Robinson, T.E. and Berridge, KC. (1993) The neural basis of drug craving an incentive-sensitization theory of addiction. Brain Res. Rev. 18, 241-291. Stewart, J., deWit, H. and Eikelboom, R. (1984) The role of unconditioned and conditioned drug effects in the self-administration of opiates and stimulants. Psycho]. Rev. 91, 251-268. Victoria-Estrada, A. (1993) Risikosituationen bei Personen mit Alkoholabhangigkeit und Alkoholmissbrauch [Situations of Risk in Individuals with Alcohol Dependence and Alcohol Misuse]. Doctoral Dissertation, University of Cologne.
Drug and Alcohol Dependence 41 (1996) 75-79
Excessive drinking situations in German alcoholics: replication of a three-factor model used for North Americans A. Victorio-Estradaa,b,*, R.F. Mucha”,“, E.R. Stephand “Institute of Psychology, University of’ Cologne, Herbert-Lewin-Strasse 2, D-50931 K&z, Germany bCenter of Psychological Service, National Autonomous University of Mexico, Mexico City, Mexico ‘Institute for Medical Psychology, University qf Tiibingen, D- 72074 Tiibingen, Germany ‘Institute qf Businessand Social Psychology, University of Cologne, Herbert-Lewin-Strasse, D-50931 Kiiln, Germany Received 8 June 1995; accepted 7 March 1996
Abstract German alcoholics were examined for a profile of drinking situations which was found recently in North American alcoholics. A mixed sample of alcoholics (N = 336) receiving in-patient rehabilitation treatment for alcoholism was administered a validated translation of the English version of Annis’s Inventory ofDrinking
Situations (IDS). Analysis of the 100 items in the questionnaire,
using principal component methods and varimax rotation, revealed the same three factorial scales seen in the previous study in North American
alcoholics (Cannon et al. (1990) Addictive Behaviour 15, 265-269).
Drinking
was most likely to take place in
situations (1) of negative states, (2) of positive states, and (3) where personal control was tested. Although there are large differences in the level and social acceptability of drinking between Central Europe and English-speaking data confirm the cross-cultural validity of basic determinants of drinking.
countries, the present
Keywords: Drinking situations; Cross-cultural; Alcoholics; Questionnaire; Motivation
1. Introduction
in which alcoholics show excessive drinking may be important for understanding the maintenance of drinking and relapse in abstinent individuals (Annis et al., 1987). However, there is still no clear idea whether the drinking situations reported by alcoholics are representative of identifiable, fundamental processes of addiction or simply reflect social/cultural organisation in the environment of the alcoholic. Recently, a population of American alcoholics were asked about the situations in which they drank excessively using an inventory of drinking situations (Annis et al., 1987). Using principle component analysis with varimax rotation of the data, it was found that there were three situations that give rise to drinking behaviour: negative states, positive states and testing of personal control (Cannon et al., 1990). The situations
* Corresponding author, Pacific0 159-201, Los Reyes, Coyoacan, 04330 Mexico D.F., Mexico.
However, comparison of the findings of Cannon et al. (1990) to other data suggests that drinking situations
may be actually specific to the population tested. In a study of proximal determinants of relapse in Germans, testing personal control, unpleasant feelings, work problems and conflict with a partner were identified as reasons (Pfeiffer et al., 1988); there had been, however, no indication that Germans drink excessively in positive states. Other studies failed to find any proximal relapse determinants in German drinkers (Rist et al., 1989). Prominent differences exist between American and other English-speaking and German cultures, regarding frequency and quantity of alcohol drinking, levels of abstinence, acceptability and availability of alcohol (Cockerham et al., 1989); for example, the national per
capita consumption of alcohol in Germany is at least 50% higher than in countries like Canada, England, Australia and the United States (Junge, 1993). Therefore, looking at the data reported in the literature, it may not seem surprising that the drinking situations reported by German alcoholics should be different from those of North Americans. This would suggest socio-
0376-8716/96/$15.00 0 1996 Elsevier Science Ireland Ltd. All rights reserved
PII SO376-8716(96)01241-O
16
A. Vicrorio-Estrada
et al. II Drug und Alcohol Dependence 41 (1996) 75 79
cultural influences in the fundamental processesof alcoholism. A major problem with comparisons across cultural groups is that parallel tests are not usually carried out in the different populations. Indeed, the findings of Cannon et al. (1990) have not been tested in a wide range of alcoholics so at the present time it is not possible to draw any firm conclusions concerning differences in the drinking situations reported by different populations of alcoholics. It is, for example, not clear how drinking situations examined by Cannon et al. (1990) correspond to the proximal determinants of relapse studied by Pfeiffer et al. (1988) and Rist et al. (1989). Moreover, it was only recently recognised in the English-speaking literature that alcohol and drugs may be consumed for their positive reinforcing effects rather than due to their negative reinforcing effects (Stewart et al., 1984; Robinson and Berridge, 1993). Therefore, failure to recognise positive states as important for drinking during preparation of the German studies may have lead to a failure to include such drinking determinants in the test instruments. Finally, only females had been examined in a study that failed to find any proximal relapse determinants in German drinkers (Rist et al., 1989); in the study by Cannon et al. (1990) only males were used. It is known that females tend not to drink in social situations where alcohol would be expected to augment pleasure (Olenick and Chalmers, 1991). Therefore, the purpose of the present study was to replicate the study carried out by Cannon et al. (1990) in a population of German alcoholics. This was possible since the same instrument used in the American study was recently translated and tested for its applicability to the study of drinking situations in German subjects (Victoria-Estrada, 1993). To the extent that the results of Cannon et al. (1990) reflect fundamental processesof alcoholism, it was predicted that the use of the same test instrument should yield the same significant profile of drinking determinants. If the cultural context plays a role in determining drinking situations in alcoholics, then one would expect to confirm differences in the data of the German alcoholics as opposed to North Americans. The present report represents data described more fully elsewhere (Victoria-Estrada, 1993). 2. Methods 2.1. Subjects A total of 336 alcoholics was tested in in-patient rehabilitation treatment. Alcoholics in treatment from 11 different centres were asked to participate in the study, and participation was entirely voluntary. The
data collection was completed over a 16-month period. There were 38 questionnaires not correctly completed. The study comprised 20% women; the average age of the sample was 42.7 years (SD. = 8.7; average male age = 42.9, average female age = 41.8 years), with an average education level of 9.1 years of schooling (S.D. = 2.1). In addition, 47% of the subjects were married, and 26% were unemployed. The average self-reported duration of problem drinking was 6.3 years (SD. = 5.9), and the average weekly intake was estimated to be 1.4 litres of pure alcohol (S.D. = 1.1). It was also found that 34% reported an episode of in-patient treatment within the previous 12 months. 2.2. Instruments A validated German version of the IDS was used along with a short questionnaire about demographic information and drinking patterns (Victoria-Estrada, 1993). The IDS is a questionnaire with 100 items contributing to eight subscales designed to assessthe frequency of heavy drinking in the last year in situations described by the item. The items are scored from 0 to 3 (Annis et al., 1987; Victoria-Estrada, 1993). The scales and the number of items contributing to each scale (in parentheses) are as follows: Unpleasant Emotions (20), Physical Discomfort (lo), Pleasant Emotions (lo), Testing Personal Control (lo), Urges and Temptations (lo), Conflict with Others (20), Social Pressure to Drink (lo), and Pleasant Time with Others (10). 2.3. Procedure In accordance with the procedures of Cannon et al. (1990), questionnaires with more than five omissions, with lack of variability in their answers, and those answered for the second time were excluded from the study. All 100 items from IDS were included in the factor analysis using (SPSS).Missing items were substituted with the mean and only loadings greater than 0.5 were considered. Then, the correlation matrix was analysed with principal components. The scree plot revealed three factors as the largest ones; they explained 43% of the variance. The three largest factors were extracted and varimax rotated. Also in accordance with recommendations by Cannon et al. (1990), any derived factors were to be used as a basis for the construction of new factorial scales; Cannon et al. (1990) criteria were also used. These scales were then evaluated for their reliability using tests of internal consistency (Anastasi, 1990) and for correlation with appropriate data from the original loo-item test. The new factor scales were then used to examine the determinants of drinking in different subgroups of subjects in the test population, using tests for correlation and repeated measures ANOVAs and t-test (with Bonferroni adjust-
A. Victoria-Estrada
et al. /Drug
and Alcohol Dependence 41 (1996) 75-79
77
Table 1 Psychometrical characteristics of the three scales for drinking situations Scales
Items
Item-difficulty
Item-total correlations
Cronbach’s alpha
Negative States Positive States Testing Personal Control
15 15 10
0.39-0.69 0.43-0.58 0.24-0.53
10.62-0.75 1X66-0.89 0.55-0.80
0.94 0.96 0.92
Minimum and maximum values of item difficulty and item-total correlations are shown.
ments for the alpha levels of 0.016 for the first comparison and 0.025 for the second).
3. Results Analysis of the data indicated a factor consisting of 37 items from the subscales ‘Negative Emotional States’, ‘Conflict with Others’ and two items from ‘Physical Discomfort’. This factor explained 27.6% of the variance and was named Negative States (NS). The second factor consisted of 30 items from the subscales ‘Pressure to Drink’, ‘Pleasant Times with Others’ and ‘Positive Emotional States’; it explained 14.4% of the variance and was named Positive States (PS). The last factor was made up of the 10 items from the subscale ‘Testing Personal Control’ and explained 5.2% of the variance; it was also called Testing Personal Control (‘W. The three factors were used as a basis for the construction of a new factorial scale. The first 1.5 items with the largest loadings from the first and second factors of the factor analysis were used to construct two new factorial scales; for the third scale the 10 items from the last factor of the factor analysis were used. The new scales had a total of 40 items as follows: Negative States (NS) using 21, 24, 3 1, 40,41,47, 51, 54, 64, 71, 74, 81, 87, 94 and 97; Positive States (PS) using 8, 13, 19, 28, 39, 48, 49, 63, 68, 69, 73, 79, 88, 89 and 99; and Testing Personal Control (TC) using all items. The results indicate acceptable internal consistency, variety of item difficulty and differentiation accuracy. The correlation between the 100 items from the whole questionnaire and the 40 items from the scaleswas high (r = 0.96) (Table 1). The scores from the 40 items were then analysed to assessthe relative importance of the three factors as determinants for drinking. To make the data from the three scalescomparable they were first weight adjusted: the scores from the first two scaleswere divided by 1.5. Results of a repeated measures ANOVA revealed an effect for the scales (F = 28.6, P < 0.001) and an interaction of sex by scale (F = 10.2, P < 0.001). Test of the correlation between the three scales were all significant: NS and PS r = 0.27, NS and TC r = 0.19, and PS and TC r = 0.37 (P < 0.001).
Further tests revealed that the scores on the three scales differed significantly from one another (Table 2) with the largest effect seen between NS and TC (t = 10.05, P < 0.001) and the smallest for NS versus PS (t = 3.48, P < 0.001). In addition, it was found that the relation of the scores on the different scales were different for men and women (Table 2). For men the scores on the NS and PS scaleswere significantly higher than for TC (t = 9.43 and 8.18, both P < 0.001). For women, however, the scores for NS were significantly higher than PS (1 = 4.38, P < 0.001) and TC (t = 3.59, P < 0.001). It was confirmed that the men and women differed significantly on the scale PS (t = 3.25, P < 0.01).
4. Discussion The present study found that on a standard test for drinking situations (Annis et al., 1987) the factorial structure of a German sample of alcoholics conformed very well with the structure reported for a sample of North American male alcoholics (Cannon et al., 1990). Scales were found which corresponded to drinking in situations (1) of negative states, (2) of positive states, and (3) where personal control was tested. The factors, their structure and their explanation of variance are almost identical in the present and previous study. There are a number of other similarities between the
Table 2 Means (S.D.) of the three scales for drinking situations for the group as a whole and for men and women Scales
Total (N = 336)
Men (n = 269)
Women (n = 67)
Negative States Positive States Testing Personal Control
17.02 (6.38)
17.03 (6.43)
17.01 (6.19)
15.34 (8.11)
16.10 (7.76)
12.27 (8.83)
12.21 (7.32)
11.95 (7.30)
13.27 (7.37)
Minimum and maximum values of item difficulty and item-total correlations are shown.
78
A. Victoria-Estraa’a
et al. / Drug and Alcohol Dependence 41 (1996) 75-79
present study population and that of the previous report (Cannon et al., 1990, p. 266), such as age (42.7 vs. 44.3 years), and marital status (47% vs. 43% married). Although only 32 items of the German scales were the same as the 40 items of the Cannon study, this should be considered high since the German sample also included women. Moreover, the subjects in the American study were recruited from a more uniform population (Veterans). Unfortunately, Cannon et al. (1990) did not report data for their sample regarding percentage of unemployed subjects, duration of problem drinking, weekly intake of ethanol, and percentage of subjects with inpatient treatment in the previous year. Thus, we are not able to compare both samples on these variables. In addition, the data in the present study were further evaluated to determine the test characteristics of the three derived scales. It was noted that the obtained factorial scales showed high internal consistency and were highly correlated with the original queitionnaire. These results confirmed the reliability and validity of the new factorial scales. Comparison of the scales to one another revealed differences in response to each scale. NS were the most important determinant for drinking, followed by PS and then TC. It was further indicated that there were sex differences in the responses of different scales. In men both NS and PS were associated with drinking and both were more important than TC as a determinant of drinking. These findings resembled the pattern obtained by Cannon et al. (1990) with their male-only sample. In women, on the other hand, NS were more important for drinking than PS and TC. Consistent with this, Olenick and Chalmers (1991) reported that female alcoholics tend to drink alone rather than in social situations, as is seen in female normal drinkers and male alcoholics. It is likely that the social situations would be associated with pleasure-enhancing effects of alcohol. Unfortunately, the relatively small group of female alcoholics in the present study made it difficult to test whether PS were indeed a significant factor in the women drinkers alone. Accordingly, our results replicate and extend the findings from Cannon et al. ,(1990). The data provide a cross-cultural validation of a pattern of empirical determinants to drink. It is seen that drinking situations involve (1) negative emotions and social conflict, (2) positive emotions, positive social interactions and social pressure to drink, and (3) situations regarding testing personal control. This pattern of determinants to drink is comparable to that previously thought by Litman et al. (1977) to represent high risk situations for relapse in a sample of British alcoholics as well as to the pattern of high-risk situations found by Isenhard (1991) with a sample of American male veterans alcoholics tested with the IDS and using a different factorial procedure.
Alcoholics appear, therefore, to show a similar pattern of risk situations for drinking irrespective of prevailing cultural differences in drinking patterns in the general population. Moreover, the results of the present study are consistent with the North American study despite many differences in the level of alcohol consumption on both sides of the Atlantic. In addition, in-patient rehabilitation treatment programs in Germany involve durations of many months compared to only a few weeks in North America (Feuerlein, 1990). Accordingly to the similar profile of situations where excessivedrinking takes place seenin alcoholics in both Germany and North America, despite major cultural and attitudinal differences with regard to alcohol consumption in the society at large, cultural factors may not have an obvious influence in those alcoholics presenting for treatment. However, the size of our sample, and the limitations of the instrument we tested, i.e. limited item content, prevent us from extending our interpretations to other populations of drinkers. Further, the present study does not in any way question the role of sociocultural factors as an important determinant of drinking in other sectors of the drinking population at large. These factors may be involved in the initiation of alcohol and substance consumption, in the consumption of levels that do not lead to dependence, or in other behaviours associated with alcohol (e.g., sexuality, aggression).
Acknowledgements We wish to thank Dr Helen Annis for allowing us to use the IDS. This research is supported by the Deutscher Akademischer Austauschdienst (DAAD). Preparation of the manuscript was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG Mu 1136/2-l).
References Anastasi, A. (1990) Psychological Testing, Macmillan, New York. Annis, H.M., Graham, J.M. and Davis, C.S. (1987) Inventory of Drinking Situations (IDS) User’s Guide, Addiction Research Foundation, Toronto. Cannon, D.S., Leeka, J.K., Patterson, E.T. and Baker, T.B. (1990) Principal components analysis of the Inventory of Drinking Situations: empirical categories of drinking by alcoholics. Addict. Behav. 15, 265-269. Cockerham, W.C., Kunz, G. and Lueschen, L. (1989) Alcohol use and psychological distress: a comparison of Americans and West Germans. Int. J. Addict. 24, 951-961. Feuerlein, W. (1990) Treatment for alcoholics in German speaking countries. Br. J. Addict. 85, 353-356. Isenhard, C.E. (1991) Factor structure of the Inventory of Drinking Situations. J. Subst. Abuse, 3, 59-71.
A. Victoria-Estrada
et al. / Drug and Alcohol Dependence 41 (1996) 75-79
Junge, B. (1993) Alkohol. In: Jahrbuch Sucht ‘94 [Adiction Yearbook ‘941(Htlllinghorst, R., Wtinschmann, B. and Bermeister, J., eds.), pp. 81-99. Neuland, Geesthacht. Litman, G.K., Eiser, J.R., Rawson, N.S.B. and Oppenheim, A.N. (1977) Towards a typology of relapse: a preliminary report. Drug Alcohol Depend. 2, 157-162. Olenick, N.L. and Chalmers, D.K. (1991) Gender-specific drinking styles in alcoholics and nonalcoholics. J. Stud. Alcohol, 52, 325330. Pfeiffer, W., Fahrner, E-M. and Feuerlein, W. (1988) Soziale Anpassung und Riickfallanalyse bei ambulant behandelten Alkoholabhangigen [Social adjustment and relapse analysis in dependent alcoholics in out-patient treatment]. Suchtgefahren 34, 357-368. Rist, F., Watzl, H. and Cohen, R. (1989) Versuche zur Erfassung von
19
Rtickfallbedingungen bei Alkoholkranken [Experiences on the characterization of conditions of relapse in alcoholics]. In: Riickfall und Riickfall-prophylaxe [Relapse and Relapse Prevention]. (Watzl, H. and Cohen, R., eds.), pp. 126-138. Springer, Berlin. Robinson, T.E. and Berridge, KC. (1993) The neural basis of drug craving an incentive-sensitization theory of addiction. Brain Res. Rev. 18, 241-291. Stewart, J., deWit, H. and Eikelboom, R. (1984) The role of unconditioned and conditioned drug effects in the self-administration of opiates and stimulants. Psycho]. Rev. 91, 251-268. Victoria-Estrada, A. (1993) Risikosituationen bei Personen mit Alkoholabhangigkeit und Alkoholmissbrauch [Situations of Risk in Individuals with Alcohol Dependence and Alcohol Misuse]. Doctoral Dissertation, University of Cologne.