A Shortened Version Was Published with the Same Title, Contemporary Drug Problems 29:619-648

Total Page:16

File Type:pdf, Size:1020Kb

A Shortened Version Was Published with the Same Title, Contemporary Drug Problems 29:619-648

A shortened version was published with the same title, Contemporary Drug Problems 29:619-648, 2002.

CAN ALCOHOL EXPECTANCIES AND ATTRIBUTIONS EXPLAIN WESTERN EUROPE'S NORTH-SOUTH GRADIENT IN ALCOHOL'S ROLE IN VIOLENCE?1

Robin Room and Sandra Bullock Centre for Social Research on Alcohol and Drugs, Stockholm University, Stockholm, Sweden

Introduction Alcohol's position in the culture differs from one society to another (Room & Mäkelä, 2000). One dimension of this variation is between what have been described as "wet" and "dry" societies (Room, 1989). Although the distinction can be made more generally, it has often been stereotypically described in terms of the difference between northern European "spirits" or "beer cultures" and southern European "wine cultures". At the "dry" end of this contrast, alcohol is held apart from everyday life as a special commodity for special contexts; drinking has traditionally been sporadic, often at festivals or weekends, with a high proportion of drinking occasions involving intoxication. At the "wet" end of the contrast, drinking is a part of everyday life (at least for men), and frequently accompanies meals. Overtly intoxicated behaviour is reported to be rare, presumably reflecting both the tolerance built up by a regular consumer, and a cultural banalization of the psychopharmacological properties of an everyday item of consumption. The contrast is often also characterized in terms of differing histories of social responses to drinking (Levine, 1992). Although there were temperance organizations in both Italy (Cottino and Morgan, 1985) and France (Prestwich, 1988) around 1900, for instance, they were relatively small- scale, with a constituency confined to the elite. Germany's temperance movement, though it reached further into working-class circles, also had limited scope (Roberts, 1984). In Britain (Harrison, 1971) and other English-speaking countries, and north of the Baltic (Johansson, 2000), temperance became an important social movement, substantially reducing the rates of social and health problems from drinking in the 80 years or so before 1930. The contrast of "wet" and "dry" cultures is not without problems, particularly if the frame of reference is expanded to a global perspective (Room et al, 1996; Room and Mäkelä, 2000). Even within western Europe, the dichotomy is somewhat problematic. Is it primarily a reflection of the distinction between "temperance cultures" (Levine, 1992) -- societies where the temperance movement was very strong -- and others? In that case, the "dry" side of the dichotomy would presumably include Finland, Sweden, Norway, the U.K. and Ireland. Alternatively, the "dry" side could be defined as societies in which spirits drinking predominated, which would limit it to Finland, Sweden and Norway until recent decades, and to no country in western Europe nowadays. Or is the crucial differentiation whether wine is drunk everyday with meals? In that case, the "dry" side might include all western European countries except Greece, Italy, Spain, Portugal, and France. A further complication is that in recent years there has been considerable convergence in levels of drinking in Europe; alcohol consumption levels have fallen sharply in the wine countries of the South, and risen in most countries to their north (Leifman, 2001b). At this point, for instance, Italy's per-capita consumption is lower than Ireland's. There has also been some homogenization in terms of beverage choice: spirits drinking has declined in the north, while wine and beer drinking have risen; and wine drinking has declined in the south, while beer drinking has increased. However, despite these trends, qualitative differences in drinking patterns between different parts of Europe seem to persist; Simpura (2001) notes that many of these "seem very persistent, and immune to change, even over decades". 1 Prepared for presentation at the 27th annual Alcohol Epidemiology Symposium, Kettil Bruun Society for Social and Epidemiological Research on Alcohol, Toronto, Canada. May 28-June 1, 2001. Thanks are due to Barbara Leigh for assistance in defining the survey items and to Thor Norström, Jussi Simpura and others of the European Comparative Alcohol Study for the opportunity to include the items. Data collection for the ECAS surveys was financed by the European Union, the Swedish Public Health Institute (FHI), and the Finnish National Institute for Social Welfare and Health (STAKES). The authors’ work is supported by the FHI and the Swedish Social Research Council (now the FAS).

1 A hypothesis that drinking plays a stronger causal role in violence in "dryer" than in "wetter" cultures has long been latent in the literature (see, for example, Christie, 1965). The most developed tradition of research testing this hypothesis has been through auto-regressive time-series analyses correlating changes in per-capita alcohol consumption with changes in indicators of violence, in particular homicide and suicide rates. Lenke (1990) set out to test the hypothesis specifically, comparing time-series analyses of the influence of levels of consumption on rates of homicide in Sweden and in France. He did find a lower influence in France. Norström (1995) found a similar difference in time-series analyses of the influence on alcohol consumption levels on suicide. On the other hand, the results of such analyses have not always fit the hypothesis: for homicide, Norström (1988) did not find a significant effect in either "wetter" Denmark or "dryer" Finland, while for suicide similarly strong effects have been found for "wetter" Hungary and "dryer" Sweden (Norström, 1995). Recently, this style of time-series analysis was carried out on a more extended basis in the European Comparative Alcohol Study (Norström, 2001; Norström et al., 2001) of 14 western European countries (the current European Union minus Greece and Luxembourg plus Norway). The countries were trichotomized into the 4 high-consumption wine countries (France, Italy, Portugal and Spain), the 3 low-consumption countries in which spirits drinking used to predominate (Finland, Sweden and Norway), and a third in-between category. The relation of changes in the homicide rate to changes in per-capita consumption was indeed strongest in the low-consumption countries and weakest in the high-consumption countries, although the relation was significant in all three groups of countries (Norström et al., 2001:93). Much the same pattern held for mortality from accidents. The relation of the suicide death rate to alcohol consumption levels was significant primarily in the low- consumption countries. These results generally support the hypothesis that alcohol plays a stronger causal role in violence, and also in casualties generally, in northern than in southern European countries. However, the results in detail were not always consistent with this overall pattern. For instance, the connection of homicide with per-capita consumption was as strong in Spain as in the low-consumption countries, while the coefficients for suicide were lower in Finland than in Belgium. Furthermore, cirrhosis mortality also showed a stronger relation to alcohol consumption in the low- consumption countries. This result would not have been predicted by conventional contrasts of "wet" vs. "dry" societies (Room, 1989). Behind the hypothesis about differences between northern and southern Europe in the relation of alcohol consumption to violence lie a variety of possible mechanisms. We spell out several such here, which could operate individually or in combination. I. A given amount of consumption can be distributed differently. In particular, it may be distributed relatively equally in smaller amounts on different days of the week, or may be concentrated in larger amounts on just one or two days, for instance at the weekend. Drinking more on an occasion may raise the risk of violence. II. Cultures may differ in the upper limit on the amount consumed on a heavy drinking occasion, and heavier amounts may carry a greater risk of serious violence. While there may well be a relation between this factor and a culture's position on (I), cultures could differ on whether extremely heavy drinking is ever considered acceptable without differing much on other aspects of drinking patterns. III. The behaviours which are accepted as going with drinking may differ. As MacAndrew and Edgerton (1969) noted, there is much evidence from ethnographic studies that cultures differ substantially in drunken comportment. At one extreme, they report that some cultures expect no differences between intoxicated and sober behaviour. Drinking in southern European wine cultures is sometimes described in such terms -- that a drinker should struggle to show no effects of alcohol on his behaviour, no matter how much he has had to drink. IV. Cultures may differ in how seriously problematic behaviour while drinking is regarded -- in the degree of tolerance for the "within limits" behaviour which MacAndrew and Edgerton (1969) propose the culture accepts, and in the extent of social sanctions on behaviour which goes beyond those limits. V. There may be variations between cultures in the cognitive recognition of a causal role of

2 intoxication in bad behaviour. Levine (1983) proposes, for instance, that, although colonial Americans disapproved of drunkenness, they did not recognize a link between it and violence. In his interpretation, the cultural construction of a link came with the temperance movement. The cognitive recognition itself carries consequences. From of a temperance movement perspective, of course, the hope was that those accepting the link would forswear drinking. Other potential consequences include an "expectancy effect", where cognitive recognition of the link becomes a self-fulfilling prophecy; and the potential for intoxication to serve as an excuse for bad behaviour (see below). VI. Intoxication may serve as an excuse for bad behaviour. Logically, if no causal link was recognized between intoxication and bad behaviour, intoxication should not be available as an excuse for bad behaviour. However, it is quite possible to consider intoxication a cause of violence and at the same time to give it little or no excuse-value. This was the position of a majority of respondents in a survey in Ontario (Paglia and Room, 1998). Most often, they justified holding the drunken person responsible on the grounds that the drinker chooses to drink despite knowing it might result in bad behaviour (Room, 1998). In the present paper, we focus on the mechanisms numbered III through VI on this list. As background, however, we first summarize what is known from the literature relevant to I and II. Some data is available from other studies on differences in drinking patterns (Simpura and Karlsson, 2001; Leifman, 2001a). In general, despite some convergences, it remains true in northern Europe more beer is consumed and in the south more wine is consumed. In addition, in the north drinking is more likely to occur on one or two occasions per week, while in southern Europe people are more likely to drink nearly every day. Swedish, Finnish and British drinkers report drinking a higher quantity per occasion than Franch, German and Italian drinkers. However, there is not a clear north- south gradient in total number of heavy drinking occasions per year (Leifman, 2001a). Less comparative data is available across Europe on how much drinking is defined as qualifying as drunkenness. Cameron et al. (2000) took a semantic approach to the issue of what is denoted by drunkenness; but the differences they found between languages and sites in terms of psychological and behavioural effects of drunkenness are not easily interpreted. When 15-year-olds across Europe were asked in the 1995 ESPAD study how intoxicated a person of their sex and age would be after drinking 5 drinks in a row, there was little difference in average scores between northern and southern Europe. In many countries, about the same proportion reported being intoxicated as reported drinking 5 or more drinks in a row 3 or more times in the last 30 days, but the proportion reporting intoxication was notably lower in some countries scattered cross Europe -- Norway, Ireland, Italy and Malta (Andersson et al., 1999). In a qualitative comparison of what Spanish and Finnish young adults reported about their experiences with drinking and drunkenness, Pyorälä (1995) found that, contrary to expectations from the literature that "drinking in a wine country is socially regulated and drunkenness is unusual", drunkenness coexisted with sociable drinking amoung young adults in both Finland and Spain. "It offers a possibility of flight from the rules of sociability in [both] cultures." One potential avenue to testing mechanisms III to VI is through studies of alcohol expectancies -- asking respondents what they expect to happen after drinking. Little previous work has been done, however, with comparative data from a variety of western European societies. One such study is by Lindman and Lang (1994), although its generalizability is limited by the fact that data was collected from a convenience sample of about 100 university students in each participating society. From answers to a list of items on expectations of greater aggression "after many drinks", Lindman and Lang derived a summary index. The results on this index were not in any clear gradient from northern to southern Europe. Among western European societies included in the study, the Spanish reported the highest expectations of violence, followed in order by Belgium, Finland, Italy and France. Students from Spain and Italy were less likely than students from the other societies to think that people should be held responsible for their behaviour while intoxicated. The 1999 ESPAD survey of 15-year-olds across Europe included items on expectancies concerning six potentially positive consequences of drinking, and six negative (Hibell et al., 2001:293). If we compare results for two countries from each of three parts of western Europe, Italian and French students were considerably less likely to expect positive consequences than students from Denmark, the United Kingdom, Sweden and Norway, and a little more likely to expect

3 negative consequences. Differences were not great on the proportion expecting that they would "do something I would later regret", with Italian students reporting this most often and French students least often. However, French and Italian students were strikingly less likely than students from the other four countries to report expecting to "feel happy" or to "have a lot of fun" after drinking.

The present study The present study is a first effort to study the potential role of the mechanisms outlined in III through VI above. The opportunity for this pilot study arose in the context of the European Comparative Alcohol Study to commission telephone surveys of the general adult population in six European countries: Finland and Sweden, representing the tier of northern European "low consumption" countries: Italy and France, representing southern European "high consumption" wine cultures, and the United Kingdom and Germany, from the tier in between (Leifman, 2001a). Data Collection. In each country data were collected via telephone interview using an interview schedule in the country’s primary language. Approximately 1000 interviews were conducted in each country. Samples were selected to be representative of the country population between 18 and 64 years of age. Once contact was made with a randomly selected phone number, the eligible individual with the next birthday in the household was selected to participate. Interviews were conducted during the spring of the year 2000. Sampling Procedures and Data Weighting. Samples were ultimately weighted to reflect country composition; however, individuals were sampled and weighted by different factors within each of the countries. In France, Sweden and the UK samples were drawn from all non-mobile phone numbers. In the remaining three countries, the sample was stratified by region, and Germany and Italy also stratified by the degree of urbanization. In Finland, the sample also included mobile telephone numbers. Cases were weighted by age, sex and region in Sweden, Finland, Italy, and the UK to match the distributions in the population aged 18 to 54 years within those countries. In Germany and France, cases were weighted by age and sex only. A preliminary analysis of the weighted sample showed that there were differences in the age distribution between the national samples. The sample from Italy was slightly older than those of Sweden, Germany, Finland and France (p=0.0544). All analyses were conducted both with and without age standardization. Since there were no significant differences in the findings, the results reported here are not age-standardized. The choice to use non- standardized results was made to allow comparison with other papers from the study which provide further information on drinking patterns (Leifman 2001a), informal social control (Hemström, 2001) and adverse alcohol consequences (Ramstead 2001). Response Rates. Response rates differed widely between the countries. The net response rate has been calculated as the number of completed interviews divided by the net sample (including only the numbers where contact was made, an eligible respondent resided and the subject either participated or refused). The highest net response rate was obtained by Sweden (75.4%), followed by Finland (60.2%), France (53.9%), Italy (46.6%), the UK (41.4%) and Germany (41.1%). More conservative (gross) estimates of response rates are between 5 and 10% lower than the net estimates; for a more detailed presentation of the study response rates see Leifman (2001a). Measures. Given the multiple purposes of the survey, and its limited scope, only five items for this analysis could be included. For the first four, the response categories were: strongly agree, somewhat agree, somewhat disagree, strongly disagree. 1. It doesn't matter how much you drink as long as you don't show the effects. This item was conceived of as a test of cultural differences in III above. Respondents agreeing with the item were seen as indicating an expectation that behaviour while intoxicated should be the same as behaviour while sober. The hypothesis was that there would be greater agreement with this item in southern than in northern Europe. 2. If two friends get angry with each other while drinking, it should not affect their relationship when they see each other again. This item was seen as a test of IV above. Respondents agreeing with it are stating an expectation that behaviour while drinking should be accepted rather than sanctioned. To disagree is to disallow the notion of drinking marking a "time out" from sober life, at least when it comes to verbal aggression. It was hypothesized that there would

4 be more agreement with this in northern than in southern Europe. 3. When someone is drunk, they should not be considered as responsible for their actions as when sober. This item was seen as a test of VI above. The hypothesis was that there would be more agreement with this in northern than in southern Europe. Note, however, that this hypothesis is in conflict with the findings of Lindman and Lang (1994). 4. Anyone might become violent if they have too much to drink. This item was seen as a fairly direct statement of the cognitive recognition in V above. It was hypothesized that there would be more agreement with this in northern than in southern Europe. Here, the findings of Lindman and Lang (1994) were equivocal. 5. If a man gets drunk, enough so that he is slurring his words, how likely is it to make him violent -- very likely, somewhat likely, somewhat unlikely, or very unlikely? This item was seen as an alternative formulation testing V above, and thus hypothesized to receive more agreement in northern Europe. The purpose of including the phrase "enough so that he is slurring his words" was to establish a more or less constant level of physical incapacitation, to minimize the potential cultural differences in the meaning of "drunk" or "too much to drink". Unfortunately, the fieldwork agencies in Italy and France omitted the phrase in the final version of the translated instrument, so that the item in those samples becomes artefactually closer to the fourth item. We suspect that the omission of the phrase in these two countries may in itself be a relevant piece of data -- that the translators dropped the phrase because they did not see it as fitting well with the level of drinking implied in their translations of "drunk" ("ayant suffisamment bu" and "è ubriaca"). These questions were asked towards the end of a survey asking about amounts and patterns of drinking and of alcohol purchasing, and experience of problems related to drinking. Other variables used in the present analysis include three demographic variables: respondent's gender, age (trichomotized at 18 to 29, 30 to 49, and 50 to 64), and education (whether the respondent has a university education). Drinking pattern variables included whether the respondent drank the equivalent of a bottle of wine on an occasion at least once a month, whether the respondent drank at least twice a week, whether beer was the respondent's most frequent beverage, whether wine was the respondent's favourite, whether the respondent drank alcohol at lunch on any of the last seven days, whether in the same period the respondent drank at dinner, and whether in the same period s/he drank at a restaurant or bar or at home but not at a meal. Variables on experience with drinking problems included whether the respondent had tried in the last 12 months to influence a person in any of eight categories to drink less, and whether the respondent reported any of eight problems with their own drinking during the past 12 months. Analysis. Initially, results from inter-item correlations are presented for the five alcohol- aggression items, along with the results of a principal component factor analysis. Following this, results of bivariate chisquare and multivariate logistic regression analyses are presented for each item in succession. Multivariate regression analysis is used to explore the interplay of demographics, drinking patterns, experience of drinking problems, and country of residence in predicting the drinking attitudes items. In these analyses, all six national samples were treated as a single sample, to explore the extent to which national differences in attitudes went along with national differences in drinking patterns and other variables. Finland was taken as the reference category for the other countries, and ages 50-64 as the reference category for age. For all other variables, the reference category was the side of the variable not shown in the table. Three models are shown in each table: with the national site variable alone, with all other variables included in the analysis, and with both of these together. Other models were run (but are not shown) with only the demographic variables and only the demographics plus the drinking patterns; odds ratios in these were not very different from those in the combined analyses which are shown. Because of the variant wording on the last item in France and Italy, only the first four items are analyzed using multivariate techniques.

Results Patterns of correlation between the items. Pearson's correlations were computed in each sample for each pair between the five items (results not shown). In all six samples, there is a moderately high correlation between the fourth and fifth items. The correlation ranges between .22 for the United Kingdom and .33 for Finland where the "slurring" phrase was included; the correlations

5 are notably higher where it was omitted (.45 for Italy, .50 for France). As previously found in data from Canada (Paglia & Room, 1998), the correlation between the belief that alcohol causes violence and agreeing that a drunken person is less responsible for their actions was relatively low in all the samples. In none of these cultures does the belief that alcohol causes violence have much relation with the idea that drunkenness might serve as an excuse for bad behaviour. In Finland (.26) and to a lesser extent in Italy (.17), there was some correlation between accepting drunken irresponsibility and feeling that a display of anger while drunk should not affect a relationship. Interestingly, though, however logical this connection might seem, the relationship tailed away to almost nothing in the other samples. Other correlations between the items were relatively low -- .16 or less, and in many cases close to zero. A Principal Components factor analysis was performed in each sample. With a cut-off of Eigenvalue 1.0, two factors were found in each sample, one composed of the last two items (Eigenvalue 1.25 to 1.54), and the other of two or three of the first three items (Eigenvalue 1.11 to 1.23). There were some differences in the relationships of items to the factors. In the United Kingdom and Finland, the first factor was close to being a general factor, while in Sweden and France the first three items had negative relations with a factor composed of the last two items. In Finland, Germany, Italy and the United Kingdom, the last two items had negative loadings on the second factor, while the loadings were positive in France and Sweden. Despite these differences in detail, the results of the factor analyses suggest a considerable similarity in the interrelationship of responses in each sample. The regularity of the first factor reflecting the last two items and the second factor the first three was increased in the Varimax rotation. The correlation matrices and factor analyses suggest that the structure of relationships between the five items is quite similar in each of the six societies.

"It doesn't matter how much you drink so long as you don't show the effects". The first column of figures in Table 1 shows the percentage in each national sample reporting they "strongly" or "somewhat agree" with this item (excluding "no answers" from the base). The numbers in each table category, on which the percentages are based, can be found in Table 10. Only a minority of respondents agree with this item in any of the national samples. The two samples with the highest rate of agreement were Finland and Italy, while Germany and France showed the lowest rate of agreement. The hypothesis of a north-south gradient in reponse to this item was thus not supported. In Finland and Sweden, and to a lesser extent in France, men were more likely than women to agree with this item. This difference was accentuated among those who drank a bottle of wine or equivalent on an occasion at least once a month, with frequent heavy drinkers more likely to agree than less frequency heavy drinkers. In Finland, Sweden and Germany, younger and older respondents were more likely to agree with this item than middle-aged respondents, while in France, Italy and the U.K. there was little difference by age. There is no obvious way to interpret the pattern of responses. While the intention in framing the item was to emphasize the issue of norms on showing or cloaking effects from heavy drinking, it is possible that some respondents oriented their answers primarily around the first phase, "it doesn't matter how much you drink". In any case, the pattern of responses gives little indication of a systematic difference between northern and southern Europe on whether effects from heavy drinking should be kept cloaked. Table 2 shows results with the three logistic regression models for "It doesn't matter how much you drink as long as you don't show the effects". The results for the models with the national sites alone parallel those in Table 1. The third model shows that being male, drinking relatively heavily each month, and drinking in restaurants, bars, or not at mealtimes are associated with greater agreement with the attitude item, while being middle-aged is asssociated with less agreement. The odds ratios for the other variables are not significant. Results in the second model, controlling for national site, do not change much, though frequent drinking now shows marginal significance. The

6 impact of controlling for the added variables on the odds ratios for national sites is modest; responses from the UK and Sweden are no longer significantly different from responses from Finland, while Italy' divergence from Finland is modestly increased.

"If two friends get angry with each other while drinking, it should not affect their relationship while sober". Each national sample split fairly evenly in responses to this item (Table 3). The proportions agreeing with this are highest in Germany and the United Kingdom, and lowest in France and Finland. Again, the hypothesis of a clear north-south gradient is not supported. French and Italian males are somewhat more likely than females to agree with this item; elsewhere, there is little difference between the genders. Frequency of heavy drinking episodes is not associated with respondents agreement with this statement. In Finland, older respondents are more likely than younger to agree with this item, while in France, there is more agreement from younger respondents. Elsewhere, the differences by age are fairly small. Table 4 shows the results of parallel analyses for "If two friends get angry with each other while drinking, it should not affect their relationship while sober". As Table 3 would lead us to expect, respondents from Italy, Germany, the UK and Sweden are all more likely to agree with this than respondents from Finland, with the divergence particularly strong for Germany and the UK. In the total sample, males and those reporting one or more alcohol-related problems are more likely to agree, while younger and middle-aged respondents and the university educated are somewhat less likely to agree than their reference groups. These relations are largely unaffected in the combined model, although influencing others to cut down becomes marginally predictive on not agreeing. Controlling for these other variables makes little difference in the results by national site; the divergence of British from Finnish responses actually becomes stronger.

"When someone is drunk, they should not be considered as responsible for their actions as when they are sober". There is a considerable range in the balance of national responses to this item (Table 5), with French respondents least likely to agree (11%), while nearly half of Italian respondents (48%) agree. Respondents from the United Kingdom and Finland rank next in agreement, followed by respondents from Sweden and Germany. Women are somewhat more likely than men to agree concerning drunken irresponsibility in Finland and Italy, while in the other samples there is little difference by gender. Except to a limited extent among males and females in Italy and females in France, those who regularly drink heavily themselves have about the same mixture of attitudes as others. Older respondents in Finland, and older and middle-aged respondents in Italy, are more likely to agree tp drunken irresponsibility than younger respondents. Responses did not differ significantly by age in the other four samples. We thus did not find a clear north-south gradient in responses to this item. Our results agree with Lindman and Lang's (1994) in this, and in the relatively high rate of agreement by Italians that drunk persons are less responsible for their actions. Multivariate results for "When someone is drunk, they should not be considered as responsible for their actions as when they are sober" are shown in Table 6. As Table 5 would lead us to expect, respondents in Italy and the UK are more likely to agree with this than Finnish respondents, while French and German respondents are much less likely, and Swedish somewhat less likely, to do so. Those drinking relatively heavily each month, those drinking frequently (marginally), and those influencing others to cut down are somewhat more likely than others to agree with this item, while the younger and middle-aged, the university-educated, and those reporting their own alcohol problems are less likely to agree. Apart from reducing the influence of fequency of drinking and having alcohol problems to insignificance, adding in the national site dimension makes little difference to these results. Controlling for the other variables also makes little difference in the results by national site.

"Anyone might become violent if they drink too much". Agreement with this item was quite high (Table 7), ranging from 45% of Finnish respondents to 73% of Italian respondents. About two-thirds of British and French respondents agreed with this item, while a little over half of Swedish

7 and German respondents did. There was thus no clear north-south gradient, and the patterns of Finnish and Italian response were starkly opposite to the hypothesis. While Lindman and Lang (1994) also found no clear north-south gradient, the patterns of response of their Finnish and Italian respondents were fairly close. In all six samples, females were somewhat more likely than males to agree with this item. Heavier drinkers among males were less likely than other males to agree in Britain, German and France, but did not respond very differently in the other samples. Heavy drinking women were less likely to agree than other women in France, but somewhat more likely to agree in Italy. Differences in response by age were relatively small in most samples, although there was less agreement with the item among younger respondents in France and middle-aged respondents in Britain. In Table 8, multivariate results are shown for "Anyone might become violent if they drink too much". Respondents at all other national sites were more likely than Finnish respondents to agree with this, with a particularly sttrong divergence for Italy. Males, the university educated, those drinking relatively heavily in the last month, and those with their own alcohol problems are all less likely to agree with this statement than their reference groups, while those with wine as their most frequent beverage and those who have suggested others cut down are more likely to agree with it. Adding in the national site dimension did not change this picture much: favouring wine and having alcohol problems both became non-significant, while those who drink at dinner were now significantly less likely to agree with the statement. Controlling for these other variables made little difference to the results by national site, though the odds ratio for Italian respondents diminished a little.

"If a man gets drunk, (enough so that he is slurring his words), how likely is it to make him violent? -- very or somewhat likely". In the four samples where the parenthetical statement was included, slightly smaller proportions of respondents agreed with this item than with the previous one (see Table 9). In France and Italy, slightly larger proportions agreed with this item. Otherwise the patterns of response to this item very much resembled the patterns for the previous item. Again, Italians were most likely and Finns least likely to agree that violence after drinking was likely, and everywhere women were more likely to agree than men. There were minor differences in the age distributions, with lower agreement for older as well as middle-aged British respondents, and little difference in responses by age in France.

Discussion The answer to the question in our paper title is, "not easily". On none of the five items which we examined in this paper was there a clear north/south gradient. Agreement with the idea that "anyone might become violent if they drink too much" was quite strong in all six samples, but strongest in Italy and weakest in Finland -- against the direction which would fit with the findings of a greater role of alcohol in violence in northern Europe. Italian respondents, again, were the most likely to agree that a drunk person is not as responsible for their actions as a sober one -- with the greatest contrast her being provided by those from another "wine culture", France. In agreement with this, the French respondents are also least likely to agree that behaviour while drunk should not count between friends afterwards -- this time joined by the Finnish respondents. While agreement was not high with the idea that "it doesn't matter how much you drink as long as you don't show the effects", there was more agreement on this from Finland, Italy and the UK than elsewhere -- again, not a north/south split. Controlling for demographics, drinking patterns, and experience of alcohol problems did not change the general shape of these findings. How are these results to be understood? The very different response rates of the samples might make a difference. But the pattern of responses does not seem to fit the gradient between higher and lower response rates. That the responses in each country generally are not hugely different between different demographic categories also puts in question how much different response rates could explain. Another possibility is differences in the meanings of the items in the different languages and cultural contexts. This explanation loses strength from the fact that the general

8 structure of response in each of the samples, in terms of the relations between the items in correlation matrices and factor analyses, seems quite similar. Differences of nuance could be affecting the frequencies responding to each response category, but it seems unlikely that an item had a totally different meaning in one culture from another. Another piece of evidence concerning the validity of the findings is the fact that other comparative studies across Europe have also not found patterns which fit an easy north/south split. Lindman and Lang's results (1994) are closest to the topical area of this paper, but Pyorälä's comparison (1995) of Spain and Finland and the findings of the ESPAD studies (Andersson et al, 1999; Hibell et al, 2000) are also relevant -- as are the mixed findings from other parts of the present study (Leifman, 2001; Hemström, 2001; Ramstedt, 2001). One theme in current southern Europe discussions of drinking and drunkenness is that patterns there have been changing, and younger drinkers are drinking in what is often called a more "Anglo- Saxon" style. The studies by Pyorälä, Lindman & Lang, and ESPAD are all on young populations, and so have no way of testing this hypothesis. But the findings in Leifman's (2001) and the present analysis do not point to particular differences in responses between younger and older respondents in southern Europe, and raise some doubt about this way of understanding the picture. If methodological artefacts and cultural convergence are not sufficient to explain the patterning of our results, we are forced back onto difficult conceptual territory. We may need to start again to develop an understanding of how drinking norms and social control of drinking work in southern European cultures. From the present study and others there are several challenges to the picture of southern European drinking which is often presented in English-language and Scandinavian sources, which, as Olsson (1990) has pointed out, often has elements of a projected fantasy. It is often said that people in wine cultures learn to drink in a controlled way at an early age, at the family dinner table, and that because of this they don't get drunk. But the results from the ESPAD surveys and from Leifman (2001) cast doubt on this. People in southern Europe drink more frequently in between the heavier drinking occasions than in northern Europe, but there does not seem to be a systematic north/south split in terms of prevalences of fairly regular heavy drinking. Social control around drinking in wine cultures is also usually thought of as internalized, in terms of self-control. Yet, in the present data-sets it is the Finns and Swedes (along with respondents from the UK) who are most likely to report regretting things they've said or done while drinking (Ramstedt, 2001), and it is the Italians who are most likely to have tried to influence others to drink less (Hemström, 2001). Underlying many hypotheses about cultural differences in drinking is a half-hidden assumption that handling drinking and problematic drinking is more effortless and unself-conscious in some societies than in others -- and in particular, in the wine cultures. The data-set we have been analyzing calls this assumption in question. Heavy drinking occasions occur in southern and well as in northern Europe, and minimizing the harm from them does not seem to be not something that happens without attention or effort. We are still left with the puzzle of explaining the clear gradients between northern and southern Europe in the time-series analyses in terms of the attributable violence and trouble per litre of alcohol. Our analysis does not provide an answer to this puzzle. Part of the answer, indeed, may lie at a very simple level. Some of the southern European consumption is accounted for by the extra low-risk weekday occasions of drinking there (compared to Sweden or Finland), when an alcoholic beverage is drunk with a family meal or some like occasion. Perhaps, when these low-risk occasions are subtracted from the total volume of drinking, the trouble per litre is fairly similar between north and south in the remaining, heavier drinking occasions.

References Andersson, B., Hibell, B., Balakireva, O., Davidaviciene, A.G., Elekes, Z., Muscat, R., Nociar, A., Sabroe, S. and Veresies, K., (1999) The concept of "drunkenness" among young people in eight major European cities, presented at the 25th Annual Alcohol Epidemiology Symposium, Kettil Bruun Society, 31 May - 4 June. Cameron, D., Thomas, M., Madden, S., Thornton, C., Bergmark, A., Garretsen, H. and Terzidou, M. (2000) Intoxicated across Europe: in search of meaning, Addiction Research 8:233-242. Christie, N. (1965) Scandinavian experience in legislation and control. In: National Conference on

9 Legal Issues in Alcoholism and Alcohol Usage, pp. 101-122. Boston University Law- Medicine Institute, Boston. Cottino, A., and Morgan, P. (1985) Italy. Pp. 83-92 in: M. Grant, ed., Alcohol Policies. Copenhagen: WHO Regional Publication, European Series No. 18. Harrison, B.H. (1971) Drink and the Victorians: The Temperance Question in England 1815-1872. Pittsburgh: University of Pittburgh Press. Hemström, Örjan (2001) Informal alcohol control in six European countries, presented at a conference on Alcohol Consumption and Harm and Policy in the European Union, Public Health Institute, Stockholm, 17-18 March. (http://www.fhi.se/pdf/Hemstrom2.pdf) Hibell, B., Andersson, B., Ahlström, S., Balakireva, O., Bjarnason, T., Kokkevi, A. and Morgan, M. (2000) The 1999 ESPAD Report: Alcohol and Other Drug Use among Students in 30 European Countries. Stockholm: CAN. Johansson, L. (2000) Sources of the Nordic solutions, pp. 17-43 in: P. Sulkunen, C. Surtton, C. Tigerstedt and K. Warpenius, eds., Broken Spirits: Power and Ideas in Nordic Alcohol Control. Helsinki: Nordic Council for Alcohol and Drug Research, NAD Publ. No. 39. Leifman, H. (2001a) A comparative analysis of drinking habits in six EU countries in the year 2000, presented at a conference on Alcohol Consumption & Harm & Policy in the European Union, Public Health Institute, Stockholm, 17-18 March. (http://www.fhi.se/pdf/leifmanecsur4.pdf) Leifman, H. (2001b) Homogenisation in alcohol consumption in the European Union, Nordic Studies on Alcohol and Drugs 19 (English Supplement):15-30. Lenke, L. (1990) Alcohol and Criminal Violence -- Time Series Analyses in a Comparative Perspective. Almqvist & Wiksell, Stockholm. Levine, H.G. (1983) The good creature of God and the Demon Rum: colonial American and 19th century ideas about alcohol, crime, and accidents. Pp. 111-161 in: R. Room and G. Collins, eds., Alcohol and Disinhibition: Nature and Meaning of the Link. Rockville, MD: NIAAA, Research Monograph No. 12. Levine, H. (1992) Temperance cultures: concern about alcohol problems in Nordic and English- speaking cultures, pp. 15-36 in: M. Lader, G. Edwards & D.C. Drummond, eds., The Nature of Alcohol and Drug Related Problems. Oxford etc.: Oxford University Press. Lindman, R.E. and Lang, A.R. (1994) The alcohol-aggression stereotype: a cross-cultural comparison of beliefs, International Journal of the Addictions 29:1-13. MacAndrew, C. and Edgerton, R. (1969) Drunken Comportment. Chicago: Aldine. Norström, T. (1988) Alcohol and suicide in Scandinavia, British Journal of Addiction 83:553-559. Norström, T. (1995) Alcohol and suicide: a comparative analysis of France and Sweden, Addiction 90:1463-1469. Norström, T., ed. (2001) "Alcohol and mortality: the post-war experience in the EU countries", Addiction 96, Supplement 1, February. Norström, T., Hemström, Ö., Ramstedt, M., Rossow, I., and Skog, O.-J., Mortality and population drinking, pp. 84-100 in: Norström, T., ed., Alcohol in Postwar Europe: Consumption, Drinking Patterns, Consequences and Policy Responses in 15 European Countries. Copenhagen: WHO Regional Office for Europe, February. Olsson, B. (1990) Alkoholpolitik och alkoholens fenomenologi. Alkoholpolitik – Tidskrift för nordisk alkoholforskning 7:184-194. Paglia, A. and Room, R. (1998) Alcohol and aggression: general population views about causation and responsibility, Journal of Substance Abuse 10:199-216. Prestwich, P. (1988) Drink and the Politics of Social Reform: Antialcoholism in France since 1870. Palo Alto, Calif.: Society for the Promotion of Science and Scholarship. Pyorälä, E. (1995) Comparing drinking cultures: Finnish and Spanish drinking stories in interviews with young adults, Acta Sociologica 38:217-229. Ramstedt, Mats (2001) Alcohol consumption and the experience of adverse consequences -- a comparison of six European countries, presented at a conference on Alcohol Consumption and Harm and Policy in the European Union, Public Health Institute, Stockholm, 17-18 March. (http://www.fhi.se/pdf/Ramstedt.pdf) Roberts, J.S. (1984) Drink, Temperance, and the Working Class in Nineteenth Century Germany. Boston: Allen and Unwin.

10 Room, R. (1989) Responses to alcohol-related problems in an international perspective: characterizing and explaining cultural wetness and dryness. Presented at an international conference, San Stefano Belbo, Italy. (http://www.bks.no/response.htm) Room, R. (1998) "Nobody poured it down their throat": popular views on responsibility for intoxicated behaviour, presented at the 38th International Congress on Alcohol, Drugs and Other Dependencies, Vienna, Austria, August 16-20. Room, R. Janca, A., Bennett, L.A., Schmidt, L. and Sartorius, N., with others (1996) WHO cross- cultural applicability research on diagnosis and assessment of substance use disorders: an overview of methods and selected results, Addiction 91:199-220. Room, R. and Mäkelä, K. (2000) Typologies of the cultural position of drinking, Journal of Studies on Alcohol 61:475-483. Simpura, J. (2001) Trends in alcohol consumption and drinking patterns: sociological and economic explanations and alcohol policies, Nordic Studies on Alcohol and Drugs 18 (English Supplement): 3-13. Simpura, J. and Karlsson, T. (2001) Trends in drinking patterns among adult population in 15 European countries, 1950 to 2000: a review, Nordic Studies on Alcohol and Drugs 18: (English Supplement): 31-48.

11 Table 1. “It doesn’t matter how much you drink as long as you don’t show the effects”. Percent Responding “Strongly ” or “Somewhat Agree” in Six Country Sub-Samples by Sex, Age, and Regular Heavy Drinking

Country Total a Sex b Age Group (years) b Heavy Drinking c **** Male Female 18-29 30-49 50-64 Males Females % % % % % % % % Finland 29 36 21 **** 38 21 32 **** 42 ** 24 Sweden 22 27 16 **** 27 20 19 * 36 *** 11 *

UK 24 25 22 26 22 26 27 24

Germany 17 18 15 18 13 22 ** 22 22

France 16 19 13 * 15 15 19 28 ** 27

Italy 25 26 24 23 24 28 28 29

* p<=0.05, ** p<=0.01, *** p<=0.001, ****p<=0.0001 a The significance test is for the variation in % agreement across country. b Significance is tested in each national sample between males and females, and across age group in % agreeing with the statement. c A indicator of significance in the male or female column of heavy drinking indicates that those who drink heavily once a month or more (an equivalent to a bottle of wine at a sitting or more) are significantly different from those who do not drink heavily in their likelihood of agreeing with the statement.

Table 2. “It doesn’t matter how much you drink as long as you don’t show the effects” Multivariate Odds of Agreement with the Statement

Model 1 Model 2 Model 3 OR 95% CL OR 95% CL OR 95% CL Sex Male (vs. Female) 1.30 1.11, 1.50 1.27 1.09, 1.47 Age Group 18 to 29 (vs. 50-64) 1.03 0.85, 1.24 0.96 0.80, 1.15 30 to 49 (vs. 50-64) 0.75 0.63, 0.88 0.73 0.62, 0.86 University Education Yes (vs. No) 0.63 0.52, 0.74 0.64 0.53, 0.75 Bottle of wine/equiv. on an occasion Monthly + (vs. Less) 1.43 1.19, 1.70 1.60 1.35, 1.89 Frequency of drinking 2+ times/week (vs. Less) 1.21 1.02, 1.41 1.14 0.97, 1.32 Beer is favourite 1.10 0.93, 1.28 1.10 0.94, 1.28 Wine is favourite 0.94 0.79, 1.11 0.88 0.74, 1.04 Drink at lunch 1.19 0.99, 1.42 1.11 0.92, 1.32 Drink at dinner 0.93 0.79, 1.09 0.91 0.77, 1.06 Drink: rest., bar, non-meal 1.11 0.94, 1.29 1.17 1.00, 1.35 # People Influenced to cut down 0.94 0.85, 1.02 0.94 0.85, 1.03 Experienced 1+ alcohol Problems 0.93 0.76, 1.11 0.92 0.79, 1.06 Country (vs. Finland) France 0.47 0.37, 0.58 0.51 0.39, 0.65 Italy 0.83 0.67, 1.00 0.68 0.52, 0.89 Germany 0.50 0.39, 0.61 0.60 0.46, 0.77 UK 0.78 0.63, 0.95 0.86 0.68, 1.08 Sweden 0.69 0.56, 0.84 0.81 0.62, 1.03

12 Table 3. “If two friends get angry with each other while drinking, it should not affect their relationship while sober”. Percent Responding “Strongly ” or “Somewhat Agree” in Six Country Sub-Samples by Sex, Age, and Regular Heavy Drinking

Country Total Sex Age Group (years) Heavy Drinking **** Male Female 18-29 30-49 50-64 Males Females % % % % % % % % Finland 43 44 42 37 40 54 *** 42 44 Sweden 52 54 50 51 51 55 51 43

UK 63 65 61 63 60 67 68 64

Germany 60 62 59 55 63 61 60 66

France 39 44 35 ** 46 37 35 ** 48 39

Italy 56 60 52 ** 57 53 60 60 51

* p<=0.05, ** p<=0.01, *** p<=0.001, ****p<=0.0001 a The significance test is for the variation in % agreement across country. b Significance is tested in each national sample between males and females, and across age group in % agreeing with the statement. c A indicator of significance in the male or female column of heavy drinking indicated that those who drink heavily once a month or more (an equivalent to a bottle of wine at a sitting or more) are significantly different from those who do not drink heavily in their likelihood of agreeing with the statement.

Table 4. “If two friends get angry with each other while drinking, it should not affect their relationship while sober” Multivariate Odds of Agreement with the Statement

Model 1 Model 2 Model 3 OR 95% CL OR 95% CL OR 95% CL Sex Male (vs. Female) 1.22 1.07, 1.37 1.19 1.06, 1.34 Age Group 18 to 29 (vs. 50-64) 0.88 0.74, 1.02 0.82 0.70, 0.95 30 to 49 (vs. 50-64) 0.87 0.75, 0.99 0.87 0.75, 0.99 University Education Yes (vs. No) 0.78 0.68, 0.89 0.80 0.70, 0.90 Bottle of wine/equiv. On an occasion Monthly + (vs. Less) 0.95 0.81, 1.10 0.95 0.82, 1.10 Frequency drinking 2+ times/week (vs. Less) 1.05 0.92, 1.20 1.11 0.97, 1.25 Beer is favourite 1.04 0.91, 1.17 1.06 0.93, 1.20 Wine is favourite 1.03 0.89, 1.17 1.08 0.94, 1.22 Drink at lunch 0.91 0.78, 1.05 0.87 0.74, 0.99 Drink at dinner 0.98 0.86, 1.12 1.50 0.92, 1.19 Drink: rest., bar, non-meal 1.05 0.92, 1.18 1.10 0.97, 1.24 # People Influenced to cut down 0.93 0.86, 0.99 0.96 0.89, 1.03 Experienced 1+ alcohol problems 1.26 1.07, 1.47 1.37 1.21, 1.54 Country (vs. Finland) France 0.86 0.72, 1.02 0.97 0.79, 1.18 Italy 1.70 1.42, 2.02 1.86 1.48, 2.33 Germany 2.01 1.68, 2.40 1.98 1.59, 2.44 UK 2.21 1.84, 2.64 2.47 2.01, 3.02 Sweden 1.43 1.19, 1.70 1.34 1.08, 1.66

13 Table 5. “When someone is drunk, they should not be considered as responsible for their actions as when they are sober”. Percent Responding “Strongly ” or “Somewhat Agree” in Six Country Sub-Samples by Sex, Age, and Regular Heavy Drinking

Country Total Sex Age Group (years) Heavy Drinking **** Male Female 18-29 30-49 50-64 Males Females % % % % % % % % Finland 31 28 35 *** 29 27 41*** 29 41 Sweden 23 24 22 25 20 25 24 19

UK 36 35 36 37 33 38 34 40

Germany 20 20 19 22 20 16 23 22

France 11 11 10 9 12 12 14 19

Italy 48 45 51 *** 37 50 54 *** 54 * 59

* p<=0.05, ** p<=0.01, *** p<=0.001, ****p<=0.0001 a The significance test is for the variation in % agreement across country. b Significance is tested in each national sample between males and females, and across age group in % agreeing with the statement. c A indicator of significance in the male or female column of heavy drinking indicated that those who drink heavily once a month or more (an equivalent to a bottle of wine at a sitting or more) are significantly different from those who do not drink heavily in their likelihood of agreeing with the statement.

Table 6. “When someone is drunk, they should not be considered as responsible for their actions as when they are sober” Multivariate Odds of Agreement with the Statement

Model 1 Model 2 Model 3 OR 95% CL OR 95% CL OR 95% CL Sex Male (vs. Female) 0.92 0.79, 1.05 0.91 0.79, 1.03 Age Group 18 to 29 (vs. 50-64) 0.80 0.66, 0.95 0.76 0.64, 0.90 30 to 49 (vs. 50-64 0.85 0.72, 0.99 0.86 0.74, 1.00 University Education Yes (vs. No) 0.70 0.59, 0.81 0.70 0.60, 0.81 Bottle of wine/equiv. on an occasion Monthly + (vs. Less) 1.24 1.04, 1.47 1.41 1.20, 1.65 Frequency drinking 2+ times/week (vs. Less) 1.06 0.91, 1.23 1.17 1.01, 1.34 Beer is favourite 0.87 0.75, 1.01 0.86 0.74, 0.98 Wine is favourite 1.01 0.85, 1.18 0.98 0.84, 1.14 Drink at lunch 1.18 0.99, 1.39 1.14 0.97, 1.34 Drink at dinner 0.93 0.79, 1.08 0.91 0.78, 1.05 Drink: rest., bar, non-meal 1.00 0.89, 1.19 1.09 0.94, 1.24 # People Influenced to cut down 1.13 1.04, 1.22 1.19 1.10, 1.28 Experienced 1+ alcohol problems 0.97 0.80, 1.15 0.80 0.69, 0.91 Country (vs. Finland) France 0.26 0.20, 0.33 0.29 0.22, 0.37 Italy 2.04 1.69, 2.44 1.97 1.55, 2.51 Germany 0.54 0.44, 0.66 0.66 0.51, 0.83 UK 1.21 1.00, 1.46 1.29 1.04, 1.60 Sweden 0.65 0.53, 0.79 0.76 0.59, 0.96

14 Table 7. “Anyone might become violent if they drink too much”. Percent Responding “Strongly ” or “Somewhat Agree” in Six Country Sub-Samples by Sex, Age, and Regular Heavy Drinking

Country Total Sex Age Group (years) Heavy Drinking **** Male Female 18-29 30-49 50-64 Males Females % % % % % % % % Finland 45 40 51 *** 50 43 46 43 * 48 Sweden 54 48 60 **** 52 55 55 43 65

UK 66 62 70 ** 72 60 71 *** 55 ** 70

Germany 53 49 57 * 57 54 47 * 36 ** 52

France 65 60 71 *** 59 67 71 ** 46 *** 54 *

Italy 73 70 76 * 67 75 74 71 87 *

* p<=0.05, ** p<=0.01, *** p<=0.001, ****p<=0.0001 a The significance test is for the variation in % agreement across country. b Significance is tested in each national sample between males and females, and across age group in % agreeing with the statement. c A indicator of significance in the male or female column of heavy drinking indicated that those who drink heavily once a month or more (an equivalent to a bottle of wine at a sitting or more) are significantly different from those who do not drink heavily in their likelihood of agreeing with the statement.

Table 8. “Anyone might become violent if they drink too much” Multivariate Odds of Agreement with the Statement

Model 1 Model 2 Model 3 OR 95% CL OR 95% CL OR 95% CL Sex Male (vs. Female) 0.73 0.64, 0.82 0.75 0.66, 0.84 Age Group 18 to 29 (vs. 50-64) 1.02 0.86, 1.19 1.07 0.91, 1.25 30 to 49 (vs. 50-64) University 0.97 0.84, 1.11 0.99 0.86, 1.13 Education Yes (vs. No) 0.68 0.59, 0.77 0.66 0.58, 0.75 Bottle of wine/equiv. on an occasion Monthly + (vs. Less) 0.79 0.67, 0.91 0.80 0.69, 0.92 Frequency of drinking 2+ times/week (vs. Less) 0.90 0.78, 1.03 1.04 0.91, 1.18 Beer is favourite 0.97 0.85, 1.10 0.92 0.80, 1.04 Wine is favourite 1.11 0.96, 1.27 1.16 1.01, 1.33 Drink at lunch 1.08 0.92, 1.25 1.13 0.97, 1.30 Drink at dinner 0.88 0.76, 1.00 0.95 0.83, 1.08 Drink: rest., bar, non-meal 0.96 0.84, 1.09 0.92 0.81, 1.04 # People Influenced to cut down 1.13 1.04, 1.22 1.16 1.07, 1.25 Experienced 1+ alcohol problems 0.87 0.73, 1.02 0.70 0.62, 0.79 Country (vs. Finland) France 2.26 1.88, 2.70 2.10 1.70, 2.57 Italy 3.19 2.64, 3.84 2.68 2.11, 3.39 Germany 1.35 1.12, 1.60 1.39 1.12, 1.72 UK 2.32 1.93, 2.77 2.45 1.99, 3.01 Sweden 1.41 1.18, 1.67 1.45 1.16, 1.79

15 Table 9. “If a man gets drunk, (enough so that he is slurring his words), how likely is it to make him violent”. Percent Responding “Strongly ” or “Somewhat Agree” in Six Country Sub-Samples by Sex, Age, and Regular Heavy Drinking

Country Total Sex Age Group (years) Heavy Drinking **** Male Female 18-29 30-49 50-64 Males Females % % % % % % % % Finland 39 36 43 * 43 35 42 29 ** 42 Sweden 42 38 46 ** 43 40 44 30 ** 44

UK 63 59 66 * 69 62 59 54 * 62

Germany 42 38 48 ** 48 42 37 * 30 * 44

France 71 66 76 *** 71 73 68 51**** 71

Italy 84 81 86 * 82 87 80 * 86 93

* p<=0.05, ** p<=0.01, *** p<=0.001, ****p<=0.0001 a The significance test is for the variation in % agreement across country. b Significance is tested in each national sample between males and females, and across age group in % agreeing with the statement. c A indicator of significance in the male or female column of heavy drinking indicated that those who drink heavily once a month or more (an equivalent to a bottle of wine at a sitting or more) are significantly different from those who do not drink heavily in their likelihood of agreeing with the statement.

Table10. Base Numbers for Odd Numbered Tables (weighted n)

Country Total N Sex Age Group (years) Heavy Drinking Male Female 18-29 30-49 50-64 Males Females n n n n n n n Finland 1004 512 492 273 462 269 250 81 Sweden 989 505 484 303 396 290 184 58

UK 984 501 483 255 466 263 254 139

Germany 1004 519 485 272 472 260 115 36

France 995 496 499 331 417 247 119 35

Italy 1000 498 502 249 421 330 104 58

16

Recommended publications