This paper is available in Estonian, here, translated by Adrian Pantilimonu, and available in French, here, translated by Mathilde Guibert.

      Cultural control of drinking by universal moderation or differential access 

      Ron Roizen*

      Paper presented at the annual meeting of the Society for Cross-Cultural Studies, Westin Hotel, El Paso, Texas, August, 1988.

      *Alcohol Research Group, Institute of Epidemiology and Behavioral Medicine, Medical Research Institute of San Francisco, 1816 Scenic Ave., Berkeley, CA 94703, U.S.A.

      The prospects for a new generation of cross-cultural hypotheses and analyses of drinking behavior quietly improved a good deal over the past decade because of the emergence of a new and still largely untapped store of data.  These data come from a growing series of broadly comparable social surveys focused on drinking behavior and carried out in a variety of cultural settings.  The trend began in 1970-1979 with, first, a four-country Scandinavian drinking survey carried out in Iceland, Norway, Sweden, and Finland (Makela, 1984, 1986) and, second, a three-culture study coordinated by the World Health Organization and carried out in community-level samples in Mexico, Scotland, and Zambia (Rootman, 1983).  The WHO study -- called the "WHO Community Response to Alcohol-Related Problems" project -- spawned two additional comparable and nearly contemporaneous surveys in Canada (Smart, 1980) and in the U.S. (Roizen, 1981).  More recently, surveys incorporating at least part of the WHO questionnaire have been done in Spain (Caetano and Martinez, 1987), in Japan (Clark, 1986), in a second region of Mexico (Caetano and Medina-Mora, 1986), and among Spanish-surname subgroups in the U.S. general population (Caetano, 1986).  An August 1986 conference in Washington, D.C. -- which drew participants from researchers involved in the original WHO project, from researchers at work on subsequent WHO-like projects, and from others interested in this emerging  resource -- bespeaks the continuing vitality of this new tradition of research.

      These studies offer a tonic to cross-cultural research not only because they are new and unfamiliar but because survey data bring a new kind of information to the territory, thus forcing new modes of thinking upon us.  And although these survey data are not as richly textured as ethnographic reports nor as comprehensive in their coverage of the world's cultures as the Human Relations Area Files, they nevertheless offer the field a valuable new window on cross-cultural variation and a methodology that nicely complements the more familiar brands of cross-cultural data.  For example, survey data almost invariably require an observer/analyst interested in cross-cultural variation to pay equal heed to variation within each cultural setting as well.  Cross-cultural hypotheses trying to make their way in this sort of data environment must somehow manage to incorporate both cross-cultural and within-cultural variation into their explanatory agenda -- no small duty.

      Today, I would like to illustrate some of the charms of this sort of data by drawing on WHO-study data from the three original settings and the U.S. data, which in fact derive from a survey of Contra Costa County California in 1979.  I have two goals.  One is to illustrate how ordinary survey data can be used to examine structural and normative issues, issues that are usually reserved for other kinds of data and analysis styles.  The other is to offer one or two conjectures on cross-cultural theory borrowing from these same data.  Before we begin, however, a word or two should be said about the data sources.  My survey data comes from seven separate samples:  one in California, one representing the Lothian (Edinburgh) region of Scotland, two representing two kinds of communities -- one urban and the other rural -- in Mexico, and, finally, three samples representing an urban residential, periurban, and rural settlement in Zambia.  Technically speaking, it is problematic to combine the samples drawn within a given country into a single sample, but that has been done nevertheless.  Similarly, you should be aware that these data do not reflect the national populations of these settings but merely specific communities therein.  Data, moreover, have been weighted everywhere besides the U.S. in order to bring the samples into equality across the genders.

      Cultural Permission To Drink:

      WHO respondents in Scotland, Mexico, Zambia, and the U.S. were asked to estimate much drinking would be "all right" for males and for females of four specific, hypothetical or test ages, 16, 21, 40, and 60.  Four pre-coded answers were offered:  (1) no drinking, (2) one or two drinks only, (3) enough to feel the effects but not drunk, and (4) "getting drunk is sometimes alright."  The data I will be presenting are the frequency distributions for the whole samples in each setting; they tell us not the answers of males or of females regarding these age/sex statuses but instead the answers of the whole population about males and about females.

      To begin, let us consider the normative patterns relating to the cultural control of access to drinking in general, a question we can address by initially collapsing together all three pro-drinking response categories.  The Scottish (see Chart 1) and U.S. (Chart 2) "permission" patterns were very similar.  Both showed little difference in permission rates between the sexes (though females ran slightly below males), and both suggested an age-plateau form with high -- and even nearly universal -- permission granted for the three test ages twenty-one and over.  Simple as they are, these curves seem to hint at three kinds of connections between drinking norms and the wider cultural environment -- these links appearing as cultural premiums (1) on majority age, (2) on universalism or egalitarianism, and (3) on legalism.  Concerning majority age, the normative pattern we see is consistent with a high cultural premium on a single, consensual, and officially-sanctioned "threshold" age of majority, one governing the right to drink as well as other adult prerogatives.  The flatness of the curves across the three "adult" ages and the similarity of the curves for the sexes suggests a high cultural premium on universalism or egalitarianism.  Finally, I see legalism hinted in the close agreement between popular opinion and official laws governing the right to drink.  Legalism, in turn, suggests that the state holds and articulates the issue of cultural approval for drinking in the Scottish and U.S. settings.

      The Mexican (Chart 3) and Zambian (Chart 4) permission patterns were quite different both from each other and from the universalistic and age-plateaued pattern of the Scottish/U.S. data.  The Mexican data showed a substantial divide in cultural permission across the sexes.  The Zambian data, on the other hand, showed wide variation in permission across test ages.  Notice, for example, that all four female age-statuses in the Mexican sample are accorded lower rates of cultural permission than three of the four male age-statuses (Chart 3).

      In Chart 6 I have ordered the Zambian results from highest to lowest cultural permission level (first 40-year-olds, next 60-, next 21-, and, finally, 16-year-olds).  Notice in this chart that the results follow a clear-cut pattern of age stratification:  40-year-old males and females ranked 1st and 2nd respectively in cultural permission levels; 60-year-old males and females ranked 3rd and 4th; 21-year-old males and females ranked 5th and 6th; and, finally, 16-year-old males and females ranked next-to-last and last.  To be sure, no matter what the age-status, Zambian males were accorded higher levels of permission than females (even considerably higher for 21-year-olds), but more importantly, females in each age-status group were consistently accorded a permission level at least as high as that accorded males in the next-lower test-age group.  Permission levels in the Zambian case, then, can fairly be said to be stratified primarily by age whereas permission in the Mexican case was stratified primarily by sex (see Chart 5).  In both the Mexican and the Zambian cases, moreover, permission to drink appears to involve a more gradual or graded cultural license in place of the sharply divided pattern at majority-age suggestion in the Scottish and U.S. permission norms.

      The Mexican and Zambian normative patterns suggested neither the majority-age premium, nor the universalism, nor the legalism hinted in the Scottish/U.S. pattern.  On the contrary, in both Mexico and Zambia the data reveal that the right to drink is much more strongly tied to one's ascribed status, or, in other words, to bases of social stratification more characteristic of traditional societies.  Mexican and Zambian permission norms, then, seem to have 'stained-out' (to use Duster's [1983, p. 326] evocative term) the ascribed-status bases of each society's broader social stratification systems, much as a biologist's dye stains tissue on a microscope slide.  Mexico's gender-based and Zambia's age-based stratification systems are not in and of themselves particularly surprising.  Mexico, of course, is well known for its machismo men and modest women, as Africa is known as home to many age-based systems of social stratification.  Yet the striking difference between the two bases of stratification -- gender and age -- raise interesting questions.  What are the chief differences between ascription based on age and on gender (La Fontaine, 1978)?  How is it that permission-to-drink norms reflect the underlying architecture ascribed stratification in each cultural setting?  What does the apparent parallelism between drinking permission and stratification tell us, on the one hand, about drinking and, on the other, about stratification?  What are the chief differences in drinking practices, attitudes, and problems between universalistic and ascriptive social structures?

      Finer-grain puzzles are also suggested in these findings.  What does it mean, for example, that in Mexico 21-year-olds (of either sex) were granted higher permission levels than like-sexed 60-year-olds, whereas in Zambia the reverse was true -- 60-year-olds were granted higher levels 21-year-olds? Did the gradual, non-plateaued form of Mexican and Zambian permission norms imply that adulthood itself is likewise defined in an incremental or cumulative way? Why did the norms applying to 60-year-olds decline in levels of permission from those of 40-year-olds in Mexico and in Zambia as well as (if only in a hinted-at form) in the Scottish and U.S. data?

      Permission to Drink and the Power of the State:

      If the permission-to-drink normative patterns hinted at different levels of involvement by the state in the regulation of drinking, then just such a relationship was unexpectedly lent support elsewhere in the WHO data, once more nicely illustrating the potential for survey data to address structural societal variables. As the study's title suggests, the WHO researchers were interested in community responses to alcohol-related problems. Its survey questions in this territory happened to include a series of four vignettes about hypothetical alcohol-related dilemmas. This series provided data on the perceived severity of alcohol-related problems in the community, on perceptions of such problems' perceived prevalences, and -- most relevant here--on preferred kinds of social control responses. (Data on only three of the four vignettes will be considered here.)    The first vignette described a man who had drunk so much that "he falls down on the road and cannot get up" (MANFALL); the second described a man who has on two occasions struck his wife when he was drunk (HITWIFE); the third described a man "who spends so much money on drinking that there is not enough food for his family" (NOFOOD).

      Included in the battery of questions asked about each vignette were three questions asking respondents to evaluate the appropriateness of the involvement of various people or agencies in helping or controlling the man in question. For example, the questions asked (1) whether the man's relatives ought to help him, (2) whether a bystander on the street should do something, and (3) whether the police or appropriate authorities should be called in.

      This series was introduced into the survey questionnaire in order to monitor norms and attitudes regarding the social handling of one or another sort of drinking-related problem.  Yet its results also painted a clear picture of the relative ascendancy of the differential involvement of the state -- represented by the "police/authorities" involvement -- across the Scottish, Mexican, and Zambian samples. Thus, it is worth briefly examining the different Patterns of control reported in these three setting.  In the Zambian samples (see Chart 14A) roughly 90 percent agreed that relatives should help in all three vignette situations.  Concerning bystander involvement, 90 percent supported bystander involvement for the MANFALL situation and from 75-80 percent supported it for the HITWIFE and  NOFOOD situations. Less than half the samples supported the involvement of the police or appropriate authorities in the MANFALL and HITWIFE situations, but almost three-quarters supported such involvement in the NOFOOD situation. Involvement by police/authorities ranked third -- behind relatives and bystanders -- for all three situations. The Scottish results (Chart 14B) were quite different.  There, the vignette situations were seen as highly differentiated and sorted into two groups -- the MANFALL situation, on the one hand, and the HITWIFE and NOFOOD situations, on the other. For all situations, involvement by relatives was supported by 80-85 percent of the sample. A like percent supported bystander intervention in the MANFALL situation, but involvement by police/authorities in this situation was supported by less than half the sample.  The HITWIFE and NOFOOD situations -- both of which involve third-party victims of the central actors drinking -- were not seen as appropriate occasions for the involvement of bystanders by roughly three-quarters of the sample; in these, police/authorities was approved by over three-quarters of the sample. The patterns suggest an alternativity relationship between bystander and police/authorities involvements and a clear division of responsibility between the benevolent interventions (of a bystander) and the potentially punitive interventions (of police/authorities) -- with state involvement contraindicating involvement by the citizen on the street. Clearly, the state's monopoly of police power and the differentiation and definition of its role is much greater in the Scottish than in the Zambian case. Interestingly, the Mexican pattern of responses (Chart 14C) appeared to fall between the Scottish and Zambian patterns.

      Charts 14D, E, & F present the same data as 14A, B, & C this time contrasting each cultural setting with respect to a single vignette situation. Chart 14D reports the MANFALL situation, in which all three cultural settings reveal quite similar response patterns -- high on relative and bystander involvement and low on police/authorities involvement. In the HITWIFE situation, however, Chart 14E throws the issue of state control in the Scottish setting into sharp relief.  Here we see that the Zambian response pattern to HITWIFE is not much different from the response pattern to MANFALL.  The Scottish pattern shows a clear preference for the involvement of police/authorities and only low support for bystander involvement.  In the Mexican samples bystander and police/authorities involvements are roughly equally supported by about two-thirds of the sample. Finally, in the NOFOOD situation (Chart 14F), the Zambian pattern shows more support for police/authorities involvement, but such involvement appears to be added to high support levels for involvement by bystanders rather than substituting for it. In the Scottish and (though less so) in the Mexican case, as noted already, involvement by police/authorities is perceived more as an alternative to bystander involvement.  These patterns, then, give us a clear if only cursory glimpse of the normative expectations surrounding relatives, citizens on the street, and civil authorities or police in regard to exemplary problematic situations.  They illustrate the marked difference in the expectations regarding state control across the three cultural settings examined and lend a measure of support to the idea that the growth of the state's power is probably structurally associated with the universalistic normative patterns evidenced in the earlier permission-norm data we examined.

      Cultural Permission To Drink Enough to Feel Alcohol's Effects:

      Though the Scottish and U.S. norms governing access to drinking suggested a universalist form, it is well known, of course, that even in these settings drinking behavior is strongly associated with (or marked by) gender and age variables.  Indeed, the WHO data from this same study reaffirmed this long-standing truth.  For example, fully 80-85 percent of Scottish and U.S. young males (18-29) reported drinking at least as often as once a week; whereas among older females (50+) the proportion drinking that frequently dropped to only 25-35 percent.  Concerning such relationships, the draft WHO monograph noted that "when we examine frequent drinking, in all four cultural settings the distributions for males of all ages fall entirely above the distributions of their female compatriots."  What light might the normative data shed the normative arrangements for heavier drinking?  Let us return data from the age/sex drinking norms series, this time with that series' responses dichotomized so as to divide  respondents who would permit "feeling the effects" or sometimes getting drunk" to a hypothetical status from those who suggested "no drinking" or only one or two drinks maximum.  This dichotomy will tell us more about the normative structuring of higher-quantity drinking in each setting.

      The Scottish (Chart 7) and American (Chart 8) patterns are interesting both for their similarities and their differences. First, it is apparent in both that permission to drink is not equivalent to permission to drink enough to feel alcohol's effects. In both sites, too, the permission distributions now taken on more shape -- that is to say, deviated from the "flat plateau" age-structure evidenced in the earlier permission-to-drink curves. This suggests more of a culturally sanctioned age-based hierarchy with respect to higher-quantity drinking. In both settings and for both sexes, moreover, the apparent age-hierarchies evidenced in this new shape are the same: first, 40-year-olds, next 21-year-olds, next 60-year-olds, and, finally, 16-year-olds -- this, a status with very low approval for higher-consumption drinking. Interestingly, the age-based hierarchies for both Scottish and U.S. females reveal a wider gap in permission levels between 40-year-olds and 21-year-olds than occurs between 21-year-olds and 60-year-olds. The same relative spacing is in evidence for Scottish males, if more weakly, but disappears for U.S. males.

      Perhaps the most striking result of the Scottish and U.S. comparison, however, is the higher permission levels enjoyed by Scottish males in contrast with either Scottish females, U.S. males, or U.S.  females. To take a More than 60 percent of the Scottish sample approved feeling alcohol's effects for a  40-year-old male, whereas the comparable approval levels for 40-year-old Scottish females was a little less than 50 percent, and for U.S. 40-year-old males and females roughly 40 percent.  (The differences, here, are better illustrated in Chart 11.)  These results imply that the hierarchy of higher-consumption permission in the Scottish sample is ordered both by sex and by age whereas in the U.S. data -- though small gender differences permission rates are evident -- the permission hierarchy is geared to age alone.

      The Mexican (Chart 9) and Zambian (Chart 10) data reveal different patterns. In the Mexican sites, permission to feel the effects is very low for females of all four ages. Indeed, females are by and large simply not permitted to drink enough to feel alcohol's effects -- the highest approval level was won by 40-year-old females and that amounted to only about 15 percent.  For approval ran from about 30 percent, for 21- and 60-year-olds to as much as 45 percent for 40-year-old males. Notice that although 21-year-old males received a ten-percentage point higher approval level than 60-year-olds for drinking, per se, approval levels for feeling the effects were roughly equal for the two ages. Sixteen-year-old males received approval from only about 5 percent of the sample, a figure that appears to be quite low until it is contrasted with the approval levels granted females.

      Zambia's pattern is different yet again.  The striking finding here is the high levels of approval granted "feeling the effects" for 40- and 60-year-old males and females. On balance, the data for Mexico, Scotland, and the U.S. have shown that approval levels for feeling the effects among the three statuses 21-years or older run about half as high as approval levels for any drinking -- a little higher for Scots, a little lower for the U.S. and Mexican samples. In the Zambian case, however, the ratio of approving effects to approving drinking jumps to roughly 70 percent. In the Zambian case, then, approval of drinking is almost equivalent to approval of feeling alcohol's effects -- suggesting the possibility of quite different structural arrangements for drinking.

      Some Speculations:

      Here, I want to suggest some broad theoretical ideas that have percolated up from these various findings over the years.

      It is intriguing to contrast the U.S. and the Zambian patterns with respect to permission-to-drink and higher-consumption permission. The U.S. norms suggested, of course, a pattern that permitted drinking universally to all adult statuses but imposed a normative obligation to restrict drinking to moderate levels. In the Zambian case permission to drink is a differentially allocated but the norms on consumption are much weaker.  I have called these two models of normative control "universal moderation" (UM) and "differential access" (DA) -- and it will be noted that they provided the title of my presentation today. My hunch is that these two models of normative control may carry broad implications for the structuring of drinking behavior and offer an interesting line of connection between drinking behavior, on the one hand, and other dimensions or aspects of culture.

      The responses we have examined refer to "ideal" norms rather than actual conduct. We have seen, however, that even ideal norms in UM normative settings like the U.S. and Scotland pertain more to permission to drink per se than to heavier drinking.  Drinking per se, then, appears to be symbolically associated with the universalistically defined rights and responsibilities of adulthood in these settings, but heavier drinking is symbolically associated with a separate stratum of cultural meaning, and one drawing on more traditional status arrangements.  I was particularly impress, for example, by the close agreement in heavier-consumption norms between U.S. and Mexican males (see Chart 12). This result may, of course, be pure coincidence, but it nevertheless highlights the interesting possibility that drinking norms, particularly in less traditional cultures, occupy a point of tension or stress between two competing normative systems.

      The difference between the U.S. and Scottish heavier-consumption norms is also interesting.  If the structure of drinking's norms were tied firmly to the level of modernization in a society, then we would generally expect Scottish and U.S. norms at both permission levels to be the same. The standard indicators of modernization theory are, after all, quite broad and would give us no reason, I suspect, to call either the U.S. or Scotland the more modernized of the two.  It follows, then, that the observed differences in U.S. and Scottish  heavier-consumption norms suggests that modernization-type variables are not the only dimension of culture at work in shaping drinking's normative structure.  This, I think, is a telling possibility and may open up an interesting analytical approach to teasing out of survey data which aspects of culture, after all, are tied most securely to drinking. Just such findings, as it happens, can prove helpful in interpreting some of the other data collected in such surveys.  For example, indicators of "alcohol dependence" in the Scottish and U.S. samples revealed strikingly similar frequencies in the two samples across six of the seven items used.  The seventh item, however, showed U.S. males and females reporting much higher frequencies than the Scots (see Chart 13).  This item, as it happens, asked if one had felt that he or she should stop drinking or cut down in the past year -- a measure, perhaps, of both self-recrimination and the "dryness" of drinking norms. Could it be that the more permissive normative structure observed for Scottish men explains this curious difference in symptom frequencies. But if so, why do Scottish females differ from U.S. females on this measure as much or more than Scottish males differed from U.S. males?

      I have not begun to tap the full analytical possibilities of these data.  Yet I think these various findings and conjectures will give a foretaste of the kinds of analyses and new questions that may follow from the growth of cross-cultural survey data in this field.


      Caetano, Raul, (1986). "Patterns and problems of drinking among U.S. Hispanics," pp. 143-186 in Report of the Secretary's Task Force on Black and Minority Health, Vol. VII, Chemical Dependency and Diabetes, Rockville, MD: U.S. Department of Health and Human Services, 1986.

      Caetano, R., and Martinez, R.M., Alcohol use in Madrid and among U.S. Hispanics, Berkeley: Alcohol Research Group, report C47, 1987.

      Caetano, R., Medina-Mora, M.E., "Immigration, acculturation and alcohol use: A comparison between people of Mexican descent in Mexico and the U.S. 1986," Berkeley: Alcohol Research Group, report C46, 1986.

      Clark, W.B., "Reports on facial flushing among alcohol drinkers in Japan, in Hawaii and in California," presented at the International Symposium on Patterns of Alcohol Use and Misuse Among Different Populations, Washington, DC: 11 Aug - 15 Aug, 1986.

      Duster, T., "Commentary," pp. 326-330 in Room, Robin, and Collins, Gary (eds.), Alcohol and Disinhibition: Nature and Meaning of the Link, Research Monograph No. 12, DHHS Publication No. (ADM) 83-1246, Washington, D.C.: USGPO. 1983.

      La Fontaine, J.S. (ed.), Sex and Age as Principles of Social Differentiation, London: Academic Press, 1978.

      Makela, K., "Permissible starting age for drinking in four Scandinavian countries," Journal of Studies on Alcohol 45:522-527, 1984.

      Makela, K., "Attitudes towards drinking and drunkenness in four Scandinavian countries," Annals of the New York Academy of Sciences 472:21-32, 1986.

      Roizen, R., The World Health Organization study of community responses to
      alcohol-related problems; Annex 41: A review of cross-cultural findings, Geneva: World
      Health Organization, 1981.

      Rootman, I., Community response to alcohol-related problems: Phase II. Final report. Geneva: World Health Organization, March, 1983.

      Smart, 1980 -- reference missing.