Genetic theories make little sense out of the enormous differences in alcoholism rates between social groups--like the Irish and the Jews--at opposite ends of the continuum in incidence of alcoholism (Glassner and Berg, 1980; Greeley et al., 1980). Vaillant (1983) found such ethnic distinctions to be more important than inherited tendencies toward alcoholism for determining clinical outcomes like a return to controlled drinking. In addition, the incidence of alcoholism is influenced by social class (Vaillant, 1983) and by gender--so much so in the latter case that theories of inherited alcoholism have been limited solely to men (Öjesjö, 1984; Pollock et al., 1984).

These sociocultural-gender differences have provoked a good deal of theorizing, some of it quite imaginative. Milam and Ketcham (1983) suggest that it is the duration of exposure to alcohol that determines a cultural group's alcoholism rate, since evolutionary selection will eliminate those susceptible to alcoholism. However. while metabolic differences and variations in sensitivity to alcohol have been found among ethnic and cultural groups (Ewing et al ., 1974; Reed et al ., 1976), these group differences have not been found to predict alcohol misuse (Mendelson and Mello, 1979). The most striking case of divergent cultural patterns of drinking in the face of prominent racial reactions to alcohol is the pattern established by the Chinese and Japanese Americans on the one hand, and the Eskimo and American Indian groups on the other. Drinking in these groups is marked by a distinctive facial reddening and accelerated heart beat, blood pressure and other circulatory system measures, as well as by acetaldehyde and other alcohol metabolism abnormalities. However, the Chinese and Japanese Americans have the lowest alcoholism rates of all American cultural groups and the Eskimos and American Indians have the highest such rates (Stewart, 1964).

Vaillant (1983) suggested a modified cross-generational selection process to explain the large difference in the appearance of alcohol dependence between his college and his core-city sample: the lower incidence of dependence in the college group could be due to the economic and social failures of fathers of alcoholics that made it less likely their children would enter college. However, in explaining his extremely strong finding of ethnic differences in alcoholism, Vaillant relied on standard interpretations of how different cultures view alcohol and socialize its use. What makes Vaillant's reference to genetic determinism for his social-class results more surprising is his overall recommendation that: "At the present time, a conservative view of the role of genetic factors in alcoholism seems appropriate" (p. 70)

Vaillant (1983) was led to such conservatism by a number of his data. Although he did find that subjects with alcoholic relatives had three to four times the alcoholism rates of those without traces of familial alcoholism, this result appeared in the absence of the statistical controls needed to separate genetic and environmental causality. When Vaillant examined differences between those with alcoholic relatives who did not live with them and those with no alcoholic relatives as a kind of environmental control, the ratio of the incidence of alcoholism was reduced to 2 : 1. There could also be additional environmental factors besides this one of immediate modeling effects of drinking that could reduce this ratio even further. Indeed, the Vaillant study disputes the alcoholism concordance rates that have been found in genetically similar and environmentally dissimilar populations which recent genetic models presuppose.

Other data fail to support biological inheritance of alcoholism. Gurling et al. (1981), when comparing MZ and DZ twins, found that the nonidentical pairs showed a higher pairwise concordance rate for alcohol dependence. This British group has also presented a comprehensive critique of the twin and adoption studies (Murray et al., 1983). Regarding Goodwin and his colleagues' (1973) seminal discovery of an alcoholism inheritance among adoptees, Murray et al. noted that the investigators' definition of alcoholism was unique, including a low cutoff in the amount of consumption (daily drinking, with six or more drinks consumed 2 or 3 times a month) combined with reported loss of control. The definitions in Goodwin et al.'s study are crucial since control adoptees (those without biological-alcoholic relatives) were more often problem drinkers than were index adoptees (those with biological-alcoholic relatives)--a finding which was reversed for subjects identified as alcoholics. Murray et al. commented: "Could it be that Goodwin's findings are simply an artifact produced by the threshold for alcoholism accidentally dividing heavy drinkers in the index and control groups unevenly?" (p. 42).

Murray et al. (1983) point out that such definitional issues frequently raise questions in the genetic studies. For example, Schuckit et al.'s (l972) finding--that half-siblings with an alcoholic-biological parent who were reared by nonalcoholic parents showed a heightened risk of alcoholism--defined alcoholism as "drinking in a manner that interferes with one's life." This seems a better description of alcohol misuse than of alcoholism. In other words, this study identified genetic transmission of alcoholism in a category for which Goodwin et al. (1973) had rejected it. Consider also that Cadoret and Gath's (1978) finding of genetic determination in adoptees held only for a primary diagnosis of alcoholism, and that a larger group of subjects with a secondary diagnosis of alcoholism came entirely from among those without alcoholic-biological parents. These shifting definitional boundaries actually enhance the statistical likelihood of uncovering alcoholic inheritance in each study.

Vaillant addressed himself particularly to the notion, first put forward by Goodwin (1979), that inherited alcoholism marks a distinct and separate variety of the disease. This is, of course, a reworking of the A.A. (1939) version of alcoholism. Working against this view of alcoholism--and its updated models of inherited sex-linked differences in alcoholism etiology and of a special variety of alcoholism characterized by inherited ASP--are findings that the same socially based differences in alcoholism rates pertain as well for less severe gradations of alcohol misuse. That is, those same ethnic, social class and gender groups that have a high incidence of problem drinking (Cahalan and Room, 1974; Greeley et al., 1980) also display a high incidence of alcoholism (Armor et al., 1978; Vaillant, 1983). It simply strains scientific credulity to imagine that the same factors which act in a socially mediated way to determine alcohol misuse also operate through separate genetic paths to influence alcoholism. Moreover, epidemiological studies such as Vaillant's and the Cahalan group's have always found more severe forms of alcohol dependence to merge imperceptibly and gradually with lesser degrees of problem drinking, so that a distinct, pathological variety of alcoholism does not stand out along a population curve of those who have drinking problems (Clark, 1976; Clark and Cahalan, 1976). Collations of measures of neurophysiological impairment likewise describe a smooth distribution of data points (Miller and Saucedo, 1983).