A Review of Studies of Drinking Patterns in the United States Since 1940

Abstract

Objective:

This article reviews changes in drinking patterns in the general U.S. population since the first appearance of the Quarterly Journal of Studies on Alcohol in June 1940.

Method:

Contents of the three iterations of the journal (Quarterly Journal of Studies on Alcohol, Journal of Studies on Alcohol, and Journal of Studies on Alcohol and Drugs) were reviewed along with additional prominent key articles on the subject known to the author.

Results:

Early articles in the field tended to focus on alcoholics and alcoholism, usually with a biological emphasis. The earliest “scientific investigation” of drinking patterns that appeared in iterations of JSAD reported data based on a survey in 1946, and the earliest report of trends in drinking patterns compared the 1946 data with data from 1963. Since then, there have been numerous evaluations of drinking patterns in the general U.S. population.

Conclusions:

There have been major changes over the past 75 years in the manner in which drinking patterns are measured and summarized. Perhaps the biggest change was the introduction of probability sampling. Additional major changes continue to occur, responding to technological changes (e.g., the use of computers, cell phones, and broadband connections) in how people interact with their environment.

This article reviews the evolution of studies of drinking patterns in the general U.S. population since the first appearance of the Quarterly Journal of Studies on Alcohol (QJSA) in June 1940. As used in this overview, the term drinking patterns refers broadly to drinking behaviors including prevalence, the amount consumed, and quantity–frequency measures. This article provides a brief sketch of the relevant history of such studies; space constraints do not allow for an in-depth analysis of what has been learned about drinking patterns. In general, only broad-based studies (mostly national) are considered; thus, studies conducted in only a single state are excluded.

Method

Contents of the three iterations of the journal (Quarterly Journal of Studies on Alcohol [QJSA], Journal of Studies on Alcohol [JSA], and Journal of Studies on Alcohol and Drugs [JSAD]) were reviewed, with special attention to articles that reported data on national patterns of alcohol use and reports describing findings on quantity and frequency measures. In addition, results from prominent national data sets were evaluated regarding similar data from other venues. The findings are divided into four arbitrary epochs in the evolution of the relevant methodological approaches.

Results

Early phases of the field: 1940 to the 1970s

In 1943, Bacon presented some foundations for a sociologic study of drinking behavior and provided interesting insights into approaches for studying drinking. For example, in discussing the value (or lack thereof) of questionnaires, Bacon believed that information generated to date regarding personal use of alcoholic beverages often carried “too much affect” (p. 444) and argued for a more qualitative approach to the sociologic study of drinking.

Subsequently, what was perhaps the first semimodern article that attempted to describe drinking patterns among a more representative sample appeared in 1947 by Riley and Marden. Those data on drinking patterns came from a national survey that was focused on public attitudes toward alcohol and alcoholism. The authors stated, “These data yield a new over-all estimate of the incidence [sic; this would currently be considered “prevalence,” that is, the proportion of the population who were drinkers] of drinking in 1946, and delineate for the first time, we believe, the social pattern of drinking in the United States” (Riley and Marden, 1947, p. 265). It is interesting that the survey methodology was only briefly described in a footnote, which said, in part, that the survey was “Conducted for Rutgers University by the National Opinion Research Center of the University of Denver,” and that “The interpretations are the authors' and do not necessarily reflect the opinion of any of the sponsoring groups” (p. 265). No information was provided on how or when the survey was conducted, and the article did not present the number of participants.

In this work, respondents were placed into three broad and not well-defined categories: (a) abstainers, (b) occasional drinkers, and (c) regular drinkers, with the latter defined as those who drink at least three times a week and all other drinkers designated as occasional drinkers. Three tables presented a cross-tabulation of these categories with (a) the size of place of residence (five categories from farm to more than 1,000,000), (b) the person's economic level (prosperous, average, poor), and (c) their religious affiliation (Protestant, Catholic, Jewish). No information was provided on how the variables were ascertained or operationalized, and a single figure provided the “incidence of drinking” by Gender × Age Group. Differences in drinking status by education (less than a high school education vs. others) were cited in text and by place of birth of the respondents' parents and grandparents. An interesting omission was the region of the country where the respondent resided. Among the conclusions drawn by the authors was that changes in society in postwar America were likely to have contributed to increased proportions of drinkers that might have related to more women taking on less traditional roles, as well as a trend in the United States toward a more urban and a more educated population.

Seventeen years after Riley and Marden's 1947 article, Mulford (1964) published results of what might be considered today as a more representative national survey of patterns of beverage alcohol use, noting that, “The only previous scientific investigation of the nation's drinking patterns [the Riley and Marden study] was conducted nearly a generation ago” (p. 634). It appears that the articles published in what was then QJSA were fairly representative of the state of the art in scientific descriptions of patterns of drinking at the time. Mulford commented that some studies of more localized state populations had been done (Mulford in Iowa [Mulford and Miller, 1959], Maxwell [1952] in Washington, and one major community survey in California [Knupfer et al., 1963]) of drinking practices, but it does not appear as if much was published during this epoch regarding national drinking patterns, the focus of this review. Table 1 provides the first data on trends in the U.S. population that were the product of “scientific investigations.”

Table 1.

Percentage who drink in the United States, 1946 and 1963, by selected social characteristics (adapted from Mulford, 1964)^a

	1946 (n = 2,677) %	1963 (n = 1,509) %
Total	65	71
Sex
Men	75	79
Women	56	63
Education
Completed high school	70	79
Less than completed high school	62	60
Age, in years
21–25	73	78
26–35	75	78
36–45	68	78
46–55	58	66
>55	50	59
Religion
Catholic	79	89
Protestant	59	63
Jewish	87	90
Community size
> 1,000,000	77	79
50,001–1,000,000	72	70
2,500–50,000	61	73
Rural nonfarm	57	60b
Rural farm	46

Author’s notes: Estimates of percentages of drinkers in 1946 are based on responses to the question: “Would you mind telling me which of these you sometimes drink?” Response possibilities included a variety of alcoholic beverages. Drinkers were defined as those who reported use of any beverage (Riley and Marden, 1947). Estimates of percentages of drinkers in 1963 are based on affirmative responses to the question: “Do you ever have occasion to use alcoholic beverages, such as liquor, wine, or beer, or are you a total abstainer?” (Mulford, 1964).

^aSome changes have been made to conform to current journal style;

^b1963 figure is for all rural.

Mulford (1964) also compared his results with some data from Gallup polls that asked, “Do you ever have occasion to use alcoholic beverages, such as liquor, wine, or beer, or are you a total abstainer?” Mulford reported that this rather imprecise Gallup question revealed an increase from 65% to 71% in the percentage of U.S. residents who were drinkers between 1946 and 1963, with much of the increase attributable to a 12% rise among women in respondents in the 21–25 years age group. (Drinkers were defined as those who responded “yes” to the question.)

The Mulford (1964) article also reported prevalence by “social segments,” including region of country, gender, age, education, community size, marital status, income group, occupational status, and religious affiliation. These elements were important as the data indicated that the prevalence of drinkers in different regions of the country “has not previously been published” (p. 639). Nine different regions were reported, with a very wide range of prevalence of drinkers, from 33% in the East South Central to 88% in the Mid Atlantic. (Mulford [1964] used the same question that Gallup used to define drinkers.) Consistent with more recent studies, Mulford found that the prevalence of drinkers was highest among men; those ages 21–39; the higher educated; those in larger cities; the unmarried; those in higher income and occupational status groups; and among Jews, Catholics, and Lutherans. Despite the advances in methodology, although differences between demographic characteristics were noted, there was no mention of tests of statistical significance for these differences. Like Riley and Marden, Mulford hypothesized that as the population becomes more urban, better educated, and more prosperous, the prevalence of drinkers may be expected to increase.

The Mulford (1964) article is notable for its attempt to provide information on heavier drinking and associated problems in a national population. A quantity–frequency measure of consumption, the Q-F index, was a prominent part of how the data were presented. This measure was originally developed by Straus and Bacon (1953) and previously used by Mulford and Miller in their studies in Iowa (Mulford and Miller, 1959, 1963).

Beginning shortly after Mulford's 1963 survey, several studies of national samples were published in what was then QJSA. These were based on a National Institute of Mental Health award given to the Social Research Group of George Washington University to conduct a national survey “to describe in detail the drinking practices of Americans as accurately as possible” (Cahalan et al., 1969, p. 5). The fullscale survey used a national probability sample drawn from a representative sample of the total population of persons age 21 or older living in households in the contiguous 48 states. Data were gathered from 2,746 completed interviews representing a 90% response rate from eligible households.

The survey benefited from several smaller scale pilot studies, including a small study of 81 persons to evaluate the validity of the interview. The researchers noted that they were concerned with the extent to which misclassification of drinking behavior could result in serious distortion of reported relationships (Cisin, 1963). The great majority of interviewers were men because an earlier study found that men elicited higher rates of drinking than women (Kirsch et al., 1965), and all interviewers were nonabstainers (based on findings by Mulford and Miller [1959, 1963] that nonabstainers elicited higher rates of drinking than abstainers). Thus, the methods used in this protocol demonstrated great strides forward compared with studies published previously.

The first report of results from this national study was published in QJSA in 1968 (Cahalan and Cisin, 1968a, 1968b), with a longer version published in 1969 as Monograph Number 6 of the Rutgers Center of Alcohol Studies (Cahalan et al., 1969). The researchers included a Q-F index of the amount of alcoholic beverages consumed over a specified period. The approach estimated the amount a person drank by multiplying the approximate amount “usually” consumed on an average occasion by the reported frequency in a stated period (often a month). This scheme was borrowed from a method initially developed by Straus and Bacon (1953) and later adapted by Mulford and Miller (1959) as well as by Maxwell (1952). The new survey added a dimension of variability (which combined the most usual amount consumed and the highest amount drunk occasionally) to produce a Q-F-V index. This allowed respondents to be classified into five categories: abstainers, infrequent drinkers, light drinkers, moderate drinkers, and heavy drinkers using more precise definitions of categories that improved on prior approaches. The researchers also developed a Volume–Variability (V-V) index, which incorporated whether the respondent ever consumed five or more drinks at a sitting. Subsequently, there have been numerous attempts at improving these new and important indexes (e.g., Bowman et al., 1975; Knupfer, 1966), as described in several overviews (Greenfield and Kerr, 2008; Little et al., 1977; Mäkelä, 1996; Room, 1991). Although the 1964–1965 National Alcohol Survey (also referred to as the National Survey of Drinking Practices; Cahalan et al. 1969) used very different questions and methodology than previous national studies, the proportions of drinkers and nondrinkers were similar to those reported by Gallup (in a February 1966 poll) and Mulford (1964).

The Cahalan et al. monograph (1969) included a rudimentary discussion of age versus generational differences in alcohol use, although the concept of an Age-Period-Cohort decomposition of effects was not yet fully developed. Cahalan et al. concluded that the data indicated that drinking rates declined with age. Later developments in the study of patterns of drinking have been concerned with the decomposition of age-period-cohort effects (see, e.g., Kerr et al., 2011). Overall, the 1969 monograph provided extensive tables showing the Q-F-V classifications by gender, age, race, index of social position, family income, occupation of chief breadwinner, educational level, marital status, presence of children in household, region of country, urbanization, place of birth, father's country of origin, national identity, religion, church attendance, and various combinations of the variables.

Although viewed primarily as a cross sectional study, the 1964–1965 survey was also used to provide a baseline for longitudinal evaluations. A subsample of 1,359 respondents from the 1964 survey (called N1) were reinterviewed in 1967 (N2), and results were reported in a book that focused not so much on drinking patterns as on problem drinkers (Cahalan, 1970). A third survey (N3) interviewed a supplementary sample of 978 men ages 21–59, again with a focus on problem drinkers. Data from N2 and N3 samples were combined and provided the basis for another monograph (Cahalan and Room, 1974).

In summary, the early work in this field reflected primarily qualitative evaluations. An ever more sophisticated series of studies used quantitative approaches to document multiple aspects of alcohol intake patterns. The latter culminated in the work of Cahalan and colleagues that in many ways paved the way for future cross-sectional and longitudinal studies that keep the alcohol field informed about alcohol intake patterns and their change over time.

Single stage and multistage cross-sectional studies since the 1970s

A series of cross-sectional studies grew out of the work of Cahalan and colleagues. Over the years, the design of these investigations has evolved, but much of the work harkens back to the methodological innovations made in the 1960s, with the investigators working to provide as much continuity as possible. Reflecting the increasing costs of in-person surveys, many of the more recent surveys have been conducted by telephone, making it difficult to replicate the 90% response rate of Cahalan's original study.

Between 1970 and 1975, the National Institute on Alcohol Abuse and Alcoholism (NIAAA) funded Louis Harris and Associates and the Opinion Research Corporation to evaluate the effects of NIAAA's public service advertisement campaigns, but these were focused primarily on attitudes rather than drinking patterns (e.g., Harris and Associates, 1971). Also in the 1970s, four national surveys of drug use were funded by the National Institute on Drug Abuse (NIDA), work that included some alcohol use questions (Abelson and Atkinson, 1976; Abelson and Fishburne, 1976; Abelson et al., 1977; Fishburne et al., 1980). This ongoing series of cross-sectional surveys was expanded by the National Household Survey on Drug Abuse (with the title of the survey changed to the National Survey on Drug Use and Health in 2002), funded initially by NIDA and later by the Substance Abuse and Mental Health Services Administration (e.g., Office of Applied Studies, 2002, 2003). There have been major advances in methodology across these studies, including changing from paper-and-pencil to computer-assisted surveys, an expansion to allow state-level estimations, and the addition of a monetary incentive. These and other methodological changes have required researchers to be cautious in assessing trends over time (Kennet and Gfroerer, 2005).

Additional nationally representative cross-sectional survey data can be found in the NIAAA-funded National Longitudinal Alcohol Epidemiological Survey (e.g., Dawson et al., 1995) that focused primarily on disorders associated with alcohol intake. This study used the NIAAA Alcohol Use Disorders and Associated Disabilities Interview Schedule-DSM-IV (AUDADIS-IV) to generate diagnoses of alcohol abuse and dependence but also gathered data on alcohol use patterns.

Other national surveys have incorporated alcohol use pattern measures. The National Health Interview Survey (e.g., Williams et al., 1986) was initiated by The National Health Survey Act of 1956 (Public Law 652, 84th Congress), which provided for a continuing survey and special studies regarding the health of the population. This is a large, nationally representative survey conducted continuously since 1957, administered through the Centers for Disease Control and Prevention (CDC), and it includes a small number of questions about alcohol use. These surveys are conducted by CDC's National Center for Health Statistics and use in-person interviews; like many continuing surveys, methods and procedures have changed over time, with more recent surveys using computer-assisted techniques. Although only a small number of questions about alcohol are typically included, in 1988 a large module of questions relevant to alcohol was included. This allowed for a number of in-depth studies going beyond patterns of use, for example, a study of prevalence of DSM-III-R alcohol abuse and dependence (Harford and Grant, 1994). The strength of this series is that it allows for examining differences in a large cross-sectional study that incorporates many demographic subgroups. For example, Breslow and Smothers (2004) studied the drinking patterns of older Americans (age 60 and above) by combining data from 1997 to 2001. One important finding was that differences in patterns of quantity and frequency were obscured by the use of a composite quantity–frequency measure.

Another relevant CDC study, the National Health and Nutrition Examination Survey is a complex, stratified, multistage probability sample of the civilian, noninstitutionalized population residing in the United States. This study, which has been conducted on a continuous basis since 1999, allows for some estimates of some alcohol-related behaviors, such as the percentage of the population who exceed recommended drinking guidelines (Guenther et al., 2013).

Some efforts have focused on measuring patterns of alcohol use in nationally representative samples of specific segments of the population. NIAAA funded a major initiative in the mid-1970s examining alcohol use and attitudes in a nationally representative sample of students in grades 7 through 12 (Rachal et al., 1975, 1976). Also in the mid-1970s, NIDA funded a study focused on high school seniors in which the primary focus was on illicit drugs, although measures of alcohol use and related attitudes and behaviors were included. Viewed as both a cross-sectional protocol in any one year and as a multistage investigation, the Monitoring the Future study by Johnston et al. (2012) continues today, conducting annual surveys in much the same format as in the original. It, thus, provides some of the best data on changes in drinking, albeit on a restricted segment and without great depth in the questions. Some major findings from this series of annual surveys have been the documentation of drug (and alcohol) intake patterns and attitudes since the 1970s to the present, including information regarding alcohol patterns across 12th graders in the United States. Regarding alcohol, between 1991 and 2012 the lifetime prevalence of alcohol use by grade 12 changed from 88% to 69% (Table 5-5a) [88% to 69%, Table 4-1a] ever having been drunk went from 65% to 54% (Table 5-5a) [65% to 54%, Table 4-1a] with additional data available on prior 30 day and daily use patterns (Johnston et al., 2013a). Other surveys have also focused on younger populations including a national school-based survey of 9th- through 12th-grade students carried out on a biennial basis by the CDC since 1991 (e.g., CDC, 2012). Further, in 2001, the CDC established the Alcohol Program in its Division of Population Health that uses the two public health surveillance systems to monitor trends in excessive and underage drinking and to assess related health effects.

Another specifically targeted segment has been college students. The Monitoring the Future study follows 12th graders after graduation from high school, allowing for comparisons of college students with others not in college. One finding is that drinking among college students has not declined as much as others of similar age: having five or more drinks in a row at least once in the prior 2 weeks declined from 44% in 1980 to 37% in 2012 among college students, compared with a decline from 41% to 30% among non–college students (Johnston et al., 2013b). Another series that targeted college students was the College Alcohol Study that was active from 1993 to 2001 (Wechsler and Nelson, 2008). Some interesting findings from this work include the focus on recording patterns of heavy episodic, or “binge” drinking (five or more drinks per occasion for men and four or more for women), reports on the potential changes in drinking patterns after a school bans alcohol use on campus, and the ways that drinking problems at universities have adverse effects on other students (Wechsler and Nelson, 2008; Wechsler et al, 2001). Other population segments that have been specifically targeted include women (Wilsnack et al., 2006), persons in the armed forces (Bray and Hourani, 2007; Burt et al., 1980), and individuals in the labor market (Frone, 2006).

Recent longitudinal studies

In the early 2000s, the first stage of the National Epidemiologic Survey of Alcohol and Related Conditions (NESARC) was sponsored by the NIAAA. This study of a nationally representative sample of more than 43,000 U.S. adults specifically targeted disorders but also gathered useful information on patterns of alcohol intake using the AUDADIS-IV (although there were some differences from prior studies in how the instrument was implemented; Grant et al., 2004). A second wave of the NESARC was conducted in 2004–2005, and a third is currently in progress. Some notable results of these efforts include identifying predictors and correlates of drinking cessation and how they differ across the life span, detailed comparison of drinking patterns across racial and ethnic groups, transitions in and out of problematic use of alcohol, and information on how to most efficiently screen for heavier drinking and alcohol problems (Dawson et al., 2007, 2009, 2010, 2013).

In addition to the NESARC longitudinal study, several other national surveys that asked about alcohol use have a longitudinal component. One example is the National Longitudinal Surveys of Youth–1979 (NLSY79), funded by the U.S. Department of Labor, that first interviewed a nationally representative sample of 12,686 14- to 22-year-olds in 1979. Annual assessments continued through 1994, with biennial assessments since then, including questions about alcohol use in most, but not all, assessments. A second iteration of this study, NLSY97, was initiated in 1997 and has a similar design. Inclusion of a longitudinal component allows for study of the stability of alcohol consumption patterns over time (e.g., Grant et al., 1988; Kerr et al., 2002).

In summary, modern methods and statistical approaches have been used to further our understanding of the pattern of alcohol use in the same people followed over time. Additional important information can also be generated through sequential observations across populations when measured by similar approaches in different years.

Studies of cohort effects and related changes in alcohol use patterns over time

As noted above, a series of national alcohol surveys has provided a way to assess changes over time in drinking patterns among the national population of adults. Hilton and Clark (1987) analyzed change between 1967 and 1984 when the interviews had maintained some questions that were worded (almost) identically. These authors used a “Volmax” measure that distinguished nine categories of drinkers, crossing low, medium, and high volume (that is, how many drinks per month) by low, medium, and high maximums (that is, how many drinks were consumed per drinking occasion). Thus, for example, the Medium Volume–Medium Maximum category consists of those who drank between 17.5 and 44.9 drinks per month and who drank three or four drinks per occasion at least “once in awhile” but who never drank five or more drinks per occasion. The distribution of 10 categories (including abstainers) between 1967 and 1984 demonstrated few notable and consistent subgroup differences. Given the substantial cultural changes that occurred in that 17-year interval in the United States, the stability of drinking pattern is notable.

Midanik and Clark (1994) used two of the Alcohol Research Group's National Alcohol Surveys to assess changes in drinking patterns, this time using 1984 and 1990. Unlike the earlier Hilton and Clark research, Midanik and Clark reported six dichotomous measures of drinking (current drinking, current wine drinking, current beer drinking, current liquor drinking, weekly drinking, having five or more drinks on one occasion at least weekly). All six measures showed significant declines in the 6-year interval, which were mostly similar across subgroups in reflecting both an increase in rates of abstinence and some shifts toward lighter drinking. Despite a general lowering of alcohol consumption in the United States, there were no overall changes in reports of either dependence symptoms or social consequences (Midanik and Clark, 1995).

Greenfield et al. (2000) analyzed data from the Alcohol Research Group's 1984, 1990, and 1995 National Alcohol Surveys to examine trends in consumption. Five dichotomous measures of drinking, similar to those used by Midanik and Clark (1994), were examined, all of which showed sharp drops between 1984 and 1990 but remained level from 1990 to 1995. Greenfield et al. used different measures of heavy drinking than Midanik and Clark; one measure (five or more drinks ever in previous year) was relatively stable from 1984 to 1990 and from 1990 to 1995, whereas the other (five or more drinks weekly in previous year) declined from 1984 to 1990 and increased slightly from 1990 to 1995. The findings suggest little change from 1990 to 1995, following a prominent decline from 1984 to 1990.

The trends in drinking patterns cited above were reported as essentially representing trends without regard to the complexity of the varying influence of age, period, and cohort effect. The “hazards” of inferring effects of age, period, or cohort on drinking behavior had been well described as early as the 1970s (Glenn, 1977, 1981; Glenn and Zody, 1970), but it was not until the late stages of the century and the early 2000s that use of an age-period-cohort interpretation of change became common. Differentiating age, period, or cohort effects is of paramount importance because the differentiation can guide the search for underlying causes. If age effects are consistently found, then the observed differences may reflect general maturational processes. If a period effect is found (affecting all ages and cohorts more or less equally), the causal influences are likely to be contemporaneous societal forces. If a cohort effect is revealed, then the analyst needs to look at many kinds of potential explanatory factors, including potential demographic changes associated with cohort, such as cohort size (Easterlin, 1980). As noted by Kerr et al. (2009), “The decomposition of trends . . . into age, period, cohort, and demographic effects offers an important perspective on the dynamics of change in alcohol use patterns” (p. 27). Kerr et al. (2004) used the age-period-cohort paradigm to analyze data from the 1979, 1984, 1990, 1995, and 2000 National Alcohol Surveys and concluded that changing cohort demographics had significant effects on beverage-specific consumption. For example, higher educational attainment in the population over time was associated with reduced beer and increased wine consumption. The same authors (Kerr et al., 2009) later reported results using a similar paradigm based on surveys from 1979 to 2005. They found, among other things, that although alcohol consumption levels continued earlier declines for those ages 26 and up, there was an increase in those ages 18–25. The authors caution that ambiguities remain in trying to interpret what effects appear to be operating in such analyses, due to the strong assumptions typically required to achieve identification of the models, particularly the assumption of the absence of interactions among period, age, and cohort effects. A more recent effort that incorporated the 2010 survey further cautions that, given the limitations, “Analyses of other surveys should be considered for confirmation” (Kerr etal., 2011, p. 314A).

Keyes et al. (2011) echoed the need for caution in interpreting data using an age-period-cohort approach. In their review and synthesis of cohort effects in alcohol epidemiology, they found that there appears to be evidence that younger birth cohorts in the United States are more likely than older cohorts to engage in heavy episodic drinking. But Keyes et al. (2011) and Keyes and Li (2010) pointed to methodological issues that need to be considered in future research on analysis of alcohol consumption patterns.

Conclusions

There has been remarkable progress since 1940 in the evolution of studies of patterns of alcohol consumption over time. Each epoch of work and each methodological innovation have had assets and liabilities that must be kept in mind when viewing the results of the studies cited in this review.

Indeed, many methodological issues need to be considered in future research on all aspects of alcohol consumption patterns. In the past 75 years, there have been many changes in the way patterns are defined and measured, and future refinements will, no doubt, continue to be made. But perhaps the biggest changes to date have occurred in the methods used to collect data. The earliest surveys used pencil-and-paper face-to-face interviews of samples of individuals in households. Today, a range of methods are used, including medical examinations in traveling vans, computer-assisted in-home interviews, telephone interviews using both landline and cell phones, Internet-based online surveys, mobile device–based surveys, etc. Continued refinements in how questions are worded to assess consumption patterns are being made, and innovative methods such as the Timeline Followback continue to be improved (Fiellin et al., 2013). There have also been changes in how to select samples. Seventy-five years ago, quotas were used; then for many years, area probability samples were state of the art. Telephone technology could use random digit dialing for a while, but this is increasingly inadequate. The Internet and new mobile devices have changed people's lives, and they have changed how alcohol researchers have been able to examine alcohol use patterns as well. It is impossible to predict how we will be measuring alcohol use patterns 75 years from now, but it is certain that it will not be the same as it is today.