Cohort Variation

Bernard Jaworski, University of Pittsburgh
William J. Sauer, University of Pittsburgh
ABSTRACT - Consumer behavior occurs within the boundaries of a given historical period. In order to assess how history might affect consumption it is first necessary to define some of the central issues involved in such an inquiry. This paper addresses three issues which are particularly important in any period-specific inquiry. These issues include cohort versus generational analysis, intercohort and intracohort variation, and cohort adaptation. The importance of these issues are considered in relation to previous findings in consumer research.
[ to cite ]:
Bernard Jaworski and William J. Sauer (1985) ,"Cohort Variation", in NA - Advances in Consumer Research Volume 12, eds. Elizabeth C. Hirschman and Moris B. Holbrook, Provo, UT : Association for Consumer Research, Pages: 32-36.

Advances in Consumer Research Volume 12, 1985      Pages 32-36

COHORT VARIATION

Bernard Jaworski, University of Pittsburgh

William J. Sauer, University of Pittsburgh

ABSTRACT -

Consumer behavior occurs within the boundaries of a given historical period. In order to assess how history might affect consumption it is first necessary to define some of the central issues involved in such an inquiry. This paper addresses three issues which are particularly important in any period-specific inquiry. These issues include cohort versus generational analysis, intercohort and intracohort variation, and cohort adaptation. The importance of these issues are considered in relation to previous findings in consumer research.

INTRODUCTION

Since the late 1960's a number of social science theorists in subfields such as life-span developmental psychology (Baltes, Reese and Lipsett 1980), life-course sociology (Elder 1974, 1981), age stratification (Riley, Foner and Johnson 19 72), and social history (Haraven 19 78) have directed their efforts toward the study of behavior in historical context. Historical context encompasses both the historical events and the social conditions which existed/ transpired during an individual's lifetime. Recently articles have appeared in the consumer literature which have emphasized the benefits of the historical approach (cf. Rentz and Reynolds 1980, Rentz, Reynolds and Stoudt 1983). While viewing consumer behavior from this perspective may provide significant new insights to the field, care must be taken with regard to the use of this approach. This paper builds on this developing area of consumer research by clarifying some of the underlying conceptual issues involved in an analysis of historical effects. In particular we address (1) the distinction between cohort and generational analysis, (2) the variation that can occur between and within cohorts, and (3) cohort adaptations through time. Following this discussion we provide alternative interpretations of previous findings. In the final section we briefly review some of the reasons why cohort analysis has received little attention in consumer research.

COHORT OR GENERATION

The terms " cohort" and " generation" have found increasing use in the marketing literature in the past few years (cf. Rentz and Reynolds 1980, Reynolds and Rentz 1981), and although there is a tendency in both the social sciences and popular literature to equate the two concepts, there are important differences which warrant attention. In order to most effectively utilize either concept, a concise and clear understanding of their meanings and boundaries is necessary.

Early European sociologists who were the first to pursue the concept of generations employed a thirty year unit of measurement (Hannheim 1952). This unit of time was accepted for use since it marked the length of time that characterized the creation of familial descent patterns between parents and children. However, this specific method of conceptualizing generations was more reflective of the rate of change in the family structure than it was an accurate indicator of the rate of social change. Rapid social change and it's attendant modifications in values necessitate attention to much shorter time spans as significant units for the study of generations. Today we are more conscious of the fact that generations are not units fixed by the mere passage of time, but are more profitably understood as the result of the interaction of individual maturation with the socio-historical environment inhabited by the individual.

Turning to the specific definition employed by Mannheim (1952) we notice that he defined generation in terms of age groups and shared experiences:

The fact of belonging to the same class, and that of belonging to the same generation or age groups, have this in common, that both endow the individuals sharing in them with a common location in the social and historical process, and thereby limit them to a specific range of potential experience predisposing them for a certain characteristic mode of thought and experience, and a characteristic type of historically relevant action (p. 291).

Some years later Ryder (1965), in attempting to examine the utility of cohort also referred to age groups and shared experiences:

a cohort may be defined as the aggregate of individuals (within some population definition) who experienced the same event within the same time interval...Each cohort makes fresh contact within the contemporary social heritage and carries the impression of this encounter through life (p. 844-845).

Finally, during the mid 1970's Bengston and Cutler (1976) suggested the following definition of cohort:

the cohort perspective [is 1 a focus which emphasizes demographic attributes of age groups. Born during a given period of history, a particular age cohort experiences in similar ways the consequences of historical events (p. 131).

Given these nominal definitions it is not surprising that the terms have been used interchangeably. We believe, however, that the confusion that has arisen can be traced to the ambiguity of "shared experience" within a historical period. One can share an experience with a historical period such as the 1940's or 1960's in at least three distinct ways. The first of these is the cohort approach, which usually centers on some demographic characteristic shared by a group. Typically in the social science literature researchers emphasize age cohorts. This age cohort distinction is reflected in Bengston and Culter's (1976) definition mentioned above. A second type of classification emphasizes the location of a family unit. Researchers pursuing this type of analysis focus on the generational transmission of norms or values. Concerning the differences between cohort and generation, Kertzer (1983) writes "generation, then, is a relational concept bound to the realm of kinship and descent, it is not an appropriate tool for dividing societies into segments or populations into aggregates (p. 128)." In other words, generation should be restricted to family descent while cohort should refer only to the succession of individuals who pass through the social system. A third type of historical grouping has been termed the "historically conscious agents of social action" (Bengston and Cutler 1976). These unique groupings are comprised of subunits within age cohorts.

The distinction between cohorts, family generation, and "social action" cohorts is important. It is not uncommon to find researchers who divide their sample into age groupings (one, five or ten year intervals) and examine how groups differ with regard to a particular phenomenon. Differences which emerge from this analysis are often attributed to age effects. However, it is usually unclear if the authors are attempting to make a case for age, cohort, family, or within cohort effects. We hope the present discussion allows for a broader interpretation of age and historical effects.

Given this qualification, we now turn to the issue of "differential" exposure to historical events by separating cohorts into intracohort and intercohort analyses.

INTERCOHORT AND INTRACOHORT VARIATION

For the term cohort to be utilized as an explanatory concept it must organize persons into distinct groups with empirically demonstrated variation between them. Birth year is often employed as the grouping variable in intercohort analyses. The use of age as a structural variable depends on the extent to which similar experiences are shared by members comprising the age strata. This assumption is integral to the concept of an age cohort. If age does not discriminate individual's responses to broad historical change, then its reliability as an independent variable may be called into question. Chronological age has been shown to be less than compelling as a predictor of values and behavior (Wohlwill 1970) due, at least in part to the heterogeneous nature of age grouPs (Zaltman, Alpert and Heffring 1980).

Rather than abandoning age as a grouping mechanism, we suggest that researchers begin to think more about intracohort variation. The lack of intracohort awareness has arisen because there is a tendency to regard historical change as a monolithic entity that operates on a societal level. In fact, there is no socio-historical change independent of its effects on specific individuals. The cohort analyst must be prepared to specify the populations in which intercohort and intracohort variation is expected to occur. An example of the value of this approach is illustrated in a program of research by Elder (1974, 1981). Elder examined the variation of two cohorts (one born in 1920 and one born in 1929) by focusing on depression hardship (percentage loss of income) and its effects on adolescent and adult health and behavior. His results indicate that both between and within cohort variation is evident in early and late adulthood.

Elder's work demonstrates the necessity and benefits of combining both the inter and intracohort type of analysis. The difficulty arises when the researcher must decide whether to focus on inter or intracohort comparisons. Although there are no general guidelines, some insights can be gained by considering the movement o f cohorts through time. In the next section we focus on a variety o f issues related to cohort adaptations.

COHORT ADAPTATIONS

The concept of cohort is based on the premise that each cohort carries the imprint of early socialization forward in time. The traits that characterize a specific cohort are expected to be reflected in their behavioral and attitudinal patterns ,. subsequent years. Again, the power of age to differentiate the characteristics of groups and individuals over several decades of time must be critically examined. The coherent nature of a cohort is partially a function of the relative plasticity of human behavior on one hand, and the rate of social turnover on the other.

Variation exists between individuals when their lives become more different as environmental and self-initiated pressures force them to change. Whether intercohort effects continue to exist over time will depend on the amount of similarity existing between adults of similar age. One source of pressure to change results from the aging process itself, in that age norms vary (Neugarten and Datan, 1973) and thus precipitate change over the entire course of the life span.

The nature of any single cohort over time represents a problematic area for the further refinement of cohort dynamics. Ryder (1965) stated that "cohorts can be pulled apart gradually by the slow grind of evolutionary change (p. 851)." We assume a similar position with respect to this issue and urge that consideration of the nature and quantity of change in adulthood be related to the cohort perspective in a more systematic manner. The impact of experience may create a need for socialization during adulthood that mitigates the effect of earlier periods of socialization. Various researchers have demonstrated that intracohort differences are greater among older cohorts than among the younger ones (Baltes, Reese and Lipsett 1980). One explanation for this difference may be the fact that older individuals have been confronted with a wider variety of experiences than younger adults, hence 9 their life course becomes more differentiated with age (Clausen 1973). Consequently, this would create greater differences between adults of similar ages than between groups of dissimilar age.

Previous developments suggest that there may be no fundamental variation in the mechanism of change occurring at various stages of life (Gewirtz 1969). Social-learning theorists, such as Mischel (1969) for example, argue that socio-environmental conditions shape and support one's behavior and attitudes. Inherent in this conceptualization is the notion that a changing environment produces modifications in behavior. This is important for the dynamics of intercohort differentiation in that changes during middle and late adulthood may create forces which pull apart the identifiable nature of cohorts.

As mentioned above, it is necessary to examine all sources of influence if we are to be capable of explaining behavior. Among adults, the effects of past historical changes may affect one's characteristics at present, but more recent socio-historical events are also at work in shaping behavior. In fact, as an individual ages, s/he has been exposed to a greater succession of socio-historical events. It is for this reason that we suggest that a cumulative historical effect may be in operation among adults of varying ages. In other words, an individual's characteristics are molded by a life-long sequence or events, not just those experienced during childhood and earls adulthood.

In sum the issues of cohort versus generation, intracohort variation, and cohort adaptation through time provide some interesting perspectives on individual development. However, the impact of these varying perspectives and the "value added" of these distinctions has yet to be addressed in a consumer context. Rather that focus generally on how these issues relate to consumption, in the next section we reinterpret and critique studies that have not made these distinctions. It should be noted that we have selected research of the highest quality for this purpose. On many cases we are simply adding an additional interpretation to the results rather than a critique of the research.

ALTERNATIVE INTERPRETATIONS OF PREVIOUS RESEARCH

Areas of research in consumer behavior which emphasize sociological interpretations of behavior, typically, do not consider the issues raised in this paper. In order to stress the importance of these issues, the present section suggests alternative methods of analysis or reinterpretations of results previously published in the consumer area. In some cases the alternative methods may provide additional insights into the area, in others, it may significantly alter the interpretation of the findings.

Cohort Analysis

Cohort analysis was first introduced into the consumer literature by Reynolds and his colleagues (Rentz and Reynolds 1980, Rentz, Reynolds and Stoudt 1983). They proposed that additional insight may be obtained by examining age effects in three ways: age, period and cohort. Age effects represent the standard chronological interpretation of age. Period effects refer to impact of a given historical period and cohort effects relate to life in a given era often indexed by birth year. It is important to consider this pioneering effort since the results they obtained differed from previous age interpretations. In one study they note

In terms of prediction, the results of the study indicate that total soft drink consumption will not decline as a result of larger cohorts... this prediction is in direct contrast to predictions based upon age interpretations of cross-sectional data (Rentz, Reynolds, and Stoudt 19 83, p. 19)

In light of these findings it is important to critically examine age interpretations of previous findings. This need to reexamine is consistent with an earlier observation that age is typically used as a substitute for other phenomena that are less easily measured (Zaltman, Alpert, and Heffring 1980). Emphasizing only a chronological interpretation of age differences may further result in construct validity problems. That is, age is only a proxy measure for the underlying phenomenon of interest.

The interpretation of age effects is an issue which often arises in sociological interpretations of socialization, norm development and symbolism. Considering first the socialization area. Frequently one encounters studies which have various age groupings (e. g., ages 6-8, 9-12). Results from these studies are typically attributed to age differences or maturational effects (see Ward 1974 for a discussion-of these types of studies). Another often used strategy involves the selection of a specific age-groups to assess "age effects" (cf. Moschis and Churchill 1978). The principal limitation with both of these approaches is their failure to consider effects due to the historical period or to cohort differences. By employing chronological age or Piagetian models these approaches suggest that age differences are due to intellectual or maturational development. This interpretation necessarily assumes that these effects are not linked to the historical circumstances of a given era. However, the socialization process of an individual who is age six in 1984 is very different than the socialization process of a child who was age six in 1964. Hence the total effect is more than simply chronological age. Developmental models or socialization, however, are not sensitive to these effects. Cohort analysis, on the other hand, is ideally suited to assess changing patterns in behavior by isolating cohort differences.

The importance of cohort interpretations can also be applied to a very interesting symbolism study by Belk, Bahn and Mayer (1982). They examined the tendency of individuals to make inferences about others based upon their choice of consumption objects (i.e. cars). Analyzing data across six age groups, the authors uncovered several choice patterns. One finding was reported as '... consumption-based stereotyping follows a curvilinear pattern, emerging sometime between preschool and second grade and tending to decline after college age (p. 10)." Thus they interpret this result in terms of chronological age. This interpretation assumes that as college age students enter "adulthood" they will shift their perceptions to be congruent with the adult age group. Without longitudinal data or a cohort analysis, however, one is unable to untangle age, period, or cohort effects. As a result, two alternative explanations of their data are possible. One is that the effect may be due to the current historical period. That is, there is something unique about the early 1980' s that accounted for the differences between the college age and adult samples. A second, and more likely interpretation, would be that cohort effects are responsible for the outcome. Were there events or social circumstances that occurred during the lifetime of the adult sample that resulted in distinct perceptions? Clearly the adult population and college age population experienced unique histories, these unique histories or cumulative effects are cohort effects. Chronological age simply refers to a maturational concept devoid of any historical meaning. In short, we would be more inclined to interpret their findings as cohort effects not age effects.

A similar interpretation issue is evident in family decision making. This is particularly true of proponents of the family life cycle (e.g. Murphy and Staples 1979). In the Murphy and Staples (1979) revision of the FLC the middle age category includes head of household ranges which span upwards of twenty years. Since a central assumption of the FLC is homogeneity within and heterogeneity between categories, this wide age range is particularly troublesome. However, to our knowledge no one has systematically analyzed the within stage variation that may be a function of cohort effects. Isolating these effects would enhance the stability of FLC predictors by lowering the error variance within stages.

Cohort Variation

Variation between cohorts may take two forms: intercohort and intracohort variation. Since the previous discussion centered principally on the need to consider intercohort interpretations, the present section emphasizes intracohort variation.

Rentz et al (1983) emphasized the variation that can occur between cohorts. The cohort categories used in the analysis were based upon age intervals spanning ten year periods. Yet recent developments in sociology and life-span psychology question the breadth of these intervals. For example, Elder's 11974) work on within cohort variation centered on the differential historical exposure within a one year cohort. In the Rentz et al analysis, additional analyses could be performed to assess the possibility of within cohort variation. Is the same cohort effect evident for social class, urban/ rural and across each year in the 10 year interval? It might be informative to divide the sample into age cohorts defined by birth year. This would allow one to isolate shifts in consumption in age-adjacent cohorts. Shifts in consumption patterns would signal the development of new market segments and perhaps a shift in firm resources.

The Belk et al (1982) study might also benefit from a within cohort analysis. We would argue that the adult grouping may not represent a homogeneous grouping of individuals. Although social class was examined in the analysis one cannot assume that a narrow variance of perceptions exists in such a broad category. We propose that subgroups within the adult subsample might report perceptions that are congruent with the college age population. Focusing analysis within these groups might allow one to isolate intracohort variations.

Work in consumer socialization also focuses on variations that occur across age groups. However, it might be more informative to consider how socialization differs within a group in a given historical period. For example, Elder's (1974) analysis of within cohort variation in socialization emphasized the differential effects on economic loss of children. This strategy resulted in a innovative, theoretically grounded split of the cohort, moving beyond demographics to isolate indicators of childhood variation. Similar comparisons could be made in terms of consumer socialization. Are there differences between youths from employed and unemployed families? To what extent are history-linked age norms (Neugarten and Datan 1973), prevalent in the socialization process? Does childhood or adult socialization in the 1960's differ from the 1980's?

Cohort Adaptation

As cohorts move through history they become more differentiated as a result of social and cultural change, age norms, and the cumulative effects of earlier life experiences. Given the number of factors which differentiate individuals it is surprising that the "elderly" are often considered a homogeneous segment. Recently authors have begun to analyze the young-old (65-75) and old-old (75+) as two separate age groupings. Yet this approach is subject to the same criticism. The age intervals are too wide to capture homogeneous segments. In response to the wide variability in the elderly population, a more systematic attempt to develop measures of what age is "supposed" to represent was undertaken (Zaltman, Alpert, and Heffring 1980). They concluded that "chronological age and the stereotypical variables it is sometimes used to represent may not correlate closely and...age may not predict particular other phenomena very well (p. 12)." They suggested that emphasis be given to the specific components of what age represents.

A second issue related to cohort adaptation is the idea of changing age norms. Concerning age norms Neugarten and Hagestad (1976) write:

Lifetime becomes transmitted into social time, and chronological age into social age. Age classes, age grades and age status systems emerge as social constructions. The interactions between age groups is socially regulated; the allocation of persons of different ages to given social roles comes to reflect the underlying age-status system, and age norms form a pervasive network of social control (P. 35)

In other words, being a certain age carries with it certain social expectations and behavior patterns which are defined by society relative to a given historical period. For example, having a child at age thirty four was once considered late now it is closer to the mean age. Unfortunately, very little work has been done to address age norms in the consumer literature.

A final issue concerns the cumulate life experience hypothesis. In order to enhance our understanding of history it is useful to examine cohorts across various time periods. Isolating cultural events and social changes which have occurred during the lifetime of the individual helps to explain their present consumption activities. Historical amnesia is perhaps more characteristic of researchers in scientific disciplines than it is of individual consumers.

WHY THE SLOW ADOPTION?

With the exception of the Reynolds and Rentz series, consumer researchers have been slow to adopt this perspective. Yet cohort types of analysis can be used to isolate shifts in socialization, norms, values, behavior and so forth.

Why are consumer researchers slow to adopt the cohort perspective? First, cohort forms of analysis are rooted in sociological/demographic traditions. Since most consumer research is psychological, few researchers have an interest or desire to pursue broad social-historical issues. Second, it is more convenient to explain age differences in terms of age effects. Since cohort analysis has not received much attention in the literature, few researchers would criticize a study for ignoring cohort effects. Third, applying cohort types of analysis to a data set requires additional statistical knowledge. Researchers may be unwilling to invest the time to learn the estimation procedures. Finally, findings based upon cohort effects are ripe for alternative explanations due to the number of intervening factors that may be present during any particular historical era. As a result it may be difficult to reach agreement on the nature and relative impact of historical events. Despite these limitations cohort analyses provides greater insight and understanding into the historical nature of consumption activities.

CONCLUSION

Although an attempt was made throughout this paper to discuss the implications of historical awareness, it is important to summarize the principal arguments and implications. The simplest message is that focused attention on the concepts of cohort, intracohort variation, and cohort adaptation provides a wider interpretation of factors which influence consumption.

The distinction initially made between cohorts and generations is necessary in order to establish the impact of social and cultural changes. Cohorts constructed on the basis of some shared encounter with history provides only one operationalization of historical effects. A lineage or familial generation and "social action" subgroups comprise the two other forms of historical groupings.

Variation in attitudes and behavior can occur both between (inter) and within (intra) cohorts. This variation can be traced to various individual resources and the effects of life experiences. Finally, cohorts do not experience discrete events. Rather, history leaves its imprint at each stage of life. This implies that older age-cohorts will be most differentiated in terms of life experience.

In this article we stressed the need for precise interpretations of age variation both between and within cohorts. Grouping the elderly into one consumer segment is one example of a lack of understanding of cohort variation. Perhaps more attention needs to be devoted to the "error variance" within subsegments of the population.

Finally, we have suggested some benefits of focusing on cohort movement through time. Emphasis placed on understanding individuals in historical context is not a new idea. However, it is new to consumer research and it makes a great deal of sense at a time of rapid social and cultural change.

REFERENCES

Baltes, Paul B., Hayne B. Reese and Lewis P. Lipsett (1980), "Life-Span Developmental Psychology," Annual Review of Psychology, 31, 65-110.

Belk, Russell, Kenneth D. Bahn and Robert N. Mayer (1982), "Developmental Recognition of Consumption Symbolism," Journal of Consumer Research. 9 (June) 4-17.

Bengston, Vern L. and Neal E. Cutler (1976), "Generations and Integenerational Relations: Perspectives on Age Groups and Social Change," in Handbook of Aging and the Social Sciences, R.H. Binstock and E. Shanas eds., New York: Van Nostrand Reinhold Co.. 130-159.

Clausen, John A. (1973), "The Life-Course of Individuals," in Aging and Society: A Sociology of Age Stratification, Matilda White Riley, Marilyn Johnson, and Anne Foner eds., New York: Russell Sage Foundations, 457-514.

Elder, Glen H. Jr, (1974), Children of the Great Depression, Chicago: University of Chicago Press.

Elder, Glen H. Jr. (1981), "History and the Family: The Discovery of Complexity," Journal of Marriage and the Family, (August), 489-518.

Gewirtz, Jacob (1969), "Mechanisms of Social Learning: Some Roles of Simulation and Behavior in Early Human Development," in Handbook of Socialization Theory and Research, D.A. Goslin ed., Chicago: Rand McNally and Co.

Haraven, Tamara K. (1978), "Introduction: The Historical Study of the Life Course," in Transitions: The Family and Life Course in Historical Perspective, Tamara Haraven ed.. New York: Academic.

Kertzer, David I. (1983), "Generation as a Sociological Problem," in Annual Review of Sociology, Ralph H. Turner and James F. Short Jr. eds., Palo Alto: CA: Annual Reviews Inc, 125-149.

Mannheim, K. (1952), "The Problem of Generations," in Essays on the Sociology of Knowledge, Paul Keczkemeti ed., New York: Oxford University Press, 276-322.

Mischel, Walter (1969), "Counting and Change in Personality," American Psychologist, 24, 1012-1028.

Moschis, George P. and Gilbert A. Churchill Jr (1978), "Consumer Socialization: A Theoretical and Empirical Analysis," Journal of Marketing Research, 15 (Nov. ) 599-609.

Murphy, Patrick and William Staples (1979), "A Modernized Family Life-Cycle," Journal of Consumer Research, 6 (June), 12-22.

Neugarten, Bernice L. and Nancy Datan (1973), "Sociological Perspectives on the Life-Cycle," in Life Span Developmental Psychology: Personality and Socialization, Paul B. Baltes and K. Warner Schale eds ., New York: Academic, 53-69.

Neugarten, Bernice L. and Gunhild O. Hagestad (1976), "Age and the Life Course," in Handbook of Aging and the Social Sciences, Robert H. Binstock and Ethel Shanas eds., New York: Van Nostrand and Reinhold Co., 35-55.

Rentz, Joseph O. and Fred D. Reynolds (1980), "Separating Age, Cohort and Period Effects in Consumer Behavior," in Advances in Consumer Research, Vol. 8, Kent B. Monroe ed., Washington, D.C., 596-601.

Rentz, Joseph 0., Fred D. Reynolds and Roy G. Stout (1983), "Analyzing Changing Consumption Patterns with Cohort Analysis," Journal of Marketing Research, 20 (February), 12-20.

Reynolds, Fred D. and Joseph 0. Rentz (1981), "Cohort Analysis: An Aid to Strategic Planning," Journal of Marketing, 45 (Summer), 62-70.

Riley, Matilda White, Marilyn Johnson and Anne Foner (1972), Aging and Society: A Sociology of Age Stratification, New York: Russell Sage Foundation.

Ryder, Norman B. (1965), "The Cohort as a Concept in the Study of Social Change," American Sociological Review, 30, 843-861.

Ward, Scott (1974), "Consumer Socialization," Journal of Consumer Research, 1 (September), 1-14.

Wonlwill, J. (1970), "The Age Variable in Psychological Research," Psychological Review, 77, 49-64.

Zaltman, Gerald, Mark Alpert and Michael Heffring (1980) "Issues in the Measurement and Use of Chronological Age and Nonchronological Age," Paper presented at the Association for Consumer Research 11th Annual Conference.

----------------------------------------