Considering Generations From a Lifespan Developmental Perspective

Abstract

We extend recent critiques of research on generations in the work context by proposing a differentiated lifespan developmental perspective. We accomplish this through several means: First, we define generations and trace the historical development of this concept from sociological thought experiments to their contemporary (mis)use for understanding individual-level psychological processes at work. Second, we review core principles of the lifespan developmental perspective, with an emphasis on contextual-dialectical models of human development. Third, we argue that generations are better understood from a contextualized lifespan framework that accounts for time period and history-graded developmental influences that may impact individuals’ attitudes, values, beliefs, motives, and behavior at work. Fourth, we propose a new lifespan model of generations and offer several associated propositions to serve as a theoretical guide as well as an agenda for future research. Finally, we review and critique methodologies typically used to study generational effects, outline methodological recommendations to guide future studies, and offer practical recommendations based on our model. Overall, this article contributes to better theorizing and more rigorous research and practice regarding the idea of generations at work.

Work and organizational psychology researchers and practitioners have become increasingly interested in the sociological notion of generations, along with psychological theories of lifespan developmental processes (Costanza & Finkelstein, 2015; Finkelstein, Truxillo, Fraccaroli, & Kanfer, 2015; Joshi, Dencker, & Franz, 2011; Rudolph, 2016). Curiously, these two topics have not been jointly considered within this literature. This is unfortunate, because there have been recent criticisms of the study of generations in the workplace that beg for further theoretical elaboration. In the following, we offer a perspective that integrates these diverse literatures, focusing on generations from a lifespan developmental perspective.

Our overarching goal is to extend critiques of generations research (e.g., Costanza & Finkelstein, 2015; Lyons & Kuron, 2014; Parry & Urwin, 2011) by proposing a differentiated lifespan perspective on age-related differences in workplace processes and outcomes. We aim to bolster the development of theory that appropriately distinguishes age-related factors, and propose the exploration of alternatives to generational concepts. We accomplish this through several means. First, we define generations and critically outline the historical development of this concept as it emerged and has been adopted by psychologists who study aging and work. Second, we outline key tenets of the lifespan developmental perspective, highlighting similarities and differences between sociological and psychological accounts of generational effects. Third, we explore how generations can be understood through a lifespan developmental lens, focusing on how contemporary time period and history-graded developmental influences impact individuals’ attitudes, values, beliefs, motives, and behavior. Fourth, we propose an integrative lifespan model of generations and associated propositions. Finally, we outline an agenda for future research and practical recommendations to guide future practice based upon our model.

THE GENERATIONS CONCEPT

In its common sociological conceptualization, the term “generation” refers to a group of people born during the same time span, who by virtue of their chronological age proximity have shared similar life experiences (e.g., major historical events, see Eyerman & Turner, 1998). The modern social-scientific usage of the term is most often linked to a thought experiment offered by Mannheim (1927/1952), which recapitulates a classic philosophical problem explored by Hume (1777/1985). At hand is the following question: What if a single generation lived in perpetuity, such that the concept of generations was absent? This question was asked as a means of giving perspective to the study of social change, by positing whether or not such change would ever occur if new birth cohorts never challenged the ideas of previous cohorts. This notion led to a paradigmatic shift in sociological thinking, and gave an immense amount of power to the generations concept (i.e., an aggregate phenomenon) as a catalyst for social change (i.e., also an aggregate notion).

Various functionalist lenses for defining and exploring the generations concept have been considered. As a school of sociological thought, functionalism seeks to understand order within society by focusing on social consensus and shared constructions. From a functionalist perspective, generations may be reasonably invoked as a mechanism for explaining societal dynamics. However, the notion of generations has been adopted by psychologists as an explanation for disaggregated individual-level behavior. Their argument typically follows something like, “Individuals that fall into ‘Group A’ display, on average, some characteristic that is in some capacity different from that of ‘Group B’, and we can tie this difference to a construction we have labeled ‘generations’ that differently defines these groups.” Different generational groups are typically defined by nonoverlapping age-brackets (e.g., people born between 19XX and 19XX + T are members of “Generational Group A,” whereas people born between 19YY and 19YY + T are member of “Generational Group B,” where “T” represents a time increment, typically in years).

Across systems of psychological thought, we see different perspectives on the role of generational groups for explaining behavior. Here, we explore the nuances of the lifespan developmental perspective as it relates to understanding and defining generations. We take this perspective because, arguably, work and organizational psychology has generally adopted a rather “black-box” understanding (cf. Lawrence, 1997) of the theoretical mechanism linking generations to work behavior. That is to say, generations are defined in terms of different perspectives (e.g., on the basis of different life history experiences, social contexts, broadly defined socioeconomic opportunities) each of which may lead to very different conclusions when comparing members of one generation to another. However, the overt psychological mechanisms that actually lead to these different perspectives are not explored in this literature (i.e., differences between generations are assumed to be caused by something that resides in the generational “black-box,” but this thing is not observed, measured, or evaluated—rather the mechanism is inferred from group membership formed on the basis of arbitrary age-bracketing).

Mannheim’s View of Generations

Even within sociology there are marked differences in the understanding of generations, as defined by various theoretical perspectives. For example, Mannheim viewed generations more as historical units, and suggested that there is substantial within-generation variation in responses to historical contexts that might give rise to a generation. Whereas Mannheim’s work is often cited as an explanation for the rigid means of defining and classifying individuals into generational groups, it is somewhat ironic that his own writing on the topic recognizes that there is far more variability within generational groups than modern reinterpretations may lead us to believe. Most notably, Mannheim argues that birth year proximity represent but one circumstance under which a shared behavioral consciousness may emerge.

Ryder’s View of Generations

Ryder (1965), who uses the term “birth cohorts” in lieu of “generations,” presents a different and rather simplified interpretation of the ideas offered by Mannheim. Ryder’s perspective shuns the notion of shared consciousness that Mannheim offered, and instead suggests that social change could be tied directly to clearly bounded birth cohort (i.e., age-bracketed) groupings, presuming that a shared consciousness or purpose was unnecessary to define a generational group. Arguably, the strict definition of generational categories on the basis of birth cohorts that we see in contemporary psychological research stems from the conflation of these ideas presented by Mannheim and Ryder.

Ryder’s perspective is reductionist in that it aims to simplify the task of understanding the role of generations in the development of social change. In much the same way that demographic factors such as age and gender serve as proxy indicators for differences in attitudes, traits, values, or other characteristics, so do cohorts as Ryder would define them. In many respects this is unfortunate, for although Mannheim’s view does not give rise directly to testable propositions, at least the proposed mechanism (i.e., the development of a shared consciousness among members of a generational group) hints at the possibility for empirical elaboration. Ryder’s approach has been aptly criticized as too cohort-centric (i.e., giving too much power to the idea of cohorts) and representing cohort determinism (i.e., the assumption that all members of a birth cohort experience aging in the same way, see Walker, 1993).

Riley’s View of Generations

Age stratification theory (e.g., Riley, 1973, 1987; Riley, Johnson, & Foner, 1972) recognizes that biological aging is accompanied by changes in social roles and positions over time. Society defines age-graded roles (e.g., transitions from student to worker to spouse to parent to retiree, etc.) and various norms that accompany such roles are likely to change over time. Social change can thus be explained by shifting values associated with these role-norm dynamics. Thus, age stratification theory gives substantial power to the role of generations as a unit for social change.

The birth cohort is the entity that defines generations according to age stratification theory (i.e., people born at about the same time). Such cohorts simultaneously move through a society’s age-graded systems (e.g., experience similar role transitions around the same time). Each experience presents new roles and challenges to the members of a given cohort; likewise, each cohort brings a new set of norms and values to bear on such experiences. Thus, the succession of cohorts over time represents a catalyst for social change, because different historical experiences give rise to unique influences on existing social structures (Riley, Foner, & Waring, 1988). Age stratification theory has been criticized for viewing society as a homogenous structure that similarly and consistently affects all members of a birth cohort (e.g., Novak, 2012). More generally, both Ryder and Riley’s perspectives on generations largely ignore the role of individual agency, experiences, and sensemaking processes for understanding the historical contexts that purportedly give rise to generations (e.g., experiencing national conflict may lead one person to patriotism and another to pacifism). This implied notion of “agency within structure” is reflected in the more contemporary life course perspective.

Elder’s View of Generations

Whereas the macro-level functionalist perspectives offered by Mannheim, Riley, and Ryder were aimed at explaining social change at the level of society (c.f., Mannheim does tacitly recognize the role of individual agency), Elder’s life course perspective instead focuses on the meso-level implications of social change on individual lives. The life course consists of various roles that define a biography—an individual’s accounting of their own developmental course. Life course scholars study processes at different levels of analysis that transition people into various roles, direct people along trajectories within these roles, and signal people’s exit from such roles. The life course perspective acknowledges individual, biologically driven trajectories of aging, however in the functionalist tradition, a significant emphasis is placed on the role of social structures and the embeddedness of one’s biography within social institutions. Such biographies are organized around various age-graded roles and relationships that define a normative developmental course. Life course theory further recognizes that three chronologies—historical, biographical, and social time—comprise a dynamic system that defines the interaction of individual developmental courses embedded within larger social systems (e.g., Elder, 1996; Moen, 1998). One advantage of the life course theory for understanding generations is the acknowledgement of individual developmental processes and the recognition that various contextual factors have the potential to affect developmental trajectories.

Defining and Differentiating Age, Period, Cohort, and Generations

The previous discussion makes it clear that there is remarkable variation regarding the use of the term “generation” present in the literature. Across disciplines (e.g., psychology, sociology, anthropology), a variety of different definitions, conceptualizations, and valuations of generations emerge. Highlighting this confusion, Kertzer (1983) refers to this “multivocality” as a “… liability in science” (p. 125). The definition of generations and the theory that describes their impact are intermingled to the extent that to explain one often necessitates the qualification of the other. For example, generations often represent groupings of birth cohorts that have some meaning attached to them, whereas any given birth cohort by itself can be thought of as value-free (i.e., decontextualized) generation (e.g., the cohort of people born in 2015).

The term birth cohort is used in various, often ambiguous ways in the literature. For example, Twenge and Campbell (2001) recognize: “A birth cohort is usually defined as all people born in a given year. However, the term can also be used more generally to refer to generational differences, which include larger numbers of birth years in the same group” (p. 322). On the other hand, the lifespan developmental literature typically uses the term cohort to refer to a single birth year. For example, Baltes and Reinert (1969) note: “Schaie (1965) formulated a well-explicated general model for the study of developmental problems which is based on three components: age, cohort (birthdate), and time of measurement” (p. 169). Given our argument for a lifespan developmental perspective, we take the position here that the use of the term cohort should only invoke the notion of a single birth year. It is only when we start adding meaning to artificial groupings of birth cohorts (e.g., via shared experiences that manifest across age-bracketed groupings of cohorts) that the generations construct emerges. In essence, we give the term “cohort” a “generationalized” meaning when we start ascribing significance to assumptions regarding the experiences of a particular group of people born within the boundaries of a particular arbitrary time span.

To make this discussion clearer, we adopt the following common distinctions between age, period, and cohort (Glenn, 2005): age refers to one’s biological/chronological age in years (i.e., years since birth), period refers to contemporaneous time (i.e., the current year), and cohort refers to one’s birth year (e.g., 1984). The complexities of “juggling” the effects of age, period, and cohort constitutes what sociologists have deemed, “the generation problem” (e.g., Kertzer, 1983). Differentiating these effects is important, because unless two of these three factors are defined by identical (assumed to be unmeasured) causal factors, then any conclusions drawn regarding the effect of the third may be spurious with respect to the other two. For example, in a cross-sectional research design, any correlations between chronological age and an outcome variable (e.g., job satisfaction) may be due to either age or cohort effects (i.e., because period is held constant in such designs). Thus, unless the same causal factor is driving both age and cohort effects in such an analysis, it is impossible to separate the two effects from one another, or to unequivocally attribute such an effect to either one or the other source of variance. This highlights a related issue, in that there is a simple, yet remarkably confounding linear dependency among age i, period j, and cohort k, such that

This dependency wreaks havoc on statistical models that attempt to separate age, period, and cohort effects, and reams of literature exist that have attempted to rectify this problem (e.g., Baltes, 1968; Baltes, Cornelius, & Nesselroade, 1979; Baltes, Reese, & Nesselroade, 1977; Kosloski, 1986; Schaie, 1965; see Glenn, 2005 for a review).

The introduction of the “shared experiences” component of generations bears some consideration here as well. This argument for the construction of generational groups rests on the idea that developmental experiences early during one’s lifespan (i.e., sociocontextual factors) are more important for the formation of stable traits (e.g., personality, attitudes, values) than later life experiences. Another way to put this, is that research concerning generations has assumed that there is diminishing intraindividual variability in traits over time (i.e., increased rigidity and decreased plasticity) associated with the process of aging. Such “crystallization” results in decreasing age-graded differential susceptibility to contextual influences. This manifests such that young individuals are more susceptible to contextual influences by virtue of flexibility in, for example, the capacity for socialization. This argument allows for the ability to tie birth year (i.e., cohort) to the shared experiences argument, and permits contemporaneous influences (i.e., period) to be ignored.

This argument likewise implies a certain level of age-related heterogeneity in the effects of such contextual influences that serve to define generations. Indeed, operationalizations of generational groups that are found in the literature assume agreement among the constituents within a “generation” (i.e., generational groupings defined by a range of birth cohorts exist at a higher level of analysis). This assumption suggests that there is more variability between generational groups than within, and this logic opens up several possible opportunities for misleading conclusions to be drawn based on improper conceptualizations of these different levels of analysis (cf. Klein & Kozlowski, 2000). For example, assuming that individuals within generational groups share the assumed characteristics of the group itself represents an ecological fallacy (i.e., erroneous inferences made about the nature of individuals on the basis of their group membership; Robinson, 1950; Thorndike, 1939). Moreover, because inferences regarding “generational differences” are often made on the basis of aggregated individual level data, the possibility for atomistic fallacies is present (i.e., erroneous inferences made about the nature of groups on the basis of relationships observed among individuals; Ostroff, 1993). Mirroring this, Walker (1993) suggests that assuming generational groups defined by clusters of birth cohorts possess higher-level meanings may lead one to fall into a “cohort trap” (p. 144) that is defined in terms of the potential for both ecological and atomistic faults in thinking.

The classical and oversimplified solution to the age–period–cohort problem is to ignore period effects, on the basis of the crystallization and ratification argument (Ryder, 1965). This argument offers that period effects can be ignored, as most attitudes, perceptions, and trait-like individual differences are formed early in life. This is the same logic applied by contemporary cross-temporal meta-analysts (e.g., Twenge, Konrath, Foster, Keith Campbell, & Bushman, 2008). For example, Gentile, Wood, Twenge, Hoffman, and Campbell (2015) suggest that to accomplish the isolation of cohort effects, age and period effects are necessarily and justifiably confounded in cross-temporal meta-analysis. Despite this argument, it is easy to find examples of macro-level contemporaneous contextual effects (i.e., period effects) that can manifest as behavioral, attitudinal, or affective dynamics for people of all ages at work, regardless of birth year. Table 1 summarizes several empirical studies that have considered the role of period effects on work-relevant outcomes.

Furthermore, at a more micro-level, research on trait plasticity points to the within-person malleability of certain personality traits across time (e.g., Park & Reuter-Lorenz, 2009; Roberts, Walton, & Viechtbauer, 2006; Roberts, Wood, & Smith, 2005; Staudinger, 2015). Other work has similarly considered dynamics in wellbeing across the lifespan, pointing to age-related effects (i.e., nonlinear U-shaped dynamics in happiness, life satisfaction, depression, and anxiety) that are irrespective of cohort effects (e.g., Blanchflower & Oswald, 2008). These lines of research call into question the fundamental assumptions of the crystallization and ratification argument as applied to the study of cohort effects in cross-temporal methodologies. Moreover, the application of such methodologies is subject to self-criticism by those applying them (i.e., the recognition of the limitations of confounding cohort and period effects; e.g., Twenge et al., 2008, p. 894). Period effects are also being recognized by more recent cross-temporal work, which has largely ignored them in the past (e.g., Leckelt et al., 2016).

Another solution to the age–period–cohort dependency problem is to artificially split and combine birth year cohorts into generational groups (e.g., Davis, Pawlowski, & Houston, 2006; Dilworth & Kingsbury, 2005; Hess & Jepsen, 2009). This splitting and combining procedure removes some of the dependency associated with age, period, and cohort effects by dividing age into categories on the basis of birth year. This technique has the advantage of providing an illusion of theoretical sophistication, and a great deal of effort has been applied to constructing (i.e., naming and classifying general characteristics of) generational groups. This is the statistical procedure through which the adoption of the “shared experiences” argument manifests in much of the psychological literature that addresses generations. Artificial bifurcation has the added advantage of synthetically changing the nature of age-relevant variability present in one’s data. On the one hand, this adds to the misconception that generational groups are meaningful, and on the other, furthers the divide between reality and the inferences drawn from our data (e.g., MacCallum et al., 2002; Rudolph, 2015). Curiously, studies adopting this methodology rarely attempt to estimate age or period parameters; the simplification of cohort effects by themselves is often sufficient to justify conclusions regarding generational effects.

The preceding review has been critical of the application of sociological notions of generations to the study of individual-level behavior within psychology. This critique leaves us with an important question to consider next: “If traditional sociological conceptualizations of generations are such a poor match to psychological theories, what can we do about it?” In the next sections, we introduce the lifespan development perspective as a reasonable and defensible theoretical framework for understanding age, period, and cohort-like ideas within psychological research on work and aging.

THE LIFESPAN DEVELOPMENTAL PERSPECTIVE

The lifespan developmental perspective is an integrative, meta-theoretical, and multidisciplinary framework. It aims to describe and explain stability and change in experiences and behavior across time, and to optimize developmental processes (Baltes, 1987; Baltes, Lindenberger, & Staudinger, 2006; Baltes, Reese, & Lipsitt, 1980; Baltes, Staudinger, & Lindenberger, 1999). Lifespan researchers are interested in identifying general principles of development (i.e., intraindividual change), interindividual differences in development, and the extent and antecedents of intraindividual malleability in development. Baltes (1987) summarized seven key propositions of the lifespan perspective (Table 2). Although all seven propositions have implications for understanding generations from a lifespan perspective, propositions four (i.e., historical and sociocultural embeddedness) and five (i.e., the paradigm of contextualism) are most important in this regard, and will therefore be considered further.

Historical/Sociocultural Embeddedness

Despite its focus on describing, explaining, and modifying individual development (i.e., ontogenesis) the lifespan perspective is not mono-theoretical and person-centered, but is rather a multi-level, multi-disciplinary, and person–environment interaction framework (Baltes, 1987; Lerner, 1996). Even critics acknowledge that historical, sociocultural, and other contextual factors are well-integrated within the lifespan perspective (Dannefer, 1984). The proposition of historical and sociocultural embeddedness acknowledges that development is influenced not only by biological factors (e.g., genetics), but also by historical and evolutionary contexts (i.e., the time period during which an individual’s development unfolds), changes in sociocultural conditions (i.e., structural factors such as the economy, education, and medical care), and interactions of person and contextual factors. In describing developmental contextualism, Lerner (1996) emphasizes reciprocal influences of changes in multiple levels of organization. Thus, individual development is not only a product of contextual influences, but also as a process of actively shaping developmental and contextual factors at multiple levels (e.g., within families or organizations, see Lerner & Busch-Rossnagel, 1981).

The importance of historical and sociocultural factors to the lifespan perspective was informed by two lines of research. First, sociological research demonstrated mutual relationships between individual development and intergenerational continuity and change (Elder, 1994). Specifically, this research demonstrated that individuals’ development both influences and is influenced by biological and sociocultural intergenerational exchanges (e.g., fertility, parenting, and grandparenting, but also intergenerational distributions of resources at a societal level). Second, this research found evidence for strong birth cohort effects on developmental outcomes, such as cognitive abilities and personality characteristics (Baltes, 1968; Riley et al., 1972; Schaie, 1965). In particular, Schaie’s (1994, 1996, 2013) research on adult intellectual development found that fluid cognitive abilities (e.g., working memory, processing speed) typically decreased with age; at the same time, there were substantial improvements in these abilities across successive birth cohorts (Gerstorf, Ram, Hoppmann, Willis, & Schaie, 2011; Schaie, 2013). Elder and colleagues (Elder, 1974; Elder & Liker, 1982) demonstrated that historical events (i.e., the Great Depression) experienced in childhood were associated with adults’ personality. Similarly, Nesselroade and Baltes (1974) showed that changes in adolescents’ personality were influenced by both age-related factors as well as historical contexts (i.e., the Vietnam War). Beyond these seminal works, however, empirical studies of cohort effects on individual development are rare; research has instead adopted a methodological or descriptive understanding of such phenomena (Baltes et al., 1979).

The lifespan perspective assumes that cohort effects must be explained using a multidisciplinary approach. For instance, Baltes (1987) acknowledged that “… classical psychological theory has little to offer when it comes to interpreting the substantive meaning and origin of cohort effects …. The fields of cultural anthropology, historical sociology, and historical medicine may prove to be more relevant” (p. 620). Possible explanations for cohort effects in intellectual development may involve continuous improvements in education, health and medical care, and the increasing complexity of work and home environments (Baltes, 1987). Cohort effects on certain outcomes (e.g., cognitive ability) should be stronger when the mechanisms involved (e.g., complex work environments) are specific and theoretically relevant for the outcome of interest (Zacher, 2015b). For instance, individuals’ attitudes toward job security may be more influenced by the extent to which jobs were available when they first entered the labor market than by other contemporaneous events (e.g., protests against a war).

Baltes and colleagues (1979) argued that the way researchers understand potential cohort effects depends on their disciplinary background, theoretical orientation, and research questions. Specifically, cohort effects can be treated in four different ways. First, they can be treated as error. Most psychologists interested in basic and normative processes of child development would treat cohort effects in this way (i.e., they would be generally neglected). Second, cohort effects can be treated as transitory, historical irregularities that temporarily disturb normative developmental processes. Third, cohort effects can be conceived as systematic quantitative changes in levels of an outcome without a general change in underlying mechanisms. Finally, cohorts can be treated as a substantive theoretical mechanism or process variable, which requires explication of the “... form and nature of cohort change that is judged to be developmental, the need for such concepts as stages or transitions in representing cohort change, and the type of explanatory mechanisms involved in producing cohort change” (Baltes et al., 1979, p. 80).

Most psychologists would treat cohort effects in one of the first three ways, assuming that cohort effects and sociocultural changes are generally not relevant for those domains of psychological development established during genetic and cultural evolution (Baltes, 1987). For instance, Baltes (1987) argued that in developed countries, cohort effects on cognitive development are rather small due to the stability of the biocultural architecture. In contrast, sociologists are more likely to adopt the fourth approach (Baltes & Nesselroade, 1984; Dannefer, 1984), which suggests that findings regarding individual development cannot be generalized to other historical epochs. Indeed, sometimes cohort effects are used to negate the existence of any developmental regularities and to argue for pure historicism (Baltes, 1987). According to Baltes and Nesselroade (1984), “… the cohort concept legitimately has a different predominant status in sociology than it has in psychology. Our prediction was that sociologists would have a tendency to view cohort effects as an essential aspect of any sociological theory (representing cohort as a theoretical process variable), whereas many psychologists, due to their primary emphasis on individual-level paradigms and the search for invariant behavioral laws, would generally opt for the first three types” (pp. 844–845).

The Paradigm of Contextualism

The paradigm of contextualism assumes that individuals have to process, react to, and act upon normative age-graded, normative history-graded, and non-normative influences that co-determine development (Baltes, 1987). Normative age-graded influences include person (i.e., biological) and contextual (i.e., environmental) determinants of development that most people encounter as they age, such as biological maturation, deteriorating senses, and declining physical strength, as well as very common socialization events (e.g., school entry, marriage, birth of children, retirement). Thus, these determinants are strongly related to age and are predictable in terms of their onset and duration. Importantly, the use of the term normative is understood in a statistical-descriptive, not value-based prescriptive sense (Baltes & Nesselroade, 1984). History-graded influences include person and contextual factors that are linked to a specific historical period (e.g., the Great Depression, World War II) that may differentially influence people’s development. Finally, non-normative influences include person and contextual determinants of development that are rather uncommon and whose manifestation is unique to each individual. Non-normative influences are not closely related to ontogenetic or historical development and are rather unpredictable (e.g., career changes, divorce, job loss, severe illnesses, accidents, death of significant others). Together, the dynamic and interactive impact of these three influence systems leads to stability and change, as well as multidimensionality and multidirectionality in individual development (Baltes, 1987).

Lifespan theory assumes that there are interindividual differences (e.g., based on gender, social class, ethnicity) in the experience and effects of these developmental influences, including the two normative influence systems (Baltes & Nesselroade, 1984). Moreover, Baltes and colleagues (1980) argued for the existence of general lifespan profiles that describe the relative influences of age-graded, history-graded, and non-normative influences on individual development across the lifespan (Figure 1). First, the influence of normative age-graded factors is U-shaped across the lifespan, with a strong influence in childhood, decreased impact in adolescence, and increased importance in adulthood and old age. With regard to intellectual development, Baltes and colleagues argued that there is less biological and cultural stabilization in adulthood and old age, which are characterized by more plasticity and potential for active control by individuals, compared to earlier life phases, which are characterized by much more biocultural regularity.

Second, the strength of normative history-graded determinants takes an inverted U-shape across the lifespan, with a peak in adolescence and early adulthood, and weaker influences in childhood and old age (Figure 1). Baltes and colleagues (1980) argued that adolescence and early adulthood are phases in which the relationship between the individual and society as well as intergenerational dialectics are particularly salient, and in which individuals—influenced by the sociocultural environment—lay the foundation for their adult life. Finally, Baltes and colleagues speculate that the strength of non-normative influences increases linearly across the lifespan. Important life events become more important predictors of development over time due to age-related declines in evolutionary-based genetic control over development and the increased multidirectionality, heterogeneity, and plasticity in developmental outcomes at higher ages. Although Baltes and colleagues emphasized the relative strengths of developmental influences across the lifespan, at no point are these sources irrelevant to individual development. For instance, Schaie’s (1979) cohort-sequential research suggests that cohort effects in intellectual functioning exist across different ages. Moreover, Baltes and colleagues pointed out that these profiles are prototypical and that their depiction does not take into account the interactions or transactions among the three systems of influences.

GENERATIONS FROM A LIFESPAN DEVELOPMENTAL PERSPECTIVE

Lifespan developmental researchers do not use the generations concept as introduced by sociologists such as Mannheim. When they use the term, they typically refer to individuals’ birth year cohorts (Baltes, 1968), but they do not make assumptions about broader categories of individuals based on rather arbitrary collections of birth years and shared life experiences of people born within certain timespans. The lifespan perspective defines development (i.e., intraindividual change) and interindividual differences in such intraindividual trajectories, but it does not categorize individuals based on their birth years or shared life experiences into broader generational groups. Consistently, Kosloski (1986) argued that defining a “generational cohort” by a shared time period does not automatically imply that individual members of that cohort shared common experiences, and that such cohort effects are meaningless.

The lifespan perspective does accept that historical and sociocultural contexts can impact individuals. However, it assumes that such contextual factors impact experiences and behavior at the individual level, not as shared generational effects. Even though the lifespan perspective suggests that history-graded effects are “normative,” it does not make assumptions about effects on collective experiences and behaviors of generational groups. Instead, the lifespan perspective argues that events and experiences associated with a historical period constitute factors that can impact each individual’s developmental outcomes and interact with age-graded and non-normative factors.

In the revised version of his classic general developmental model, which aims to methodologically disentangle age, period, and cohort effects on development (Schaie, 1965), Schaie (1986) criticized time-based cohort concept used in developmental psychology. He argued that the conceptualization of age, period, and cohort as chronological indices (i.e., years) is problematic, because they are “empty constructs” that are mere correlates of substantive factors and thus cannot have causal effects (Birren, 1999; Schaie & Hertzog, 1985; Wohlwill, 1970). Schaie (1986) suggested a theoretical framework in which cohort and period indicators are decoupled from calendar time and re-conceptualized as more meaningful variables. In this framework, the meanings of cohort and time period are broadened to have more explanatory value for individual development, such that a cohort is defined as “the total population of individuals entering the specified environment at the same point in time” (Schaie & Hertzog, 1985, p. 92) and time period is re-defined as “historical event time.”

For cohorts, the time of entry does not have to be birth, and can include time markers related to biocultural development, such as puberty, parenthood/grandparenthood, menopause or societal markers such as workforce entry, marriage, or retirement (Schaie, 1986; see similar life course sociology arguments by Diewald & Mayer, 2009; Mayer, 2009; Settersten & Mayer, 1997). More recently, the motivational theory of lifespan development has discussed cohort-defining events and processes as age-graded opportunity structures (Heckhausen & Shane, 2015; Heckhausen, Wrosch, & Schulz, 2010). The framework further suggests that there can be history-graded cohorts (e.g., such as being member of the initial staff of a start-up organization), as well as non-normative cohorts characterized by events such as divorce, infectious disease, or other social group memberships. Thus, cohorts are re-defined as an interindividual difference variable (Schaie, 1986). The uncoupling of period effects from calendar time involves identifying important historical events, and assessing the timing and duration of their greatest influence. Period is therefore re-defined as an intraindividual change variable (Schaie, 1986).

Integrative Lifespan Model of Generations

One of the themes we have argued for in the preceding review is that traditional sociological perspectives on generations are too deterministic and reductionist for understanding psychological phenomena concerning work and aging. It should be clear that, when applied, such perspectives are typically used to defend predictions at a higher level of analysis without clear or reasonable justifications (i.e., aggregate social phenomena explaining individual level behavior). Although recent work has advanced generations theory away from traditional sociological perspectives (e.g., Joshi, Dencker, Franz, and Martocchio, 2010; Rudolph & Zacher, 2015), no attempt has been made to integrate broader lifespan developmental perspectives into a testable theoretical framework. To coalesce these ideas and the theory reviewed earlier, we argue that a more contemporary model for understanding generations must be grounded in the traditions of lifespan developmental contextualism. To achieve this, our integrative lifespan model of generations can be viewed as a parallel argument to the contextual-dialectic model of lifespan development proposed by Lerner and Busch-Rossnagel (1981). The tenets and concepts of the lifespan development perspective should supplant outdated sociological notions to guide thinking and research. Thus, the propositions discussed here and summarized in Table 3 should serve as a foundation for the application of such lifespan perspectives to the study of generations at work.

The core argument of the contextual model of lifespan development is the constructivist notion that people are both the product and producers of their own developmental course. This idea leads to the assumption that, at the individual level of analysis, the influence of age-graded and historical/contextual influences are inherently codetermined and thus inseparable. This sentiment is reflected in earlier work by Baltes (1979, p. 2), who suggests:

As development unfolds, it becomes more and more apparent that individuals act on their environment and produce novel behavioral outcomes, thereby making the active and selective nature of human beings of paramount importance. Furthermore, the recognition of the interplay between age-graded, history-graded, and non-normative life events suggests a contextualistic and dialectical conception of development is the reflection of multiple forces which are not always in synergism, or convergence, nor do they always permit the delineation of a specific set of endstates.

Given this, the most fundamental proposition of our integrated lifespan model of generations is that the influence of historically graded and sociocultural context variables occurs at the individual (i.e., person) level of analysis, and not as a manifestation of shared phenomena:

Proposition 1. Historical and sociocultural contexts impact experiences and behavior at the individual level, not as shared generational effects.

Proposition 1 suggests that future research must focus on individual-level indicators of historical and sociocultural influences that have been hypothesized to give rise to generational effects. Offering that such effects manifest at the individual level opens up myriad psychosocial explanations for observed historically graded and sociocontextual influences on behavior. For example, one might argue for investigating differences in perceptions, attitudes, and values at the person level that can be tied to individual experiences and life histories—what Bronfenbrenner (1993) would refer to as an ecological developmental system (Moen, Elder, & Lüscher, 1995). Indeed, this idea represents an important departure from prior research, which has focused on “black-box” explanations for such effects (e.g., Gursoy, Maier, & Chi, 2008; Wong, Gardiner, Lang, & Couon, 2008). Directly measuring individual-level influences should aid in unpacking this black-box.

The notion that historical and sociocultural experiences serve as intraindividual influences is further supported by ideas derived from various contextual models of lifespan development (e.g., Baltes, 1979; Ford & Lerner, 1992; Lerner & Busch-Rossnagel, 1981; Moen et al., 1995). Such models highlight another facet of our argument against the prototypical conceptualization of generations. Just as we have argued that age, period, and cohort effects are empirically confounded, the notion of age-graded, history graded, and non-normative life events are phenomenologically confounded. That is to say, such mechanisms of developmental change are both functionally and theoretically inseparable. More specifically:

Proposition 2. Developmental contextualism implies that age, period, and cohort effects are codetermined and inherently inextricable.

The direct implication of Proposition 2 is that researchers must abandon time-based operationalizations of generations. Past research concerning generations at work has ignored this idea by conflating age, period, and/or cohort (e.g., Jurkiewicz, 2000; Lyons, Duxbury, & Higgins, 2007) making it impossible to extract meaningful conclusions regarding these individual effects from such studies. Given that time-based operationalizations of age, period, and cohort effects are inseparable and serve merely as proxies for actual psychological and developmental process, a corollary proposition follows from Proposition 2. Namely, and consistent with the arguments levied by Schaie (1986), a contextualized understanding of these processes necessitates the reconceptualization of time-based age, period, and cohort effects as measurable or observable psychological variables:

Proposition 3. A contextualized understanding of individual lifespan development necessitates alternative operationalizations of age, period, and cohort effects.

Given Proposition 3, two subordinate propositions can also be explicated on the basis of the ideas presented by Schaie (1986). Both of these propositions describe methodological concerns for defining and developing empirical investigations of such contextualized phenomena under the assumption that age, period, and cohort effects must be re-conceptualized (i.e., decoupled from calendar time). Thus:

Proposition 3a. Cohort effects are defined in terms of interindividual differences.

Proposition 3b. Period effects are defined in terms of intradindividual changes.

More specifically, Proposition 3a suggests that, in lieu of time-based operationalizations, cohorts should be operationalized as interindividual differences (i.e., as psychological individual differences). Additionally, consistent with Schaie (1986), Proposition 3b suggests that period effects should be defined in terms of individual-level dynamics (i.e., as with-person changes over time that can be tied to a specific occasion and place, life experience, or condition). Bearing these propositions in mind, let us explore methodological issues and recommendations for the implementation of this integrative, contextualized lifespan model of generations.

METHODOLOGICAL ISSUES AND RECOMMENDATIONS

To understand how this integrative lifespan model of generations can be applied to future research endeavors, and to address the propositions raised herein, it is necessary to discuss methodologies for studying such effects. This discussion will begin with a consideration of traditional developmental methodologies, and then consider new recommendations in line with the propositions raised earlier. Baltes and colleagues (1977) outline features, strengths and weaknesses of developmental designs, a summary of which is provided here. Generally, procedures for estimating cohort effects have been quasi-experimental in nature. Though we acknowledge these quasi-experimental methods, we also propose new ways of considering age, period, and cohort-like effects that better match the notion of contextualism that we propose here, including both updated quasi and traditional experimental methods (Table 4).

Quasi-Experimental Methodologies

Cross-sectional methods

Single time-point methodologies have typically been applied to study generational effects in work contexts (Costanza et al., 2012). In cross-sectional designs, people of varying ages are sampled on one occasion. Cross-sectional designs are of little utility for studying generations, because one cannot separate age, period, and cohort effects. Because data are collected at one time point in such designs, period effects are held constant (e.g., 1990; see example in Table 4, Panel A), whereas age and cohort variables are perfectly confounded with one another. Thus, cohort and age effects cannot be unequivocally separated in such analyses. In fact, such designs may lead to spurious conclusions regarding the nature of age effects in the presence of true cohort effects (Figure 2).

As suggested previously, one way to alleviate this dependency is to bundle cohorts into larger “generational groups.” In doing so, such groups are constructed on the basis of clustering birth cohorts in temporally ordered increments (e.g., D’Amato & Herzfeldt, 2008; Cennamo & Gardner, 2008; Leiter, Jackson, & Shaughnessy, 2009). The practice of splitting and recombining continuous variables in this way is frowned upon (e.g., MacCallum, Zhang, Preacher, & Rucker, 2002), and can lead to incorrect assumptions regarding the influence of cohort effects in the presence of actual age effects (Rudolph, 2015). This procedure (partially) removes the dependency between age and cohort (i.e., generations are aggregated cohorts in such a procedure). However, this does not lead one to make more valid inferences regarding the effect of cohort versus age. There are other issues with cross sectional methodologies that further limit the inferences one can draw from such designs. For example, single time point designs cannot inform hypotheses regarding age-related changes, nor individual stability.

Longitudinal methods

Longitudinal designs are likewise limited in their ability to study generations. In a single-cohort longitudinal developmental design, people born at the same time are sampled repeatedly across the span of several successive years (see Table 4, Panel B). This is the traditional developmental design of intraindividual change. In such designs, age and period are necessarily confounded, whereas cohort is held constant across time. Thus, no inferences regarding cohort effects are tenable in such designs, as the methodology constrains one to study a single cohort over time. In contrast, cross-sequential longitudinal designs start with a cross-sectional sample, and subsequently add a longitudinal sequence to each cohort. Though popular and nearly ubiquitous in the work and aging literature, cross-sequential methodologies are likewise problematic for the study of generations. This is because age and cohort effects are initially confounded by the cross-sectional nature of the baseline measurement, and subsequent observations cannot serve to tease such effects apart.

Cross-temporal methods

Apart from cross-sectional methods and longitudinal methods, cross-temporal or time-lag methods, including cross-temporal meta-analysis (e.g., Twenge, Campbell, Hoffman, & Lance, 2010), try to circumvent the linear age, period, and cohort dependency by isolating cohort effects and disregarding period effects. For example (Table 4, Panel C) one could compare outcomes (e.g., job attitudes) from samples of similarly aged respondents collected at multiple different points in time (e.g., job satisfaction of 40 year olds in 1990 vs. the job satisfaction of 40 year olds in 2010). In doing so, differences could be tied to either cohort or period effects, as age is held constant in such comparisons. The crystallization and ratification argument is then applied to justify overlooking period, and favoring support for cohorts as primary explanatory variables. Of note, period effects are equally likely in such cases, because of the confounding of cohort and period.

Longitudinal-sequential methods

Whereas simple longitudinal designs are of little value for studying generations, more complex longitudinal designs, including cohort-sequential designs and time-sequential designs can more appropriately models age, period, and/or cohort effects. Longitudinal-sequential designs combine elements of cross-sectional, cross-temporal, and longitudinal designs into single methodologies. Cohort-sequential designs involve sampling people of different birth cohorts repeatedly across a span of time. Such designs are essentially comprised of two or more overlapping longitudinal studies. Considering the example in Table 4, Panel D, three cohorts born in 1950, 1960, and 1970 are surveyed three times across (i.e., baseline, 10 years, 20 years) across a 20-year span.

There are distinct advantages to the use of cohort sequential designs compared to the other developmental methodologies reviewed earlier. For example, because different measurements are collected between cohorts over time, age is not automatically confounded with cohort (e.g., cohort comparisons of 40 year olds in 1990 and 2000 are acceptable in such a design). Perhaps of equal importance, we can also compare people to their own baseline levels, which cannot be achieved via cross-temporal methods. This allows for the direct modeling of within-person change relative to other factors in such models. Related to this, the staggering of baseline measurements (i.e., samples are drawn independently from different cohorts) means that longitudinal comparisons (i.e., differences in developmental trajectories) are meaningful between cohorts. Cohort-sequential designs also allow for cross-sectional comparisons, as different age groups are measured at each time point. Finally, because similar age groups are represented across different measurement occasions, cross-temporal/time-lag effects can also be represented (e.g., comparing 40 year olds in 1990 vs. 2000). It is important to note, however, that because period effects are confounded with cohort effects, that this design does not allow for one to draw strong conclusions regarding contemporaneous time of measurement effects.

A related methodology, the time-sequential design, consists of two or more cross-sectional studies conducted at different measurement occasions. The advantages of the time-sequential design largely mirror that of the cohort-sequential design. However, whereas the major strength of the cohort-sequential design is its ability to unconfound age and cohort, the strength of time-sequential designs is to unconfound age and period. The example in Table 4, Panel E shows one such plausible time-sequential design. Compared to cohort-sequential designs, there is some flexibility in sampling strategies that can be adopted in time-sequential designs. Specifically, the samples employed by such designs can be independent of each other (e.g., different cross-sectional samples collected at each time point) or the same across measurement occasions.

Proposed Quasi-Experimental Methodologies

The preceding review hints at the fact that separating age, period, and cohort effects within a single study design is difficult if not impossible without some other restrictions imposed upon one’s data (e.g., artificially grouping birth year cohorts). Our contextualized lifespan perspective suggests the need for a different approach to these issues. Consistent with Schaie, (1986), Little (2013) suggests that age, period, or cohort effects can be viewed as potential contextual variables that can be measured and modeled in one’s analyses. Specifically, including theoretically justifiable and well-operationalized alternative measures of such effects allows one to gauge the amount of variance that can be accounted for and/or control for their influences, statistically. Indeed, reasonable alternative (i.e., non time-based) operationalizations can be independently conceptualized for age, period, and cohort effects (see Table 5 for a summary).

For example, alternative measures of age beyond chronological age might be considered (e.g., age/age group identity, subjective age, self-perceptions of aging, attitudes toward aging, awareness of age-related changes, see Diehl et al., 2014). Still other proxy variables concerning functional (e.g., health) or cognitive capacities associated with successful aging at work (e.g., Rowe & Kahn, 1997; Zacher, 2015a) may be considered as alternative age operationalizations. In terms of period effects, various alternative indices at different levels of analysis may be considered. At the individual level, one might measure the influence of contemporaneous life events to address person-specific period effects. Such events could be subjectively assessed (e.g., “To what extent have major events changed your life in the past 12 months?”), constructed through biographical narratives (e.g., McAdams, 2006), or more objectively scaled (e.g., the Life Events Inventory; Cochrane & Robertson, 1973). At higher levels of analysis, research has considered contemporaneous economic conditions (e.g., unemployment rates) as an index of contextual factors that influences job insecurity (e.g., Keim, Landis, Pierce, & Earnest, 2014). Other research has tied contemporaneous unemployment rates to differences in narcissism (Bianchi, 2014) and job satisfaction (e.g., Bianchi, 2013). Indeed, similar national indices of wellbeing may also prove fruitful for addressing period effects in future analyses (e.g., OECD Indicators, Gallup-Healthways Wellbeing Index).

In terms of alternatives to cohort effects, rather than estimating such effects from birth year, one might consider collecting an index of generational identity as an indicator of cohort membership (Finkelstein, Gonnerman, & Foxgrover, 2001; Weiss, 2014; Weiss & Lang, 2009; 2012). Still another strategy would be to collect measures that reflect a person’s response, attitudes, or opinions surrounding a particular event, movement, or phenomena that that is suspected to give rise to the shared consciousness that defines generations. A related idea involves construing cohorts differently, for example around life transitions (e.g., post-secondary education to initial career) or perceived career stages (e.g., “time to retirement”). Construing such “developmental deadlines” as meaningful psychological time indices has a well-founded precedence in the lifespan developmental literature (e.g., Freund, 1997; Heckhausen & Tomasik, 2002; Haase, Heckhausen, & Köller, 2008).

Beyond alternative operationalizations, the application of the proposed quasi-experimental methodologies requires recognition of how these effects manifest at different levels of analysis. Fundamentally, questions of aging require the investigation of within-person (i.e., intraindividual) changes over time. Given our proposed model, the redefinition of period and cohort effects requires the explication of person-level effects (i.e., interindividual differences) that manifest between-persons, but can also take the form of cross-level effects (e.g., individual differences that may buffer or augment within-person processes over time).

Proposed Experimental Methodologies

As suggested previously, generational identity is one reasonable alternative operationalization of cohort effects for the study of generations (e.g., Joshi et al., 2010). This argument implies measuring the strength of one’s generational identity, along the lines of Finkelstein (e.g., Finkelstein et al., 2001) or Weiss (e.g., Weiss, 2014). Another reasonable possibility for conceptualizing this idea is through the manipulation of generational identity via momentary activation and priming of generational characteristics (e.g., Bargh & Morsella, 2010; Eschleman, 2016). Related age-based characteristics (e.g., age stereotypes) can be primed via such activation frameworks. For example, Stein and colleagues (2002) demonstrated that priming negative age stereotypes can impair memory task performance. Moreover, related research suggests that the embodiment of age-related stereotypes has long-term influences on intraindividual health and well-being outcomes (e.g., B. R. Levy, 2003; B. Levy, 2009). A related experimental approach involves the manipulation of time horizons (e.g., future time perspective, see Carstensen, 1993, 1995). For example, research that has experimentally manipulated temporal deadlines has suggested that both goal focus and contents shift with the perception of limited versus limitless timelines (see Carstensen, Isaacowitz, & Charles, 1999 for a review). Additionally, such manipulations need not be direct, as suggested by evidence from natural experiments concerning the dynamic interplay between contemporaneous period effects (i.e., the September 11th, 2001 terrorist attacks and the SARS outbreak of 2003) and socioemotional goals. Indeed, experiencing sociocultural events that increase the salience of the fragility of one’s life increases motivation to derive emotional meaning from life, regardless of one’s chronological age (Fung & Carstensen, 2006).

PRACTICAL RECOMMENDATIONS AND CONCLUDING THOUGHTS

The goal of this article was to offer an extension of recent critiques of research on generations in the work context by proposing a differentiated lifespan perspective. Here, we have argued that this differentiated approach better accounts for the individual-level manifestations of such effects that psychologists are interested in investigating. Although we have been critical of past attempts to empirically distinguish “generations” and to identify “generational differences,” it bears noting that practitioners have arguably latched onto these idea more vigorously than academic researchers. We suspect that the application of the generations concept to organizational practice stems from the fact that this concept serves as a heuristic that helps distill an otherwise complicated issue (e.g., intra- and interindivdiual variability) into smaller pieces that are easier to understand and communicate. It may also be that generationally based explanations are offered as an attempt to veil overt ageism (Rudolph & Zacher, 2015). Given our review and critique, we next offer some specific recommendations that attempt to integrate the propositions raised here with the needs of practice. Consistent with our integrative lifespan model of generations, we suggest that both research and practice would benefit from a moratorium on time-based operationalizations of generations as units for understanding complex dynamics in organizational behavior. In addition to this central idea, we see three additional areas in which practice may be informed by our model.

First, practitioners would be wise to adopt a lifespan perspective on work and aging, which recognizes that continuous developmental influences far outweigh the potential for categorical intergenerational differences. This perspective compels practitioners to recognize that there is a great deal of diversity both within and between generations. Second, the practice literature on this topic emphasizes the need to train managers to recognize intergenerational differences, and the need to actively manage generational differences as a matter of personnel practice (e.g., Baldonado, 2008; Eisner, 2005; Kapoor & Solomon, 2011). We suggest that the focus of these efforts be shifted to a broader recognition of aging as a process, toward understanding broader implications of workforce aging, and to recognizing the dynamic interplay between age and normative life stage considerations (e.g., shifting dynamics in work and family demands over time; see Zacher, Rudolph, & Reinicke, in press). There is also a need to recognize that attitudes, values, beliefs, motives, and behavior may shift predictably with age, and not as a function of generationally based effects. Organizational initiatives should be designed to address lifespan dynamics in such psychological processes, rather than tailoring interventions to specific generational groups (e.g., Langan, 2012; Tulgan, 2016). Finally, we suggest that practitioners must be cautious in interpreting the results of research on generations given the limitations of such works noted here. Policy recommendations (e.g., Hershatter & Epstein, 2010; Twenge, 2013) should not be based on the very tenuous conclusions that have been drawn thus far regarding generational differences.

In conclusion, our lifespan model of generations, and associated propositions and recommendations should serve as a guide for more rigorous organizational research and practice concerning the idea of generations. Moreover, this work stands as a reminder for the need to better educate researchers and practitioners regarding the limits and pitfalls of generational thinking. Our hope is that a differentiated lifespan perspective on generations helps to clarify and support future work that seeks reasonable means of conceptualizing and understanding generation-like explanations for individual-level psychological processes and outcomes in the work context.

REFERENCES

Author notes