Fuzzy But Useful Constructs: Making Sense of the Differences Between Generations

Abstract

While the research on generational differences has continued to grow, there are still questions about whether “generation” is a useful construct. In this study, generational boundaries or cut-offs are examined using a large, nationally representative sample of high school seniors whose attitudes and work values were assessed over time. Compared to Boomers and GenX’ers at the same age, Millennials were less likely to endorse social values (such as making friends at work) and more likely to endorse leisure values (such as vacation time). For workplace settings, Millennials were less likely than GenX’ers to favor owning their own business or working for a large corporation and were less likely than Boomers to favor working at a social service organization. However, while mean-level generational differences were apparent, there were no clear cut-offs between generations. Instead, the trends were more gradual and linear, suggesting that generations might be best conceptualized as fuzzy social constructs. Considerations for the use of generations as a meaningful construct are discussed.

The topic of generations has become both important and controversial in the organizational literature (Costanza, Badger, Fraser, Severt, & Gade, 2012; Costanza & Finkelstein, 2015). As the aging population in the workforce (i.e., Baby Boomers) is retiring, and the younger population is entering the workforce, organizations want to better understand potential differences between these generations and how to motivate, reward, and lead them in today’s workplace (Cennamo & Gardner, 2008; Twenge, Campbell, Hoffman, & Lance, 2010). However, one of the largest issues in the research on generations is whether “generation” is even a useful construct.

In this article, we discuss a specific variant of this question: Do generations, assuming they are meaningful social constructs, have clear boundaries? For example, is the difference between Gen X and Millennials seen clearly at the single birth year that separates those two groups—say, 1981—or is the difference more gradual such that those born in 1979 and 1982, the first a GenX’er and the second a Millennial, are actually more similar on psychological variables than those born at different ends of GenX—say 1965 and 1979? To borrow a term from biology, are generational differences clinal (i.e., on an incline or gradient)? That is, do generations exhibit gradual differences over across a continuum rather than sharp, bright differences that allow for clear categorization?

The idea of categories versus continua or dimensions has long been debated in other areas of psychology. For example, in discussion of whether sad individuals should be categorized as clinically depressed/non-depressed, or whether they should be placed on a continuum of depressive symptomatology, research suggests that depression is a dimensional, not a categorical (i.e., taxonic), construct (Hankin, Fraley, & Lahey, 2005; cf. Prisciandaro & Roberts, 2009). For most psychological variables, research primarily supports a dimensional approach. In a recent large review, the suggestion is that, based on the best studies, approximately 14% of the psychological constructs investigated are better described as categories than continua (Haslam, Holland, & Kuppens, 2012). Thus, it is expected that generational differences are more likely to follow a dimensional pattern with smooth rather than discrete transitions between them.

In this study, we will examine this issue directly by investigating how much the mean differences between generations are represented by clear delineations as opposed to gradual, continuous changes in two sets of variables relevant to the organizational literature. Specifically, we examine attitudes about desired workplace and work values in a cross-temporal, nationally representative sample of 12th graders.

WHAT ARE GENERATIONS?

First, a note on terminology. Throughout the manuscript, we use generational labels (e.g., GenX, Millennials) to describe specific generations. We want to be clear—especially given the arguments in this manuscript—that these generational labels are by their very nature imprecise for describing the construct. When we use the term generation, we are referring to specific groupings of birth cohorts described below (a birth cohort is everyone born in a certain year). These are based on self-reported birth dates. We do not intend to suggest that these generational groupings are monolithic psychosocial entities. Our usage of the terms is done for (a) ease of presentation, and (b) consistency with common discourse on the topic.

That said: Generations describe a group of individuals born during contiguous birth years who experience similar cultural contexts. These individuals both experience a shared historical and cultural environment and also shape culture in their own way (Gentile, Campbell, & Twenge, 2013; Mannheim, 1952). For example, Baby Boomers were shaped by the demographic boom after WWII, the rising economy of the Eisenhower years, and the spiritual awakening started by those from the previous generation who popularized Eastern Mysticism and psychedelics. The Baby Boom, in turn, shifted the culture. For example, student protests changed the structure and experience of universities and ideas about the Vietnam War.

Generations are usually defined as being around 17–20 years in length. Part of that is biological; it is about the time it takes for a human to mature and reproduce. It is also historical. In their seminal historical analysis of generations, Strauss and Howe (1991) identified this time frame as roughly correct for mapping cultural changes between generations. And it is arguably economic, reflecting cycles of development and decline (Kondratieff, 1979; Thompson, 2007). However, this definition of generations is also a social construction, similar to race and culture. As such it represents something generally agreed upon but not firmly established.

THE THEORETICAL GROUND OF GENERATIONAL FUZZINESS

There are several theoretical reasons to expect generational fuzziness, especially in areas of values, attitudes, and traits. These reasons derive from multiple disciplines, especially sociology and psychology, some of which are directly focused on generations and some which are related.

To start, traditional sociological theory largely places generational changes in the context of historical shifts. It is still an open question whether dramatic shifts (e.g., WWII, 9/11, the Pill) or more gradual shifts matter more. And, of course, even dramatic shifts take time to work through the culture. For example, Pearl Harbor, which was a sudden, dramatic cultural effect, took time to spread and influence citizens throughout the United States as the country mobilized for war.

More recently, however, some sociological theorists have been more direct about the importance of generational identity in distinguishing generations from birth cohorts. This argument can perhaps be oversimplified as “birth cohort plus shared generational identity equals generation” (e.g., Alwin & McCammon, 2007). This theoretical approach is novel in that it adds the importance of social identity models to generational work (e.g., Tajfel & Turner, 1979). This latter approach is consistent with a somewhat “fuzzy” view of generational change, demanding both historical and identity shifts. To provide a specific example, when the band The Who released their hit single My Generation in the mid-1960s, they were making an argument for generation as both an age cohort—“hope I die before I get old”—and an identity—“talking ‘bout my generation.” In this context, the term “generation,” however, did not apply to every member of the Boomer birth cohorts, but instead applied only to those who identified with the generational identity. Importantly, this generational identity likely spread through the populace over time. In this example, it started in England and then spread to the United States. This spreading cultural phenomenon was so powerful it was dubbed the “British Invasion.”

The case for generational fuzziness is also evident in several psychological models. We focus on four of these, which we label cultural, persuasion, personality, and social contagion. Cultural models focus on generations as phenomena linked to cultural change (e.g., Gentile et al., 2013) and assume that culture and individuals mutually constitute and reinforce each other (e.g., Markus & Kitayama, 2010). As a result, cultural change will lead to generational changes in areas of identity and these changes will, in turn, affect culture. This process is complex and very challenging to model statistically with existing data. Research comparing Asian immigrants to the West compared to Asian citizens in Asia and Caucasians in the West, for example, suggest that the self becomes more individualistic in Western cultural context, but this type of transition is neither full acculturation nor immediate (e.g., Heine, & Lehman, 1997). There is no work on acculturation that we are aware of that suggests the process is immediate; instead, it appears to be a longer-term process. Assuming this pattern holds with acculturation to cultural change, the changes should also be somewhat slow and incomplete with changes gradual and linear rather than abrupt and clear.

There are other reasons that thinking of generations as cultural constructs would suggest a more gradual change. First, culture influences people at different points during their lifetime. Culture is likely the most influential during adolescence and emerging adulthood as the individual moves from parental control into full adulthood. For example, narcissism may be shaped by unemployment rates during emerging adulthood (Bianchi, 2014; Leckelt et al., 2016). However, culture influences us from the day we are named (Twenge, Dawson, & Campbell, 2016). This suggests a stronger influence of culture during childhood and adolescence but continued cultural impacts throughout life, which should produce gradual rather than very clear boundaries between generations. Second, culture is not evenly applied geographically. Cultural changes occur in different parts of the country at different times. In the case of the United States, fashion and cultural trends often start at the coasts and then move across the country. Given this, it is plausible that generational differences will appear earlier in some regions than others. Third, there are also possible inter-individual age differences in cultural effects. That is, not everyone reacts to culture in the same way and at the same time. In the marketing literature, the term early adopters is often used to describe individuals who adopt products, especially technology, before others (e.g., Vishwanath, 2005). There may be an analog of this in cultural adoptions, so that as new cultural forms take hold, certain people with adopt them early and others will wait. For example, support for same-sex marriage changed relatively quickly during the 2000s and 2010s, but the shift occurred over 10–15 years rather than in 1 year (McCarthy, 2015). Some people changed their opinion early in this period, while others took longer, slowly increasing the average response to this change.

Although the variation in location and personality in cultural adoption could lead to gradual generational boundaries on many variables, there may be sharper cut-offs between generations linked to experiences of historical events. For example, those who were of military age during WWII had a very different experience than those who were not. The event was so powerful that it might have affected veterans similarly across people and places. Likewise, those who experienced 9/11 during their formative years, Millennials, had a different experience than those who were younger and not aware of it. However, this caveat aside, we believe a cultural model of generational change suggest a dimensional view of generations is more appropriate.

Regarding broad attitudinal changes, the primary psychological models focus on the construct of persuasion. However, even strong and concerted efforts at mass persuasion, such as those regarding smoking and public health, take time. Also, even direct and targeted interventions, although effective over time, resulted in an incomplete change in tobacco use (Rose, Hamilton, Colwell, & Shipley, 1982). One challenge is that most persuasion models trade strength for breadth. Mass persuasion via traditional media, public health campaigns, and so on is thought to be relatively weak in terms of changing the views or behaviors of any one person (Petty & Cacioppo, 1996). Thus, it can take years of these campaigns to effect widespread attitudinal change (e.g., Wakefield, Flay, Nichter, & Giovino, 2003). It seems plausible that the timeframe has been shortened as a result of the development and growth of social media, but still the change is not immediate.

Regarding personality, there is a growing awareness that personality can change, which stands in opposition to the early models of Freud and James. However, personality change is seldom simple or dramatic. Regarding contextual factors, major life changes (e.g., entering the workforce, marriage) are associated with changing personality (Roberts & Mroczek, 2008). This has been largely considered as part of a broader social investment model of personality that suggests relationships between social investment in social roles (work, family, religion, and volunteerism) and one’s personality (Roberts, Wood & Smith., 2005).

Another approach focuses on the spread of psychosocial variables (i.e., contagion, e.g., mood, body mass, smoking) through social networks (e.g., Hill, Rand, Nowak, & Christakis, 2010). The study of social networks finds that changes, while they do occur, are neither sudden nor absolute. Specifically, contagion-like processes in social networks (a) occur in clusters—not everyone in the network is influenced, and (b) associations among people on these variables decrease rapidly and usually are generally uninformative after three degrees of separation—they do not meaningfully extend much farther than that (Christakis & Fowler, 2013). This is said with the caveats that (a) online networks might accelerate certain transmissions, such as the effects of context (e.g., weather) on psychological mood (e.g., sadness) (Coviello et al., 2014) and (b) that key relationships might even change across birth cohorts. For example, even genetic associations with key psychosocial outcomes may differ across birth cohorts (Rosenquist et al., 2014).

In total, from the models reviewed, there is a general agreement that changes in values, attitudes, or personality traits across generations will take time and be gradual, with some individuals changing sooner, some later, and some not at all.

GENERATIONS IN THE WORKPLACE

Generational differences are an increasingly important, but also confusing, topic within the workplace. Baby Boomers are retiring, and businesses are increasingly interested in hiring and retaining Millennials so they can successfully work with and learn from the older generations while both groups are part of the workforce. Unfortunately, both the academic and popular literatures have come to conflicting conclusions about whether generational differences in work values exist (e.g., American Management Association, 2014; Costanza et al., 2012; Gentry, Griggs, Deal, Mondore, & Cox, 2011; Lyons & Kuron, 2014; Twenge et al., 2010) and, if so, what they are. For example, several sources maintain that Millennials are more focused on altruistic values than the generations before them (Alsop, 2008; Lancaster & Stillman, 2010), while other sources find little evidence for this (Twenge et al., 2010) or find even lower levels of empathy among Millennials (Konrath et al., 2011). Some believe Millennials are more interested than previous generations in being entrepreneurs and owning their own business (Asghar, 2014), while some empirical reports find instead a decline in young people owning their own business (Simon & Barr, 2015).

To our knowledge, generational differences in workplace setting preferences have not been previously examined using time-lag data. When considering the setting in which they would prefer to work, individuals may consider factors such as the size of the organization or the type of industry. Understanding generational and time period differences in preferences for work settings would be helpful for recruiting and retaining young employees. Past results showing Millennials’ higher extrinsic and lower intrinsic work values (e.g., Twenge et al., 2010) suggests they would find working in a social service organization (i.e., as a social worker) less attractive and working in a large corporation more attractive. However, in a time-lag study comparing Millennial and Boomer college students on the Holland careers typology, Millennials expressed a greater desire to pursue social careers, often described as those that help others (Bubany & Hansen, 2011). Thus, generational differences in workplace preferences are uncertain.

Some of the discrepancies among studies can be attributed to methodological issues. Most of the systematic studies on generational differences in the workplace (e.g., Cennamo & Gardner, 2008; Davis, Pawlowski, & Houston, 2006; Jurkiewicz & Brown, 1998; Wong, Gardiner, Lang, & Coulon, 2008; for reviews, see Costanza et al., 2012 and Twenge et al., 2010) used measurements taken only at one point in time. Using a cross-sectional design does not allow the researcher to separate age/career stage differences and generational differences (Schaie, 1965). Thus, any differences found in these studies could be due to generational or cultural change, but they could also be due to Millennials’ younger age and limited career experience. Cross-sectional studies are often considered the weakest form of evidence for generational differences (Lyons & Kuron, 2014). As a result, some researchers have cautioned against organizations implementing human resource practices that attempt to address generational differences in the workforce, given that most studies in the area are cross-sectional. In their recent meta-analysis, Costanza et al. (2012) concluded that generational differences were either nonexistent or small. However, all of the studies in the meta-analysis were cross-sectional, and more than half were dissertations or unpublished manuscripts.

Additional methodological concerns may also interfere with the analysis of generational differences. First, precision in measuring historical changes is relatively low. At best, we have data collected annually. At that level of detail, at least with attitudes and traits, gradual changes typically occur. With more fine-grained resolution, we might see more dramatic changes at certain discrete historical points. For example, daily or weekly polling data shows relatively large changes in attitudes towards political candidates following discrete historical events (e.g., convention “bounces”). That said, the multi-year shifts across generations with annual data suggest that if key events are involved, there are typically many of them. In addition, most studies use general questionnaires (e.g., personality, core attitudes, and happiness) in over-time surveys. From an individual differences perspective, these constructs should not change very much over a short period of time because traits are designed to be global and stable. Along these lines, there is some work linking cultural primes with self-concept, especially around individualism, but these effects do not mimic cultural change, which itself is gradual. Furthermore, much of this priming work has been called into question (e.g., Pashler & Wagenmakers, 2012). Even with an incredibly dramatic event such as the opening of Japan by Admiral Perry or the Spanish conquest of much of what is now Latin America, traditional cultural practices and values are still evident in both these places centuries later (Simonton, 1997). This suggests, at least anecdotally, that major historical changes are not enough to fully change a culture.

THE PRESENT RESEARCH

This article examines generational and time period differences in preferred job characteristics/values and industries/workplaces using a time-lag design. This design saamples respondents at the same age but born at different times (e.g., 18 year old born in 1980 and an 18 year old born in 1990), which reduces the potential confound of age/career stage present in cross-sectional studies. Although the time-lag method cannot separate the effects of time period (cultural shifts affecting everyone of all ages) and generation (changes affecting only the young and persisting as they age), the elimination of age effects means that time-lag studies can zero in on cultural changes—whether they are due to generation or time period (Campbell, Campbell, Siedor, & Twenge, 2015).

Previous research using this time-lag design method found that Millennials surveyed in 2006 valued extrinsic work values more, and intrinsic work values less, than 1976 Boomers (Twenge et al., 2010). Pre-recession Millennials reported wanting both more money and status and less time spent at work, a difficult combination that reinforced the idea that Millennials were entitled (Twenge & Kasser, 2013). However, the recession may have served as a “reality check,” altering preferences as younger Millennials spent their formative years witnessing layoffs and unemployment. Thus, Millennials may no longer be higher in extrinsic and leisure values compared to Boomers. Similarly, previous research showed that Millennials rated social values as less important; for example, Millennials rated making friends at work lower than Boomers did (Twenge et al., 2010). The growth of computer-mediated communication such as texting and social networking sites has reduced the limitations of physical boundaries on people’s social relationships (Myers & Sadaghiani, 2010) and in turn, may have altered the social dynamics at work. The greater use of mobile technology and social media is providing Millennials with readily available connections to their friends outside of work, making the workplace as a place to develop social relationships less important. As technology continues to advance, this may lead to additional decreases in the desire for social rewards from the workplace.

The specific goal of the present research is to determine whether generational differences exist and whether the boundaries between generations are gradual or abrupt. Based on past research and theory, we predict that generational differences will be evident but generational differences will be gradual rather than abrupt. We do this by examining cross-temporal changes in work values and desired work settings. Generational differences can be tested by comparing average values from different generations; the nature of generational changes can then be examined visually by plotting the year-to-year changes.

METHOD

Sample

Monitoring the Future MtF (Johnston, Bachman, & O’Malley, 2006) has surveyed a nationally representative U.S. sample of high school seniors (12th graders) every year since 1976. The survey uses a multi-stage random sampling procedure to select high schools and then students to complete the survey. The participation rate of schools is between 66% and 80%, and the student participation rate is between 79% and 83% (Johnston et al., 2006). The analyses in this paper include 1976–2014. N’s range from 105,661 to 108,335 depending on the item. These surveys sample a like-age group at different points in time; thus, any differences are not due to age, and we can be more confident that they are due to generation/time period and not factors such as maturation.

Measures

Work values were measured on a five-point Likert scale, grouped into five factors with satisfactory or near-satisfactory internal reliability, and based on the factor analysis and measurement invariance analyses of this dataset in Twenge et al. (2010): for Leisure an example item = “A job where you have more than two weeks’ vacation,” (alpha = 0.68); for Intrinsic an example item = “A job which is interesting to do,” (alpha = 0.67); for Extrinsic an example item = “A job which provides you with a chance to earn a good deal of money^,” (alpha = 0.72); for Altruistic an example item = “A job that gives you an opportunity to be directly helpful to others,” (alpha = 0.65) and for Social an example item = “A job that gives you a chance to make friends,” (alpha = 0.59). (Note: this final alpha is sub-optimal which might limit precision in comparing yearly values.) Refer to Table 1 for all items for each of the five work values.

Workplace setting preference items, answered on a four-point Likert scale from “not at all acceptable” to “desirable” included: “working in a large corporation,” “working in a small business,” “working in a government agency,” “working in the military service,” “working in a school or university,” “working in a police department or police agency,” “working in a social service organization,” “working with a small group of partners,” “working on your own (self-employed).”

Analytic Strategies

We analyzed the data in two primary ways. First, we grouped participants into generations by the year when they completed the survey. Using the generational cutoffs of Strauss and Howe (1991), those in their last year of high school in 1966–1978 are Boomers, 1979–1999 Generation X, and 2000–2014 Millennials. Second, we examined the data by year. This allowed us to compare the patterns of data represented by the generations to the specific years to visually interpret if the boundaries are distinct and bright or unclear.

Given the large sample sizes, virtually all differences are statistically significant. Thus we focus primarily on effect sizes and patterns in the data.

RESULTS

First, we compared work values (e.g., extrinsic, social; see Table 2; Figure 1). These analyses showed notable generational differences. The largest differences appeared in social and leisure values, with Millennials favoring social values less (d = −0.29) than Boomers while favoring leisure values more (d = 0.25). However, these differences are gradual changes rather than sharp breaks between generations. We can see this clearly from the yearly data (see Figure 1). For example, social values decline throughout the GenX and Millennial years in a fairly linear fashion so that, for example, early GenX’ers are higher in social values than later GenX’ers, and late GenX’ers and early Millennials look fairly similar. Leisure values increase through the years of GenX and decline slightly through the Millennial years, rather than suddenly rising at any one cutoff. Averaged across the generations, the differences are significant, but the trends are linear or curvilinear rather than sudden.

We next examined generational and time period differences in the desirability of different workplaces. Despite common views of Millennials as entrepreneurial, they are actually less likely to say they would like to own their own business, especially compared to GenX’ers (d = −0.21; see Table 3). Similar to the shifts in work values, however, this shift is gradual. The desirability of owning one’s own business declines throughout the Millennial generation (see Figure 2). Compared to Boomers at the same age, Millennials were less likely to want to work in a social service organization and more likely to want to work at a large corporation, a school or university, or a government agency (see Table 3). Regarding effect sizes, Millennials are modestly less willing than Boomers (d = −0.16) and slightly less likely than GenX’ers (d = −0.03) to want to work in a social service organization. Millennials are modestly more willing than Boomers (d = 0.13) and less willing than GenX’ers (d = −0.18) to work in a large corporation. Millennials are modestly more willing than Boomers (d = 0.14) and GenX’ers (d = 0.11) to work in school or university. All of these changes are gradual rather than sudden (see Figure 2).

DISCUSSION

To investigate generational boundaries, we examined generational differences in work values and workplace setting preferences in a large, nationally representative sample of high school seniors administered since the 1970s. We approached the data by grouping people into generations and comparing generational averages and then examining yearly changes in these variables. It is clear from the results that, at least on these work-related attitudes and values, there are indeed generational differences, but there are no abrupt or clear cut-offs between generations. The trends in the data tended to be relatively linear and/or be somewhat curvilinear. There was no evidence of a staircase pattern that would imply a clear generational boundary.

Differences due to age are a useful analogy. Few would argue that age differences do not exist, but few would also argue that the difference between being 19 versus 20 is hugely impactful. However, the difference between being 20 versus 29 is impactful. Nevertheless, it is common to group people by age (e.g., “people in their 20s,” “twenty somethings.”) Similarly, generational differences exist, and grouping people into generations is useful even though the differences tend to build over birth years—just as developmental differences build with age.

These findings are consistent with the various sociological and psychological models discussed previously. Whether we consider classic sociological models where generational change is driven by dramatic historical events, a cultural model where generational change follows cultural change, or a more psychological model such as persuasion, generations should transition gradually in most cases. Any change that happens to a generation (i.e., series of contiguous birth cohorts) is likely to impact individual members of the generation unevenly, and further, be spread throughout the generation unevenly. This makes the possibility of abrupt generational change unlikely. Of course, we cannot logically rule out abrupt changes from these data. There could be dramatic shifts between generations on certain variables, but these abrupt generational shifts would likely need to be driven by: (a) significant causal forces that impact all members of a generation simultaneously, or (b) significant forces that are transmitted through the generation in a rapid, viral fashion. For example, a major technological breakthrough that led to free energy or superhuman artificial intelligence would potentially have a stunning effect on generational work attitudes and values; but this kind of rapid technological shift is rare.

Implications for Research and Practice

These results, along with past findings regarding generations, bring to mind several issues. We focused on whether or not generation is a valuable construct and argued that generations are useful (Campbell et al., 2015). Generational labels describe large, socially defined groups that differ in meaningful ways. However, generations are very broad social constructs that lack precision. Especially at the beginnings and ends of generations, the differences between generations are likely to be minimal. But like other constructs such as age and race, not having clear categories does not preclude the existence of meaningful differences across the continuum. Generational cut-offs should usually be useless for comparing people at generational cusps on most traits, attitudes, and abilities. The difference based on birth cohort between individuals born in, say, 1964 and 1965 should be irrelevant in most cases. However, the 1955 and 1975 birth cohorts should show differences.

Regarding future research, there are three major issues. First, we need to develop and test models that show the transmission of variables of interest (e.g., attitudes, values, personality, etc.) through cohorts and cohort members. For example, we need to be able to model something like the shift in work values not just as an annual mean, but as something that appears in a small group of individuals and then spreads unevenly through same age and younger cohorts. What we can view now in the data contain a great deal of variance. There are mean-level generational differences, and we can see individual scores in some data sets (like MtF), but we cannot see the spreading of values between people, or why any specific person came to have the value that he or she has. In short, we cannot measure specific cultural, historical, or social impacts on cohorts’ members.

One way to do this would be to measure the hypothetical historical, cultural or social impact on the individual (e.g., how much were you affected by 9/11? How much do you use a cell phone? How much are you influenced by the values of your friends?) With these additional variables, we could increase understanding of generational transmission as well as increase prediction (although self-report would be only a first step, with other data such as peer report, social media content, and so on perhaps better). Also, we need to test more complex models that include not just generational cut-offs but theoretically relevant, historical, or cultural events in the model. Finally, it is time to consider prospective work. All the work we have is based on historical data. A more rigorous test would be to set models and predict future changes.

For applied purposes, these findings suggest that generational membership, especially near purported cut-offs, is a crude marker of work attitudes or values—at least the ones that were measured here. If organizations are interested in specific values or attitudes in the workplace, selecting on those values themselves, rather than a loose correlate like generational membership, would be preferable.

There could also be differences that correlate with age but reflect hiring date. For example, in many organizations such as universities, new hires will typically be more qualified than older hires because they have been trained in the latest and most up to date skills, methods, and tools. This is a byproduct of increasing selectivity. At the same time, employees of all ages share many things in common, such as an interest in the work of the organization. Although there is no comparative data we know of, it is possible that people in specific organizations differ more from those in other organizations than they differ by generation. Perhaps to gain some precision over the broad construct of generations researchers could group individuals into smaller birth related “micro-generations,” such as 5-year birth cohorts or even single-year cohorts. However, the trade-off would be ease of use and breadth of the concept of generations. For example, describing Baby Boomers in an organization is more inclusive, although less precise, than describing those born between 1950 and 1955.

This is an issue that occurs in other areas of organizational psychology as well. For example, using the term “white” to describe anyone of historical European descent misses the nuances that could be retained by describing a group as “Irish American” or “White Southerners.” Similarly, personality traits are most often understood using broad models like the Big 5. These are useful, but describing people at the facet level would be more precise.

We do not have an optimal answer to this quandary. Largely, what is optimal depends on context. With research, it might be best to describe five year cohorts as a compromise, or, better, present data year by year as well as by generation. In the workplace, the generational labels are likely to stick, but these should be acknowledged as large, general groupings rather than precise descriptors. Given the data, it is likely more useful to consider the exact year in which an employee was born rather than just his/her generational category.

Limitations and Future Research

There are several limitations to this research worth noting. First, we used a limited set of items to compare generations. There could still be clear generational differences with bright-line cutoffs on other psychological traits. We have not seen them in the data we have analyzed over many different traits (Twenge, 2014), and do not suspect this is the case on theoretical grounds, but they still might be present. Second, these data were from a large sample of individuals when they were 18 years old and in their senior year of high school. Although these individuals may have some work experience, they are very unlikely to be full-time employees. This limits the generalizability of the data to work attitudes that are largely predictive. If we were able to examine cross-temporal data with full-time employees, we might find clearer differences between groups. Nevertheless, these surveys do provide some insight into what individuals across several generations thought when they are seniors in high school. Again, these data capture attitudes about work preferences and values for all three generations when they are at this young age. Finally, we want to again be clear that there can be and are significant cross-temporal generational differences. As just two examples, attitudes toward working mothers have shifted d = 0.75 since the 1970s (Donnelly et al., 2016), and attitudes toward gays and lesbians more than d = 1.00 (Twenge, Campbell, & Freeman, 2012). Both of these trends, in addition to the shifts in work attitudes discussed here, have had an impact on the workplace. However, even these examples of large changes do not show bright generational cut-offs.

As the older generation continues to remain in the workforce, even coming out of retirement in many cases as a result of failed retirement savings or an increased life expectancy, researchers will need to search out datasets on these birth cohorts. The focus of generational differences as a research topic increased over the last decade and the data sets that have been used are aimed at younger individuals rather than across all ages. This makes comparisons challenging and an important undertaking for future research.

With generational research, methodological issues are typical limitations. In this case, we were unable to examine cohort/generation and period effects separately. With more complex data sets, (e.g., sets that included different ages for each cohort), other methods such as IRT could be used to partially tease apart cohort/generation, time period, and birth cohort (e.g., Twenge, Campbell, & Carter, 2014). There may be other methodological advances that would help research on generations when the appropriate datasets are available such as forms of growth curve analysis (e.g., Ram & Grimm, 2015). Our hope is that organizational researchers interested in studying generations think about developing ideal surveys now that can be continued every year and eventually used by future researchers.

Finally, there are some interesting patterns in the data that are worthy of future research. For example, workplace preferences appear to converge over time. This reflects a long-term drop in interest in working in large organizations (e.g., large corporations, government) and a simultaneous increase in interest in small companies. Although these data do not directly speak to the meaning of this change, it appears that there was a decline in the “lifetime employment” big company model starting in the mid-1980s coupled with a rapid increase in interest in small businesses, perhaps start-up businesses, starting with the Great Recession. But this is a question for future research with additional data points.

CONCLUSIONS

We found that generations differ on workplace-related attitudes such as desired workplace and work attitudes and values. These generational differences do not reflect abrupt shifts, but more gradual trends that can be modeled as linear or curvilinear. Research and practice in the area on generations need to be cautious when describing significant and sudden differences between generations. However, going to the other extreme and saying that generational differences do not exist is also not accurate. Generational differences often emerge from cultural change and just because cultural change is often linear and gradual does not mean it does not exist. Thus, it is important for organizations to consider more than simple generational membership when trying to understand workers.

REFERENCES

Author notes