Work Motivation Is Not Generational but Depends on Age and Period

Schröder, Martin Article — Published Version Work Motivation Is Not Generational but Depends on Age and Period Journal of Business and Psychology Provided in Cooperation with: Springer Nature Suggested Citation: Schröder, Martin (2023) : Work Motivation Is Not Generational but Depends on Age and Period, Journal of Business and Psychology, ISSN 1573-353X, Springer US, New York, NY, Vol. 39, Iss. 4, pp. 897-908, https://doi.org/10.1007/s10869-023-09921-8 This Version is available at: https://hdl.handle.net/10419/313167 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Vol.:(0123456789) 1 3 Journal of Business and Psychology (2024) 39:897–908 https://doi.org/10.1007/s10869-023-09921-8 ORIGINAL PAPER Work Motivation Is Not Generational butDepends onAge andPeriod MartinSchröder1 Accepted: 10 October 2023 / Published online: 6 November 2023 © The Author(s) 2023 Abstract Many argue that work motivation varies with year of birth, suggesting the utility of generational labels such as Z, Y, X, or Baby Boomer. This article tests this generational hypothesis by using multilevel regressions with data from 584,217 individuals sampled by the Integrated Values Survey in 113 countries over more than 30 years. The results show that the importance of work first increases and then decreases with an individual’s age and that the importance of work tends to decrease for everyone with the passing of historical time. These age and period effects make later-born generations seem work averse, while birth year and thus generational membership hardly explain work motivation after accounting for age and period effects. The article also tests whether ten other work-related attitudes can be explained through generational membership. The results suggest that the historical time period and an individual’s life course explain work motivation and work attitudes better than generational membership. This life course and historical explanation is suggested as an alternative to the generational hypothesis of work motivation and attitudes that prevails in much of the literature. Keywords APC analysis· Generation Z· Generation Y· Generation X· Babyboomer· Attitude change· Work motivation· Life course· Karl Mannheim Introduction Over the twentieth century, a number of purported generations have been identified, notably the so-called (1) World War Generation (born approximately 1925–1945), (2) the Baby Boomer (born 1945–1965), (3) Generation X (1965–1980), (4) Generation Y (1980–2000), and (5) the later-born Generation Z (for an overview, cf. Costanza etal., 2012; similarly, cf. Mahmoud etal., 2021; Twenge & Campbell, 2012; Kowske etal., 2010). These generations are said to hold very different attitudes. Popular culture describes Generation Y as being particularly lazy (Time, 2013; also cf. Gallup, 2016). Part of the academic literature agrees, arguing that members of Generation Y are “not ready to sacrifice their private life for work” (Kultalahti & Liisa Viitala, 2014: 576f.; also cf. Twenge etal., 2010), “express a weaker work ethic, a greater desire for leisure and less work centrality” (Twenge & Campbell, 2012: 9) and are “less likely than older generations to be ‘engaged’ in the workplace” (Milkman, 2017: 9; also cf. Meriac etal., 2010). A sprawling management literature therefore gives advice on how to handle “lazy” and “unengaged” millennials (Deal & Levenson, 2016; Ubl etal., 2017). However, is an individual’s year of birth and associated generational membership indeed related to his or her work motivation? Does this relationship hold after taking into account that individuals change their attitudes over their life course (age effects) and that all members of society change their attitudes with the passing of historical time (period effects)? In other words, can work motivation truly be explained by a person’s birth year, or is it better explained by how old a person is and in which historical time period he or she is surveyed? This article answers this question by showing that after accounting for period and age effects, widely assumed generational effects disappear. While later-born generations may indeed appear work-shy, this is merely because later-born generations are surveyed later (when everyone sees work as less important due to a historical time period effect) and because later-born cohorts are younger (when everyone has always been less inclined to work due to an individual life Additional supplementary materials may be found here by searching on article titlehttps:// osf. io/ colle ctions/ jbp/ disco ver * Martin Schröder martin.sc[email protected] 1 Saarland University, Faculty ofHuman andBusiness Sciences, 66123Saarbrücken, Germany 898 Journal of Business and Psychology (2024) 39:897–908 1 3 course effect). Similar results are found for ten additional work-related variables, which means that generational membership cannot explain work attitudes, while age and time period can. The following section reviews how generations are classically conceptualized. It then shows how empirical studies have failed to do justice to the classical theory of generations, as they either confuse cohort with age and period effects or simply neglect age and period effects. This failure has led empirical studies to mistakenly conclude that work motivation varies with generational membership. The classic definition of a generation comes from Karl Mannheim (1952 1928: 298), who argued that generations come into being where individuals “are in a position to participate as an integrated group in certain common experiences.” Thus, the first precondition for a generation to come into existence is that a birth cohort has lived through shared circumstances that influenced its members similarly during their impressionable years (Ryder, 1965; Rudolph etal., 2021). Mannheim’s (1952 1928: 298) idea behind this was that “[e]arly impressions tend to coalesce into a natural view of the world […] even if the rest of one’s life consisted in one long process of negation and destruction of the natural world view acquired in youth, the determining influence of these early impressions would still be predominant.” This implies that events that happen while a birth cohort is approximately 15 to 25 years old have a durable effect on this particular birth cohort, but not on other birth cohorts, which have either not yet reached their impressionable years or already passed them. Mannheim wrote his foundational text “The Problem of Generations” shortly after World War I. He was thus under the impression that birth cohorts that lived through the horrors of the Great War were durably scarred by this experience, while birth cohorts that were either too young or too old to have been marked by this experience remained unaffected. While truly life-changing events such as war service may durably mark a particular birth cohort and thus turn it into a generation, Mannheim also made clear that other birth cohorts never experience such life-changing events during their impressionable years and thus never turn into a generation marked by a common life experience that leads to intragenerationally similar attitudes (Mannheim, 1952 1928: 309). In contrast, the literature has postulated the advent of a new generation every 10–20 years, often failing to specify which “common experiences” (Mannheim, 1952 1928: 309) or “same event” (Ryder, 1965: 845) have allegedly influenced members of birth cohorts during their impressionable years so that they can be considered a new generation. Others have argued that Generation Y was formed by having witnessed September 11 and the Iraq War, by having grown up with computers, cell phones, internet, and social media (Milkman, 2017; Twenge etal., 2010; Weber, 2017). Even the bankruptcy of Enron Corporation (Twenge & Campbell, 2012) and the introduction of the Harry Potter book series (Gierzynski & Eddy, 2013) are said to have influenced Generation Y both durably and distinctly. However, some have wondered whether a company bankruptcy or a book series are truly swaying enough to influence one particular birth cohort’s attitudes relative to all others. Why, these authors ask, should events influence the attitudes of only one particular birth cohort, rather than those of all birth cohorts or of no birth cohorts at all (Campbell etal., 2017; Costanza etal., 2012; Rudolph etal., 2021)? The first critique against the concept of a generation is thus whether a particular birth cohort has indeed lived through circumstances during their impressionable years that were exceptional enough to influence this and only this particular birth cohort. If, contrary to this, not only one but all birth cohorts are similarly influenced by a historical event or time period, then one cannot speak of a generational effect; rather, one must indeed speak of a historical period effect that influences all members of society similarly, instead of turning one particular birth cohort into a generation whose attitudes differ from all others (Mannheim, 1952 1928; Ryder, 1965; Luo & Hodges, 2022). Second, the experience of a cohort must have influenced it for the remainder of its life; otherwise, one can only speak of a transitory age effect, which does not durably differentiate one generation from all others but simply differentiates the young from the old (Ryder, 1965; Campbell etal., 2017; Costanza etal., 2012; Milkman, 2017; Kiley & Vaisey, 2020). On this second account, the generational literature has also been criticized for not taking age effects seriously. Notably, studies that merely sample data from one time point cannot possibly differentiate what is often described as a generational effect from an age effect because, at each particular point in time, everyone of the same age was born at the same time and thus belongs to the same cohort. To understand why it is therefore impossible to disentangle the effects of age, period, and cohort without repeated measurements of each, consider Table1, which shows how cohort membership in each year results directly from a combination of birth year and interview year. Imagine that a sample from the year 2000 is analyzed (first column of Table1). As Table1 shows, everyone in this sample who is 20 years old must logically have been born in 1980. Thus, any work motivation found for a person in this group could be either due to the effect of being born in 1980 (cohort effect) or from the combined effect of being 20 years old (age effect) in the year 2000 (period effect). Similarly, with measurements from only the year 2000, it is impossible to tell if work motivation results from being born in 1975 (cohort effect) or from the combined effect of being 25 years old (age effect) in the year 2000 (period effect) because no one in 2000 is 25 years old without being born in 1975. 899Journal of Business and Psychology (2024) 39:897–908 1 3 This so-called age-period-cohort problem is only solvable when one has measures of individuals from multiple cohorts at multiple times and multiple ages. Because then one not only has a cohort that is 20 years old in the year 2000 but also a cohort that is 20 years old in the year 2005 and so on, which allows to separate the effect of age (everyone changing their opinion as they grow older, irrespective of time period and cohort) from period effects (everyone in society changing their attitude with historical time, irrespective of their age and cohort) and to distinguish both effects from generational differences (individuals having different attitudes because of their birth year, irrespective of age and time period). However, existing studies often use data from only one cohort, from only one age group, or from only one time period. These studies can therefore not disentangle the three effects. Yet this does not stop them from making sweeping generalizations about generations. Kultalahti & Viitala (2014: 576) compared members of different birth cohorts at merely one point in time, still, they argue that “catering for work-life balance was a very critical factor for our sample of Generation Y.” Their results might, however, also be due to Generation Y having been relatively young when Kultalahti and Viitala sampled them, and young people might generally care more about worklife balance; thus, the authors’ results would be due to age rather than due to generational effects, which they hold responsible. Similarly, Jobe (2014: 306) argued that “Boomers have strong beliefs in work for work’s sake.” However, Baby Boomers might have been at the peak of their careers when they were interviewed by Jobe; thus, what appeared as an effect of birth cohort and thus generational may have indeed been an effect of age. Because of this structural indeterminacy, researchers such as Cucina etal. (2018: 259f.) have cautioned that cross-sectional designs are “a major limitation to generational research […] since age and generation are confounded.” Some scholars suggest, however, that even after age and period effects are properly accounted for, generational influences on attitudes remain important; they argue that “if we want to make our best guess (net of age) about what a person thinks […] we would be better off knowing what year the person was born than what year we are observing them” (Vaisey & Lizardo, 2016: 8). Others generalize this argument, claiming that “Generational differences are real and useful” (Campbell etal., 2015: 324). From this, one can derive the following hypothesis, which this article aims to test: Generational Hypothesis (1): Birth year (birth cohort orpurported generation) explains work motivation, irrespective ofage andhistorical time period As discussed above, an alternative to this generational hypothesis is that age effects rather than generational membership determine individual attitudes. Studies that have not taken the latter into account might merely show that those who are older think differently than those who are young (Meriac etal., 2010); this might also explain why Generation Y’s predecessors have been described as similarly lazy when they were young (Smola & Sutton, 2002), whereas now—when they are older—they are seen as the hardworking predecessors of Generation Y (Mahmoud etal., 2021). Seeing a purported generation as being lazy when it is young and hard-working when it is middle-aged is more compatible with an age rather than a cohort effect. This leads to the second hypothesis, which is as follows: Life Course Hypothesis (2): Age (over anindividual’s life course) explains work motivation, irrespective ofbirth year andhistorical time period As mentioned above, another part of the literature fails to meet Mannheim’s further requirement by not considering period effects. Notably, studies neglect period effects when they compare today’s young to those of yesteryear, thus holding age constant, to then argue that “recent generations were progressively more likely to value leisure at work” (Twenge etal., 2010: 1119, 1131; also cf. Twenge & Campbell, 2012; Brandt etal., 2022). However, merely showing that today’s adolescents are different from those of yesteryear could just as well indicate that everyone has changed their opinion with the passing of historical time, rather than indicating generational change (Campbell etal., 2015: 325; Arnett, 2014: 195). The possibility that historical time periods may explain what is often seen as generational differences leads to the third hypothesis, which is as follows: Table 1 Age-period-cohort table Year 2000 2005 2010 2015 2020 Age 20 1980 1985 1990 1995 2000 25 1975 1980 1985 1990 1995 30 1970 1975 1980 1985 1990 35 1965 1970 1975 1980 1985 40 1960 1965 1970 1975 1980 900 Journal of Business and Psychology (2024) 39:897–908 1 3 Historical Hypothesis (3): Year ofmeasurement (historical time period) explains work motivation, irrespective ofage andbirth year Although age (Hypothesis 2) or period effects (Hypothesis 3) might therefore explain what is often understood as generational differences (Hypothesis 1), many scholars continue to argue for fundamental differences between generations, thus upholding Hypothesis 1 despite alternative explanations (Mahmoud etal., 2021; Campbell etal., 2015). Others caution, however, that “the relatively sparse empirical research published on Millennials is confusing at best and contradictory at worst” (Deal & Levenson, 2016: 191), urging that before any claims about generations can be made, one must account for the confounding influence of age and period, contrary to what existing research has done (Kowske etal., 2010; Costanza etal., 2012; Rudolph etal., 2021). While this has been done for work satisfaction (Cucina etal., 2018; Kowske etal., 2010), it has not been done for work motivation (Costanza etal., 2012: 381). Especially lacking are studies of work motivation that extend beyond US student and industry samples (Campbell etal., 2015: 325). Particularly, existing studies bemoan that “no known studies have examined generational differences in PWE [protestant work ethic] in European nations” (Zabel etal., 2017: 311). This lack of cross-cultural evidence is problematic because generations are said to be formed through shared experiences, and these are likely to differ across countries. In spite of this, US generational labels are often simply applied to other countries (Campbell etal., 2015: 325), which is clearly nonsensical, as some of the events that purportedly gave rise to generations in the USA simply did not take place in other countries (Rudolph etal., 2021: 949). This makes it an act of faith to assume that alleged generations, whose existence is already in doubt for the USA, can be readily found in other countries (however, cf. a study using German data, Schröder, 2018, 2019). To test whether generations can be found cross-nationally, this article uses data on multiple cohorts sampled at multiple time periods and multiple ages to disentangle the effect of age, period, and cohort on work motivation and other work-related attitudes cross-nationally for the first time. Data andMethod Data The Integrated Values Survey merges the World Values Survey and the European Values Study into one large dataset, providing data from 584,217 individuals sampled in 113 countries between 1981 and 2022. Cohorts born before 1925 are omitted, as they were mostly retired during the sampling period, making it difficult to gauge their work motivation. This leaves a sample with a mean age of 42 (sd = 16.1, min = 15, max = 89), of which 49.9% are male, with an average survey year of 2006 (sd = 10.1, min = 1981, max = 2022). Birth cohorts are defined through 5-year intervals of common birth years, following the usual boundaries in the literature as mentioned above. While these typically classify 15-year birth cohorts as belonging to the same generation, proponents of the generational thesis argue that “[t]here is no agreed on year for the beginning or end of generations” (Campbell etal., 2015: 323). Thus, using 5-year birth cohort intervals gives cohort effects a better chance to emerge than lumping 15-year birth cohorts into one group. For example, if the first Generation X members born 1970–1974 would find work very important, while the last Generation X members born 1980–1984 find work less important, then grouping them all as one Generation X would find no effect overall, while grouping them in 5-year birth intervals would show the respective positive and negative effects that would otherwise cancel themselves out. If, conversely, all members of a 15-year birth cohort find work more important, then each of the 3 × 5-year birth cohorts that comprise this purported generation would also find work more important. Thus, using finer-grained birth cohorts brings no information loss but instead adds detail. This approach gives purported birth cohort effects a better chance to emerge, even if these are not always neatly arranged along the exact starting and end dates of purported generations. So why not use ever smaller birth cohorts, such as distinguishing each year as a birth cohort? This would make no sense theoretically, as, e.g., being born in 1998 compared to 1999 compared to 2000 cannot possibly be conceptualized as a generational difference and because statistically, with ever smaller birth cohorts, sample sizes eventually become too small. Table2 shows how many cases exist per cohort. One generation must be used as a reference category, with which other generations are compared. Substantively, it does not matter which group is the reference, as, e.g., arguing that Gen Z has a lower work motivation than the Baby Boomer is the same as arguing that the Baby Boomers have a higher work motivation than Gen Z. Although there might be an interest in comparing the “newest” Gen Z birth cohort to all others, Table1 shows that only 3210 Gen Z individuals are in the sample, and these individuals are at most 22 years old, as Gen Z was born starting in 2000, while the most recent data are from 2022. Because there are so few Gen Z individuals and because they have not yet completed their formative years, Gen Z is not a good reference category. Instead, the Gen Y birth cohort born 1990–1994 is used for this purpose, as it contains 24,313 individuals, which makes for a more robust reference category. 901Journal of Business and Psychology (2024) 39:897–908 1 3 The main variable of interest is whether purported generations find work “very important” (= 1), “rather important” (= 2), “not very important” (= 3), or “not at all important” (= 4). These numbers are inversed, so that larger numbers indicate a larger importance of work. Table3 shows the descriptive data for the main variables used. Method As mentioned above, cohort, age, and period cannot be linearly separated, as every cohort has the same age at the same time. Therefore, unambiguously assigning influences to cohort, age or period is a “futile quest” (Glenn, 1976: 900; Ryder, 1965; O’Brien, 2011). This is only a problem, however, if age, period, and cohort are measured categorically, as in the example above. The age-period-cohort problem is unproblematic, when trying to measure whether a categorically defined birth cohort has certain attitudes irrespective of continuous age and period trends that affect all cohorts similarly (Mason & Wolfinger, 2001; Pampel & Hunter, 2012; Chauvel & Schröder, 2014; Bell, 2020; Fosse & Winship, 2019a; Schröder, 2018, 2019). This means that by controlling for the effect of having been asked later in historical time and by controlling for the effect of being one year older in the individual life course, the effect of having been born in a certain year can be separated from the effect of the progression of historical time and of individual age. The effect of belonging to a purported generation, in other words, becomes identifiable relative to the effect of individual aging and the effect of societal changes in work attitudes. The results section first presents models that explain attitudes through cohort membership, controlling for only age and age squared. Modeling age as a continuous rather than categorical variable makes sense, as age effects are inherently continuous (Bell, 2020). Using not only age but also age squared makes sense, as many life effects are not simply continuous but first go in one direction and then in the opposite direction. For example, work may be more important when one starts working life and then eventually become progressively less important as one approaches retirement. The historical timeperiod is modeled as a continuous variable (year of survey). The intuition behind this approach is that attitudes may change gradually over time. Thus, for example, work attitudes in 2020 should largely result from attitudes in 2019, more than, say, from attitudes in 1995. For this reason, age and period are modeled as continuous rather than categorical variables. Contrary to age and period, cohorts are conceptualized as categorical, as the argument for distinguishing generations is that one generation is precisely not a continuation of the preceding one. For example, Generation Y is not understood as a continuation of the preceding Generation X but rather as something categorically new. Indeed, the literature’s claim is that a new generational label is warranted because each generation is categorically different. Calculating the effect of having been born during a certain time after controlling for age and age squared indicates whether birth cohorts find work more or less important, irrespective of age-related changes in work motivation that affect all cohorts similarly. Additionally, controlling for the timeperiod shows whether purported generations hold different opinions, not only irrespective of their age but also irrespective of when they have been surveyed in historical time. Models other than the standard model proposed herein can be used to calculate cohort effects. However, these models are less useful for the task at hand. Notably, the so-called hierarchical age period cohort (HAPC) models (Yang & Land, 2006) try to minimize unexplained variance by clustering individuals by age, period, and cohort. However, such models arrive at systematically incorrect results, such as suggesting that formal education declines with age, which is impossible (Chauvel & Schröder, 2015; O’Brien, 2017; Bell Table 2 Cases per cohort Birth cohort name Birth year Cases Percent Cum. WW (“World War generation”) 1925–1929 14,819 2.40 2.40 WW 1930–1934 20,260 3.28 5.69 WW 1935–1939 27,756 4.50 10.19 WW 1940–1944 33,708 5.46 15.65 BB (“Baby Boomer”) 1945–1949 42,601 6.91 22.56 BB 1950–1954 51,912 8.42 30.97 BB 1955–1959 58,292 9.45 40.42 BB 1960–1964 63,204 10.25 50.67 BB 1965–1969 60,779 9.85 60.52 X (“Generation X”) 1970–1974 60,094 9.74 70.27 X 1975–1979 55,098 8.93 79.20 X 1980–1984 47,757 7.74 86.94 Y (“Generation Y”) 1985–1989 39,414 6.39 93.33 Y 1990–1994 24,313 3.94 97.27 Y 1995–1999 13,624 2.21 99.48 Z (“Generation Z”) 2000–2004 3,210 0.52 100.00 Total 616,841 100.00 Table 3 Descriptive data for variables Variable Obs Mean Std. Dev. Min Max Year of birth (cohort) 616,841 1964 17 1925 2004 Age 616,841 42 16 15 89 Year of survey (period) 616,841 2006 10 1981 2022 Important in life: Work 584,217 3.5 .74 1 4 902 Journal of Business and Psychology (2024) 39:897–908 1 3 & Jones, 2018; Lois, 2019). The problem of such models is that their assumptions are made “under the hood”—that is, an algorithm decides how variation is assigned to cohorts, age, and period—leaving it unclear why certain results come about, which has led critics to argue that authors and users of these algorithms “do not really know why their model works (because it often does not)” (Bell & Jones, 2018: 797). Rather than hiding assumptions within an algorithm, the literature urges scholars to make the assumptions behind age-period-cohort models apparent (Bell, 2020; Fosse & Winship, 2019a, 2019b). For the calculations below, and in accordance with the hypotheses above, these assumptions are simply that everyone might change their opinions (1) both linearly and nonlinearly with the progression of age over the life course and (2) linearly with the progression of historical time. However, I use an HAPC model as a robustness test to ensure that its results do not contradict the main results, which they do not (see Online Annex “OA5: HAPC” for a discussion and results). Last, the dependence of attitudes within countries needs to be accounted for. Multilevel regressions inflate standard errors accordingly. For these multilevel regressions, I cluster individuals in countries and calculate random intercepts, which is the simplest way of clustering. Rather than calculating random slopes, I also calculate everything on a per country basis to test whether the results presented below exist in individual countries as well. This approach is also substantively important for understanding whether generations exist in some countries but not in others (for results, see “OA4: Different countries”). The main results present linearly scaled marginal effects. However, using odds ratios yields substantively similar results (see “OA7”). The code used to obtain the results is available online, and the data are publicly available; thus, all calculations are replicable. Results How Is theYear ofBirth Related toWork Motivation? The first model shown in Table4 explains work motivation through cohort membership while controlling for age, measured as the effect of becoming 10 years older (Age) as well as age squared (Age*Age). The second model additionally controls for the timeperiod (Period) by accounting for the effect of 10 years of elapsed historical time. Ten-year increments have been chosen simply for convenience, as the effect size of 1 year is very small, making it easier to interpret the effect of the passing of 10 years of age and calendar time. In Model 1, Country_SD indicates that the importance of work deviates by 0.2 from country to country, while Individual_SD indicates that the standard deviation between individuals is 0.7. This means that only approximately 7% of the variation in the importance of work lies between countries, while 93% of the variation lies between individuals (intraclass correlation coefficient = 0.22/(0.22 + 0.72)). Substantively, Model 1 shown in Table1 (and its plotted marginal effects in the left-hand graph of Fig.1) shows that before accounting for time period (the historical time when respondents were surveyed), Generation Y’s predecessors indeed seem to have assigned more importance to work. For example, the birth cohort of 1990–1994 (Generation Y) assigned work an importance of 3.58 out of 4 possible points. In contrast, the first birth cohort of 1925–1929 assigned works an importance of 3.85. The importance of work thus indeed seems to decline with each successive birth cohort. This fits the literature’s assumption that later-born birth cohorts, particularly Generation Y and Z, are less motivated to work. Table 4 Multilevel regression models for importance of work Generations: WW, World War, BB, Baby Boomer + p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001 Important in life: work (1) (2) WW1925 0.27*** (4.69) 0.01 (0.21) WW30 0.21*** (3.93) −0.02 (−0.48) WW35 0.16** (3.17) −0.05 (−1.13) WW40 0.14** (3.19) −0.06 (−1.29) BB1945 0.12** (3.03) −0.06 (−1.38) BB50 0.10** (2.97) −0.05 (−1.46) BB55 0.09** (2.91) −0.05 (−1.39) BB60 0.08** (2.91) −0.04 (−1.26) BB65 0.06*(2.35) −0.04 (−1.61) X1970 0.04*(2.03) −0.04+(−1.68) X75 0.03+(1.80) −0.03 (−1.42) X80 0.03*(2.15) −0.01 (−0.69) Y1985 0.02*(2.16) −0.00 (−0.00) Y90 0.00 (.) 0.00 (.) Y95 −0.01 (−0.50) 0.01 (1.05) Z2000 −0.03 (−1.13) 0.01 (0.34) Age+10 years −0.05*** (−4.86) −0.01 (−1.02) Age+10*Age+10 −0.05*** (−12.85) −0.04*** (−12.85) Period+10 years −0.04*** (−3.57) Constant 3.58*** (112.34) 3.69*** (125.03) Country_SD 0.20*** (−27.20) 0.20*** (−27.24) Individual_SD 0.70*** (−19.33) 0.70*** (−19.33) AIC 1,247,165.41 1,247,127.53 BIC 1,247,390.97 1,247,364.36 ICC_level_2 .071 .071 Countries 113 113 Observations 584,217 584,217 903Journal of Business and Psychology (2024) 39:897–908 1 3 However, Model 1 shown in Table1, and its visualization shown in the left-hand graph of Fig.1 only control for a cohort’s age. These apparent cohort effects can therefore be due to two reasons. First, later-born birth cohorts might indeed find work less important, irrespective of when they have been surveyed, which would be an actual cohort effect. Second, however, it is important to consider that one’s birth cohort is strongly correlated with the historical time period during which an individual was interviewed (r =.42, sig < .001). In other words, later-born cohorts have on average been surveyed later, when everyone may find work less important. In this case, what appears as a cohort effect would truly be a period effect. It would then only appear as if successive birth cohorts find work less important, while the actual reason for declining work motivation is that everyone finds work less important over time (a period effect). To distinguish between these two possibilities, Model 2 shown in Table2 controls not only for individual age but also for time period. The right-hand graph of Fig.1 visualizes the corresponding cohort effects, which are now controlled not only for the progression of an individual’s age on work motivation but also for the progression of historical time on an individual’s work motivation. Controlling for age and period shows that almost all birth cohorts find work similarly important. Thus, contrary to the predictions of the literature that later-born generations find work less important due to a cohort effect, Model 2 shown in Table1 and the right-hand side of Fig.1 show that actual cohort differences hardly exist when cohorts are asked about their work motivation at the same age and the same historical time. This is possible due to the significant and substantively negative period effect of Model 2 shown in Table1, which indicates that with every 10 years of calendar time, work is seen as 0.04 points less important by all birth cohorts. Thus, with every 10 years of passed calendar time, the importance of work decreases for all cohorts of all ages by a roughly similar magnitude as the difference between two generations. Because later generations have been asked later in historical time, they appear more work averse. However, as the lower level of work motivation is related to the time at which work motivation was queried rather than being related to when an individual was born, this is a period effect instead of a birth cohort effect that could be interpreted as generational. To show this explicitly, “OA2” in the Online Annex models the importance of work as a function of the historical time period. The graphs thus show how the importance of work tends to decrease after 1990 even before controlling for cohort effects. However, after controlling for cohort effects, 3.85 3.79 3.74 3.72 3.7 3.68 3.67 3.66 3.64 3.62 3.61 3.61 3.6 3.58 3.57 3.55 3.5 3.6 3.7 3.8 3.9 WW1925 WW30 WW35 WW40 BB1945 BB50 BB55 BB60 BB65 X1970 X75 X80 Y1985 Y90 Y95 Z2000 Not controlling forperiod 3.7 3.66 3.64 3.63 3.63 3.63 3.64 3.65 3.65 3.65 3.66 3.68 3.69 3.69 3.7 3.7 3.5 3.6 3.7 3.8 3.9 WW1925 WW30 WW35 WW40 BB1945 BB50 BB55 BB60 BB65 X1970 X75 X80 Y1985 Y90 Y95 Z2000 Controlling forperiod Average marginal effects, relative to GenY(birth cohort 1990) Generation/birth cohort Generations:WW=World War,BB=Baby Boomer Important in life:Work Fig. 1 Importance of work before controlling period effects (left) and after (right) 904 Journal of Business and Psychology (2024) 39:897–908 1 3 the period effect of declining work motivation remains almost the same. Thus, period effects explain why later-born cohorts find work less important (because they have been asked later in historical time, when everyone finds work less important). Conversely, cohort effects do not explain that everyone finds work less important with the passing of historical time. Put even more simply, period effects remain after controlling for cohort effects, but cohort effects do not remain after controlling for period effects. This outcome implies that studies which claim that work is unimportant for Generation Y assume a cohort effect, when truly what they show is a period effect. Substantively, this means that members of Generation Y, Z, X, or any other generation are not work averse due to when they were bornand thus not as a generation. Instead, everyone is becoming work averse over time, and later-born cohorts simply fit into this trend of declining work motivation, which affects all birth cohorts similarly. As later-born individuals have been asked later in historical time, they at first glance appear to find work less important due to when they have been born, when in reality, they find work less important because they have been asked later in historical time, i.e., when everyone finds work less important. In addition, age effects are much stronger than cohort effects. To illustrate this, the graph under “OA2” in the Online Annex shows a typical “importance of work” age trajectory after controlling for period and cohort effects. This shows how during an individual’s life cycle, the importance of work typically increases from 3.42 around age 15 to 3.66 around age 40 and then declines to 2.72 around age 85. Thus, the importance of work varies by as much as 0.94 points over the life cycle, irrespective of cohort membership or period. In comparison, no single cohort is more than 0.06 points apart in how important it considers work to be after controlling for age and period. The importance of work thus varies approximately 15 times more during an individual life cycle than the difference between any two birth cohorts can account for. Both age and period effects therefore predict the importance of work far better than any generational difference. This disconfirms the Generational Hypothesis (H1) in favor of the Life Course Hypothesis (H2) and the Historical Hypothesis (H3). Simply put, an individual’s age and historical time period explain work motivation, while his or her generational membership does not. Different Specifications andExplanations Age-period-cohort analysis is sensitive to model specifications. It is therefore particularly important that results are robust across different specifications. First, generational differences might be found beyond the specific variable of work motivation analyzed in the main regression. Regressions shown in the Online Annex, therefore, analyze additional work attitudes, such as the subjective importance of (1) leisure time, (2) good work hours, (3) the opportunity to use initiative, (4) generous holidays, (5) thinking that you can achieve something, (6) having a responsible job, (7) having a job that is interesting, (8) having a job that meets one’s abilities, (9) having pleasant people to work with, and (10) meeting pleasant people as important aspects of one’s job. Similar to work motivation, the first eight aspects seem to change (in this case, become more important) from one cohort to the next before controlling for the periodof measurement. However, after controlling for the historical time period, no cohort effects remain for any of these variables. Only the importance of “generous holidays” increases slightly from one cohort to the next after controlling for the historical time period. The importance of having pleasant people to work with even slightly decreases from one cohort to the next after controlling for time period. The last variable, which measures the importance of “meeting pleasant people” as an important job aspect, does not even display a clear cohort effect before adjusting for the period and still does not do so afterward (for all results and their discussion, see “OA3: Further variables”). These findings support the main result by showing that beyond the specific tested variable of work motivation, cohort membership does not explain other types of work attitudes after accounting for the effects of the historical time period. This means that individuals change their work attitudes with age and over historical time periods, but not on a generational basis. Second, some scholars argue for “a global generation [with] a global consciousness” (Edmunds & Turner, 2005: 564f). This can be seen as a strong assumption since generational trends might differ from one country to another, as countries have different cultures and different events that might have influenced birth cohorts. That cohorts may differ between countries has already been handled by using multilevel regressions, which cluster individuals within countries, rather than assuming that all individuals pertain to the same homogenous sample. However, it remains unclear whether generations exist in some countries but not in others. This is why “OA4” in the Online Annex displays separate results for each of the 34 countries that participated from the earliest to the latest IVS waves and has at least 100 individuals in the base category of the 1990–1994 birth cohort. These individual country results, as shown in “OA4” in the Online Annex, support the main result fairly uniformly; of the 34 analyzed countries, only Austria, Chile, Mexico, and Turkey show a cohort-based decline in work motivation after accounting for period effects. In other countries (such as, notably, the USA), no cohort trend exists after accounting for time period, while in yet others (such as Germany), there is even a cohort trend of increasing work motivation after accounting for time period.