Exploring Intergenerational Attitudes in Portuguese Work Contexts Using Age Stereotypes

ABSTRACT An age-diverse sample of Portuguese workers (n = 474) reported their preference for several stereotypical personality traits of different generations in coworkers and team leaders. The results show that, overall, middle-aged traits were preferred in both cases. Only younger or highly educated participants preferred young as much as middle-aged traits. Respondents occupying a supervising role did not differentiate among age traits. These results suggest that age discrimination levels among work peers are moderate, vary widely with sociodemographic characteristics and organizational positions, and are strongly influenced by social beliefs about the “right” age for work, the middle period.


Introduction
Work contexts are becoming increasingly age-diverse with the well-known aging of populations in the Western world and the need to delay the retirement age.This has been a well-documented tendency in many Western countries since the beginning of the millennium (e.g., Chybalski & Gumola, 2021).Older workers have remained or want to stay in organizations, changing the usual human context of workplaces with their more significant presence.In the past, especially in the 1990s, many workers retired early, so older workers (i.e., those aged 60 and above) represented a minority of organizations' personnel.Such small numbers of older employees have led to the ascription of low status to these workers in organizations, overlooking their unique abilities and stimulating age stereotypes that are at odds with the goals of most work organizations (e.g., Posthuma & Campion, 2009).Some research has also shown that younger workers feel discriminated against, in some areas of activity, in payment and benefits, promotions, and training (cf., De la Fuente-Núñez et al., 2021, for a review).Indeed, it is difficult to avoid the stereotypes associated with each of these life stages and the intergenerational conflicts that may arise in all age-diverse contexts (Joshi et al., 2011(Joshi et al., , 2010)).
Since age identities may interact with work identities, such as a team's or department's, influencing the organizational climate, it is now more important than ever to understand the new dynamics of work relations emerging with the increased age diversity (Joshi et al., 2010;Kunze & Boehm, 2013;Ng & Parry, 2016).
Numerous studies have focused on discrimination against older workers by employers and managers (cf., Cebola et al., 2021, for a review).However, to our knowledge, research has yet to analyze the cross-attitudes of work peers of different cohorts.Thus, the present study attempts to fill this gap, further including the middle-aged cohort, on which there is virtually no research, suggesting that they are, as Finkelstein et al. (2013, p. 635) put it, "the norm around which the younger worker and the older worker are to be compared."Indeed, research has justifiably focused on cohorts that experience discrimination in work contexts; the shortage of data on middle-aged workers seems to indicate that, compared with other cohorts, they hold the characteristics that are more valued at work.
Given that organizations generally have a formal hierarchy, the analysis of peers' attitudes must consider the possible variations due to their occupation of supervisory roles.Although supervisors are, by definition, closer to operatives than top managers and administrators, they often must endorse the management positions, traditionally entailing a preference for younger workers to the detriment of older workers (e.g., Karpinska et al., 2013;Mulders, 2020).
Simultaneously, intergenerational attitudes may differ depending on situations and contexts; they include multiple and differently valued dimensions.For instance, Netz and Ben-Sira (1993) measured cross-generational attitudes in nonspecific contexts, considering aspects such as efficacy, autonomy, and acceptability, and obtained different results for each dimension.Research in work contexts has also noticed this double evaluative standard, namely the beliefs about older workers being gentle and, at the same time, slow and outdated (e.g., Finkelstein et al., 2013;Posthuma & Campion, 2009).Thus, it is essential to account for this multidimensionality when analyzing peers' relationships in age-diverse work contexts.
Like some other attributes, for instance, gender or ethnicity, age is likely to motivate social desirability issues in judgments and evaluations.Although informal communication has plenty of prejudiced references against older adults (McCann & Giles, 2002), people may purposely act in a nondiscriminatory fashion to present themselves as tolerant and fair.Stereotypical traits have been used to indirectly measure opinions and feelings about social groups to avoid such social desirability motives in responses.For instance, Kleissner and Jahn (2020a) recently employed them in an "implicit association" methodology -time to respond to positive or negative stereotypical traits after being exposed to pictures of older or younger workers.Thus, we decided to use age-stereotypical traits as a proxy for age labels to assess respondents' attitudes toward peers free of self-presentational motives (cf., Levy & Banaji, 2002).
The present study was conducted with Portuguese workers residing in the northwestern region (one of the most populated in the country).It comprised a pilot study and a main study.The pilot test aimed to obtain a list of personality traits associated with workers of the younger, middle-aged, and older cohorts.The most frequent and distinctive traits for each cohort were then retained to be used in the main study, in which an age-diverse sample of workers rated the traits of preferred coworkers and leaders.This procedure aimed to achieve a nonobtrusive measure of attitudes, avoiding age category labels, which tend to produce socially desirable responses.Furthermore, to account for the multidimensionality of intergenerational attitudes in work contexts, we asked the participants to rate age traits both in preferred coworkers and in team leaders, attempting to assess attitudinal differences due to horizontal or vertical relationships in work contexts (Collins et al., 2009;Perry et al., 1999).
Given the novel nonobtrusive method used, the present study is essentially exploratory.However, we could formulate three hypotheses based on social identity theory and previous results obtained with conventional methods.We expect an overall preference for middle-aged traits (Hypothesis 1) based on the widespread belief that the skills and competencies required for good work performance are more often found in middle-aged individuals than in individuals of other age groups (Finkelstein et al., 2013).At this stage of life, people have the energy and dynamism to innovate and face challenges, but they have the experience and knowledge to avoid errors and leave aside projects that are too risky.On the other hand, younger people are dynamic and bold but lack practice (De la Fuente-Núñez et al., 2021;Rabl, 2010), whereas older ones are experienced but afraid of new solutions (Posthuma & Campion, 2009).
Hypothesis 2 is based on social identity theory (Ellemers & Haslam, 2012;Tajfel & Turner, 1986), according to which, when social categories are salient, people tend to see themselves and others more as members of their respective groups than as individuals.One concomitant of this process is in-group bias, or in-group favoring behavior, namely expressing a preference for in-group members and their distinctive attributes to the detriment of those of the outgroup and, in some circumstances, depreciating the out-group's members and their features.Thus, although age labels are not mentioned in the study, we may conjecture that respondents identify with their own age group's traits and prefer them to others, producing an effect like in-group bias (e.g., Kleissner & Jahn, 2020b).
Finally, in Hypothesis 3, we expect an interaction with the respondents' role in the organization.Since line managers and supervisors are prone to endorse the administration's points of view, which traditionally tend to prefer younger members to the detriment of older ones (e.g., Karpinska et al., 2013;Mulders, 2020), we may expect this bias to emerge as well so that these respondents will prefer younger traits more than those not occupying such roles.

Design
The pilot test consisted of a small survey following a 3 × 3 mixed design in which the between-participants factor was the participant's age group, and the within-participant factor was the target age group.The data were collected at the beginning of 2019 after the Department Ethics Committee had approved the project.

Participants
There were 19 participants, 11 (58%) of whom were female, with ages varying between 22 and 65 years old; M = 42.63,SD = 11.49.Of the sample, 58% had graduated from college, 16% had completed high school, and 26% had finished primary school (nine years).Their job experience varied from 1 to 47 years; M = 22.47, SD = 14.54.All the respondents participated voluntarily in the study.

Measures
The questionnaire began with sociodemographic questions followed by the instruction, "Please write in the lines below the seven personality traits that, in your opinion, best characterize people in each of the three following age groups: 20-35 years old; 35-50 years old; and 50-65 years old."The respondents were then asked to evaluate each of the previously written traits on scales ranging from 1 = very negative to 5 = very positive, with 3 = neutral.

Procedure
We requested the human resource department of a medium-sized firm in the Porto region to ask for volunteers among its employees to participate in a quick poll (developed by researchers from the Psychology Department of the University of Porto) that would take place in their workplace on the next day.On the following day, we contacted the volunteers, handing them, individually, the questionnaire to complete and put in the human resource department's mailbox.We also used social networks to invite people to participate in a brief poll and shared with the volunteers the link to Google Docs, where the questionnaire was located.

Analysis
We used a repeated-measures ANOVA to test the differences in positivity among the three age groups.

Results
We first categorized the respondents according to age: six were 24-35 years old, six were 36-47, and seven were 52-58.The two genders were evenly represented in the categories; χ 2 < 1.
Most respondents complied with the instruction, producing an average of 19.95 words per respondent.The 379 responses were then analyzed by three independent judges, who aggregated some of them according to semantic similarity (only words considered equivalent by the three judges were aggregated).The final wordlist consisted of 228 traits (Table 1 displays the more frequently reported ones).The final wordlist had an overall positivity mean of 3.36 and SD of 0.51 (cf.original databases at https://osf.io/fv3ud/).
To detect possible biases in the evocation of the traits, we averaged the respondents' evaluation of the traits evoked for each age group, forming three age indices, and analyzed them through a repeated-measures ANOVA.The test yielded a non-significant effect, F (2, 36) = 2.64, ns, indicating that the traits evoked for the three age groups had similar positivity levels.

Design
The main study consisted of a survey following a 3 × 3 × 2 design in which the between-participants factor was the participants' age group, the withinparticipant factors were the three traits' age groups, and the two measures were coworker and leader.The data were collected between 2019 and 2021.
The Department Ethics Committee approved the design and materials.

Participants
The sample comprised 474 respondents, of whom 53% were female, with ages varying between 20 and 65 years old, M = 41.59,SD = 12.46.We categorized them according to three age groups: younger (20-35 years old), n = 161, with an average age = 27.24,SD = 4.10; middle-aged (36-50), n = 159, with an average age = 42.10,SD = 4.24; and older (51-65), n = 154, with an average age = 56.08,SD = 3.97.Given the time of data collection, we may consider that the younger group roughly corresponds to Millennials, the middle-aged group to Generation X, and the older group to tardy Baby Boomers and early members of Generation X (cf., Costanza et al., 2012).
Regarding the respondents' education level, 41% had graduated, 32% had completed high school, and the remainder had finished primary school (9 or 6 years, depending on their age).The sample thus had a sociodemographic profile equivalent to the pilot test sample.Only 7% were not working (although they had worked previously).The number of working years varied from 6 months to 49 years; M = 21.47 years, SD = 13.75.About one-third of the respondents (30%) reported having supervisory functions in the organization, such as coordinator, supervisor, team leader, or line manager.In addition, 32% of the sample worked in the public sector.Using the National Occupation Classification, 32% had intellectual or scientific occupations, 29% were professionals of protection or security or sellers, 11% were administrative clerks, 10% were qualified industry technicians, 8% were directors or similar, and the remainder were machine operators, intermediate technicians, military, or non-qualified workers.

Measures
To build the questionnaire, we began by selecting 18 traits, six for each age group, from the pilot test list (cf., Table 1 -traits in bold).This selection combined four successive criteria: (1) traits were among the most frequently evoked for each age group; (2) traits were more frequent in the age group in question than in any other group; (3) traits had been evoked by respondents of at least two different groups; and (4) there were no differences in positivity among the three groups of traits.The 18 selected traits were mixed in a list to be presented in the questionnaire.Thus, the questionnaire consisted of the initial socio-demographic questions followed by the requirement to report the extent to which each of the 18 words on the list described persons whom they liked or would like to work with on a 9-point scale (1 = totally no; 9 = totally yes).Next, we asked the participants to think of capable work team leaders and report the extent to which each of the 18 words described that person.For half of the sample, the task order was reversed.

Procedure
We requested the administration of one high school, one university department, and two medium-sized companies to provide information about the study to their staff.The e-mail also included the questionnaire link and asked respondents to share it with their acquaintances.We also used social networks to invite participants and asked respondents to share the link with their acquaintances.Only participants aged 20-65, who were currently working or had worked before and resided in the northwestern region, were retained.Several questionnaires (26%) were handed on paper to individual respondents, especially older respondents, to obtain an even number of respondents in the three age groups.

Analysis
We used principal component analysis (PCA) and Cronbach's alpha to examine the internal consistency of the sets of traits.The three groups of traits' ratings for coworkers and team leaders were analyzed through successive mixed ANOVAs with the ratings of the three sets of traits as withinparticipant factors (Age Traits) and respondents' age, gender, and so on as between-participants factors.When required, age was introduced as a covariate.In all cases, we used Bonferroni adjustment for multiple pairwise comparisons, and when non-sphericity of data occurred, we considered the Greenhouse-Geisser correction.

Results
Factor analyses of the six sets of traits (3 age groups × 2 measures) revealed that a single factor was extracted in all cases except for the sets of younger traits, in which two components were produced in both measures: one aggregating "Curious," "Spontaneous," and "Dreamers" and the other aggregating "Creative," "Dynamic," and "Persevering."Thus, we decided to analyze only the traits of the second component in the statistical analysis because they had higher frequencies in the pilot test than the others while keeping a reported positivity level like that of other groups' traits (average positivity = 3.97).Furthermore, to maintain the same criterion in the other scales, we analyzed only the three traits with higher loadings: for the middle-aged, we retained "Committed," "Stable," and "Sensible" (average positivity = 4.10); for older people, we retained "Calm," "Organized," and "Assured" (average positivity = 4.00).The six scales presented good reliability levels, with Cronbach's alphas varying between 0.74 and 0.84, allowing the averaging of their scores.
In sum, middle-aged traits received the highest ratings for coworkers and team leaders when considering the whole sample.Not hypothesized but consistent with the literature, both measures rated older and younger traits identically.

Respondents' age group
To check Hypothesis 2, we conducted two mixed ANOVAs of the three sets of trait ratings in coworkers and team leaders, respectively, with the Respondent's Age Group as a between-participants factor.The tests revealed a significant interaction in the coworker measure, F (4, 942) = 4.41, p = .002,η p 2 = .02,but not in the team leader measure, F (4, 942) = 1.61, ns.
In sum, Hypothesis 2, which assumed that in-group bias would lead participants to prefer their age group traits, was only partially confirmed.We considered in-group bias when groups diverged significantly from the whole sample's results, overrating their own group's traits.The results show that, regarding preferences for coworkers, the younger cohort had preferences that fit an in-group bias pattern -members of the younger group diverged from the whole sample by overrating their traits.For middle-aged respondents, the ingroup bias would be to differentiate their traits even more than the whole sample.Instead, they diverged from the whole sample, rating their traits similar to those of the older group; therefore, the in-group bias hypothesis was not confirmed among this cohort.Finally, older respondents did not differentiate among sets of traits, thus diverging from the whole sample.However, this divergence was not aimed at favoring their age group relative to the others; consequently, Hypothesis 2 was also not confirmed in the older group.

Respondents' role
Concerning Hypothesis 3, we repeated the mixed ANOVAs with the Respondents' Role as the between-participants factor.Given that supervisors were, on average, older than non-supervisors, t (472) = 5.92, p < .001,we had to control for respondents' age, introducing their raw age as a covariate into the analysis.
The ANCOVA yielded a significant interaction in the coworker measure, F (2, 942) = 6.50, p = .002,η p 2 = .01;in the team leader measure, F (2, 942) = 2.20, ns.Decomposing the significant interaction, we found differences among respondents who did not occupy a supervisory role, F (2, 656) = 9.41, p < .001,η p 2 = .03,but not among those occupying this role, F (2, 284) = 1.17, ns.Pairwise comparisons revealed that the respondents who did not occupy a supervisory role, in line with the whole sample's results, preferred middleaged traits in coworkers to either young traits or old traits (both p < .001)and did not differentiate between these two (ns) (cf., Figure 2).Thus, Hypothesis 3 was not confirmed; that is, occupying a supervisory position in the work context does not entail the endorsement of the management's traditional preference for younger workers' stereotypical attributes, such as creativity and dynamism.Contrarily, supervisors were fairer in their preferences, not distinguishing among age traits.

Respondents' gender
The ANOVAs yielded a nonsignificant Age Traits × Gender interaction, F (4, 1888) = 1.11, ns, indicating that male and female respondents' preferences did not differ.

Respondents' education level
Raw education levels were aggregated into two groups -1 = high school or lower, and 2 = higher education -and the new variable (Education Level) was introduced into mixed ANOVAs as a between-participants factor.As those with more schooling were younger than those with less schooling, t (472) = 7.11, p < .001,we controlled for respondents' age by introducing this variable as a covariate.
Regarding the ascription of traits in team leaders, the main effect of Age Traits was only significant in the higher education subsample, F (2, 386) = 6.45, p = .003,η p 2 = .03;in the lower education subsample, F (2, 386) = 2.20, ns.Pairwise comparisons showed that the higher education subsample rated middle-aged traits similarly to younger traits (ns) but higher than older traits (p < .001),and the latter two were not distinguished (ns).The lower education sample did not differentiate among traits for team leaders.
In sum, respondents with lower education tended to prefer middle-aged and older traits to younger traits.In contrast, respondents with higher education preferred middle-aged and younger traits to older traits.

Respondents' employment sector
Participants' employment sector was unrelated to their age; t (472) = 1.30, ns.The mixed ANOVA yielded a marginally significant Age Traits × Sector interaction in the coworker measure, F (2, 944) = 2.65, p = .08,and a significant interaction in the team leader measure, F (2, 944) = 7.60, p < .001,η p 2 = .02.Decomposing the interaction of the coworker measure revealed that Age Traits were rated differently by participants of the two sectors, the public sector, F (2, 300) = 8.37, p < .001,η p 2 = .05,and the private sector, F (2, 644) = 4.90, p = .01,η p 2 = .02.In the public sector, middle-age traits were rated similarly to older traits (ns) but higher than younger traits (p < .001).In the private sector, the inverse occurred; middle-aged traits were rated similarly to young traits (ns) but higher than old traits (p < .001).The decomposition in the team leader measure also revealed a significant Age Traits × Sector interaction, both in the public sector, F (2, 300) = 9.67, p < .001,η p 2 = .06,and in the private sector, F (2, 644) = 9.31, p < .001,η p 2 = .03.Pairwise comparisons showed the same patterns of preference for the coworker measure: the public sector's preference for older traits and the private sector's preference for younger traits.

Discussion
The pilot test showed that the personality traits associated with workers of each of the three generations differ considerably but do not differ widely from those obtained in other Western cultures (e.g., Finkelstein et al., 2013;Posthuma & Campion, 2009).
The main study showed that, even though the attitudes were measured indirectly, middle-aged workers were preferred overall as coworkers and as leaders, as predicted in our Hypothesis 1, thus confirming the perceived higher suitability of this cohort, currently Generation X, for work contexts.We also predicted that, beyond this general result, participants would display in-group favoritism, a tendency to prefer their own age group's traits (Hypothesis 2).This result emerged only among younger participants, who appreciated their age traits as much as those of middle-aged workers.As a case in point, the younger generation (also called Y or Millennials) has been described as highly self-centered (Twenge et al., 2008), which is consistent with self-serving behavior when age groups are at stake.However, even though this notion may help explain our results, as Weber and Urick (2021) observed, the association of specific personality traits with generations tends to oversimplify a rather complex phenomenon and should be considered cautiously.Indeed, one could also conjecture that, rather than being self-centered, younger participants were following the current societal trends that value creativity and originality above other human characteristics more closely than the other cohorts.
The remaining groups of participants did not display such in-group biases.For instance, older participants were the most unbiased, showing an even appreciation of the three sets of traits.Lastly, middle-aged respondents appreciated old traits as much as their traits while devaluing young traits.One possible explanation for this result is that middle-aged workers feel closer to the older cohort as they approach this phase of life and may automatically feel more sympathy for it.At the same time, a positive appreciation of the younger generation may not be so binding because, as Garstka et al. (2004) noted, this cohort is perceived as a transition to a more valued one, the middle-aged cohort.Its members do not feel age discrimination as intensely as those of the older cohort do (cf., Rabl, 2010).Accordingly, people may feel they do not have to show special treatment to the younger group.The opposite reasoning applies to the older group; that is, some discretion is due given that, in general, this group cannot expect to benefit from an improved status in the future; thus, it may be perceived as more harmful to devalue older workers than younger workers.
Nevertheless, North and Fiske (2016) found that younger workers were more prone to treating older coworkers unresponsively, at least in some dimensions, than any other cohort.The authors further found that such a result is mediated by "succession beliefs," that is, the more unyielding young workers' conviction that older workers should retire to make way for younger generations, the more they withheld scarce resources (like training) from older workers.These findings could also explain why, in our results, younger participants rated old traits much lower than other traits (see, Figure 1).
In sum, considering the above set of results, our Hypothesis 2 was not confirmed (except in the case of the younger cohort), indicating the existence of many other work-related factors which overwhelm the general tendency to favor one's group attributes.As Lyons et al. (2019) pointed out, generational identity in the workplace is a complex construct affected by proximal influences, like those of the immediate workgroup or the organizational culture, and by more distal ones, such as external influences of the activity sector and the workers' professions as well as of societal beliefs and norms related to age (cf., also Weber & Urick, 2021).This means that generational identities in the workplace are mutable and may even remain unexpressed when, for instance, organization or work team identities are maintained saliently.Thus, such complexity of generational identity in workplaces may help to explain why ingroup favoritism did not emerge in our results since this behavior derives from members' need to maintain a highly distinctive group identity (Ellemers & Haslam, 2012).
The results also showed variations related to respondents' belonging to social groups other than age.This was the case for respondents with higher education, who appreciated younger traits as much as middle-aged ones, thus devaluing older traits, and for respondents employed in the public sector, who appreciated older traits as much as adults' traits while devaluing younger traits.In the first case, one may conjecture that respondents with higher education were more attracted to the traits, per se, that are stimulated in higher education, such as creativity and dynamism, than to the generation they were denoting.However, this association might be challenging to confirm.Otherwise, public workers' preference for older traits could be explained by the type of activities performed in this sector, which are generally different from those of the private sector.Indeed, the activities accomplished by the public sector -administrative, education, health, and so on -are considered more essential and are less subject to competition than those accomplished by the private sector, at least in Portugal.We may speculate that the different characteristics of the sectors may result in the public sector's higher appreciation of older workers' organization and stability along with the private sector's higher appreciation of younger workers' creativity and dynamism, which are traits suitable for competitive contexts.These results suggest that age cohorts may be differently appreciated across work contexts and by peers with different sociodemographic characteristics.However, it should be remembered that there were no significant differences according to participants' gender in our data.
The results were identical when comparing the ratings of traits for coworkers and team leaders in the sample; middle-aged traits were preferred to those of the other cohorts in the two measures.The exact matching results of the measures were found when participants' gender, education level, and employment sector were included.However, when participants' age group was considered, the differences among sets of traits were significant in the coworker measure but not in the team leader measure, suggesting that leadership-specific criteria might have affected this set of ratings.Indeed, as several researchers have observed, the reciprocal attitudes in the supervisor-subordinate dyad are always complex, but they become even more complex when the dyad does not correspond to the conventional older supervisor-younger subordinate type (e.g., Collins et al., 2009;Perry et al., 1999).Power issues related to age and organizational role status and the status incongruence of less traditional dyads may be faced very diversely by individuals independent of their age cohort.We could thus speculate that significant variations among participants in this matter might have blurred the patterns of differences observed in the coworker measure.
Regarding the differences between those holding a supervisory role and the other respondents, the results showed that the former were fairer than operatives, suggesting that supervisors trusted that both younger and older workers could make equally significant contributions, appreciating the dynamism of the former as well as the experience of the latter.These results refute our Hypothesis 3, which, due to the inexistence of consistent direct evidence, was drawn from indirect but more confirmed evidence on employers' tendency to prefer younger to older workers (Karpinska et al., 2013;Mulders, 2020).However, some research has shown that many supervisors, working more closely with their subordinates than top managers, refuse to become involved with human resource politics and informally disregard some directives, which may harm their relationship with subordinates (cf., Knies et al., 2015;Leisink & Knies, 2011).The results may also be viewed as the beginning of a turnoff in administrations' traditional positions toward older workers and the adoption of a receptive attitude toward the benefits of age diversity (Lagacé et al., 2019) and/or the awareness of the problems that arise with an age-discriminating climate in the workplace (Kunze & Boehm, 2013).
Regarding the methodology of the study, it is essential to note that the use of stereotypes to measure intergenerational attitudes implies that the results are restricted to workers who embody their age-stereotypical traits: the responses concern traits, not coworkers or team leaders.Although stereotypical traits, by definition, are more prevalent in the respective social group than in others, members of other groups may hold them as well (Rioux & Mokounkolo, 2013;Weber & Urick, 2021).The nonobtrusive method of assessing attitudes attempts to circumvent socially desirable responses and is especially useful for predicting subtle forms of discriminatory behavior (Levy & Banaji, 2002;McCann & Giles, 2002).However, everyday intergenerational relationships in the workplace encompass both intended and unintended behavior; consequently, the present results are less helpful in understanding the social dynamics that derive from intentional behavior.Even so, it is essential to recall that subtle age biases derive from associative structures that determine behavior whenever conscious processing is less effective due to hastiness, emotive or bewildering circumstances, and so on (Kleissner & Jahn, 2020a).Moreover, although more difficult to discern, subtle biases are generally judged as harmful as explicit discrimination since they are more difficult to question (Levy & Banaji, 2002).

Limitations
Despite the exciting results, the present study has several limitations to be addressed in future investigations.Perhaps the most critical limitation involves the potential biases introduced by the few stereotypical traits.Although we applied special care in selecting traits for each group regarding their frequency, distinctiveness, and evaluation, inaccuracies could have been prevented by using more traits.Employing a larger sample in the pilot test to guarantee a greater consensus about the stereotypical traits and their respective values would also help to improve this methodology.
There are also limitations concerning the generalizability of the results.The sample of the main study was reduced, considering this type of research's goals, and biased in terms of the sector (the public sector was overrepresented) and education level (more than 40% were graduated), which, as shown by the present results, are significant moderators.Even though the age preferences of these two overrepresented groups are opposed, canceling each other out (so that the results of a more balanced sample should not change significantly), these biases should be considered in future studies to improve sample generalizability.The validity of the present study is also limited to the Portuguese working population.Thus, a significant future contribution would be to achieve cross-population generalizability, i.e., to ascertain whether this pattern of results applies in other countries and cultures.

Implications for theory, policy, and practice
Overall, middle-aged traits were rated more highly for coworkers and considered more appropriate in team leaders than any other cohorts' traits.However, the slight differences observed between means suggest that the age-discriminatory attitudes among coworkers are less severe than those of many employers and administrators.Older workers' traits were appreciated by some groups of respondents, suggesting that they are not viewed as the "poor relative," and younger workers' traits were appreciated like those of middle-aged workers by other groups of respondents.Nevertheless, the existence of specific education on human development, especially for the younger cohort, to become more aware of both the diverse ways in which the same things are perceived at different life stages and the different experiences people have throughout their lifespan would further enhance the understanding of peers from other generations (Graf & Bartlett, 2020).Such a goal could also be achieved through continued contact between workers of different cohorts, for instance, in intergenerational cooperation, which promotes familiarity and interpersonal knowledge and moves beyond age stereotypes (e.g., Joshi et al., 2011Joshi et al., , 2010;;Lyons et al., 2019).In times of increased age diversity in workplaces, promoting intergenerational contact and cooperation, with an emphasis on common goals, should be an important goal of organizational management in the building of a healthy intergenerational climate, which, as recent evidence has shown, is directly related to work satisfaction (e.g., Firzly et al., 2021;Lagacé et al., 2019).

Conclusions
The present study provides further knowledge on age discrimination in work contexts by focusing on peers' cross-generational attitudes.Not denying the importance of employers, who are the ultimate agents of recruiting and laying off, conceding, and withholding promotions, as well as allocating training, the age-based attitudes of work peers and the resulting behaviors are prominent aspects of the building of a beneficial intergenerational climate.Moreover, including middle-aged workers (mainly from Generation X), whom research has generally overlooked, enriches the knowledge on this topic by informing about their views and how the other cohorts see them.Finally, the nonobstructive methodology allowed the examination of implicit intergenerational attitudes, essential factors of subtle discriminatory workplace behaviors.

Figure 1 .
Figure 1.Preferences of traits in coworkers as a function of respondents' age group.

Figure 2 .
Figure 2. Preferences of traits in coworkers as a function of responder's role.

Table 1 .
More frequently reported traits for each target age group with corresponding positivity mean (Pos.).
Traits in bold were used in the Main Study.Traits with * were effectively analyzed.