ABSTRACT
Most Western societies face the challenge of steadily ageing workforces. In recent decades, research on ageing has intensively focused on the subjective age concept to understand the challenges and risks of increasingly ageing workforces. Nevertheless, the subjective age construct is subject to several conceptual uncertainties, namely, regarding its stability and potential work-specific drivers of subjective age. We address these limitations by a) investigating the stability of subjective age in a worker sample, and b) identifying work-specific drivers (e.g., negative work events, positive work events, work stress) of subjective age perceptions. Building on social identity and lifespan theories, we test our conceptual assumptions with an online sample of 168 U.S. employees, applying growth curve modelling in a daily diary study over one workweek. Results indicate that subjective age is a mutable construct and varies between- and within-person in the course of a workweek. We identify positive work events and work stress as between-person drivers and negative work events as a within-person driver of subjective age. We discuss theoretical implications of these findings as well as consequences for practitioners.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. The authors use the term “subjective age bias” (SAB) for what we call relative subjective age. Nevertheless, we use the term relative subjective age in this paper, as this is the more common expression (e.g., see Kotter-Gruehn, Neupert et al., Citation2015; Kunze et al., Citation2015).
2. The authors postulate subjective age and age-group identification as moderators in the relationship between negative work events and daily affect and cognitive engagement.
3. Relative subjective age represents the difference score between chronological age and subjective age, showing if a person feels older or younger than his/her chronological age. Negative values (low relative subjective age) indicate the number of years a person feels younger than the chronological age, while positive values indicate the number of years a person feels older than the chronological age.
4. We chose the above-mentioned payment method and system following the recommendation of several articles analysing MTurk samples and their motivation (e.g., Berinsky, Huber, & Lenz, Citation2012; Buhrmester, Kwang, & Gosling, Citation2011; Paolacci, Chandler, & Ipeirotis, Citation2010). Accordingly, we set a compensation amount that is over the average amount paid on MTurk. Furthermore, we considered the bonus procedure in order to increase the likelihood of ongoing or frequent participation.
5. Negative values (low daily relative subjective age) indicate the number of years that a person feels younger than the chronological age at a specific day, while positive values indicate the number of years that a person feels older than the chronological age at a specific day.
6. To understand the within- and between-person variance, we estimate one model with the linear- and one model adding the quadratic time coefficient. Results show that the quadratic time coefficient is significant (µ = .24; p ≤ .001) and shows better model fit [log likelihood (5) = −2226.18; AIC = 4462.37] compared with the model, including only the linear time coefficient [log likelihood (4) = −2231.10; AIC = 4470.20; ∆likelihood-ratio χ² (1) = 9.84, p ≤ 01; ∆AIC = 7]. Therefore, we run all further analyses with the quadratic time coefficient.
7. We test if the random-coefficient model (Model 3, ) is superior to the random-intercept model (Model 2, ) with a significant model improvement [∆lr χ² (2) = 23.20; ∆AIC = 19.20; p ≤ .001]. Consequently,
we apply random coefficient modelling for all further analyses.
8. Robustness checks indicate that findings are consistent when calculating with or without control variables (chronological age, gender, daily subjective physical health, daily subjective mental health, daily pain, daily positive affect, daily negative affect, daily life satisfaction). Results are available upon request.
9. MTurk is an efficient and cost-saving tool to obtain access to a large number of potential participants, which received increasing attention in the last decade for empirical research (Berinsky et al., Citation2012; Buhrmester et al., Citation2011; Goodman, Cryder, & Cheema, Citation2013). We chose this platform because it offers a reliable sample of the U.S. workforce population, independent from organization affiliation. Buhrmester et al. (Citation2011) characterize MTurk as more representative as normal online samples, offering reliable results comparable with results from tradition assessment methods (Goodman et al., Citation2013; Paolacci et al., Citation2010).