Skip to main content
Log in

The Benefits of Work: A Meta-analysis of the Latent Deprivation and Agency Restriction Models

  • Original Paper
  • Published:
Journal of Business and Psychology Aims and scope Submit manuscript

Abstract

Despite conclusive evidence about the positive impact of working relative to unemployment for psychological well-being, there remains much uncertainty about why working relates to improved well-being. Two theoretical perspectives that have often been contrasted in examinations of this question are the latent deprivation model and the agency restriction model. The former emphasizes (latent) psychological benefits of work (time structure, collective purpose, social contact, social status, and enforced activity), asserting that lower well-being in unemployment is due to the deprivation of access to these benefits. The latter emphasizes the monetary (manifest) benefit of work, asserting that it is the financial strain caused by lacking income that is primarily responsible for restricting agency and lowering well-being in unemployment. Here, we integrate these theories with a meta-analysis based on 90 primary studies/sources, 1147 effect sizes, and 69,723 people. Results support a unified account of these theories: employment provides access to all of these psychological and monetary benefits of work, and each benefit is significantly associated with lower psychological distress and higher life satisfaction. The monetary benefit was especially strongly related to life satisfaction. Meta-analytic structural equation modeling revealed that the benefits (except for enforced activity) fully mediated the effect of employment status on psychological distress; in contrast, only collective purpose, social status, and financial strain partially mediated the effect of employment status on life satisfaction. We discuss the implications of these findings for future research and for individuals, organizations, and policymakers to improve the experience of employment and mollify the harms of unemployment.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2

Similar content being viewed by others

Notes

  1. Unemployment rates in the USA over the past year (early 2022–early 2023) have been the lowest in over 50 years. At a higher rate, obviously many more than 5.7 million people would be experiencing unemployment. Also, this figure excludes another 5.3 million people not in the labor force who currently want a job.

  2. Of note, Warr’s (1987) vitamin model extended and elaborated Jahoda’s and Fryer’s work. This theory proposed several environmental characteristics (i.e., the benefits of work) that accounted for well-being in non-linear ways (De Jonge & Schaufeli, 1998). However, given its focus on job characteristics, which is somewhat outside the scope of the current paper that examines employment and unemployment, and the limited number of empirical studies compared to those that have examined Fryer and Jahoda’s work (Wood & Burchell, 2018), we do not focus on this perspective in the current paper. We return to Warr’s framework in the Discussion section.

  3. Please contact the primary authors for all correlations.

  4. In testing for mean differences in access to the latent and manifest benefits between unemployed and employed groups, sources had to meet the first two inclusion criteria, reporting information for unemployed and employed groups and measuring at least one of the benefits. In testing for the relationship between the benefits and well-being, sources had to meet the latter two criteria, reporting information on at least one of the benefits and at least one well-being variable (i.e., psychological distress or life satisfaction).

  5. Some empirical work has also examined out of the labor force (OLF) individuals in addition to unemployed and employed individuals. We do not include OLF individuals in this meta-analysis as our focus is on unemployment and employment, and also due to the high degree of heterogeneity across OLF categories (e.g., college students, retirees, homemakers) that would render comparisons conceptually unclear. Moreover, there was an insufficient number of effect sizes for each OLF subgroup to be sufficiently power bivariate analyses.

  6. Several studies examined the latent benefits as a single or unitary construct with a unidimensional measure (e.g., Brief et al., 1995; Šverko et al., 2008). We coded nine studies that met our inclusion criteria but did not include them in analyses due to the very small number of effect sizes provided for mean difference and correlational analyses.

  7. An anonymous reviewer raised the point that ESS samples may systematically differ from the other samples given differences in data collection. To investigate this possibility, we conducted a subgroup analysis to determine whether confidence intervals of each correlational estimate from ESS samples overlapped with zero and would therefore give rise to substantively different conclusions versus the confidence intervals of the mean estimate (i.e., based on all studies). The results can be found in the online supplement (Appendix S2). As seen there, half of confidence intervals between the ESS samples and non-ESS samples do overlap. But, the meta-analytic correlations based on ESS samples also tended to be smaller in magnitude than estimates based on non-ESS samples. This may be due to the ESS samples coming from non-published sources, to the in-person interviews of ESS data collection that potentially altered responses (especially for negatively valenced variables, such as financial strain and psychological distress), and/or to differences in item content between these measures and commonly used measures of the benefit and well-being variables.

  8. In cases where single-item scales were used to measure focal variables, we followed previous meta-analyses and coded alpha as .70 (Hülsheger & Schewe, 2011; Wanuous & Hudy, 2001).

  9. Two sets of studies reported relevant effect sizes from the same datasets: Selenko and Batinic (2013), Selenko et al. (2011), and study 2 in Batinic et al. (2010); and study 1 in Batinic et al. (2010) and Paul and Batinic (2010). We coded effect sizes from the studies in these two sets that reported the results based on the most complete set of data (i.e., the largest sample size). We contacted an author of these studies to ensure we were not capturing the same data or effect sizes (i.e., we were capturing independent effect sizes from these sources). Also, note that the study that we did include had reported switching the labels for two LAMB measures, so we coded those effect sizes according to their original labeling to ensure that all effect sizes of a given construct were consistent.

  10. For papers that did not provide effect sizes or all the necessary information to compute them, we e-mailed the authors of the source for missing information, given the source was not from before 2010. Only one author responded and provided the requested information. However, upon closer inspection, this primary study did not meet the inclusion criteria.

  11. For mean comparison analyses of access to the benefits, we recoded a variable’s valence (when necessary) using the scale’s reported midpoint and range. For example, if a study used a negatively valenced measure of time structure (indicating deprivation of time structure rather than access to time structure), ranging from 1 to 5 with a midpoint of 3, and reported a mean of 2.8, we recoded the mean to 3.2. Cohen’s d values were then re-calculated accordingly. Where this information was not reported, the source was unable to be included in the mean comparison meta-analysis.

  12. The difference in independent samples versus sources pertained to the instances where a source may have included more than one study, meaning that within a single source there may have been two or more independent samples that could be included in the meta-analyses.

  13. We allowed the residuals of the mediators to covary in this model. Global model fit (e.g., χ2 value and associated fit indices) was perfect as the model was just identified.

  14. We note that job characteristics theory discusses variables that are similar to the vitamins, which we referenced earlier in the paper when discussing implications for managers, but we elected to discuss the vitamin model due to its relevance for both employed and unemployed people as it is largely an extension of latent deprivation theory (Wood & Burchell, 2018).

References

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to John A. Aitken.

Ethics declarations

Conflict of Interest

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Aitken, J.A., Cannon, J.A., Kaplan, S.A. et al. The Benefits of Work: A Meta-analysis of the Latent Deprivation and Agency Restriction Models. J Bus Psychol (2023). https://doi.org/10.1007/s10869-023-09920-9

Download citation

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s10869-023-09920-9

Keywords

Navigation