Pay-for-performance, job attraction, and the prospects of bureaucratic representation in public organizations: evidence from a conjoint experiment

ABSTRACTDoes pay-for-performance – a hotly debated compensation scheme for incentivizing public service efficiency – induce inadvertent heterogeneity in job attraction that is counteracting the prospects of bureaucratic diversity and representation? Using data from a pre-registered conjoint experiment among US residents (n = 1,501), we examine whether pay-for-performance (compared to fixed pay) affects attraction to a public service job differently across race, gender, and age. Contrary to theoretical expectations, we find that pay-for-performance does not diminish attraction to a public service job within or between demographic groups. In fact, we find indications that pay-for-performance may enhance job attractiveness among individuals identifying with minority racial groups.KEYWORDS: recruitmentpay-for-performancerepresentative bureaucracyequitysurvey experiment Disclosure statementNo potential conflict of interest was reported by the author(s).Supplementary materialSupplemental data for this article can be accessed at https://doi.org/10.1080/14719037.2023.2245841Notes1. Prolific validates country of residence by requiring verification with a non-VOIP phone number with the appropriate country code (https://participant-help.prolific.co/hc/en-gb/articles/360021786394). The quota sampling for race was accomplished using Prolific’s prescreening feature; we posted two identical studies within the Prolific recruitment system – with one study limiting enrolment to White participants, the other limiting enrolment to non-White participants.2. https://doi.org/10.1257/rct.7632-2.0.3. https://doi.org/10.7910/DVN/NSHT68.4. Respondents’ race, gender, and age are treated as binary variables (White vs. non-White; man vs. woman; younger vs. older (median: 31 years old)). For estimating the interaction effect for race and gender, we dropped from our sample respondents who selected ‘prefer not to answer’ or ‘non-binary’.5. Thus, in terms of null hypothesis testing, we cannot reject that pay-for-performance has no effect on job attraction. But can we reject that pay-for-performance has a true effect of substantive interest? Equivalence testing (Lakens Citation2017) with the smallest effect size of interest (SESOI) set at .05—the same value we use for power analysis based on Schuessler and Freitag (Citation2020)—reveals a non-significant equivalency result for the effect of pay-for-performance. As shown in Figure 1 by the 90% confidence interval for pay-for-performance, we cannot reject that the effect of pay-for-performance on job attraction may be at least as extreme as an increase in job attraction of 5% points (although we can reject a decrease in job attraction of 5% points or more). Thus, equivalence analysis suggests that pay-for-performance has no substantial negative effect on job attraction, but the results are inconclusive regarding a potential positive effect of substantive interest.6. Similarly, auxiliary analyses do not support that the interactive effects of pay-for-performance and racial minority status are moderated by the mode of performance evaluation (Online Appendix C, Table A2, model 3), as we might expect if pay-for-performance was perceived as potentially discriminatory when based on subjective performance appraisals.7. Equivalence tests (Lakens Citation2017) – again with the SESOI = .05—reveal significant equivalency results for gender and age, but a non-significant equivalency result for race (see Online Appendix D). Thus, we can reject that the true moderating effects of gender and age on the impact of pay-for-performance on public service job attraction are at least as extreme as ±5% points. However, we cannot reject that pay-for-performance increases job attraction by 5% points or more for racial minority individuals relative to White individuals.