Person:

Linos, Elizabeth

This dissertation presents three empirical studies on how to improve human capital in the public sector. In reverse chronological order, the essays ask who is attracted to public sector jobs (Paper 3); consider who can actually get a public sector job (Paper 2); and evaluate how current civil servants can be more effective at doing their job (Paper 1). In doing so, the dissertation presents tools that public managers can use to improve human capital within the constraints of government.

Paper 1 focuses on the rising trend of teleworking options in the US federal government. Using eight years of rich administrative data on 10,000 employees at the United States Patent and Trademark Office (USPTO), I evaluate the impact of instituting a program that allows employees to work from home full time, not only on those who telework, but also on their peers who remain in the office. I find considerable impact on organizational effectiveness. Teleworkers display increased retention and two divergent effects on output: while productivity per hour of examination time goes down, teleworkers examine more applications per quarter because they spend more of their workday examining cases and fewer hours in meetings. Interestingly, having more teleworking peers increases productivity for those who stay in the office, but it also increases their sick leave and attrition, a proxy for burnout. I hypothesize that this is due to increased monitorability and task visibility in the office. This study shows that teleworking programs affect organizations beyond their direct impact on teleworkers and that these additional effects must be incorporated into the overall evaluation of whether teleworking works for organizations.

Paper 2 considers how to improve diversity in the civil service, with a focus on the police. The paper reports the results of a randomized controlled trial (RCT) conducted in cooperation with a UK police force that was experiencing a disproportionate drop in minority applicants at one stage in its assessment process, the Situational Judgment Test (SJT). Drawing on insights from the literatures on stereotype threat (Steele & Aronson, 1995), belonging uncertainty (Walton & Cohen, 2007), and values affirmation exercises (Harackiewicz et al., 2014), we redesigned the wording on the email inviting applicants to participate in the SJT. The results show a 50 percent increase in the probability of passing the test for black and minority ethnic applicants in the treatment group; the intervention had no effect on white applicants. Therefore, the intervention closed the racial gap in the probability of passing the test without lowering the recruitment standard or changing the assessment questions.

Paper 3 considers how to increase the number and diversity of people who are attracted to the police in the first place. The study presents the results of a field experiment that varied job advertisements on a postcard. The results suggest that, contrary to popular wisdom, public service motivation (PSM) messages are ineffective at attracting candidates that would not have applied anyway. Rather, messages that focus on the challenge of being a police officer are twice as effective at attracting new candidates, bringing in 16 out of the 70 new applicants compared to eight who received a PSM message. Non-white individuals disproportionately respond to messages that focus on the challenge and career advancement. Overall, messages that focus on the personal benefits of applying to a job are three times as effective at getting individuals to apply as messages that focus on serving the community, without an observable loss in applicant quality.

Despite growing investments in evidence-based policymaking, a significant gap remains between the production of policy-relevant research and the use of evidence to impact policy. This article argues that overcoming this production-adoption gap requires reconceptualizing evidence adoption as a series of behavioral and organizational hurdles that determine whether evidence is useful, usable, and, ultimately, used. Drawing on a large interdisciplinary literature, we develop a unified framework and highlight key gaps in our understanding of each of these critical stages for bringing evidence to scale. We conclude by outlining a research agenda for advancing our understanding why a production-adoption gap persists and knowledge of “what works” to increase evidence adoption – which, together, are essential for ensuring that the evidence-based policymaking movement meets its promise of improving societal outcomes.

Health worker shortages are straining healthcare systems worldwide, yet there is limited causal evidence on how to recruit healthcare workers across national borders. In collaboration with Danish healthcare providers, we randomly assign 110,000 European nurses to receive one of 10 different digital messages (or no message), encouraging them to apply for a nursing job in Denmark. We selected interventions using an approach we call “community-led intervention design” (CID) that systematically balances academics’, practitioners’, and healthcare workers’ expertise. Contrary to expert predictions and prevailing policy guidance, messages highlighting pay decreased overall engagement by 7.7% (1.7 percentage points), positive responses by 6.3% (1.4 pp), and applications by 19.0% (0.15 pp) compared to baseline. Messages emphasizing social support and job benefits outperformed pay messages, increasing engagement by 8.1% (1.67 pp) and 9.6% (1.97 pp) respectively. Evidence from an additional online experiment (N = 1,799) suggests that highlighting pay may reduce credibility and signal a less desirable job in dimensions such as work-life balance or prosociality. This megastudy has far-reaching implications for international labor market decision-making, healthcare workforce policy, and hypothesis generation in behavioral science.