Yoav Goldstein - Research

Work in Progress

Heuristic Self-Evaluation in High-Stakes Decisions: Field of Study and Long-Term Earnings (in revision, new draft coming soon)

older version: Learning from Round Test Scores: University Field Choices and Career Outcomes

This paper examines how uncertainty about one’s ability interacts with cognitive heuristics to shape high-stakes educational choices and career paths. Using administrative data, I exploit a regression discontinuity at a round-number threshold on the national university entrance test. Crossing the threshold on the first test substantially increases the likelihood of pursuing high-return STEM degrees, even though it lies below the admission cutoffs for these programs. This behavior is consistent with left-digit bias: students interpret the score’s left digit as a meaningful signal, leading them to improve their scores through retesting, apply to, and ultimately enroll in STEM programs. In subsequent years, they are more likely to work in the tech sector, earn higher wages, and reach the top of the income distribution. Leveraging this behavioral discontinuity, I estimate large causal returns to STEM degrees among students uncertain about applying. The findings suggest that heuristic self-evaluation can lead to persistent disparities in economic opportunity that are unrelated to actual ability.

Changing Minds: How Academic Fields Shape Political Attitudes with Matan Kolerman-Shemer (draft is available upon request)

Education has become a central source of political division, yet causal evidence on how educational experiences shape political ideology remains limited. This paper examines one key dimension—academic field specialization, a defining feature of postsecondary education. We link entry and exit surveys for 310,000 students at 477 U.S. four-year colleges (1990–2019) to estimate the effects of fields of study. Using selection-on-observables with rich controls—including baseline ideology and intended major—we estimate plausibly causal effects of major choice. Humanities and social sciences shift ideology left by approximately 0.09 standard deviations (4 percentage points) compared to natural sciences, while economics and business shift it right by 0.14 standard deviations (6 percentage points). Sensitivity analysis shows that unobserved selection would need to be unrealistically strong to overturn the results. Quasi-experimental evidence from shifts in field availability reinforces the causal interpretation. Fields also affect policy positions, especially on economic issues, and shape behaviors such as voting and protesting. Life goals shift as well: economics and business increase the priority placed on financial prosperity relative to other fields. The evidence points to teaching and academic content, rather than peer influence or earnings expectations, as the main driver of these effects. Given current enrollment patterns, field specialization in postsecondary education contributes to both political polarization and gender gaps in political attitudes.

Gifted Children Programs’ Short and Long-Term Impact: Higher Education, Earnings, and the Knowledge-Economy with Victor Lavy (in revision, new draft coming soon)

older version: NBER w29779

This paper examines the short-run and longer-term effects of studying in gifted classes in high schools. Our results show that GCPs have tiny effects on high school academic achievement but substantially influence university outcomes. This influence is manifested in the choice of field of study, a higher incidence of double majors, and an increased likelihood of pursuing advanced degrees. Interestingly, participation in gifted classes does not affect earnings or employment in knowledge-based sectors, implying that gifted children do well in the labor market regardless of their participation in a special class. Finally, participation in gifted classes does not affect the likelihood of marriage or having children. Still, it positively affects the spouse’s “quality” driven by marriages between gifted students and their classmates. We discuss potential mechanisms by relating our findings to the literature in psychology about gifted children.

Published and Forthcoming Papers

Challenging Encounters and Within-Physician Practice Variability with Gabriel Chodick, Ity Shurtz and Dan Zeltzer

Forthcoming at The Review of Economics and Statistics

We examine how physician decisions are impacted by difficult cases - encounters with newly diagnosed cancer patients. Using detailed administrative data, we compare primary care physicians' decisions in visits that occurred before and after difficult cases and matched comparison cases by the same physicians on other dates. Immediately following a difficult case, physicians increase referrals for common tests, including diagnostic tests unrelated to cancer. The effect lasts only for about an hour, and is not driven by patient selection or schedule disruption. The results highlight difficult encounters as a source for variability in physician practice.

Realization of Low Probability Clinical Risks and Physician Behavior: Evidence from Primary Care Physicians with Gabriel Chodick and Ity Shurtz

Forthcoming at The American Journal of Health Economics

We examine whether primary care physicians alter their clinical decision-making following realizations of low probability risks among their patients -- events of colon cancer diagnoses. Relying on comprehensive administrative visit level data from a large Israeli HMO, we find that physicians increase their patients' use of colonoscopy tests substantially during the first three months following such events and that in the subsequent twelve months the effect dissipates. We find no indication that older -- typically more experienced -- physicians are less responsive to Events, nor do we find that physicians' response varies by their gender. Considering that in our setting, it is unlikely that colon cancer diagnoses convey information in the traditional sense, these results indicate the existence of attention effects. Unexpected events induce a stronger physician response and the increase in the use of colonoscopy tests is more pronounced among patients that are similar to the patient involved in the event, in line with recent work on cognitive mechanisms and memory where surprises and associative memory play a key role in decision making. We also do not find evidence that the quality of colonoscopy tests decreases with the increase in their use, suggesting that in response to events, physicians induce adherence to colonoscopy tests among their at-risk patients rather than merely lowering the cut-off risk level of prescribing colonoscopy tests.