CSAP Predoc: Meta-Reanalysis of Experimental Political Science Research
POSITION FILLED
This position is part of The Center for the Study of American Politics (CSAP) Pre-Doctoral Fellows Program, which is affiliated with the Tobin Center/Economics Pre-Doctoral Fellows Program. The CSAP predoc program provides a high-quality education and training in quantitative political science research for individuals who are considering pursuing a Ph.D. in political science or a closely related discipline. CSAP predocs are invited to participate in all Tobin Center predoc activities. Information about the CSAP predoc program is available at this link.
FACULTY SUPERVISOR:
POSITION DESCRIPTION:
The goal of this project is to produce generalizable answers to social scientific questions through ``meta-reanalysis,'' a design in which original study datasets are re-analyzed using standardized tools and the resulting re-analyses are formally synthesized through meta-analysis. Building on the Coppock's previous meta-analytic work on persuasion, candidate choice, and discrimination, this project seeks to establish the external validity of research findings in select literatures, including for a start, the experimental record on the effects of group cues and perspective taking.
The key contribution of this project is the approach to the generalizability problem, the concern that the results of any particular study might not travel to other contexts and times. The lab's design- based approach is grounded in the principles of commensurability, transparency, and collaboration. Within a research literature, we start by seeking the set of designs that target the same estimand with similar sampling, assignment, and measurement procedures, i.e., the commensurable research designs. We insist on transparency by only including studies whose replication data we can obtain or reconstruct, allowing us to apply the same theoretically-appropriate reanalysis procedures to all studies. Obtaining generalizable answers from any corpus of data requires deep domain knowledge in order to appropriately model heterogeneity within and across research sites, which is why our approach centers scholarly collaboration with substantive experts. This project will allow us to estimate whether a research finding holds generally or if it varies from place to place -- and if it varies, what might explain why.
REQUISITE SKILLS AND QUALIFICATIONS:
The main tasks will fall into three categories:
1. Data cleaning and preparation. All-data related tasks will be done in R, in the "tidyverse" style, so familiarity with R and tidyverse coding is preferred.
2. Design-aware literature review. We only include primary studies that adhere to particular designs, so exposure to and familiarity with experimental designs in political science is preferred.
3. Data collection via communication with primary study authors. We need to ask primary study authors to share their data with us, which requires a polite and prompt emailing style.