Philip Stark, Professor

Open (1) Look who's talking: Are women and URM faculty candidates interrupted more frequently?

Open. Apprentices needed for the fall semester. Enter your application on the web beginning August 15th. The deadline to apply is Monday, August 27th at 9 AM.

A recent study finds that women and underrepresented minority (URM) candidates for faculty positions are interrupted at higher rates than non-URM male candidates (Blair-Loy et al., 2017. Soc. Sci., 6, 29 doi:10.3390/socsci6010029). This project will quantify interruptions of job talks in the College of Engineering at Berkeley and assess whether candidates' gender and ethnicity matter.

Review video-recorded faculty job talks to transcribe the number, nature, and duration of interruptions. Perform statistical analysis to quantify interruptions and to test whether race and gender matter, using permutation tests. Help draft a paper for publication.

Qualifications: Familiarity with R or Python, LaTeX, and git.

Weekly Hours: 6-8 hrs

Open (2) Conjoint Analysis when the Assumptions Do Not Hold

Open. Apprentices needed for the fall semester. Enter your application on the web beginning August 15th. The deadline to apply is Monday, August 27th at 9 AM.

Conjoint measurement is a method for combining more than one kind of measurement into a single scale. It was introduced by Tukey and Luce in 1964. It became the basis of conjoint analysis, used to analyze consumer choice and predict what consumers will buy. In turn, conjoint analysis has become a tool for estimating damages in civil lawsuits. Conjoint measurement rides on a number of assumptions, including the "cancellation axiom" that are not guaranteed to hold in any particular situation. This project examines what happens if conjoint analysis is applied to choices when preferences do not satisfy the cancellation axiom.

Coding and testing a Bayesian implementation of choice-based conjoint analysis in python. Generating synthetic examples that do not satisfy the cancellation axiom, and running the code on those examples.
Literature research on conjoint analysis.
Drafting a manuscript about the results.

Qualifications: Fluency in Python; strong software engineering skills, including testing and integration; strong written English skills.

Weekly Hours: 6-8 hrs

Related website: http://www.stat.berkeley.edu/~stark