Autores: Jorge Dávalos, Angela Figueroa

Resumen

We evaluate the reliability of Student Evaluations of Teaching (SET) within the context of a developing country, using administrative data at the student level from Universidad del Pacífico in Lima, Peru (2017-2022). We focus on addressing biases that affect SET's validity, particularly the overlooked issue of sample selection bias, which we correct using observational student-level data. Our analysis builds on two different SET surveys implemented on freshmen students during the period of analysis. The statistical model includes fixed effects, instrumental variables, and sample selection modeling akin to Heckman's approach. Our results reveal that not correcting for non-participation may overestimate the correlation between SET scores and student grades by about 50 percent and that the relationship between SET scores and students’ actual learning is statistically insignificant when correcting for all other sources of (endogeneity) biases.

Expositor

Jorge Dávalos

Director de Investigación del Partnership for Economic Policy (PEP) Network y es profesor asociado a tiempo parcial en el departamento académico de Administración de la Universidad del Pacífico. Su investigación se centra en la modelización estadística en diversas areas de las ciencias sociales, incluidas el emprendimiento femenino, la economía laboral, economía agrícola, medio ambiente, y educación superior. Ha sido consultor para agencias internacionales, incluyendo Naciones Unidas-OIT, el Banco Mundial, y otras entidades financieras. Posee un doctorado en econometría por la Universidad de Ginebra, Suiza.