Ability Peer Eﬀects in Small Study Teams (link)
Job Market Paper
While team homework assignments are a widely used teaching tool, evidence on the relationship between a student’s knowledge obtained from teamwork and her team’s ability composition remains scarce. My identiﬁcation strategy relies on within-student variation in achievement across two similar courses, of which only one has team assignments. This variation allows me to establish a causal eﬀect of a team’s ability composition on individual achievement. I establish this link in a setting where students can self-select their peers potentially based on social ties, which have been found to often profoundly aﬀect peer interactions. Based on their performance in the previous year, I classify students to be either regular or very high ability. I ﬁnd that the share of very high ability peers has a statistically signiﬁcant and sizable negative eﬀect on regular students, whereas very high ability students are seemingly (but not statistically signiﬁcantly) aﬀected positively. The result suggests that encouraging students to form homogeneous ability teams might increase their individual performance.
peer achievement spillovers, social networks, post-secondary education, homework assignments, free-riding
I21, I23, J24, L23
Renata Rabovič & Pavel Čížek
Revise & Resubmit at Journal of Econometrics
To analyze data obtained by non-random sampling in the presence of cross-sectional dependence, estimation of a sample selection model with a spatial lag of a latent dependent variable or a spatial error in both the selection and outcome equations is considered. Since there is no estimation framework for the spatial lag model and the existing estimators for the spatial error model are either computationally demanding or have poor small sample properties, we suggest to estimate these models by the partial maximum likelihood estimator, following Wang, et al. (2013)'s framework for a spatial error probit model. We show that the estimator is consistent and asymptotically normally distributed. To facilitate easy and precise estimation of the variance matrix without requiring the spatial stationarity of errors, we propose the parametric bootstrap method. Monte Carlo simulations demonstrate the advantages of the estimators.
asymptotic distribution, maximum likelihood, near epoch dependence, sample selection model
C13, C31, C34
Work in Progress