Peer Effects with Sample Selection

This paper studies peer effects in the presence of sample selection. Nonrandom selection in group participation is fundamental in social interactions studies. With peer effects and simultaneity, individual selection bias leads to a group-level selection bias. We propose to include correction terms for both selection biases and use them to solve the reflection problem. We apply our method to study peer effects in an online training program in China. We document significant peer effects and selection bias in the duration of lecture attendance among the trainees. Ignoring sample selection would yield misleading results of the peer effects