Misclassification in linear-in-means models: Theory and application to peer effects estimation
研究了线性均值模型中因变量和自变量的误分类如何导致同伴效应估计偏误，并给出一个基于估计值比值的简单方法帮助研究者判断偏误方向。
This paper investigates how misclassification affects the estimation of peer effects in linear-in-means (LIM) models. We formally show that the peer effect estimate is biased by an ”own effect” due to error in the group average and a smearing effect due to misclassification in the individual regressor. The direction of the total bias depends on how misclassification is distributed across groups. We develop a simple heuristic based on the ratio of the estimated peer effect to the estimated individual effect to help researchers identify the likely direction of the total bias. We verify the theoretical predictions in the setting of peer effects of students with learning disabilities on other students’ performance.