Evaluation of Imputation of Covariates in an Impact Analysis With Regression Adjustment

In an impact analysis using random assignment, researchers often deal with missing values in both the covariates and the outcome variables of regression models. Clearly rigorous methods are needed to impute missing values in the outcome variables to minimize the potential bias in impact assessments. When imputation is applied to covariates of the regression analyses, the effect of imputation is less clear on impact analyses. This paper assesses this effect, using a random assignment evaluation of the Growing America Through Entrepreneurship (GATE) program. Two outcome variables used in the original evaluation are modeled against a set of 10 covariates, a treatment indicator, and variables associated with the site of the evaluation. Impacts are assessed with different types of missingness in the covariates with values imputed using mean imputation and sequential hot deck.