Application of Multiple Imputation in Dealing with Missing Data in Agricultural Surveys: The Case of BMP Adoption

Missing-data problems are common in farmer surveys but are often ignored in the literature. Conventional methods to address missing data, such as deletion and mean replacement, assume that data are missing completely at random, which rarely holds. This study compares these approaches to the multiple imputation method, which produces different parameter estimates. The mean replacement method increases the central tendency of data, leading to more significant but smaller coefficients than the other methods. We recommend using both the deletion and multiple imputation methods to deal with missing data; results generated by the mean replacement method may not be as reliable