Constructing a Scale of Female Contributions to Agriculture: Methods for Imputing Missing Data

Abstract

Five methods are described for estimating missing data in cross-cultural codes. The reliability of these alternative methods is assessed in the context of constructing a scale of total female contributions to agriculture. Simulations are used to explore the potential effects on scale reliability of larger amounts of missing data in the same data set. The imputed scales show surprisingly high reliability, even when more than half of the original data points are missing.