Imputing Missing Data: A Comparison of Methods for Social Work Researchers

Abstract

Choosing the most appropriate method to handle missing data during analyses is one of the most challenging decisions confronting researchers. Often, missing values are just ignored rather than replaced with a reliable imputation method. Six methods of data imputation were used to replace missing data from two data sets of varying sizes; this article examines the results. Each imputation method is defined, and the pros and cons of its use in social science research are identified. The authors discuss comparisons of descriptive measures and multivariate analyses with the imputed variables and the results of a timed study to determine how long it took to use each imputation method on first and subsequent use. Implications for social work research are suggested.

A generalized item response tree model for psychological assessments

“…The APA Task Force on Statistical conference (Wilkinson & Statistical Inference, 1999) recommended that researchers report patterns of missing data and use statistical techniques to address the missing data problems. However, adequate reporting and handling of missing data is often ignored in practice (Peng et al, 2006;Saunders et al, 2006;Schlomer et al, 2010).…”

Predicting Savings From Adolescence to Young Adulthood: A Propensity Score Approach

“…To accurately complete missing data, multiple imputations use information from the observed variables as well as the missing data. The Markov Chain Monte Carlo method was performed to create five completed, or imputed, datasets with no missing data (Saunders et al, 2006;Schafer & Graham, 2002). The results were then pooled across the five imputed datasets to reduce bias in the estimations of parametric statistics (Saunders et al, 2006).…”

“…The Markov Chain Monte Carlo method was performed to create five completed, or imputed, datasets with no missing data (Saunders et al, 2006;Schafer & Graham, 2002). The results were then pooled across the five imputed datasets to reduce bias in the estimations of parametric statistics (Saunders et al, 2006). The descriptive results, bivariate test statistics, beta coefficients, standard errors (SE), odds ratios (OR), and the R 2 values reported in the results were obtained from the averaged, pooled results across the five imputed datasets (Rubin, 1987;Saunders et al, 2006).…”

The effect of misssing data handling methods on goodness of fit indices in confirmatory factor analysis

“…For example, it could be difficult for an elderly person to finish the questionnaire by reason of age (a measured Köse 209 variable) but not because of his or her level of depression (the outcome being measured) (Saunders et al, 2006).…”

“…Academic journals expect from the authors to take appropriate steps to properly handle missing data, but most articles do not give necessary attention to this isue (Sterner, 2011). "Small" percentages of missing values are less problematic but there is no common definition of "small amount of missing data" in the literature (Saunders et al, 2006).…”