ALAN C. ACOCK Oregon State University Working With Missing Values

Less than optimum strategies for missing values can produce biased estimates, distorted statistical power, and invalid conclusions. After reviewing traditional approaches (listwise, pairwise, and mean substitution), selected alternatives are cov-ered including single imputation,multiple imputa-tion, and full information maximum likelihood estimation. The effects of missing values are illus-tratedforalinearmodel,andaseriesofrecommen-dations is provided. When missing values cannot be avoided, multiple imputation and full informa-tion methods offer substantial improvements over traditional approaches. Selected results using SPSS,NORM,Stata (mvis/micombine), andMplus are included as is a table of available software and an appendix with examples of programs for Stata and Mplus. Traditional approaches for working with miss-ing values can lead to biased estimates and may either reduce or exaggerate statistical power. Each of these distortions can lead to invalid conclusions. Missing values are endemic across the social sciences (Juster & Smith, 1998), and family studies is no exception. King, Hopnaker, Joseph, and Scheve (2001) found that about 50 % of the participants in political survey data have missing values, and family research often approximates this level of missing values. Many of the major data sets that are utilized in articles appearing in family journals have serious prob-lems with missing values. This is true even for large public use data sets such as the Nationa