Best Practices for Addressing Missing Data through Multiple Imputation

Abstract

A common challenge in developmental research is the amount of incomplete and missing data that occurs from respondents failing to complete tasks or questionnaires, as well as from disengaging from the study (i.e., attrition). This missingness can lead to biases in parameter estimates and, hence, in the interpretation of findings. These biases can be addressed through statistical techniques that adjust for missing data, such as multiple imputation. Although this technique is highly effective, it has not been widely adopted by developmental scientists given barriers such as lack of training or misconceptions about imputation methods and instead utilizing default methods within software like listwise deletion. This manuscript is intended to provide practical guidelines for developmental researchers to follow when examining their data for missingness, making decisions about how to handle that missingness, and reporting the extent of missing data biases and specific multiple imputation procedures in publications.

Stigma for seeking psychological help for military sexual trauma is associated with more frequent suicidal ideation among women service members and veterans.

“…Discharge status (yes/no) was modeled as a covariate predictor of all endogenous variables and allowed to covary with the MST arbitrary codes; discharge status was the only demographic characteristic with a significant bivariate association with suicidal ideation. We used full information maximum likelihood to estimate the model using all available data, which is preferred to listwise deletion (Woods et al, 2021). Indirect effects were estimated in Mplus using the Model Indirect command, and 95 percentile confidence intervals using 10,000 bootstrap samples were generated.…”

The contributions of reading and phonological awareness for spelling in grade three isiXhosa learners

“…Participants who had at least one missing data point had lower reading and phonological awareness scores (small effect size) and lower spelling scores (medium effect size) at the start of grade 3 than participants with complete data, i.e., the data were Missing at Random. Since listwise deletion would drastically reduce the sample size and bias the estimates (Woods et al 2021), we imputed the missing data using predictive mean matching in mice (multiple imputation by chained equations) (Van Buuren & Groothuis-Oudshoorn 2011). We imputed 10 datasets with 10 iterations each and saved the final model's final iteration as the dataset we worked with.…”

The contributions of reading and phonological awareness for spelling in grade three isiXhosa learners

“…the data was Missing at Random. Since listwise deletion would drastically reduce the sample size and bias the estimates (Woods et al 2021), we imputed the missing data using predictive mean matching in mice (multiple imputation by chained equations) (van Buuren & Groothuis-Oudshoorn 2011). We imputed 10 datasets with 10 iterations each and saved the final model's final iteration as the dataset we worked with.…”