Online Program

How to Apply Missing Data Techniques in Practice (302988)

Keywords: missing data techniques

All statisticians employ techniques to handle missing data—either actively or passively. Most statistical packages default to a listwise deletion approach for cases with missing data; however, this method is only appropriate for specific data structures that are rarely found in practice. Often, data are missing for a variety of reasons (e.g., attrition, nonresponse, participant refusal, survey skip patterns, purposeful missing data designs, etc.) that require additional, but often simple, steps to reduce bias caused by missing data. This poster presentation will 1) outline assumptions about missing data structures (missing completely at random, missing at random, missing not at random); 2) provide an overview of the most common methods used in practice (deletion methods, single imputation methods, multiple imputation methods); 3) use a data set to demonstrate differences in approaches, highlighting the strengths and weaknesses of each; and 4) offer recommendations for implementing missing data techniques.