| A previous article discussed some of the causes | | | | Therefore, a case with data missing on one |
| of missing data and some of the consequences | | | | variable will be used only in steps that do not |
| of analyzing only complete cases. This newsletter | | | | involve that variable. The advantage is that the |
| will discuss some other common ways of dealing | | | | sample size for each individual analysis is generally |
| with missing data, with a discussion of their | | | | higher than with complete case analysis, but the |
| advantages and disadvantages. | | | | results are unbiased only if the data are MCAR. It |
| Available case analysis (pairwise deletion) calculates | | | | can also lead to mathematical problems in |
| each step of the analysis separately using the | | | | computing estimates of some parameters, and is |
| cases that have data available for that step. | | | | not recommended. |