From: Studying missingness in spinal cord injury data: challenges and impact of data imputation
Pattern | Definition | Example |
---|---|---|
Missingness completely at random | Values are missing independently of their true unknown value and independently of other variables | A LEMS value is missing for participant A with no underlying reason |
Missingness at random | Values are missing independently of their true unknown value but the pattern depends on other variables | A LEMS value is missing for participant A because they had a cast at the time of assessment, i.e. knowing the cast status gives information on whether LEMS value will be missing or not |
Missingness not at random | Having a value missing depends on the true unknown value | A LEMS value is missing for participant A because their injury was so severe that they could not come for the assessment, i.e. the underlying true LEMS gives information on whether LEMS value will be missing or not |