Single imputation methods are not recommended, as they depend on assumptions that are often unrealistic and frequently result in an underestimation of the variability and a spuriously low P value. Instead, apply multiple imputation if the following conditions apply:  When should multiple imputation be used to handle missing data. ## MCAR: Missing Completely At Random data Data is said to be MCAR if the missing data value is unrelated to any observed or missing data. In this case, the tendency for the data point to be missing is completely random. In other words, there is no systematic reason that makes some data more likely to be missing than others. For example, missing data as a result of a laboratory technician dropping a blood sample, or data missing from surveys lost in the mail likely occur randomly. This assumption can be tested by separating the missing and the complete cases and examine the group characteristics. If characteristics are not similar for both groups, the MCAR assumption does not hold. Unfortunately, most missing data are not MCAR. ## MNAR: Missing Not At Random data In data missing not at random (MNAR), missing values _do_ depend on unobserved values. Examples include people who are overweight and, as a result, less likely to report their weight, or patients with more comorbidities who tend to drop out of a study more readily than those with less comorbidities. ## MAR: Missing At Random data Data is considered MAR if the reason for the missing data is unrelated to the missing values but are related to some of the observed data. The tendency for the data point to be missing under this assumption is systematically associated with the observed data, but not the missing data. For example, if men are more likely to correctly report weight than women, the weight variable is considered MAR. ## References 1. [When and how should multiple imputation be used for handling missing data in randomised clinical trials – a practical guide with flowcharts](https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-017-0442-1) 2. [The Sin of Missing Data: Is All Forgiven by Way of Imputation?](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6293424/)