completecases

Complete cases are observations in a dataset for which all variables of interest are observed (i.e., have no missing values). In statistical practice, analyses often proceed with complete-case analysis, also known as listwise deletion, where any record with missing data is excluded from the analysis. The term is commonly encountered in discussions of missing data strategies and in the R function complete.cases, which identifies complete records.

In R, complete.cases returns a logical vector indicating which rows are complete. If applied to a data frame, it tests all columns; you can restrict to a subset of variables by applying complete.cases to a subset. Data can be subset with df[complete.cases(df),] or df[complete.cases(df[c('x','y','z')]),].

Advantages of complete-case analysis include simplicity and use of only observed data, avoiding imputation biases and being compatible with many standard statistical methods. However, there are notable disadvantages. It reduces the effective sample size, which lowers statistical power. It can also bias results if the missingness is related to unobserved data or the outcome, i.e., when data are not missing completely at random (MCAR). If missingness depends on observed data (MAR) or unobserved factors (MNAR), complete-case estimates may be biased.

Alternatives to complete-case analysis include multiple imputation and maximum likelihood methods (such as FIML), which use available data more efficiently under certain missingness assumptions. Pairwise deletion is another approach but can lead to inconsistent covariances and biased inference.

See also: missing data, MCAR, MAR, MNAR.