Complete-Case Analysis Sample Clauses

Complete-Case Analysis. Standard statistical methods for regression analysis have been widely developed to an- alyze rectangular data sets, where values of all variables are measured for all subjects. Therefore, in the presence of missing data, a natural solution is to discard those subjects with any missing values and conduct the analysis based on the subset of complete cases. The aforementioned method, called complete-case (CC) analysis or list-wise deletion, is a default procedure for handling missing data in many statistical packages such as R and SAS. There is a major advantage to CC analysis: it can be directly used for any kind of statistical analysis (e.g., survival analysis, longitudinal analysis) without any computa- tional methods for dealing with missing data (Xxxxxxx, 2001). CC analysis obtains some attractive statistical properties as well, relying on the missing-data mechanism. If the data are MCAR, CC analysis would provide unbiased estimates since the reduced sample of fully observed subjects is a random sub-sample of the original sample. However, the standard errors from the CC analysis will usually be larger since a sub-sample is utilized, leading to a wider confidence interval. On the other hand, if the data are MAR or MNAR, CC analysis will generally lead to biased results with few exceptions as have been dis- cussed in detail by Xxxxx (1987); Little and Xxxxx (2002). For instance, the model of generalized linear regression using CC when the missingness only depends on independent variable, provides unbiased coefficients estimates. Another case that CC analysis is valid is when we fit a logistic regression using incomplete data and the probability of missing values on any variable depends only on the value of the dependent variable but not on any of the independent variables. In that case, the CC analysis yields consistent estimates of the regression coefficients and their standard errors (Vach, 2012).