Analysis of covariance (ANCOVA)
Scientists want to account for factors other than the independent variable which may influence their dependent variable. There are situations, however, in which the investigator may not be able to control for certain factors. One option may be to have each subject complete a caffeine consumption questionnaire and then statistically adjust for the effects of caffeine consumption in their analysis. As a result, the investigator has to use daily caffeine consumption as a covariate thus reducing the error variability in their study. This type of statistical procedure is called Analysis of Covariance (ANCOVA). ANCOVA works by "holding constant" the covariate across group means. That is, for a dependent variable, each group mean has the variance associated with the covariate accounted for in the model separately, so that all group means adjusted for the influence of the covariate.