APA Dictionary of Psychology

an approach to assigning participants to treatment conditions in which meaningful discrete strata within the sample (e.g., gender, experience) are used to identify homogeneous subsamples; individuals from each subsample or “block” are then assigned randomly to the different conditions. In this way, participants are initially matched on a “blocking variable” that the researcher wishes to control. The acknowledgment of heterogeneity within the sample enables the researcher to reduce within-group variance and to use that information when evaluating treatment effects. The variable on which participants are stratified is assessed prior to the study.