Alternative Solutions to Nonequivalence between Experimental and Control Groups

This paper discusses some major threats of the nonequivalence between experimental and control groups to draw valid inferences on whether the experimental treatment works.Several alternative solutions,such as the nonequivalent control group designs and the ANCOVA statistical analysis model are suggested.The ANCOVA with a single covariate or with multiple covariates extends the elementary ANOVA by including the single or multiple pretests in the model in the form of a linear regression.Using the pretests in this way provides an adjustment for initial differences between the groups due toselection-maturation.