Adjustment of young adults with mental retardation in community settings: comparison of parametric and nonparametric statistical techniques.
The impact of four different statistical techniques on the interpretive process using data obtained from young adults with mental retardation in applied settings was evaluated. Our hypothesis was that no difference exists across levels of mental retardation for four dependent variables, jointly or separately, using (a) parametric multivariate analysis of variance, (b) nonparametric multivariate analysis of variance, (c) multiple nonparametric analyses of variance, and (d) multinomial logistic regression. The pattern of significance remained the same across the three classical and quasi-classical designs. Substantive differences were found using the fourth technique, multinomial logistic regression. The four techniques were compared using interpretational as well as statistical criteria.