DETERMINING RELATIVE IMPORTANCE OF PREDICTORS WITH THE OBSERVATIONAL DESIGN

A recent comparison of typical extreme-groups designs and observationa l designs (G. H. McClelland & C. M. Judd, 1993) showed that extreme-gr oups designs have greater power to detect interactions than do observa tional designs and that extreme-groups designs provide estimates of un standardized parameter values that have smaller standard errors than d o estimates provided by observational designs. In this study, this dis cussion is taken a step further by investigation of the advantages and disadvantages associated with inferences drawn from extreme-groups an d observational designs for the estimation of standardized effects. Ob servational designs, through accurate estimation of predictor variance s, are concluded to be superior for the purposes of standardized param eter estimation. Finally, various ways of adapting extreme-groups desi gns to better justify inferences concerning population distributions a re suggested.