The dearth of controlled studies is also relevant to our meta-analytic procedure in that in order to obtain a sufficient number of observations for meaningful analyses it was necessary to compute effect sizes that merely standardize the pre-to posttest improvement on a group-by-group basis.