Group differences and individual classification accuracy associated with letter/word oddity problems

The ability of 27 pairs of mildly mentally retarded and normally achieving 6-through 8-year-old students to apply an oddity rule to an experimental set of letter and word oddity problems was assessed. The mildly retarded group evidenced a markedly lower level of performance on all five experimental oddity problems, but they did evidence successful generalization to oddity arrays that visually were very different from those used in training. High levels (>90%) of sensitivity, specificity, positive predictive accuracy, and negative predictive accuracy were observed using only six cards. Only slightly lower (≥ 85%) values were obtained when using a cut-off score. This subset of six cards is, therefore, an excellent candidate for inclusion in a cognitive preschool screening test being developed for the early detection of children with mild mental retardation.