Predicting rehabilitation outcomes: The twisted pear revisited

Prediction of rehabilitation outcome occurs with few unifying theories or p ractices. The authors propose J. Fisher's (1959) neglected twisted pear mod el as central to understanding perplexing problems such as somatoform disor ders, mild traumatic brain injury, likelihood of therapeutic benefit from r ehabilitation interventions, and assessment of driving competence. Fisher a rgued that many psychological and biological behaviors were predictable in only one segment of the distribution. The 2 x 2 twisted pear matrix demonst rates this nonuniform scatter (heteroscedasticity), limiting acceptable cla ssification accuracy to individuals with poor-impaired predictor scores. Ou tcomes become increasingly variable as scores improve to average or better. Examples from clinical practice support the conclusion that the twisted pe ar model is a useful paradigm for rehabilitation professionals. It's hard to make predictions, especially about the future.