Abstract: | The literature on patient demographic and clinical characteristics as predictors of length of psychiatric hospital stay suggests that investigators have been unsuccessful in producing a prediction model using this information to account for any substantial criterion variance. The present study maximized predictions of length of hospital stay by using a more powerful statistical procedure and an expanded number of predictor variables. Data were obtained on 877 patients who were discharged from psychiatric hospitals in 1980. Results show that despite these improvements, the resulting proportion of variance was moderate at best. It is concluded that increases in similar information are unlikely to improve predictions. The use of untapped variables (e.g., management philosophies and politically based information) is discussed. (14 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved) |