General conditions for the success of bootstrapping models

摘要: Linear models which fit regression equations to clinical judgments. then use the fitted parts of judgments as "bootstrapped" judgments, have outperformed clinical in many tasks. Empirically, the phenomenon has been pervasive, but general conditions for success bootstrapping have never explicitly linked cross-study data. This link, combined with psychologically plausible evidence about relationships between judgmental variables, shows that will improve slightly under almost any realistic task conditions. result allows one apply bootstrapping blindly cases where criterion information is missing or vague (precisely the are useful), and be confident predictions are being improved. A simple comparison models with equal weighting also made, relative success those two not specified.