Prediction Statistics for Psychological Assessment

This workshop provides an overview of how to use evidence to estimate the likelihood of a future outcome for a specific individual, such as criminal recidivism, divorce, or college graduation. The logic of case-specific likelihoods are discussed, along with a selection of statistics that can inform the estimate of these likelihoods. Prediction statistics can be grouped into those that measure discrimination (how different are individuals with the outcome compared to those without the outcome) or calibration (how similar are the observed rates to those estimated by the prediction tool). Examples of each are discussed.

• Discuss the meaning of commonly used prediction statistics in psychology.

R. Karl Hanson, PhD, CPsych