The importance of classification tables in binary logistic regression analysis has not been fully recognized. This may be due to an over reliance on statistical software or lack of awareness of the value that computation of the proportional by chance accuracy criteria (PCC) and proportional reduction in error (PRE) statistic can add to binary logistic regression models. Case illustrations are used in this paper to demonstrate the usefulness of these computations. An overview of logistic regression is proffered along with a discussion of the function of case classifications and strategies in application of the PCC and PRE. It offers guidance for others interested in understanding how classification tables can be maximized to assess the predictive effectiveness and utility of binary logistic regression models.
White, Jeffry L.
"Logistic regression Model Effectiveness: Proportional Chance Criteria and Proportional Reduction in Error,"
Journal of Contemporary Research in Education: Vol. 2
, Article 3.
Available at: https://egrove.olemiss.edu/jcre/vol2/iss1/3