Population Cross-Validity Estimation and Adjustment for Direct Range Restriction: A Monte Carlo Investigation of Procedural Sequences to Achieve Optimal Cross-Validity

The current study employs Monte Carlo analyses to evaluate the effectiveness of various statistical procedures for determining specific values of interest within a population of 1,000,000 cases. Specifically, the proper procedures for addressing the opposing effects of direct range restriction and validity overestimation were assessed through a comparison of multiple correlation coefficients derived using various sequences of procedures in randomly drawn samples. A comparison of the average bias associated with these methods indicated that correction for range restriction prior to the application of a validity overestimation adjustment formula yielded the best estimate of population parameters over a number of conditions. Additionally, similar methods were employed to assess the effectiveness of the standard ΔR2F-test for determining, based on characteristics of the derivation sample, the comparative superiority of either optimally or unit weighted composites in future samples; this procedure was largely ineffective under the conditions employed in the current study.

Name thesis:degree_name

Master of Arts

Discipline thesis:degree_discipline

Department of Psychology

Year

2010

Author dc:creator

• Goins, David Matthew

Contributors dc:contributor

• Dr. Reagan Brown (Director), Dr. Elizabeth Shoenfelt, Dr. Anthony Paquin

• organizational psychology
• educational tests and measurements
• employment tests
• Educational Assessment, Evaluation, and Research
• Industrial and Organizational Psychology

Repository record dc:identifier

https://digitalcommons.wku.edu/theses/165

OAI identifier oai:identifier

oai:digitalcommons.wku.edu:theses-1166