Level and Pattern Relationships: Beyond Profile Analysis

• In Event: Poster Session 2
  In Poster Session: Division D Section 2 Poster Session 1

Abstract

Davison and Davenport (2002) have a two-step regression approach that defines Level and Pattern in terms of Profile Analysis that is predictive of a criterion. We take a closer look at these statistics to show that in addition to them having inherent meaning relative to score profiles and often theoretical meaning relative to a criterion, the statistics address a fundamental statistical concern. We present an example to show how mean differences can confound with structural relationships to yield spurious correlations. Level and Pattern can be used as predictors in a regression analysis to ascertain the variance attributable to mean differences, structural relationships, and/or a combination of both.

Authors

• Ernest C. Davenport, Jr., University of Minnesota

• Kyungin Park, University of Minnesota - Twin Cities

• Youngsoon Kang, University of Minnesota - Twin Cities