A Multitrait Scoring Algorithm for Rank Ordered Items

• In Event: 35.034 - Some Sharp Statistics

Abstract

Conventional assessment pairs one test with one key to infer meaning of one construct. Alternatively, situation reaction tests (SRT) can elicit multidimensional rankings from one assessment. This study sought to identify multiple trait-keys for each SRT item, each measuring a different dimension/trait. The trait-keys were identified through finding optimal correlations between the scored SRT dimension/trait with a known criterion. SRTs with 4 alternatives per item were examined in a Monte Carlo simulation under various test conditions: inter-trait correlation, sample size and number of ranking items. Dependent variables included the correlations between scored SRT traits and known criteria and corresponding CI widths. MANOVA results demonstrated that multiple trait-scores for each respondent, one per trait; can be derived simultaneously from one SRT.

Authors

• Hongyan Cui, Western Michigan University

• Brooks Applegate, Western Michigan University

• Warren E. Lacefield, Academic Software, Inc.

• David Hartmann, Western Michigan University