The Model-Based K-Index for Answer Copying Detection

• In Event: 29.082 - Roundtable Session 9
  In Roundtable Session: 29.082-15 - Methodologies for Helping to Ensure Accurate Screening and Diagnostic Decisions

Abstract

The K-index (Holland, 1996) is a well-investigated and popular statistical method for answer copying detection. However, it lacks a simple and objective process of estimating the parameter, and the probability distribution (i.e., binomial distribution) employed is unreasonable and imprecise. To improve the utilization of the K-index, this study proposed the model-based K-index (KM-index) and compared the performance of the KM-index with the K-index through a simulation study. Based on the results, the KM-index outperforms the K-index in detecting answer copying in terms of well-controlled type I error rates and higher power rates, especially when source’s ability is high. The proposal of KM-index can provide test practitioners a useful tool in answer copying detection with higher efficacy and convenience.

Authors

• Hsiu-Yi Chao, National Chung Cheng University

• Jyun-Hong Chen, National Sun Yat-sen University

• Shu-Ying Chen, National Chung Cheng University