Combining Process Information and Item Response Modeling to Estimate Problem‐Solving Ability

The action sequences of respondents in problem‐solving tasks reflect rich and detailed information about their performance, including differences in problem‐solving ability, even if item scores are equal. It is therefore not sufficient to infer individual problem‐solving skills based solely on item...

Full description

Saved in:
Bibliographic Details
Published in:Educational measurement, issues and practice issues and practice, 2022-06, Vol.41 (2), p.36-54
Main Authors: Xiao, Yue, Veldkamp, Bernard, Liu, Hongyun
Format: Article
Language:English
Subjects:
ability estimation
Adults
Bayesian analysis
Bayesian framework
Correlation
Data Analysis
Data processing
Educational tests & measurements
International Assessment
Item Analysis
Item Response Theory
Measurement
Problem Solving
process data analysis
Scores
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The action sequences of respondents in problem‐solving tasks reflect rich and detailed information about their performance, including differences in problem‐solving ability, even if item scores are equal. It is therefore not sufficient to infer individual problem‐solving skills based solely on item scores. This study is a preliminary attempt to incorporate process data analysis into the measurement of problem‐solving ability. The entire procedure consists of using information from process data as prior information for the estimation of problem‐solving proficiency in an item response model. The purpose of this study is twofold: (1) to investigate the impact of adding process information on the estimation of latent ability; (2) to examine the extent to which the ability estimates obtained from the combination model can reflect the information of the problem‐solving process. Seven problem‐solving items from the Programme for International Assessment of Adult Competencies were used. Results indicate that the inclusion of process priors enhances the correlation between proficiency estimates and process information related to the problem‐solving strategies adopted by respondents, as well as to their solution efficiency. The inclusion of process priors further reveals differences in the problem‐solving performance of respondents exhibiting the same score pattern and increases precision of latent ability estimation.
ISSN:0731-1745
1745-3992
DOI:10.1111/emip.12474