Explaining and controlling for the psychometric properties of computer-generated figural matrix items
Research output: Journal contributions › Journal articles › Research › peer-review
Standard
In: Applied Psychological Measurement, Vol. 32, No. 3, 05.2008, p. 195-210.
Research output: Journal contributions › Journal articles › Research › peer-review
Harvard
APA
Vancouver
Bibtex
}
RIS
TY - JOUR
T1 - Explaining and controlling for the psychometric properties of computer-generated figural matrix items
AU - Freund, Phlipp Alexander
AU - Hofer, Stefan
AU - Holling, Heinz
PY - 2008/5
Y1 - 2008/5
N2 - Figural matrix items are a popular task type for assessing general intelligence (Spearman's g). Items of this kind can be constructed rationally, allowing the implementation of computerized generation algorithms. In this study, the influence of different task parameters on the degree of difficulty in matrix items was investigated. A sample of N = 169 participants (all age groups) completed a set of 25 automatically generated 4 × 4 matrix items. Data collection was conducted through the World Wide Web. All items showed a good fit with the Rasch model, and item difficulty could be explained reasonably well through the implemented task parameters. The research indicated that matrix items can easily be generated using well-defined computerized algorithms. Their composite character explains item difficulty to a satisfactory degree and enables researchers to construct items with anticipated psychometric properties and Rasch model conformity. Practical advantages of these findings are pointed out.
AB - Figural matrix items are a popular task type for assessing general intelligence (Spearman's g). Items of this kind can be constructed rationally, allowing the implementation of computerized generation algorithms. In this study, the influence of different task parameters on the degree of difficulty in matrix items was investigated. A sample of N = 169 participants (all age groups) completed a set of 25 automatically generated 4 × 4 matrix items. Data collection was conducted through the World Wide Web. All items showed a good fit with the Rasch model, and item difficulty could be explained reasonably well through the implemented task parameters. The research indicated that matrix items can easily be generated using well-defined computerized algorithms. Their composite character explains item difficulty to a satisfactory degree and enables researchers to construct items with anticipated psychometric properties and Rasch model conformity. Practical advantages of these findings are pointed out.
KW - Psychology
KW - automatic item generation
KW - figural matrix items
KW - item task parameters
KW - LLTM
KW - Rasch model
UR - http://www.scopus.com/inward/record.url?scp=41549114185&partnerID=8YFLogxK
U2 - 10.1177/0146621607306972
DO - 10.1177/0146621607306972
M3 - Journal articles
VL - 32
SP - 195
EP - 210
JO - Applied Psychological Measurement
JF - Applied Psychological Measurement
SN - 0146-6216
IS - 3
ER -