J 2012

Analysis of response categories preference in spatial reasoning test using IRT Nominal Categories Model

KVĚTON, Petr, Martin JELÍNEK, Dalibor VOBOŘIL and Helena KLIMUSOVÁ

Basic information

Original name

Analysis of response categories preference in spatial reasoning test using IRT Nominal Categories Model

Edition

Ceskoslovenska Psychologie, PRAGUE 1, Academia, 2012, 0009-062X

Other information

Language

Czech

Type of outcome

Článek v odborném periodiku

Confidentiality degree

není předmětem státního či obchodního tajemství

Impact factor

Impact factor: 0.244

UT WoS

000301995000003

Keywords in English

nominal categories model; item response theory; distractor; test

Tags

Reviewed
Změněno: 17/7/2017 22:04, PhDr. Dalibor Vobořil, Ph.D.

Abstract

V originále

Analysis of response categories preference in spatial reasoning test using IRT Nominal Categories Model P. Kveton, M. Jelinek, D. Voboril, H. Klimusova Objectives. The presented study analyzes the preference of response categories in spatial reasoning test. The goal is to enhance comprehension of test behavior of respondents that determines the choice of certain response category in multiple choice items. Subjects and setting. The sample consisted of 2001 subjects (52 % females), who attended entrance exams to Masaryk university, i.e. the Learning Potential Test. Each of the items in the test consisted of one correct answer and four distractors. Five items of the Spatial reasoning subtest were analyzed. Statistical analysis. Using IRT Nominal Categories Model, for each response category conditional probabilities of their preference based on ability level estimate were computed. Results. On the basis of acquired information the authors suggested hypotheses about the roles of distractors in items. These hypotheses were further discussed in conjunction with information from test items reviewers and test-takers, who had the possibility to comment test items through official internet discussion forum. Study limitations. Suggested hypotheses about test behavior need to be further investigated.