Taiikugaku kenkyu (Japan Journal of Physical Education, Health and Sport Sciences)
Online ISSN : 1881-7718
Print ISSN : 0484-6710
ISSN-L : 0484-6710
Original investigations
Application of item response theory to a questionnaire: Use of an unfolding model for a food survey
Osamu Aoyagi
Author information
JOURNAL FREE ACCESS

2010 Volume 55 Issue 2 Pages 395-408

Details
Abstract

This study investigated the applicability of item response theory to a questionnaire survey. Parents of 288 kindergarten students were asked about their children's food tastes and eating habits. The Squared Simple Logistic Model, which is an unfolding model of the item response theory, was applied to the food survey data. Food taste parameters and individual taste parameters for food were computed using the maximum likelihood method and Newton-Raphson method. The applicability of item response theory to the questionnaire survey was examined by considering the correlations of food taste parameters and eating habits, the fitness of the model to the data, and comparison of the results with Quantification Theory Type Three. The following results were obtained:
1) Food taste parameters corresponded to the degree to which a child likes the taste of healthy/unhealthy food. Food taste parameters were significantly correlated with bad eating habits, and fitted the model sufficiently.
2) Considering the information function of the scale comprising food taste parameters, this scale was not a screening test type but rather a selective test type or a discriminating test type, and successfully discriminated among three types of eaters: “problematic,” “intermediate,” and “good.”
3) Using Quantification Theory Type Three, which, unlike item response theory, does not assume any latent traits, two sets of food taste category weights for individual data, divided randomly, were computed. The correlation between the two sets of category weights was lower than that of food taste parameters selected using item response theory. This finding indicated the advantage of item response theory over Quantification Theory Type Three from the standpoint of parameter stability.
4) Finally, to enhance the practicability of this scale, a conversion table for typical data with a high likelihood of fitting into a parameter we had already obtained, as well as showing a good fit to individual taste parameters, was devised.

Content from these authors
© 2010 Japan Society of Physical Education, Health and Sport Sciences
Previous article Next article
feedback
Top