• Fitting a Mixture Rasch Model...

Treffer: Fitting a Mixture Rasch Model to Visual Sequential Processing Memory Sub-Dimension of ASIS: The Role of Covariates

Title:
Fitting a Mixture Rasch Model to Visual Sequential Processing Memory Sub-Dimension of ASIS: The Role of Covariates
Language:
English
Authors:
Sahin, Murat Dogan
Source:
International Electronic Journal of Elementary Education. Dec 2020 13(2):277-285.
Availability:
International Electronic Journal of Elementary Education. T&K Akademic Rosendalsvein 45, Oslo 1166, Norway. e-mail: iejee@iejee.com; Web site: https://www.iejee.com/index.php/IEJEE/index
Peer Reviewed:
Y
Page Count:
9
Publication Date:
2020
Document Type:
Fachzeitschrift Journal Articles<br />Reports - Research
Descriptors:
Item Response Theory, Memory, Intelligence Tests, Children, Age Differences, Gender Differences, Goodness of Fit, Test Items, Difficulty Level, Classification, Foreign Countries
Geographic Terms:
Turkey
ISSN:
1307-9298
Entry Date:
2021
Accession Number:
EJ1285323
Database:
ERIC

Weitere Informationen

Advanced Item Response Theory (IRT) practices serve well in understanding the nature of latent variables which have been subject to research in various disciplines. In the current study, 7-12 aged 2536 children's responses to 20- item Visual Sequential Processing Memory (VSPM) sub-test of Anadolu-Sak Intelligence Scale (ASIS) were analyzed with Mixture Rasch Model (MRM). In the first phase of the study, concomitant (covariate) variables were not used. In the second phase, age and gender were added to the model, and then the two models were compared in terms of fit indices, the number of latent classes and the distribution of item difficulties in the latent classes. The results of the study suggested that there were three latent classes in both models; however, the latter model had a better fit compared to the former model. In addition, the latent classes in both models had similar characteristics, and the distributions of item difficulties in the latent classes were also quite similar in both models while they had some differences in some aspects. The sizes of identical latent classes in both models varied between 15% and 30%. The results of the current study are expected to provide a deeper insight to researchers studying measurement theory and/or intelligence measurement.

As Provided