A Comparison of Kernel Equating Methods Based on Neat Design
Abstract
Problem Statement: Equating can be
defined as a statistical process that allows modifying the differences between
test forms with similar content and difficulty so that the scores obtained from
these forms can be used interchangeably. In the literature, there are many
equating methods, one of which is Kernel equating. Trends in International
Mathematics and Science Study (TIMSS) aims to find out the knowledge and skills
gained by the fourth and eighth-grade students in the fields of mathematics and
science. TIMSS have different test forms, and these forms are equated through
common items.
Purpose
of the Study: This research aimed to compare the equated
score results of the Kernel equating (KE) methods, which are chained, and post-stratification
equipercentile and linear equating methods under NEAT design.
Methodology: TIMMS Science data were
used in this study. The study sample consisted of 865 eighth-grade examinees
who were given the Booklets 1 and 14 during the TIMSS application in Turkey.
There were 39 items in Booklet 1, and 38 items in Booklet 14. Firstly,
descriptive statistics were calculated and then the two Booklets were equated
according to NEAT design based on Kernel chained, Kernel post-stratification
equipercentile, and linear equating methods. Secondly, the equating methods
were evaluated according to some criteria such as DTM, PRE, SEE, SEED, and
RMSD.
Findings and Results: It was seen that results based on equipercentile and linear equating
methods were consistent with each other, except for a high range of the score
scale. PRE values demonstrated that KE equipercentile equating methods better
matched with the discrete target distribution Y, and distribution of SEED
revealed that KE equipercentile and linear methods were not significantly
different from each other according to DTM.
Keywords
References
- Andersson, B., Bränberg, K., & Wiberg, M. (2013). Performing the kernel method of test equating with the package kequate. Journal of Statistical Software, 55(6), 1-25.
- Akin-Arikan, Ç. (2017). Kernel Eşitleme ve Madde Tepki Kuramına Dayalı Eşitleme Yöntemlerinin Karşılaştırılması [Comparison of kernel equating and item response theory equating methods] (Unpublished doctoral dissertation). Hacettepe University, Ankara, Turkey.
- Akin-Arikan, Ç. & Gelbal, S. (2018). A Comparison of Traditional and Kernel Equating Methods. International Journal of Assessment Tools in Education, 5(3), 417-427. doi: 10.21449/ijate.409826
- Choi, S. I. (2009). A Comparison of Kernel Equating and Traditional Equipercentile Equating Methods and the Parametric Bootstrap Methods for Estimating Standard Errors in Equipercentile Equating (Unpublished doctoral dissertation). University of Illinois at Urbana-Champaign, United States.
- Godfrey, K. E. (2007). A comparison of Kernel equating and IRT true score equating methods (Unpublished doctoral dissertation). The University of North Carolina, United States.
- Grant, M. C., Zhang, L., & Damiano, M. (2009). An Evaluation of Kernel Equating: Parallel Equating with Classical Methods in the SAT Subject Tests [TM] Program. (ETS RR-09-06). Princeton, NJ: Educational Testing Service.
- Holland, P. W., & Dorans, N. J. (2006). Linking and equating. In R. L. Brennan (Ed.), Educational measurement (pp. 187–220). Westport, CT: Praeger Publishers.
- Holland, P., von Davier, A., Sinharay, S., & Han, N. (2006). Testing the untestable assumptions of the chain and post-stratification equating methods for the NEAT design (ETS RR-06-17). Princeton, NJ: Educational Testing Service.
Details
Primary Language
English
Subjects
-
Journal Section
Research Article
Authors
Publication Date
July 31, 2019
Submission Date
July 31, 2019
Acceptance Date
-
Published in Issue
Year 2019 Volume: 19 Number: 82