Journals / Ankara Üniversitesi Eğitim Bilimleri Fakültesi Dergisi / 2022 / Cilt: 55 - Sayı: 1

Investigation of Scale Transformation Methods in True Score Equating Based on Item Response Theory

Madde Tepki Kuramına Dayalı Gerçek Puan Eşitlemede Ölçek Dönüştürme Yöntemlerinin İncelenmesi

Pages
1–36
Publication Date
2022-03-31
DOI
—
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

In addition, when the RMSD values of the scale transformation methods are compared, similar results are obtained in cases where 2PLM and 3PLM are used, as the sample size and test length increase, equating errors of other methods except the SL method decrease, and It was observed that the methods had lower RMSD values in cases where the common item rate is 40% and the ability distribution of the groups is similar.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:In this research, it was aimed to compare equating errors of scale transformation methods (mean-mean (MM), mean-sigma (MS), Heabera (HB) and Stocking-Lord (SL)) in true score equating based on item response theory (IRT) under different conditions.
📊
Key Findings & Data:In line with the purpose of the study, 7200 dichotomous data sets which were consistent with two and three- parameter logistic model were generated with 50 replication under the conditions of sample size (500, 1,000, 3,000, 10,000), test length (40, 50, 80), rate of the common item (20%, 30%, 40%), type of model used in parameter estimation (two and three-parameter logistic models (2PLM and 3PLM)), and ability distribution of groups (similar (N(0-1) - N(0-1)), different (N(0-1) - N(0.5,1)) for the obtained performance of methods.
💡
Conclusion & Impact:In addition, when the RMSD values of the scale transformation methods are compared, similar results are obtained in cases where 2PLM and 3PLM are used, as the sample size and test length increase, equating errors of other methods except the SL method decrease, and It was observed that the methods had lower RMSD values in cases where the common item rate is 40% and the ability distribution of the groups is similar.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Haebara · MTK gerçek puan eşitleme ortalama-ortalama · ortalama-standart sapma

Abstract

In this research, it was aimed to compare equating errors of scale transformation methods (mean-mean (MM), mean-sigma (MS), Heabera (HB) and Stocking-Lord (SL)) in true score equating based on item response theory (IRT) under different conditions. In line with the purpose of the study, 7200 dichotomous data sets which were consistent with two and three- parameter logistic model were generated with 50 replication under the conditions of sample size (500, 1,000, 3,000, 10,000), test length (40, 50, 80), rate of the common item (20%, 30%, 40%), type of model used in parameter estimation (two and three-parameter logistic models (2PLM and 3PLM)), and ability distribution of groups (similar (N(0-1) - N(0-1)), different (N(0-1) - N(0.5,1)) for the obtained performance of methods. Common item nonequivalent groups equating design was used. R software was used for data generation and analyses. The results obtained from the study were evaluated by using equating error (RMSD) criterion. As a result of the study, considering all the conditions, it was seen that the RMSD values of the SL method were higher than the other methods, but it was seen that the MM and MS methods produced similar RMSD values. In addition, when the RMSD values of the scale transformation methods are compared, similar results are obtained in cases where 2PLM and 3PLM are used, as the sample size and test length increase, equating errors of other methods except the SL method decrease, and It was observed that the methods had lower RMSD values in cases where the common item rate is 40% and the ability distribution of the groups is similar.

Özet

Bu araştırmada, Madde Tepki Kuramı’na (MTK) dayalı gerçek puan eşitlemede, ölçek dönüştürme yöntemlerinin (ortalama-ortalama (OO), ortalama-standart sapma (OS), Stocking-Lord (SL), Haebara (HB)) farklı koşullar altında eşitleme hatalarının karşılaştırılması amaçlanmıştır. Araştırmanın amacı doğrultusunda, yöntemlerin hatalarını karşılaştırmak için örneklem büyüklüğü (500, 1000, 3000, 10000), test uzunluğu (40, 50, 80), ortak madde oranı (%20-%30-%40), parametre kestirim modeli (iki ve üç parametreli lojistik model (2PLM ve 3PLM)) ve grupların yetenek dağılımı (benzer (N(0-1) - N(0-1)), farklı (N(0-1) - N(0.5,1)) koşulları altında 2PLM ve 3PLM’ye uyumlu iki kategorili 50 yineleme ile 7200 veri seti oluşturulmuştur. Veri toplamı deseni olarak “denk olmayan gruplarda ortak madde/test (NEAT) deseni” kullanılmıştır. Veri üretiminde ve analizinde R yazılımı kullanılmıştır. Araştırmadan elde edilen bulgular, eşitleme hatası (RMSD) ölçütüne göre değerlendirilmiştir. Çalışmanın sonucunda, tüm koşullar göz önünde bulundurulduğunda, SL yönteminin RMSD değerlerinin, diğer yöntemlere göre daha yüksek olduğu görülmekle birlikte, OO ve OS yöntemlerinin birbirine benzer RMSD değerleri ürettiği görülmüştür. Ayrıca, ölçek dönüştürme yöntemlerine ilişkin RMSD değerleri karşılaştırıldığında, 2PLM ve 3PLM’nin kullanıldığı durumlarda benzer sonuçlar elde edildiği, örneklem büyüklüğü ve test uzunluğu arttıkça SL yöntemi dışında diğer yöntemlerin eşitleme hatalarında azalma oluştuğu ve ortak madde oranının %40 ve grupların yetenek dağılımının benzer olduğu durumlarda, yöntemlerin, RMSD değerlerinin daha düşük olduğu gözlenmiştir.

🔖

Cite This Article

ÇOKLUK-BÖKEOGLU, Ö., UÇAR, A., BALTA, E. (2022). Investigation of Scale Transformation Methods in True Score Equating Based on Item Response Theory. Ankara Üniversitesi Eğitim Bilimleri Fakültesi Dergisi, Cilt: 55 - Sayı: 1, 1–36. https://ajindex.com/articles/390064

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (9)

  1. Aksekioğlu, B. (2017). Madde tepki kuramına dayalı test eşitleme yöntemlerinin karşılaştırılması: PISA 2012 Fen testi örneği (Tez Numarası: 454879) [Yüksek lisans tezi, Akdeniz Üniversitesi]. Yükseköğretim Kurulu Ulusal Tez Merkezi. https://tez.yok.gov.tr/UlusalTezMerkezi/
  2. Angoff, W. H. (1971). Scales, norms and equivalent scores. In R. L. Thorndike (Eds.), Educational measurement (2nd ed., pp. 508-600). American Council on Education. Angoff, W. H. (1984). Scales, norms and equivalent scores. Educational Testing Service. Arai, S., and Mayekawa, S. (2011). A comparison of equating methods and linking designs for developing an item pool under item response theory. Behaviormetrica, 38(1), 1-16. https://link.springer.com/article/10.2333/bhmk.38.1
  3. Babcock, B., Albano, A., and Raymond, M. (2012). Nominal weights mean equating: A method for very small samples. Educational and Psychological Measurement, 72(4), 608–628. https://doi.org/10.1177/0013164411428609
  4. Baker, F. B., and Al-Karni, A. (1991). A comparison of two procedures for computing IRT equating coefficients. Journal of Educational Measurement, 28(2), 147- 162. http://www.jstor.org/stable/1434796
  5. Barnard, J. J. (1996). In search of equity in educational measurement: traditional versus modern equating methods. [Paper presentation]. ASEESA’s National Conference at the HSRC Conference Center 1996
  6. Annual Meeting, Pretoria, South Africa. Bastari, B. (2000). Linking multiple-choice and constructed-response items to a common proficiency scale [Doctoral dissertation, Massachusetts Institute of Technology]. https://scholarworks.umass.edu/dissertations_1/5557
  7. Brossman, B. G., and Lee. W-C. (2013). Observed score and true score equating procedures for multidimensional item response theory. Applied Psychological Measurement, 37(6), 460–481. https://doi.org/10.1177/0146621613484083
  8. Budescu, D. (1985). Efficiency of linear equating as a function of the length of the anchor test. Journal of Educational Measurement, 22(1), 13-20. https://www.jstor.org/stable/1434562
  9. Caldwell, L. J. (1984). A comparison of equating error in lineer and rasch model test equating method [Unpublished doctoral dissertation]. Florida State University. Cao, Y. (2008). Mixed-format test equating: Effects of test dimensionality and common-item sets (Publication No. 3341415) [Doctoral dissertation, University of Maryland-Maryland]. ProQuest Dissertations and Theses Global. Chen, H. W. (2001). Calibration of the ITBS test battery to the complete tes