HABILIDAD SINTÉTICA EN MODELOS MULTIDIMENSIONALES DE TEORÍA DE RESPUESTA AL ÍTEM

SYNTHESIZING THE ABILITY IN MULTIDIMENSIONAL ITEM RESPONSE THEORY MODELS

Un problema central asociado con los Modelos Multidimensionales de Teoría de Respuesta al Item (TRIM) es la imposibilidad de ordenar a los examinados. En este artículo, se obtienen dos índices sintéticos unidimensionales que son combinaciones lineales óptimas del vector de habilidades. Estos índices sintéticos son semejantes a la composición de referencia comúnmente usada en los modelos TRIM, pero son más fáciles de calcular. Los índices sintéticos se comparan con el parámetro de habilidad obtenido cuando un conjunto de datos multidimensionales es ajustado con un modelo TRI unidimensional.

A central problem associated with Multidimensional Item Response Theory (MIRT) Models is the impossibility of ordering the examinees. In this paper, we obtain two unidimensional synthetic indices that are optimal linear combinations of the ability vector. These synthetic indices are similar to the reference composite commonly used in MIRT models, but they are easier to calculate and interpret. The synthetic indices are compared with the unidimensional ability obtained when a multidimensional data is fitted with an unidimensional IRT (UIRT) model.

Synthesizing the Ability in Multidimensional Item Response Theory Models Habilidad sintética en modelos multidimensionales de teoría de respuesta al ítem ÁLVARO MAURICIO MONTENEGRO DÍAZ1, EDILBERTO CEPEDA2

1Universidad Nacional de Colombia, Facultad de Ciencias, Departamento de Estadística, Bogotá, Colombia. Assistant professor. Email:ammontenegrod@unal.edu.co 
2Universidad Nacional de Colombia, Facultad de Ciencias, Departamento de Estadística, Bogotá, Colombia. Associate professor. Email:ecepedac@unal.edu.co 

---

Abstract

A central problem associated with Multidimensional Item Response Theory (MIRT) Models is the impossibility of ordering the examinees. In this paper, we obtain two unidimensional synthetic indices that are optimal linear combinations of the ability vector. These synthetic indices are similar to the reference composite commonly used in MIRT models, but they are easier to calculate and interpret. The synthetic indices are compared with the unidimensional ability obtained when a multidimensional data is fitted with an unidimensional IRT (UIRT) model.

Key words: Binary response, Item response theory, Index, Multidimensional data, Synthetic estimator, Latent trait.

---

Resumen

Un problema central asociado con los Modelos Multidimensionales de Teoría de Respuesta al Item (TRIM) es la imposibilidad de ordenar a los examinados. En este artículo, se obtienen dos índices sintéticos unidimensionales que son combinaciones lineales óptimas del vector de habilidades. Estos índices sintéticos son semejantes a la composición de referencia comúnmente usada en los modelos TRIM, pero son más fáciles de calcular. Los índices sintéticos se comparan con el parámetro de habilidad obtenido cuando un conjunto de datos multidimensionales es ajustado con un modelo TRI unidimensional.

Palabras clave: respuesta binaria, teoría de respuesta al ítem, índice, datos multidimensionales, estimador sintético, trazo latente.

---

Texto completo disponible en PDF

---

References

1. Ackerman, T. (1989), 'Unidimensional IRT Calibration of Compensatory and Noncompensatory Multidimensional Items', Applied Psychological Measurement 13, 113-127.

2. Aguilera, A. & Pérez-Aguila, R. (2004), General n-Dimensional Rotations, 'WSCG SHORT Communications papers proceedings', Union Agency - Science Press, Czech Republic, , .

3. Ansley, T. & Forsyth, R. (1985), 'An Examination of the Charateristics of Unidimensional IRT Parameter Estimates Derived from Two Dimensional Data', Applied Psychological Measurement 9, 27-48.

4. Antal, T. (2007), 'On multidimensional item response theory: a coordinate free approach', Electronic Journal of Statistics 1, 290-306.

5. Baker, F. B. & Seok-Ho, K. (2004), Item Response Theory, 2 edn, Marcel Decker Inc..

6. Bock, R. D. (1972), 'Estimating Item Parameters and Latent Ability when Responses are Scored in Two or more Nominal Categories', Psychometrika 37, 29-51.

7. Bock, R. D. & Aitkin, M. (1981), 'Marginal Maximum Likelihood Estimation of Item Parameters: Application of an EM Algorithm', Psychometrika 46, 443-459.

8. Bock, R. D. & Jones, L. V. (1968), The Measurement and Prediction of the Judge and Choice, San Francisco: Holden-Day.

9. Bégin, A. & Glass, C. A. (2001), 'MCMC estimation and some Model-Fit Analysis of Multidimensional IRT Models',Psychometrika 66(4), 541-562.

10. Carroll, J. W. B. & Levine, M. (2007), 'Multidimensional Modeling with Unidimensional Approximations', Journal of Mathematical Psychology 51, 207-228.

11. De la Torre, J. & Patz, R. (2005), 'Making the Most of what we have: A Practical Application of Multidimensional Item Response Theory in Test Scoring', Journal of Educational and Behavioral Statistics 30(3), 295-311.

12. Doody, E. (1985), Examining the Effects of Multidimensional Data on Ability and Item Parameter Estimation using the Three-Parameter Logistic Model, 'The 2002 Annual Meeting of American Educational Research Association', Chicago.

13. Folk, V. & Green, V. (1989), 'Adaptive Estimation when the Unidimensionality Assumption of IRT is Violated',Applied Psychological Measurement 13, 373-389.

14. Fraser, C. (1988), 'NOHARM II: A Fortran Program for Fitting Unidimensional and Multidimensional Normal Ogive Models of Latent Trait Theory', The University of New England, Armidale, Australia.

15. Hambleton, R. K., Swaminathan, H. & Rogers, H. J. (1991), Fundamentals of Item Response Theory, Sage Publications, Newbury Park, United States.

16. Kendall, M. (1961), A Course in the Geometry of n Dimensions, Charles Griffin and Company Limited, London.

17. Kromrey, D., Parshall, C. & Chason, W. (1999), Generating Item Responses Bsed on Multidimensional item Response Theory, 'SUGI 24', SAS, , .

18. Levine, M. (2003a), 'Dimension in Latent Variable Models', Journal of Mathematical Psychology 47, 450-466.

19. Levine, M. (2003b), 'Dimension in Latent Variable Models', Journal of Mathematical Psychology 47, 450-466.

20. Mathai, M. (1999), 'Random p-Content of a p-Parallelotope in Euclideann-Space', Advances in Applied Probability 31(2), 343-354.

21. Mortari, D. (2001), 'On the Rigid Rotation Concept in n-Dimensional Spaces', Journal of the Astronautical Sciences 49(3), 401-420.

22. Peña, D. (2002), Análisis de Datos Multivariantes, McGraw Hill.

23. Peña, D. & Rodríguez, J. (2003), 'Descriptive Measures of Multivariate Scatter and Linear Dependence',Journal of Multivariate Analysis 85(2), 361-374.

24. Reckase, M. (1985), 'The Difficulty of Test Items that Measure more than one Ability', Applied Psycological Measurement 9(9), 401-412.

25. Reckase, M. (1990), Unidimensional Data from Multidimensional Data from Unidimensional Tests, 'Paper presented at the annual meeting of American Educational Research Association', Boston.

26. Reckase, M. (1997), 'The Past and the Future of Multidimensional Item Response Theory', Applied Psychological Measurement 21(1), 25-36.

27. Reckase, M. (2007), 'Multidimensional Item Response Theory', Handbook of Statistics 26, 607-642.

28. Reckase, M. (2009), Multidimensional Item Response Theory, Statistics for Social and Behavior Sciences, Springer.

29. Reckase, M. & Ackerman, T. (1988), 'Building a Unidimensional Test Using Multidimensional Items', Journal of Educational Measurement 25(3), 193-203.

30. Reckase, M., Carlson, J. & Ackerman, T. (1986), The Interpretation of the Unidimensional IRT Parameters when Estimate from Multidimensional Data, 'Annual Meeting of Psychometrics Society', Toronto.

31. Reckase, M. & Stout, W. (1995), Conditions under which Items that Assess Multiple Ablilities will be fit by Unidimensional IRT Models, 'The European meeting of Psychometric Society', Leyden, Holanda.

32. Rizopoulos, D. (2006), 'Ltm: An R Package for Latent Variable Modeling and Item Response Theory Models',Journal of Statistical Software 17(5), 1-25.

33. Sheng, Y. (2007), 'Comparing Multiunidimensional and Unidimensional Item Response Theory Models',Educational and Psychological Measurement 67(6), 899-919.

34. Sheng, Y. (2008), 'Bayesian Multidimensional IRT Models with a Hierarchical Structure', Educational and Psychological Measurement 68(3), 413-430.

35. Stout, W. (1990), 'A new Item Response Theory Modeling Approach with Applications to Unidimensionality Assessment and Ability Estimation', Psychometrika 55, 293-325.

36. Stout, W., Douglas, B., Junker, B. & Roussos, L. (1999), DIMTEST, Computer Software , The William Stout Institute for Measurement, Champaing, IL.

37. Sympson, J. (1978), A Model for Testing with Multidimensional Items, 'Proceedings of the 1977 Computarized Adaptive Testing Conference', Minneapolis: University of Minnesota, Department of Psychology, , , p. 82-98.

38. Team, R. D. C. (2008), R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0. *http://www.R-project.org

39. Walker, C. & Beretvas, S. (2003), 'Comparing Multidimensional and Unidimensional Proficiency Calssifications: Multidimensional IRT as Diagnostic Aid', Journal of Educational Measurement 40(3), 255-275.

40. Wang, M. (1985), Fitting a Unidimensional Model to Multidimensional Item Response Data: The effect of latent space misspecification on the application of IRT, Research Report MW: 6-24-85, University of Iowa, Iowa City.

41. Wang, M. (1986), Fitting a Unidimensional Model to Multidimensional Item Response Data, '', The Office of Naval Research Contractors Meeting, , , Gartlingburg.

42. Way, W., Ansley, T. & Forsyth, R. (1988), 'The Comparative Effects of Compensatory and Noncompensatory Two-dimensional Data Items on Unidimensional IRT Estimates', Applied Psychological Measurement 12, 239-252.

43. Wilson, D., Wood, R. & Gibbons, R. (1987), TESTFACT [Computer program], 'Scientific Software', Mooresville IN.

44. Yen, W. (1985), 'Increasing Item Complexity: A Possible Cause of Scale Shrinkage for Unidimensional Item Response Theory', Psychometrika 50(4), 399-410.

45. Zhang, J. & Stout, W. (1999), 'Conditional Covariance Structure of Generalized Compensatory Multidimensional Items', Psychometrika 64, 129-152.

46. Zhao, J., McMorris, R. & Pruzek, R. (2002), The Robusnetss of the Unidimensional 3PL IRT when Applied to Two-Dimensional Data in Computarized Adaptive Testing, 'The 2002 annual meeting of American Educational Research Association', New Orleans.

[Recibido en null de 2009. Aceptado en null de 2010]

---

Este artículo se puede citar en LaTeX utilizando la siguiente referencia bibliográfica de BibTeX:

@ARTICLE{RCEv33n1a08, 
    AUTHOR  = {Montenegro Díaz, Álvaro Mauricio and Cepeda, Edilberto}, 
    TITLE   = {{Synthesizing the Ability in Multidimensional Item Response Theory Models}}, 
    JOURNAL = {Revista Colombiana de Estadística}, 
    YEAR    = {2010}, 
    volume  = {33}, 
    number  = {1}, 
    pages   = {127-147} 
}