Detection of Mutually Dependent Test Items Using the LCI Test
Item response theory (IRT) is widely used for test analyses. Most models of IRT assume local independence, meaning that when the ability variables influencing the test performance are held constant, an examinee’s responses to any pair of items are statistically independent. However, many factors might cause local dependence among items. Consequently, conditional independence (CI) tests are needed among items given a latent ability variable. Hashimoto and Ueno (2011) proposed the latent conditional independence (LCI) test. While other CI tests are sensitive to dependencies of items aside from the targets, the LCI test is robust to such dependencies. However, when the two target items affect the same items, the LCI test might fail to detect local independency between the targets. The previous work of Hashimoto and Ueno (2011) is improved on to obtain a more accurate detection method.
Keywordslatent variable conditional independence test Bayesian network IRT
Unable to display preview. Download preview PDF.
- 3.Birnbaum, A.: Efficient design and use of tests of a mental ability for various decision-making problems (Series Report 58-16, no.7755-23). USAF School of Aviation Medicine, Randolph Air Force Base, Texas (1957)Google Scholar
- 4.Birnbaum, A.: Some latent trait models. In: Load, F.M., Novick, M.R. (eds.) Statistical Theories of Mental Test Scores, pp. 397–424. Addison-Wesley, Reading (1968)Google Scholar
- 5.Samejima, F.: Estimation of latent ability using a response pattern of graded scores. Psychometrika Monograph (17) (1969)Google Scholar
- 6.Samejima, F.: A general model for free-response data. Psychometrika Monograph (18) (1972)Google Scholar
- 11.Reese, L.M.: The impact of local dependencies on some LSAT outcomes. Law School Admission Council Statistical Report 95(02) (1995)Google Scholar
- 12.Sano, M.: Detecting overestimation of discrimination parameter applying mutual information. Japanese Journal for Research on Testing 5, 3–21 (2009)Google Scholar