Parameterization of Teacher-made Physics Achievement Test Using Deterministic-Input-Noisy-and-Gate (DINA) Model

Abstract

Traditional methods of test parameterization have been found defective in terms of assuming one score and not providing information on skills mastery profile of the examinees, in addition to non-estimation of the fourth parameter- slipping parameter of test items. Cognitive diagnostic modeling (CDM), specifically deterministic-input-noisy-and-gate (DINA) model is a robust procedure for measuring the skills mastery profile of the examinees, when there is no attribute hierarchy. DINA model also estimates four item parameters: discriminating, difficult, guessing and slipping which hitherto are not implemented as a whole in traditional techniques including the item response theory (IRT). Physics was chosen because it is commingled with civilization. Also, secondary school Physics teachers predominantly use teacher-made tests in their internal examinations in Nigeria.  Such instrument cannot be used to precipitate the true ability levels of the students for purposes of promotion. The data collected plus the harmonized Q-matrix were analyzed using CDM package in r software version 3.4.3 via R-Studio version 1.0.153. The results showed that: (i) eighteen items in the test fit the DINA model while thirty-two items had misfit. (ii) nine attributes were mastered by the examinees while three attributes were not mastered. It was recommended that schools’ item banks should be created by the governments for uploading psychometrically robust test items for teachers’ use.

Keywords: Physics, teacher-made test, DINA model and parameterization.