Psychometric Characteristics of the Culture Fair Intelligence Test Scale 2

Abstract

This study aims to investigate the psychometric characteristics of the CFIT-scale-2, including investigation of difficulty level, discrimination, distractor effectiveness, DIF (Differential item functioning), IRT (Item Response Theory) 2PL model, validity, and reliability. Respondents were 507 students with 245 (48%) females and 262 (52%) males, with ages ranged from 8-13 years (M= 10.88; SD = 1.35). The classical test theory (CTT) analysis showed that the item difficulty level on the CFIT-scale-2 had varying item difficulty levels, from easy to difficult. However, the test arrangement was not structured according to the suggested difficulty level from easy to medium to difficult. The discrimination of items was poor because 28 items were not included in the very good category (p > .40). In addition, 47 (25%) of the 184 distractors are ineffective. CFIT-scale-2 did not contain DIF (Adj.p > .05), and IRT analysis showed that the CFIT-scale-2 was not structured according to the difficulty pattern from easy, medium, and difficult. The CFIT-scale-2 based on IRT analysis contained 44 items or 96% with good discrimination. The results of a construct validity test using the CFA technique showed a good fit model (p < .001; RMSEA < .08; SRMR < .08), which was acceptable and supported the fit between the theoretical and empirical model. The reliability coefficient value of Cronbach's alpha was 0.88 (α > .70), indicating that the CFIT-scale-2 had good reliability, but the construct reliability was below an acceptable value (CR < .69). According to this study, the psychometric characteristics of the CFIT-scale-2 should be revised and reevaluated.