Psychometric evaluation of the KAZTEST language assessment within international measurement frameworks

Introduction National language assessments require transparent psychometric documentation to support valid score interpretation. Methods Rasch measurement was applied to operational KAZTEST data from 1,199 (listening) and 1,197 (reading) test-takers using Winsteps 5.8.3.0. Results Both components demonstrated high item reliability (0.99), satisfactory internal consistency (KR-20: 0.73 listening; 0.84 reading), and adequate model fit. Person reliability was stronger for reading (0.76) than listen