Evaluasi Psikometri Aitem Induktif Figural pada USU General Reasoning Test (UGRT)

Abstract

This research uses descriptive quantitative methods and uses a sampling technique, namely convenience sampling. The aim of this research is to evaluate figural inductive items, namely the Figural Classification (IF 1) and Matrices (IF 2) subtests on the USU General Reasoning Test (UGRT). Psychometric evaluation of figural inductive items uses Rasch modeling which includes reliability coefficient analysis, validity coefficient analysis based on internal structure, item measure analysis, Wright map, and unidimensional. This research uses secondary data, where data was obtained from P3M test participants, Faculty of Psychology, University of North Sumatra. The UGRT test was given classically to 824 samples with an age range of 15 to 60 years who were divided into 4 sets (A, B, C, D). The psychometric results obtained are that the person reliability coefficient in the IF 1 subtest produces values in sets A, B, C, and D respectively of 0.32, 0.24, 0.53, and 0.47, and the IF 2 subtest produces values in sets A, B ,C,and D of 0.25,0.28,0.43, and 0.24 respectively are categorized as weak for all sets. The item reliability coefficients in the IF 1 and IF 2 subtests are in the range of 0.94 to 0.99. Validity coefficient using 0.5 < MNSQ < 1.5; -2.0 < ZSTD < 2.0; and 0.4 < Pt Mean Corr < 0.85 in the IF 1 subtest with 36 and IF 2 subtests with 29 items meeting the fit criteria. In IF 1 there is item 2 in set B, item 9 in set C, and item 5 in set D which includes unfit criteria and IF 2 there are items 3 and 4 in set A which include unfit criteria. The item measure values obtained in IF 1 range from -2.41 to 7.13 and IF 2 range from -2.39 to 3.15. With a range of difficulty levels for items in IF 1 starting from 9 to 12 and IF 2 starting from 7 to 11. Unidimensionality in the IF 1 and 2 subtests in each set is above 20%, which means the test is adequate. The results of this study show that The validity of the dominant test is good, but it still needs to be re-evaluated, especially for the 3 items in the IF 1 subtest and the 2 items in the IF 2 which are unfit. Apart from that, it is necessary to retest with a larger and more varied number of samples to increase the reliability of the UGRT test.