Abstract
This study aimed to compare reliability coefficients on tests with dichotomous and polytomous items according to classical theory of measurement and generalizability theory, to achieve this, a compound ability test was constructed in mathematics in two forms: Dichotomous and Polytomous by using five estimation methods, which as follows: (Alpha {α}, Theta{θ}, Raju, {G}, and {φ}), Furthermore, in order to achieve the study objectives, the test was applied on two forms, each test consists of 30 items to measure students mathematical ability on the field of numbers, Algebra, Geometries and Statistics, to 300 of the sample study. the reliability coefficients of the Dichotomous and estimated according to the Classical theory on the following manner: (α= 0.87, θ= 0.85, Raju= 0.86), and according to the Generalizability theory (G= 0.87, φ= 0.86), hence, the counter coefficients of Polytomous according to the Classical theory (α= 0.95, θ= 0.92, Raju= 0.93), and according to the Generalizability theory (G= 0.95, φ= 0.94).). The results of the comparison between the coefficients according to the Cocron test and the signal/noise ratio test showed that there was a significant difference at (α =0.05) between the reliability coefficients values of the ability test on its two forms in favor of the polytomous, and the results also revealed a preference for generalizability theory in the estimation of reliability coefficients as extracted to classical coefficients.
Recommended Citation
Al-Gharaibh, Abdal hadi and bani Ahmad, Ahmad
(2023)
"Estimating Reliability on Tests with dichotomous and Polytomous Items According to the classical Theory of Measurement and Generalizability Theory: A Comparative Study,"
Jordan Journal of Applied Science-Humanities Series: Vol. 36:
Iss.
1, Article 8.
Available at:
https://digitalcommons.aaru.edu.jo/jjoas-h/vol36/iss1/8