The Effect of the Number of Alternatives and Items in Estimating the Parameters of Individuals and Items Using the Four-Parameter Parametric Model

Abstract

number of alternatives and items in estimating the parameters of individuals and items using the four-parameter parametric model, using the descriptive survey methodology. The study population consisted of all the students who study psychology and life sciences subject, it included 245 male and female students. The study sample consisted of 237 (156 male students and 81 female students) who were selected using purposive sample. The study instrument consisted of three experimental models. The validity and reliability of their content and structure were verified using internal consistency.
Individuals and items of the four-parameter model were matched using matching indicators. The parameters of individuals, items, and standard errors were estimated in classical theory (difficulty, discrimination) and item response theory (difficulty, discrimination, guessing, upper asymptote). Moreover, assumptions were achieved (unidimensionality, monotone homogeneity, item characteristic curve, item information function and test).
The results according to the classical theory have found differences between the mean parameters (discrimination) of the two and three variants, as well as between the two alternatives and the four, in favor of the three and four. Moreover, there were differences in the theory of response to the item between the parameter means (upper asymptote) for the number of items 9 and 15, and 9 and 24, in favor of 15, 24 items.