The Effect of Multiple Choice Test Number of Alternatives on The Estimation of a person's Ability and the Psychometric Properties of a Test and its Items According to Rasch Model in Item Response Theory (IRT)

Abstract

This study investigated the effect of multiple choice test
number of alternatives on the estimations of: a Person's ability, item
difficulty, and the information function of test and its items. To
achieve the study objectives, a multiple choice achievement test
consisting of 40 items in tenth grade maths was constructed. The test
took three different forms in terms of the number of alternatives: Form
One with three alternatives, Form Two with four alternatives, and
Form Three with five alternatives. The test forms were applied on a
total sample of 600 male and female students, with 200 students
assigned to each form. Data obtained for each form of the test were
analyzed separately using (BIGSTEPS) and (BILOG-MG). The study
findings revealed the following: there were no statistically significant
differences (α=0.05) among the standard error means of item in the
estimation of difficulty parameters, and no statistically significant
differences (α=0.05) among a person's reliability coefficients due to
number of alternatives.However, item reliability coefficients were
equal and there were statistically significant differences (α=0.05)
among the standard error means of a person's ability parameters, and
such estimations in a person's ability were more accurate in Form
One of the test than in Form Three, and in Form Two of the test than
in Form Three whereas such estimations were similar in Forms One
and Two. In addition to that, Form One of the test yielded more
information at the low ability level but at the medium and high ability
levels, the most information was in Form Three of the test, and there
were statistically significant differences (α=0.05) among criterion
validity coefficients in favor of Form Two of the test