The purpose of this study is to predict the answer switching behavior from the item parameters in multiple-choice test. Specifically, relationship between number of answer switching behavior and item parameters (item difficulty, item discrimination, item standard deviation and item reliability) were investigated. Multiple-choice testing is one of the testing types in which respondents are asked to select the best possible answer (or answers) out of the choices from a list (Oosterhof, 1994). Researchers have shown that Multiple-choice assessment is a useful vehicle for student assessment in core university subjects that typically have large student numbers. On the other hand, answer switching behavior is one of the common behaviors in multiple choice type test because it is affected the exam score and needs to be investigated (Wallace & Williams, 2003; Milia, 2007; Basturk, 2009).

The participants in this study consisted of 157 undergraduate level College of Education students from mid-sized universities. A Midterm exam consisted of 20 multiple-choice questions. The exam was completed in pencil on a card read by an optical marker machine. Two judges independently reviewed the exam cards to identify answer changes by noting eraser marks. On completion, the judges compared their coding and agreed on any anomalies.

The results of this study demonstrated that there is a negative correlation between number of answer switching and item difficulties. Negative significant correlation result indicated that when the item is getting difficult, number of answer switching behavior increase. For item difficulties, present study showed that there was no significant correlation between item discrimination and number of answer switching behavior. Item discrimination refers to the ability of an item to differentiate among students on the basis of how well they know the material being tested. This result shows that whether item more or less discriminate the students’ performance, it attracted similar answer switching behavior. Multiple regression analysis also conducted to see whether the item difficulties and item discrimination together would be more valid than the item difficulties alone to predict answer switching behavior. Multiple regression equation showed that when two predictor variables combined and used, they explained 34 % of the variance in number of answer switching behavior and only again item difficulties is statistically significant and valid predictor at 0.01 alpha level.