Abstract:
This study aimed at identifying the effect of item composition style in an attitudinal scale on the psychometric properties of items’ scale’s and subjects ability estimates according to item response theory. To achieve the objective of the study, five item composition styles with different number of positive and negative items were designed. The five item composition styles were extracted from a scale developed by Shraifin (2006) to identify science teachers’ attitudes towards laboratory work. The five item composition styles were administered to a random sample consisted of 500 male and female science teachers with 100 male and female teachers responding on each style. The results revealed no statistically significant differences in the means of items difficulty perimeters at (α=0.05) attributed to item composition style. Statistically significant differences in the means of items’ discrimination perimeters at (α =0.05) were found in favor of the first item response style (all items are positive). The results also revealed that information function of both the item and the scale increase as the number of positive items increases whereas, the value of the standard error decreases as the number of positive items increases. The results also indicated that the first item composition style provides more information at various trait levels. Moreover, the results revealed no statistically significant differences in the persons’ estimates of the trait attributed to the item composition style used.
