An Analysis of Sex-related Differential Item Functioning in Attitude Assessment

This study analyzes and classifies items that display sex-related Differential Item Functioning (DIF) in attitude assessment. It applies the Educational Testing Services (ETS) procedure that is used for classifying DIF items in testing to classify sex-related DIF items in attitude scales. A total of 982 items that measure attitudes from 23 real data sets were used in the analysis. Results showed that sex DIF is common in attitude scales: more than 27% of items showed DIF related to sex, 15% of the items exhibited moderate to large DIF, and the magnitudes of DIF against males and females were not equal.     

Effectiveness of Valid Mean Substitution in Treating Missing Data in Attitude Assessment

This study investigates the use of the valid mean substitution (VMS) procedure in replacing missing data in attitude assessment. Using four scales from an actual field data set, the effectiveness of VMS was compared with the multiple regression replacement procedure. The study was designed such that data were missing completely at random under two research conditions: percentage of missing items in each scale (10, 30 and 50%); percentage of missing values in each item (10, 30 and 50%). The results indicated that both procedures are similarly effective in estimating the means and standard deviations of the scales. However, VMS is superior in estimating parameters.