Effects of negatively phrased test items on internal consistency and factor structure

The aim of this research is to investigate the systematic measurement errors caused by scales consisting negatively phrased items. Meyer, Allen and Smith’s organizational commitment scale is used in order to investigate this issue. The original version of this scale (consisting of both positively and negatively phrased items) and another version in which the negatively phrased items are transformed into positive phrasing, are tested on two samples that are equal to each other in terms of gender, age and education level. Samples are formed of participants working in service sector. Findings are in parallel with previous findings suggesting that negatively phrased items cause systematic measurement errors. Item-total correlation of the negatively phrased items are found to be very low. The item-total correlation of those negatively phrased items increased when they were transformed into positive phrasing. Factor analysis of the scale which consists of negatively phrased items show that those items form a seperate factor apart from the original factor. Whereas, when the negatively phrased items are transformed into positive phrasing, the original factor structure of the scale is supported.