In order to reduce agreement bias(acquiescence) that occurs when respondents agree regardless of questions in survey, negatively- and positively-worded items are often mixed. However, methods effects can be caused from the mixed format and the effects leads to problems in test validity. Using tests with negatively- and positively-worded items, the present study examined whether theoretical factors a test tried to measure and method-effect factors could be distinguished. For this purpose, self-esteem and peer relationship scales were used for elementary, middle and high school groups respectively, and analyzed applying confirmatory factor analysis. Results showed that CTCM and CTCU models considering methods effects fitted well. However, for the self-esteem scale, methods effects appeared for negatively-worded items while for the peer relationship scale, methods effects appeared for positively-worded items. The significance of this study indicates that it is necessary to consider methods effects when exploring the factor structure of a test with negatively- and positively-worded items in a mixed format. This study also suggests further investigations regarding method effects using various tests in future studies to generalize current results.
Lee et al. (Thu,) studied this question.