Characteristics of respondents who respond differently to positively and negatively worded items on rating scales

Although evidence prevails that including positively and negatively worded items within the same scale can lead to differential response patterns, little is known about factors that predict how different these responses will be. Thus, three datasets were analysed to investigate the characteristics of respondents whose responses between positively and negatively worded items are most different. The three studies yielded two major findings: (i) responses to the positively worded items yielded statistically significantly higher means than did responses to the negatively worded items, and (ii) several characteristics were identified pertaining to those who tend to have the largest absolute discrepancies in responses between the two sets of items.