Wording effects in self-esteem scales: Methodological artifact or response style?

This article examines the analysis and interpretation of wording effects associated with the use, of both positively and negatively worded items in survey instruments. Recent research using global self-esteem scales has indicated that the strategy of including both positively and negatively worded items may introduce systematic biases that interfere with the measurement of the substantive trait, self-esteem. Although such systematic elements are "methods effects," they may not be purely artifactual in nature, and we show that this distinction between artifactual and substantive interpretations of such wording effects has important implications for the structural equation modeling (SEM) strategy that is employed. We use data from a national sample of American junior and high school students to empirically evaluate 4 research questions suggested by the literature on individual response style to the wording effects observed in self-esteem scales and other survey instruments that utilize positively and negatively worded items. Results of these analyses support the interpretation of these wording effects in terms of personality traits that have potential substantive relevance and that can be estimated and evaluated using SEM modeling strategies.