Abstract
Extreme response style is the tendency for scale respondents to use more extreme ends of a response scale (e.g., 1s and 5s on a 5-point Likert-type scale) independent of the item content. Existing research has demonstrated that such response biases potentially yield erroneous results in research findings that rely on scale responses that are susceptible to extreme responding. Potential factors that may affect such response behavior may be the format of the response options and fatigue. With the format of the response options, the current study examined whether descriptors of the ends of a scale (i.e., Completely disagree/Completely agree versus Strongly disagree/Strongly agree) or the decision to label all or only the ends of the scale options may lead to extreme responding. Furthermore, it was assessed whether fatigue (defined by the lengthening of the number of scale items) serves as a moderating factor. Study 1 was a 2x2x2 factorial design (2 response options factors and fatigue factor) and found a significant main effect for the descriptor of a scale. Specifically, more absolute descriptors (i.e., Completely disagree/Completely agree) yielded more extreme responses than did less absolute descriptors (i.e., Strongly disagree/Strongly agree). Study 2 involved examining the potential effect that extreme response style may have in mediating the relationship of
individualism/collectivism on satisfaction with life (Bettencourt & Dorr, 1997). This study found that extreme response style’s effect on the findings was negligible and did not mediate the relationship between individualism/collectivism on satisfaction so as to lead to different estimates of the effect size.
|
