Is extreme response style domain specific? Findings from two studies in four countries

Abstract

Extreme response style (ERS) may bias responses and hamper the validity of conclusions in substantive research. ERS can be controlled for by using an additional (random) sample of response style indicators (i.e., a separate, random sample of survey items). There are two options to draw response style indicators to control for ERS: from only one versus from multiple domains. In two studies (four samples in total), this paper examines the domain dependency of ERS across three domains: consumer behavior, interpersonal relationships and politics. We find in the four samples repeated evidence suggesting that ERS has a domain specific component. This finding calls into question the (often encountered) assumption that it does not matter from which domains ERS measures are drawn.