Disentangling Plausible and Implausible Straightlining: Effects of Question Format, Mode and Device

Le résumé fourni par la source

Straightlining, (nearly) identical responses in multi-item batteries, is among the most widely used indicators of satisficing. However, straightlining does not necessarily reflect low-quality responses; it can also result from genuine similar attitudes or behaviors. We differentiate between plausible straightlining batteries, where uniform responses are reasonable given specific question characteristics and content, and implausible straightlining batteries, where uniform responding is not expected and most likely signals satisficing. Using data from the GESIS Panel.pop, a mixed-mode probability-based panel in Germany, we leverage a layout change that transitioned multi-item batteries from matrix to single-item formats to study plausible and implausible straightlining. Our findings, based on mixed-effects logistic regression models, indicate that respondents are significantly more likely to straightline on plausible than on implausible batteries. While question format does not affect straightlining overall, single-item formats are associated with less straightlining on implausible batteries but more on plausible batteries. Additional analyses show that web respondents are less likely to straightline than mail respondents, and this mode effect is more pronounced in single-item formats. Smartphone respondents are more likely to straightline than desktop or tablet respondents, particularly on implausible straightlining batteries. These results highlight the importance of considering question characteristics and content when using straightlining as an indicator of satisficing to avoid flagging genuine responses as low-quality data.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.