What Drives Straightlining, Respondent, or Grid Characteristics? (303609)

Keywords: straightlining, satisficing, data quality, multilevel modeling, grid characteristics

Nondifferentiation in grid questions, also known as straightlining, happens when respondents select the same response option on all items instead of carefully answering each item individually. Although identical responses to a set of related items worded in the same direction is not necessarily a sign of careless answering, straightlining on items from different domains or on both positively and negatively worded items is considered a sign of low cognitive effort. Previous research has shown that straightlining correlates highly with other satisficing behaviors (such as speeding) and therefore is an indicator of poor data quality. Especially in online panels, where respondents might be repeatedly exposed to grid questions, straightlining is a potential issue.

Earlier research on straightlining gives emphasis on characteristics of respondents who are more likely to straightline. However, there is not enough attention paid to characteristics of the grid (e.g., position of the grid in the survey, number of items, number of response options, labeling or response options) that potentially attract straightlining. This study considers straightlining as the result of an interaction between respondent and question characteristics. Using data from the LISS online panel in the Netherlands and the German Internet Panel (GIP), we conducted multilevel modeling to identify who are more likely to straightline on what kinds of survey questions.