Florian Keusch, Ting Yan
Impact of response scale features on survey responses to factual/behavioral questions

Pp. 131-149 in: Paul J. Lavrakas, Edith de Leeuw, Allyson Holbrook, Courtney Kennedy, Michael W. Traugott, Brady T. West (Eds.): Experimental Methods in Survey Research: Techniques that Combine Random Sampling with Random Assignment. 2019. Hoboken, NJ: Wiley
[Wiley series in survey methodology]

This chapter takes advantage of an experiment on a probability online panel to examine both the main effects and the joint effects of three scale features on answers to behavioral questions using a frequency scale. The first design feature is the direction of a frequency scale. A frequency scale could descend from the highest to the lowest point. A second feature of particular relevance to web surveys is scale alignment – the presentation of scales on a computer screen. A third scale feature is the vagueness of scale labels. Survey questions measuring frequency of behaviors either ask respondents to report an exact frequency in terms of numbers (e.g. the number of times something happened in a specific time period) or to choose from a list of vague quantifiers. The chapter also looks at several respondent characteristics (age and education) and paradata (response times) that are highly indicative of respondents' likelihood to satisfice.