The effects of rating scale format on the measurement of policy attitudes in web surveys

Recent research suggests that policy attitudes should be assessed with surveys that use branched rating scales that separately measure the direction and intensity of issue preferences. However, past studies have not tested whether branched rating scales improve the quality of survey data in the context of self-administered web surveys. In the current study, I compare the effects of different issue scale formats on response distributions, survey satisficing and issue-based voting in multiparty elections. Using data from a randomized web experiment, I find that partially labeled seven- and eleven-point issue scales and fully verbalized branched scales produce very similar response distributions and comparable effects of policy attitudes on party choice. However, the findings regarding the extent of satisficing behavior are mixed. Based on the efficiency of different scale formats, scholars are encouraged to use seven- or eleven-point rating scales to measure policy attitudes in web surveys.