Pre-analysis Plan for Study of Conjoint Design: A Follow-up Experiment

This experiment is a follow-up study for Bansak, Hainmueller, Hopkins and Yamamoto (2016, doi:10.7910/DVN/WX5UXL), which examined how the number of attributes shown to respondents in a conjoint survey affected response quality. The experiment consists of two stages. Stage 1 is designed to identify a set of “filler” attributes which respondents perceive to be unassociated with the fixed set of attributes of main interest. Stage 2 investigates whether the addition of those filler attributes changes the effects of the main attributes on conjoint choice and rating tasks.