Primary Outcomes (explanation)
Preferences over political candidates is an outcome constructed by looking at 4 pairs of fictional candidates, or "Vignettes". For each candidate I include information on their name, education, job, stance on immigration, stance on redistribution, marital status and hobbies. Name, education, job, and hobbies are clustered to indicate socioeconomic status, either high SES (e.g. master's degree, chief economist, likes sailing) or low SES (e.g. professional training, mechanic, watches champion's league). Immigration stance is either for or against accepting new refugees, and redistribution stance is either for or against increasing the demands on cash benefit recipients.
The three dimensions of interest are SES, immigration stance and redistribution stance. In three of the Vignettes, candidates vary on one of the dimensions and are identical on the other two. As an example, in one Vignette on candidate is for increased immigration and the other is against, but they are both for (against) increased demands on cash benefit recipients and both high (low) SES. In the 4th Vignette, one candidate is low SES and the other is high SES, and the policy stances are identical. However, their language varies, so that the low SES criticizes the state of affairs while the high SES candidate uses more neutral language. These four Vignettes allow me to look at the effects of treatment on preferring (1) an anti-immigration candidate (2) an anti-redistribution candidate, (3) a low SES candidate (in the case where the language of the low and high SES candidate is the same, as well as in the case where the language differs). Because I hold the other two dimensions constant I consider only the effect of e.g. SES, and not SES as a possible proxy for immigration/redistribution stance.