Aggregate marginal means (MMs)

Contexts in source publication

Context 1

... (Likert scale and binary). We cluster the standard errors by respondent i. In some specifications we also use municipal FE and entropy balancing weights (see SI appendix Table C2 and Fig. B3). In the SI appendix (sections A and B), we also present more details on data collection and methods, including various covariate balance tests (Table B1 and Fig. ...

Context 2

... focus on the main findings regarding councilors' preferences using the forced choice outcome. Fig. 1 displays marginal means for proposal choice (i.e. average choice probabilities given a specific attribute level) across all levels. Given the wording of our question, forced choice and marginal means can be directly interpreted as the expected support that a policy containing this particular attribute value would receive if it were to ...

Context 3

... terms. In Fig. 2, we examine how councilors' perceived economic and socio-cultural threat posed by refugees, their ethnocentric values and their ideology might polarize policy preferences. We only focus on two of the five attributes (type of host site and who manages it) because there is not significant divergence from aggregate preferences (see Fig. 1) with respect to the other three ones. Here we note that the absence of preference heterogeneity with respect to the host site's size, in particular, further reflects the idea the proportional 1% quota is indeed perceived as the 'fair-share' allocation and is not polarized. is a clear dichotomy between left and right on the role of ...

Context 4

... We did not detect any imbalances for any of the covariates we use in the analysis. Table B1 reports those regression results. We also examined whether there is any preference for the left-hand or right-hand profile in our pair design. We did not observe any overall trends or any significant imbalances after performing this diagnostic test. In Fig. B1 we show those ...