“…This difference in the cost of sociotropic voting may explain why we observe, in both absolute and relative terms, more sociotropic than inequalityaverse voting (remember that we observed 13 (10.2%) G-subjects demonstrating inequality-averse voting, but 54 (28.1%) L-subjects voting in line with efficiency preferences; see Tables 4 and 6, first votes). Such differences may also explain why some studies find social-welfare preferences to be quantitatively more important than difference aversion (e.g, Ackert et al, 2004;Messer et al, 2010;Engelmann and Strobel, 2004), whereas others find the reverse (e.g., Bolton and Ockenfels, 2006;Sauermann and Kaiser, 2010;Höchtl et al, 2012). In F&R's thought experiment as well as in our empirical experiment, losses were distributed evenly among reform losers, while gains were evenly distributed among reform gainers.…”