I really enjoy Lane Kenworthy’s blog. He writes mostly on inequality and he tilts a little too towards advocacy for my tastes, but he always brings a lot of data to the discussion.
He posted on international opinions of inequality the other day. People say they prefer a normal distribution of income rather than one that has overall more income, but skewed towards the rich.
He suggests this means that people care more about inequality than efficiency. The problem is that people may have a some sense of diminishing returns to income in their minds when they answer these questions. There may be more income overall, but some lower income groups have lower incomes in the skewed distribution (see the forth and fifth rows in distribution E). If those groups suffer more from there losses than the rich gain, then preferences for efficiency would suggest the normal distribution over the higher-income skewed distribution.
A more careful study of social preferences — answering the question, do people care about efficiency or inequality (or reciprocity)? — is Rabin’s Understanding Social Preferences with Simple Tests (pdf) ((There’s about a billion citations to this paper at Google scholar, too. Go crazy, but be sure to report back.)). Virtues: doesn’t use survey data and it is careful to identify particular social preferences. Vices: relatively small sample size and only American and Spanish college students.
UPDATE: I got the interpretation of the diagrams a little goofed up in this post. The intuition remains the same though. My misinterpretation is evidence of measurement error inherent in surveys like these, so I’ll refrain from editing.