Why do people underestimate income and wealth inequality?

There was a recent paper in the news by Michael Norton and Dan Ariely arguing that Americans substantially underestimate the degree of income and wealth inequality in the USA. Other papers have found similar results. But why? Crowds do quite well at estimating all sorts of other quantities, but they fail dramatically on this problem, as Timothy Noah notes in a Slate piece on the Norton and Ariely paper here. More technically, we might expect from models of information aggregation that if the signals people get about the true distribution of income are unbiased, the errors should cancel out, except they do not here. So what is the source of the bias here?

Two ideas. First, maybe people estimate the distribution of income and wealth based on signals from their friends and neighbors, and they mostly associate with people like them in terms of income. Since most people also tend to place themselves somewhere in the middle of the distribution (but why? national ideology?), the estimated distribution will be more egalitarian than the true distribution. If someone were to go around publicizing information about the true distribution of income then these estimates might shift a bit, but probably not reliably; people receive signals from their friends and neighbors about the distribution of income all the time, whereas few read or care about econometric estimates of income distribution.

But perhaps a second possibility (not necessarily incompatible with the first) is that people estimate the distribution of income and wealth from signals about consumption (whether or not these are their friends); if consumption inequality is lower than income or wealth inequality (as some people suggest it is), then estimates of income and wealth inequality will also be biased downwards. Again, providing information about the true distribution of income to people is also unlikely to change these perceptions reliably, but changes in consumption patterns might (e.g., if the rich engage in more conspicuous consumption).

Do either of these accounts sound like plausible explanations? Other ideas?

Bonus query: if Wilkinson and Pickett are right that income inequality causes social and health problems via status competition over consumption, then the fact that people are systematically deluded about the true extent of inequality might be a sort of silver lining; greater awareness of inequality might induce even more social and health problems, though it might also induce more redistributive policies than currently prevail (but I wonder: beliefs about the proper degree of inequality might also adjust downwards with more accurate information, depending on how strong our natural inequality aversion actually is).