I’m normal, you’re normal
Frances Woolley at Worthwhile Canadian Initiative has another fascinating graph. This one’s based on US data from the World Values Survey, where people were asked which income decile they are in:
As she describes, it doesn’t seem to be that the survey has accidentally undersampled the poor and the rich. People seem to be deluded about where their income fits in the national distribution.
There are a number of possible explanations. At the high end, there is the fact that people with more money than you are more visible, so it’s easy to overestimate how many of them there are. At the low end, part of the explanation may be that young people whose family income has been higher in the past and will be higher in the future place themselves higher up in the distribution. The obvious example is university students, who in some ways behave more like their parents than like non-students with similar household income.
Part of it, though, is that people just overestimate how normal they are, just as you (and I) probably overestimate how many people care about statistics in the media.
Thomas Lumley (@tslumley) is Professor of Biostatistics at the University of Auckland. His research interests include semiparametric models, survey sampling, statistical computing, foundations of statistics, and whatever methodological problems his medical collaborators come up with. He also blogs at Biased and Inefficient See all posts by Thomas Lumley »
Did they rule out the possibility that people in the middle of the income distribution are more likely than the very poor or very rich to be reached by a survey, and respond to questions about income?
12 years ago
At the high-income end, they did, at least to some extent.
The proportion with college education (and higher) and the proportion in managerial positions didn’t show similar undersampling. At the low income end they may have undersampled semi-skilled and unskilled manual workers; it’s hard to tell because there’s a large (19%) “Other” category. If there is undersampling at the lower end only, it suggests that people with high income underestimate just how high it is, but people with low income are less biased.
The Canadian sample in the same survey, who were asked about explicit income ranges corresponding to the deciles had much more uniform distribution, so if it was undersampling, it happened in the US and not in Canada.
12 years ago