The chart at right shows the average credit score for people with email addresses at various Internet domains. For instance, if your email address is email@example.com, then you'd be counted in the AOL domain.
Someone at CreditKarma came up with this study. Why they would look at this correlation is just as interesting a question as the results of it.
Why on earth should people with a Yahoo email address have a (much) lower credit score than those with a BellSouth email address? I have no idea.
Anyone got a theory?
My friend Doug W. (a clear thinker who is mathematically inclined) says, in effect, “Move along, there's nothing to see here!” His explanation, which makes sense to me, though I certainly didn't catch it at first:
I’m not convinced there’s anything real here. My first gut feeling is that it’s a distribution to be expected, but there just isn’t enough info to draw any conclusions. For example, what is the standard deviation of credit scores? The range here is not that large, and they don’t say how many samples there are from each domain. Let’s take a simple model, which is probably wrong but illustrative nonetheless. There are 7 domains listed, and a total of 20,000 credit scores, so suppose that there are ~3000 per domain (certainly it isn’t evenly distributed, but the results wouldn’t chance much). The expected standard deviation given N samples is about sqrt(N), or 1/sqrt(N) as a fraction. So, for 3000 samples, you’d expect about 1/54 ~= 2% as a std deviation. So you’d expect 95% of samples to be within ~4% of the mean. If the mean is 665, and the std deviation is 15, then everything is exactly as you’d expect.