I get caught when I've been listening to music on public radio, and it ends, but the radio's still on, and I'm assaulted with Nina Totenberg.
As to the substance of the article, it's basic statistics that you have to adjust for other factors (e.g., credit history, employment record, etc.) when you seek to correlate one variable against another, e.g., loan rejection rate against race. Apparently the data in the article is unadjusted.
Is everyone rCissss? Give these people 40 acres and a mule!