Newsweek ran a story this week about the scandals rocketing through the medical research field: a huge dispute over how much corporate sponsorship of research influences outcomes. The debate centers on the conflicts of interest that result from who paid for the research and whether that influences outcomes.  But there is an interesting side note that some scientists say is potentially more important: With sponsored research, the sponsor can simply fail to report the bad results. When the drug works—publish! When the drug doesn’t seem to have much effect—don’t publish! In effect, only the good outcomes (from the sponsor’s point of view) ever see daylight.

This relates directly to Katie Porter’s post about the need for empirical research that produces original data in a field such as credit research. There are tons of data collected about customers, defaults, debt-income ratios, the role of job loss, etc. But those data are proprietary, which means the owners (the creditors) don’t have to tell anything. If they have something to say that advances their interests, e.g., during the debates over the bankruptcy law, then they can produce those data. But everything else is secret.

Original data are very expensive to develop. Without them, however, Katie has it right: our understanding of credit will be completely one-sided. We will hear only one voice—and that will be the voice of those who make money from promoting a particular point of view.

Doctors and patients are starting to worry about the effects of bias in medical research. I hope the number of people who worry about biases in what we know about credit will expand as well.