Because it potentially encourages using the data in a technically wrong way. What I mean by "wrong" has nothing to do with morality. I mean "wrong" as in people who think they can predict behavior will over-interpret the data in a technically incorrect way and try to apply that model causing real harm. That might mean giving insurance companies an excuse to over-interpret health data to increase rates or law makers legislating discriminatory laws based on bad data analysis. It gives scam artists extra ammunition to peddle their crap.