statistics

iNCREDiBLE

A pharmaceutical company wants to get a drug approved by the FDA. However, their studies appear to have no statistically significant results. The company researchers divided the data into subgroups in a number of different ways and determined significance in the subgroup. The company reported the results and the stock price rose significantly.

What do you recommend that the statistician should do?

Can someone explain to me what he should do and why he should do it in a pedagogical way?

Moonbear

Is this really intended to be a statistics question, or an ethics question?

I don't know what you've already covered in your statistics class, but one thing to consider is what would alpha be set as when determining significance if you're doing multiple post-hoc tests?

That's one direction you can think along to find an answer to this question. To get any more help, you'll have to tell us what your thoughts are on the answer so far. It will also be helpful if you let us know what type of statistics course you're taking (is it introductory level, intended for statistics majors or other science majors, etc.?) so the responses will be at the appropriate level for you.