This article gives an example that shows how it's a standard practice to figure out which people who share a common characteristic are helped by a drug in a clinical trial, and then target that group in the future. This is a "sub-group" -- and it's what drug companies do when developing a group

But, if you are a basher and are mad that the company that you are attacking was able to help 80% of the people in its clinical trial, you say that talking about subgroups is "cherry-picking."

EDIT: Tried to post the quote and it wouldn't let me, so here is a link to the whole article.

https://endpts.com/aacr18-roche-spotlights-pr...0Tecentriq