I think you fundamentally fail to understand what I just said. Proper unbiased random sampling allows you to create sub populations that tend to reflect the characteristics of a larger group, biased populations don’t share that relationship.

“Because some animals hibernate, all animals hibernate” is just as flawed as saying “Because only a small percentage of hibernate, no animals hibernate.” Instead the relationship is “Because some animals exist that hibernate, there exist animals that hibernate.”