I don't even need to run that through a model. A sample size of 10 on a population of 1 million? Hypothesis failed. And even that assumes that you selected them randomly from the population, and didn't, for example, just select 10 people who had already met your hypothesis.
I don't believe...