Mastodon Feed
Boosted by jsonstein@masto.deoan.org ("Jeff Sonstein"):
xahteiwi ("Florian Haas") wrote:
For anyone following at home, SciPy to the rescue — specifically `binomtest` from `scipy.stats`, which for `binomtest(106, 190, 0.5, alternative="greater")` returns a 𝑝-value of 0.0637. This is to say that there's a 6.37% probability to get 106 heads out of 190 tosses of a fair coin. Commonly, we'd start looking into things if that probability were below 5%.
So, even the result in the 45+ cohort isn't statistically significant; more data would be needed.