3 ms·
Yes but with traditional frequentist approaches most scientists don’t understand what’s going on under the hood of the statistical tools they’re using. There ar
by outlace 8y ago
Yes but with traditional frequentist approaches most scientists don’t understand what’s going on under the hood of the statistical tools they’re using. There are dozens and dozens of named tests like “Fisher’s exact test” and “Wilcoxon signed-rank test” and most scientists just sort of follow received wisdom about which test to use in which situation.
Bayesian methods force you to actually think about how your data and model parameters are distributed, and explicitly specify a model.
- eli_gottlieb 8y ago>Bayesian methods force you to actually think about how your data and model parameters are distributed, and explicitly specify a model. Yep. I know some very senior scientists who get a lot of mileage out of finding places where "model-free" methods are implicitly assuming a bloody stupid model, and then attacking them.