Today’s post is unusual for me in two ways - for one, it’s a Substack post but it’s a video instead of an article. For another, it’s a kind of video I don’t think I’ve made before. This looks into an aspect of doing statistical research we sorta like to think doesn’t matter, but that’s almost certainly wishful thinking. How much do the option defaults in our scientific software, set semi-arbitrarily by some coder ages ago who didn’t have our particular analysis in mind, affect our results? Whatever the answer is, it’s definitely not zero. Check it out!
Comments
No posts