Discussion about this post

User's avatar
Romeo Stevens's avatar

People strongly favor models with as many moving pieces as average free working memory size (ie 4). Untangling double counted evidence in a high factor model with various correlation between factors was prohibitive until computers.

Expand full comment
Stefan Schubert's avatar

Good post. Another reason to focus on strong evidence is that it probably correlates with *strong effects* in many cases. For instance, it is normally easier to find strong evidence that one intervention is more effective than another if the difference in terms of effects is large. But it also matters more then.

Expand full comment
9 more comments...

No posts