I recently finished
The Problem with Software by Adam Barr, which is about exactly this issue. Here are my notes:
https://www.briansnotes.io/book/the-problem-with-software/The author makes the argument for empirical research on software engineering just as you're describing, and references relevant studies.