which paper(s) to cite when stating that out-of-sample validation is better than in-sample val

Question

Validating a model (meaning, computing metrics on the prediction) using the training samples is, well, rubbish. Out-of sample validation should be preferred. This is well known by statisticians and machine learning practicionners.

However, I can hardly state it like this in a paper without a good authoritative source. Especially when the intended public is in medical field which still uses in-sample validation extensively...

What would be the best paper to cite in this case?

Answer 1

If you're happy to cite a book rather than a paper, a good reference is "An Introduction to Statistical Learning: With Applications in R" by James, Witten, Hastie, and Tibshirani. This topic is covered in Section 2.2.

A pdf copy of the book is available legally from http://www-bcf.usc.edu/~gareth/ISL/.