Can we include negative or failed experiments?

Yes—and this is strongly encouraged, as it improves hypothesis testing and model refinement.