I did read the study, and the appendix S1. I just reread it, searching for "assault" and "rape."
They define "assault" to mean "unwanted sexual contact, up to and including rape." Apparently, they measure it by the answers to Question 39, about "physical sexual harassment," which is where they got that 26% figure from.
They don't define "unwanted sexual contact." For example, Bora Zivkovic had a habit of hugging women, some of whom didn't enjoy it. That is literally an "unwanted sexual contact." It may be creepy, unpleasant or inappropriate, and it should (and did) stop. But it's not rape.
This study doesn't distinguish between unwanted hugging and forcible rape, and it doesn't break down the 26% figure into more or less violent forms of unwanted sexual contact. It doesn't even give the number of violent rapes.
If you have an unknown number of unwanted huggings, at one end of the spectrum, and an unknown number of violent rapes, at the other end of the spectrum, then that's a big grey area.
Managers get their orders from the legal department. The legal department sets rules that will give them a safe harbor from lawsuits, and not necessarily rules that are fair, reasonable, logical, or based on evidence.
This study is oversimplified. I think they're using science to advance their political agenda. That's OK, but they need better science. I'd like to see a better-quality study.