(assuming proprietary, unshareable data)
I could see it both ways:
1) If you don't use RAs, you are free to p-hack or log whichever ordinal variable you like, and no one else will ever know.
2) if you do use RAs, you can scream at them to "try again" and "check their work" whenever you don't get the results you want. They are very motivated to please you, especially if you are a Big Fish, so they may cheat on your behalf. Nice thing is that this way you have plausible deniability.
Curious what my bros think... Yes I worry about 2) a lot, especially with big names employing armies of RAs who need recommendations for grad school.