Especially with errors that can't easily be attributed to one obvious mod, many users default to removing mods one by one to find the culprit. This is inefficient and often just unnecessary. It also forces full-modlist restarts (which can take minutes), whereas starting from zero and adding mods until the error appears is far faster - still not ideal though. The gold standard here would be a binary search (aka 50/50 or binary method):
Disable half your mod list, start the game.
Error gone?
Culprit is in the disabled half - switch to the disabled half and repeat the process.
Error persists?
Culprit is in the enabled half - repeat there.
Math guarantees you'll isolate the culprit (or a tiny group of potentially remaining candidates) in 5–8 tests max, even with 100+ mods. There are some limitations to this method though, based on how modding RimWorld works:
Dependencies between mods may break if they aren't in the same half by chance.
Borderline unfeasible to identify "in-game" errors, since mod-removals will likely corrupt your save.
Therefore: The method is best used startup crashes or issues that trigger without loading a save.
Pro Tip: Use your mod manager's colour/highlight feature to mark the batches of mods you've removed each step. After pinpointing the bad mod you can then filter by colour to re-enable the previously removed mods in seconds and re-establish your prior modlist - without the troublemaker, of course.