>>3852>it is likely that people will put in crappy data into the system by mistake, or that some workplaces will try to game the system. detecting stuff like that requires statistical analysisAn example of something very useful for detecting fraud:
https://en.wikipedia.org/wiki/Benford's_law There is also a psychological resistance against repeating numbers that is stronger than what one should expect from results of throwing dice.
For reducing accidentally wrong inputs or inputs that are wrong because users are forced into wrong inputs by the constraints of the input forms like check lists. The standard input form should have at least four distinct options for any question: Yes, no, not answered (yet), does not apply.
Using insights from voting theory can be used to reduce the benefits of tactical disinformation (there is a tradeoff however in that a large dose of this robustness against tactics reduces the sensitivity to information).