Stupidity is probably the preferable option. You at least wouldn’t expect the errors to tend in a particular direction.

I think medical science had one of the higher fraud rates. The most glaring strategy that I recall was authors basically just taking a previously published study and swapping the disease of interest in the text.