I would think one of the important elements here is that the sample is so small, allowing for less fraud and less room for overlooking errors in the payments. I think the effects would be similar on a larger scale, just as a guess, but the "headline making" fraud and mistakes would draw more negative reaction than they should.
