"Likely to cheat" usually means something like "the probability of a person cheating is higher," meaning it would be based on the number of incidents, not on the amount of money.

If the poster has meant the amount, they might have said "Lower income people with cash businesses cheat more than wealthy people" (which is ambiguous) or "The IRS loses more money to low income cheaters with cash businesses" (which is unambiguous, and, when you phrase it that way, truly sounds ridiculous given the high level of wealth inequality I would expect one wealthy cheater to cheat the IRS out of more money than all of the other cheaters in the world)

Yes, your interpretation is likely the correct one. In terms of societal impact, I don't think it's the important one.

Maybe it is the important one. There's a culture of cheating among many lower-income self-employed workers, and they feel they're "entitled" to it.

That may depend on what your goals are. To put myself in the IRS' shoes, their goal is to minimize the amount of revenue lost due to fraud. In that view, the important thing to do is to go after the people who are cheating them out of the most money.

If the goal is culture change, that may be different -- but the IRS is probably the wrong entity to address it.

I will say that I have not noticed that lower-income people have any greater sense of "cheating is OK" than higher-income people. My observations are not necessarily representative of the wider culture, but it looks to me like the percentage of people who think cheating is OK doesn't change based on income.

That's exactly why I said "(Of course the return for the IRS going after the small player may not be worth it.)"