We found unhealthy pesticide levels in 20% of US produce

There's a lot of layering going on, I read both the methodology and the article written for The Guardian.

eg:

    Green beans are another example. They qualify as high-risk primarily because of a pesticide called acephate or one of its breakdown products, methamidophos. Only 4% of conventional, domestic green bean samples were positive for one or both – but their pesticide levels were often alarmingly high.

Attention is paid to the proportion of a category that is pesticide free and to the average and peak pesticide levels in those samples that test positive.

Some food categories have most samples testing positive for low levels of not very risky chemicals, other categories mostly test clean but (big but) where there are pesticides the levels are high and dangerously so.

It's done this way to get a clear picture of the risk spikes .. rather than "washing out" the risk by averaging across a large number of samples.