Reuters' investigation found 5,206 cases where people were affected by judicial misconduct over a period of 12 years, and the article noted two other investigations had found 3,500 cases and 2,251 cases each. For comparison, state courts in the USA hear approximately _one hundred million_ cases per year. [1] So there's evidence for misconduct in approximately 0.0009% of cases.
(And I'm guessing most of that misconduct was nowhere near as bad as the Marquita Johnson case, or the other examples they mentioned; they picked the most shocking examples to showcase in the article.)