Yeah, there will be some incorrect papers.

But there are errors in the other direction too. How many deleterious effects will never be proven scientifically because it's incredibly exhausting to get thousands of participants and quantify their subjective wellbeing in a controlled way?