OK, so the elephant in the room here is this: you cannot confirm a null hypothesis (in this case: "Ritalin does not work in the long term") by failing to falsif

OK, so the elephant in the room here is this: you cannot confirm a null hypothesis (in this case: "Ritalin does not work in the long term") by failing to falsify it. This study is simply not designed such that it can provide strong evidence that ritalin is ineffective in the long term. It can only fail to provide strong evidence that it is effective in the long term. Furthermore, the author refers to it as a "well-controlled study". It's not. The original 14-month study was controlled, and it showed short-term improvement. This study is a prospective follow-up. It's not a controlled trial. It's not trying to be a controlled trial. Finally, looking at the full text, they appear to have used linear modeling. That's also a red flag. Obviously, sometimes a linear model is the right choice. But most often it is used because it is the convenient choice.