Wandering Adventure Party

We used the search term “ADHD AND acetaminophen.”

That was how the studies were selected… Lol. Real robust “research” there, guys.

Jeez. What a thing to complain about. I wish we had a dishwasher.

Search for studies containing links between ADHD and Tylenol to determine if there’s a link between ADHD and Tylenol. P-hacking much? That is straight-up cherry-picking results to fit the hypothesis.

No, it doesn’t. It returns studies that contain Tylenol AND ADHD. There’s an immediate bias in favour of the hypothesis. They should be searched separately, then you would look at how many contain both, then look at how many correlate the two. Presenting only the data that correlates the two is presenting that data out of context: choosing the data to fit the hypothesis. P-hacking.

The media has done the same thing with climate change. They present for debate one scientist ‘for’ climate change and one scientists ‘against’ climate change as though there is a 50/50 chance that climate change is real, despite 99% of scientists falling on the ‘for’ side. A balanced debate would have involved 100 climate scientists with ‘1’ against and 99 ‘for’. Instead we now have people who think that climate change denial is reasonable because the data was presented in an unbalanced, or biased, way.

If you only present that data that you think is relevant then you bias the result in your favour. If the data for all studies investigating the cause of ADHD was included, and then the % including Tylenol, then the % correlating Tylenol with ADHD, you would have a very different number… A much more honest one.

Yeah… That climate change example was a bit of a stretch. I was just highlighting how easy it is to mislead people with part of the picture, rather than the whole ugly mess.

I still think that omitting studies into the cause of ADHD that don’t include Tylenol is misrepresenting the data.

If there are 1000 studies into the cause of ADHD, and only 50 mention Tylenol, then omitting the other 950 is dishonest. Let’s say 25 of the 50 find a correlation, then 25/50 is way different to 25/1000! That’s where I see the P-hacking.

Thanks for being civil, too.

Is there a non-paywall version of that link?