I spent some time with the paper Dose-dependent relationship between social drinking and brain aging after yet another random article referenced it with a statement like, "Studies have shown that any alcohol consumption at all is bad for you." In particular, Figure 2:
From this data, the authors drew the conclusion that the brain age gap, the gap between a person's chronological age and their brain age as estimated from a structural MRI, increases with increasing alcohol consumption. Such is the price we pay for publication. But if you look at the points in the figure, not only are they not normally distributed, which violates one of the assumptions of using a Pearson correlation for hypothesis testing in the first place, but, like any such distribution of points, those on the right have very high leverage. Those points on the right also represent as many as five drinks per day, which is much more than what I would consider social drinking.
I was curious about one drink per day, which is my typical consumption, and didn't want the results to be influenced by those points with very high leverage representing very different individuals. So, I uploaded the plot to WebPlotDigitizer, which helped me digitize all of the points in the plot above from its rasterized representation - I didn't feel like contacting the authors for their data and then waiting for them to respond and the digitizer worked perfectly - and loaded the x, y pairs into a Google Sheet. I then plotted the points in the range from zero to ninety drinks per ninety days, which is really from zero to one drink per day. Here are just those points with a new trend line:
The points in this range appear more normally distributed and the brain age gap now appears to be going down as we move from zero drinks to one drink per day. Our brains appear to get a little younger with one drink every day. Interesting, right?! It seems the authors drew the wrong conclusion entirely for individuals in this range. And I'd argue that one drink per day is more in line with the social drinking they reference in the title of their paper. But trends like the one above don't get you published. They don't make headlines. They're boring. But I guess I'm pretty into boring these days. Boring is what happens right before epiphany. Like the calm before a storm.