Anyway, it's definitely not true in this case. I notice the very unusual instance of people drinking soda because it's so rare these days (among my extended social circle). In some cases I _hear_ about it before I meet the person ("my bf drinks like a liter of soda a day" or whatever) -- and then meet them and, unsurprisingly, they're weirdly skinny or fat.
Overweight person A drinks diet soda because they struggle with losing weight and use it to avoid drinking calories.
Underweight person B drinks diet soda because they have body image issues and/or a mild-to-severe ED and are afraid to gain weight.
Other things might influence to a degree, but the main problem is always going to be caloric in nature.
But my "counter-evidence" would simply be me saying "Well, I don't see that". To which you would respond that it was actually I who wasn't being observant. When there is no real way to determine that. And that discussion itself is intellectually bankrupt.
My pointing out that your recollection of casual observations and your self-assessment of how well you "took the bias into consideration" is debate. I'm questioning the source of your statistics.
Because even in this study, it's from 120 people. Total. Who self-reported they had never had artificial sweeteners.
And the other obvious thing is that you are also free to ignore my opinions. I like to mention when someone is offering biased anecdotes in place of substantive discussion. In case it resonates.
I'm not claiming you should draw any universal conclusions from my anecdotes! That's absurd. But there has got to be something between 'data is from a study' and 'data is meaningless'.
The one claim I have made that you should believe, though, is that I (and others, evidently) was expecting the surprise to be that science found artificial sweeteners to have no effect, since my prior for it mattering was so high. I was surprised that this result was a surprise to anyone.
It's also a lot like the "bad toupee" effect. You think you're good at spotting toupees because you always notice when they're bad. But you don't notice when they're good, therefore they don't even factor into your dataset.
Which is why anecdotes are never data. There's no way to control the data coming in. And the filters we have on the data coming in are sometimes unknown.
I've never really seen a study on artificial sweeteners that ever propose an actual mechanism. And without fail, the study is done on rodents or rely on self-reporting to some degree or on very small sample sizes.
I don't have to present my own observations to point out flaws in your own observations. It's not a contest where "most observations" win or whatever.