I understood enough that this study found a correlation and that this was based on surveys. I thought it was an interesting finding, and concluded that this correlation should be examined more closely with more rigorous studies.
I did not go into the details of methodology and statistics and did not conclude, like you did, that this study has dubious value.
This is a trap that the public find themselves in with science reporting. Many people on HN have technical training to grasp these concepts but not understand them. I my self program, and use many of the same intellectual building blocks scientists use in the execution of my job. But I am not a scientist.
I am not a scientist is the key point because it means that I do not understand science. I know the “process” of science. I’ve read scientific papers. I’ve done toy experiments in school and in university. But I don’t understand it. To draw an analogy, Programming is a perception of reality. There are things that I do that I can never explain to management because they do not have direct experience with it. The “identity a bird in the park” XKCD comic is a meme of this concept [1] notwithstanding advances in AI research.
Like programming, science is a perception of reality. Like my management, I may have taken statistics, I know what confidence intervals are. But I have not lived the experience of building an experiment. Getting results, analyzing them, and constructing the distinctions necessary to reach an interesting and valid conclusion. If you’ve gone though that process, you know where to look for problems in a study. I and many people don’t. We will at best say, further study is needed, and at worst say that diet soda causes autism.
The public depends on experts to enter the conversation and share why things are wrong. This of course gets into the problem of “lies will make it half way around the world while the truth is still putting on its shoes”. Retractions may be made, but never perceived. This makes us vulnerable to bad faith actors employing the gish gallop and there’s not a general purpose solution to that.