Andrew Gelman is a statistician who is working for Columbia University. He also maintains a blog post where he shares his opinions about many topics, including the replication crisis in psychology and related fields like behavioral economics. He is not an expert in either field, but that does not prevent him from evaluating the research in these areas. But you do not have to read a specific blog post by him because the result is often the same. The research is not credible, sample sizes are too small, studies are selected for significance, and meta-analyses are not trustworthy. In his favorite area of statistics that uses prior assumptions to make sense of actual data, this is known as a dogmatic prior. No amount of data will reverse the conclusion that is already implied by a dogmatic prior. So, you really do not need data.
As you may have guessed, I don’t like the guy. I think he is a jerk, and that may cloud my evaluation of him. However, I do have data to support my claim that the Gelman’s statements often reflect his prior assumptions and are immune to data. He says so himself on his blog post.
After discussing some problems with a meta-analysis of nudging studies (a Nobel prize winning idea in behavioral economics), Gelman writes:
Just to be clear: I would not believe the results of this meta-analysis even if it did not include any of the above 12 papers, as I don’t see any good reason to trust the individual studies that went into the meta-analysis. It’s a whole literature of noisy data, small sample sizes, and selection on statistical significance, hence massive overestimates of effect sizes.
What are small sample sizes (some of these studies have hundreds of participants)? Where is the evidence that selection leads to MASSIVE overestimation. Gelman has no answers to such scientific questions about the evidence because he does not care about the data. His prior is sufficient to dismiss an entire literature, not just a few bad studies.
Did I cheery-pick this example? Should you trust me? To find an answer to these questions you can use AI that can read Gelman’s blog within seconds. Share one of his blog posts where he reversed a prior belief in response to empirical data. I am waiting.
The problem is not that Gelman is opinionated and shares his opinions on a blog (some people may say that is also true of myself). The problem is that he has blind followers that seem to confuse believing Gelman’s opinions with meta-science. Actual understanding of problems in science requires investigating these problems with empirical methods and draw conclusions from data; not believing in conclusions that rest on unproven assumptions.