My mental model for the difference between the two results is based on the following:
1) the studies by Krebs and Johansen are analysis based on the "National Survey on Drug Use and Health (...), randomly selected to be representative of the adult population in the United States".
2) ACX readers population is not representative of the US population, in fact, it might be skewed in some dimensions that are very relevant here.
3) there are significant differences in the fraction of each sample that report psychedelic use
3.1) in the case of Krebs and Johansen (2013, 2015), it is ~13% reporting lifetime psychedelic use, while in the subsample of ACX readers survey considered in this report it is ~100%.
One important aspect here tying this together is that I would assume ACX readers do not have the same distribution of genes associated with intelligence as the general population, and there has been evidence that there is an overlap of those genes and the genes associated with bipolar disorder (https://www.med.uio.no/norment/english/research/news-and-events/news/2019/genetic-overlap-bipolar-disorder-intelligence.html). This genetic overlap can explain a higher susceptibility of psychotic-like experiences with higher intelligence, even if there is no particular diagnose. Furthermore, by considering the multiple types of psychotic disorder, https://pubmed.ncbi.nlm.nih.gov/17199051/ has found the prevalence in the general population to be ~3%, which does not fall too far from the 4.5% that responded a firm "yes" to the survey.
To avoid such misunderstanding I edited the original comment with "subsample of ACX readers survey considered in this report (...)"