Jacob Jolij's experiment on whether random numbers seem meaningful

Courtesy of the SPR Facebook page - UKrant, a news website for the University of Groningen in the Netherlands, has a report on an interesting online experiment run by Jacob Jolij of that university:
https://www.ukrant.nl/parapsychological-...t/?lang=en

Participants were presented with ten sets of random numbers and had to say whether they saw anything meaningful in them - for example a date, a phone number or some pattern in the numbers. In fact half the numbers were pseudo-random (generated by a computer program) and half were truly random (generated by hardware by quantum tunnelling).

With nearly 300 participants, he found that they saw meaning in the true random numbers more often than in the pseudo-random numbers. The associated p value was 0.0013, so it wasn't a marginally significant effect.

(Then there's a particularly wrong-headed statistical commentary, saying that "if you have enough people participating in a certain experiment you’ll always end up with a significant p value," but that Jolij had then used Bayesian statistics, and that had shown "there is a difference"! Naturally, the Bayesian analysis produced two different answers, depending on assumptions ...)

The article says the experiment is still running, but when I tried it I had trouble going from page to page of random numbers:
https://rug.eu.qualtrics.com/jfe/form/SV...AG6hdImsbb

[Laird]

The article says the experiment is still running, but when I tried it I had trouble going from page to page of random numbers:

Likewise.

Likewise.

Of course, that makes one wonder whether the data acquisition and analysis work properly.

I emailed Dr Jolij to tell him about the problem, but received an automatic reply saying he was on leave until 2 August.

I tried and couldn't even get to the first question. Refreshed the page and it worked for 3 questions, then got hung up again. I had a somewhat meaningful number but was forced to refresh the page and then it was a different number. So yes, they have serious issues with the survey website functionality.

I never heard back from Dr Jolij, but - courtesy of David Metcalfe - the Dutch SPR has a presentation by him (in Dutch) at their Parapsychology Day this year about "Project Willekeur," which is presumably the same Randomness Project:

Likewise.

Of course, that makes one wonder whether the data acquisition and analysis work properly.

Especially when you consider that the PRNG is provided by the malfunctioning website.