Leaderboard

Actually one of the most problematic areas of designing a statistical experiment like a hypothesis test is the sample space. In other words, do the people or examples I decide to take a sample of and study accurately represent a statistical sample of the entire population that I am expecting my hypothesis to apply to? In that respect, tongues in a natural setting where the speaker was unaware they are being recorded are probably the cleanest. Depending on how you want to model the study, you might divide the samples into "church" settings where there also was interpretation / prophecy going on for all to hear, and possibly private prayer life examples too and run the statistics on both sets to see if they present any differences mathematically. I noted a couple problematic inclusions in Samarin's sample space - both of which were the recording of a medium's conversation with their spirit guide, where there was a different language involved. His inclusion of Christian and non-Christian groups is fine, especially as he is endeavoring to illustrate a non-detectable difference there. The mediums were in his non-Christian samples. I don't have access to his other resources to be able to vet other non-Christian samples. I understand what Samarin was doing there - simply collecting any available samples of known recordings of people speaking in a language they never learned. With the medium examples, though, there is never a claim of glossa going on there, simply a recorded conversation with a spirit guide. So they should be excluded from any hypothesis test. You talked to this author and confirmed this? Or is this just an ad-hominem attack?