No, the "methodology" of using a very self-selected group to represent the whole is certainly and unambiguously complete methodological bullshit. The problem is known as "confounding variables" or "common latent cause". For example, people with anxiety and depression may be more likely to join a club like Mensa.

There's certainly interesting questions to explore here, but we shouldn't settle for this extremely weak sort of study. The authors are doing the best they can, but we should take this paper as a indication that further research in this area would be fruitful, not as an accurate measurement of the nature of reality.