Thom Volker
thomvolker.bsky.social
Thom Volker
@thomvolker.bsky.social
PhD Candidate in Statistics, Utrecht University

Creates fake data for a living.

thomvolker.github.io
Reposted by Thom Volker
But if you have N=100,000 (not at all unlikely with NHANES/biobank-type data), absolutely any noise will be significant at 0.05, so it's a ludicrously large value.
November 22, 2025 at 11:52 PM
My rule of thumb is something like: If you don't know them and need something from them, it's better to be on the safe side. If you have interacted before, infer it from their communication style. If you end your emails with "Cheers, Dan", chances are I start my next email with "Hi Dan"
November 19, 2025 at 8:30 PM
Let the man be on whatever board he wants, as long as he doesn’t interact with students (or others who are dependent on him)
November 19, 2025 at 7:19 PM
Naast het humanitaire aspect (veruit het belangrijkst) is het in alle andere opzichten ook idioot. Ze kunnen de boetes niet betalen, worden vervolgens vastgezet (a 250 euro per dag) en als ze hun leven enigszins op de rit krijgen, hebben ze een schuld waar je u tegen zegt waarna ze bij nul beginnen.
November 19, 2025 at 7:17 PM
Unless I miss the point of their paper entirely, Vos & Holbert (doi.org/10.1007/s112...) argue that the "hypothetical sampling" view of frequentism is not required, but can be viewed as chance outcomes with certain proportions. This seems to alleviate these CI-interpretation issues as well.
Frequentist statistical inference without repeated sampling - Synthese
Frequentist inference typically is described in terms of hypothetical repeated sampling but there are advantages to an interpretation that uses a single random sample. Contemporary examples are given ...
doi.org
November 19, 2025 at 4:08 PM
Eerdmans nadat hij een dag de kachel uit liet: "Zie je wel, klimaatmaatregelen helpen geen steek"
November 14, 2025 at 12:06 PM
November 14, 2025 at 10:30 AM