Agreed. That’s precisely why I’ve made sure the analysis is limited to countries with more than 10,000 cases. And also why there are two separate analysis: one for total cases and the other for new cases. And last but not least, that’s also why I didn’t analyze the conformance to Benford’s law of each individual country — but instead analyzed how countries compared.

You’re right. Benford’s law can’t apply to every dataset. Except here, I’m pretty sure it does — even for each individual country.

90%+ of countries do have a distribution that seems to conform pretty well to Benford’s law (see the charts for yourself in the app). There is no logic leap here, as far as I’m concerned.