We have to be careful with benford's law. It's a math law, not a US law. I think it's supplementary, but I'm not so sure it would stand alone as evidence. But I'm not a lawyer.
Yeah but you need more than a couple of Excel bar charts by someone who clearly has no idea what they're doing.
First you need to demonstrate that your dataset actually should follow Benford's law. Generally this is only the case when you're working with data that is spread across many orders of magnitude. So populations of countries roughly follow Benford's law, but people's heights don't. With election data it depends what kinds of districts you're looking at - if they're all around the same size then the vote counts shouldn't follow Benford's law at all.
Secondly you need to do a statistical test to work out the likelihood that the deviation from Benford's law could have happened by chance. You can't just eyeball it, as the expected deviations depend on the size of the dataset. For example, there are only a couple hundred countries, so the deviations of national populations from Benford's law are actually fairly big.
This. Ward vote counts don’t scale enough orders of magnitude to expect benfords law to hold, Trumps actually do, but just because he consistently got less than 100 votes in some wards so he actually did cross a few orders of magnitude. (Nearly) All of Biden’s ward counts are localized between 100 and 1000, hence a grouping about a leading 5.
We have to be careful with benford's law. It's a math law, not a US law. I think it's supplementary, but I'm not so sure it would stand alone as evidence. But I'm not a lawyer.
CPA family member tells me it’s a real thing, is admissible in court, strong indication of fraud. Is used routinely in auditing.
Yeah but you need more than a couple of Excel bar charts by someone who clearly has no idea what they're doing.
First you need to demonstrate that your dataset actually should follow Benford's law. Generally this is only the case when you're working with data that is spread across many orders of magnitude. So populations of countries roughly follow Benford's law, but people's heights don't. With election data it depends what kinds of districts you're looking at - if they're all around the same size then the vote counts shouldn't follow Benford's law at all.
Secondly you need to do a statistical test to work out the likelihood that the deviation from Benford's law could have happened by chance. You can't just eyeball it, as the expected deviations depend on the size of the dataset. For example, there are only a couple hundred countries, so the deviations of national populations from Benford's law are actually fairly big.
This. Ward vote counts don’t scale enough orders of magnitude to expect benfords law to hold, Trumps actually do, but just because he consistently got less than 100 votes in some wards so he actually did cross a few orders of magnitude. (Nearly) All of Biden’s ward counts are localized between 100 and 1000, hence a grouping about a leading 5.