So I've stumbled across this thing called Shpilkin method, a statistical model that pretends to quantity electoral fraud. I haven't found the precise model because all I get is media talking about it, there isn't even a Wikipedia article for this. Upon primary investigation I found multiple sketchy things.
- It seems that the model is based on analysing voter turnout variation, however turnout is known to be affected by other factors such as people only caring about the biggest election
- I haven't seen the model applied an compared to any other country than Russia, what if it would detect similar voter fraud in the "democratic west"?
- The media covering this is really bad, the top results are radio free europe and some neolib pro nato french news outlet.
- Mr Shpilkin apparently got rewards from some irrelevant "Liberal Front" and also got put on the list of foreign interference personalities by Russia
All I know about statistics is that you can make numbers say whatever you want. Is there anyone with education in the field that could evaluate how valid this whole thing is?
Found this on the schmuck's wilson center bio. I don't speak academic statistics, though, so I can't comment on the details of the methods.
Just skimming the paper, the core of the argument there is:
So that's it. If you assume what you need to prove, then whatever you find is statistical proof!