I'm aware of all the criticisms of the Kanin study. None of them seem valid.
If the USA believes polygraphs are unreliable they should be banned from all usage, for all types of crimes and government screening. That hasn't happened.
Instead, people who didn't like Kanin's conclusions - and noted that many rape accusers prefer to drop their accusations rather than take a polygraph test - decided it should be banned for that specific crime only. This is not internally consistent but is consistent with the long-term trend towards trying to increase the number of rape reports.
The critics then retroactively claimed the usage of this technique made the entire study invalid, which it does not do. These people are actually trying to subtly argue something different and much more extreme - that US police statistics and records for rape accusations in that entire period are completely worthless due to the use of polygraphs and therefore no conclusions can be drawn about the accuracy of rape reports during this time at all. This also isn't the case. Attrition between rape reports and prosecution is extremely high in other western countries that don't use lie detectors, and there is attrition at every stage of the pipeline. Offering to use a polygraph and having some complainants decide to drop their cooperation entirely is just one source of attrition amongst many.
The Kanin study is utterly damning on many levels and aligns with the self-reported perceptions of police (that false accusations are very common), not just in that study town but across the USA and in other countries as well, like the UK.
> The study mostly used the investigator's judgement
That is not correct. The study only marked a rape complaint as false if the accuser stated it was false. Cases where the investigator concluded it was false were not counted.
This standard is clearly far too strict - women who made provably false claims and steadfastly refuse to ever admit it would not be counted in this statistic, thus the 40% figure for that study must be too low when using a more reasonable definition of false claim.
> Question: if you go to 100 actual false accusers ten years after that incident, and ask them "Did you falsely accuse X of rape back then?". Do you really believe that 40% would be stupid enough to tell you?
It might be worth reading the original study. It is here:
https://ia800209.us.archive.org/4/items/FalseRapeAllegations...
The answer is yes, 41% said they lied. I don't know where your 10 year requirement came from. Most cases were resolved as false far quicker than that, usually because inconsistencies in the stories surfaced very quickly. As you can see from the case files the study quotes directly, admission that the story was false is usually accompanied by an explanation of why they did it and the study shows almost all false accusations were motivated by just three reasons.
I don't know why you think this requires stupidity. Prosecution for making false rape claims nearly never happens. It is extremely safe to admit you lied. But irritating the police makes it much less safe. It makes more sense to admit you lied quickly than to drag it out and risk the police deciding to prosecute for perversion of justice.