He used voter data over the last 22 years, to track how men, women, and U voted across all repub, mostly repub, half and halfs, mostly dem, all dem.
Basically like taking a 3 number combination lock, and trying every single combination.
So based on total number of registered voters, at any given age group, for male, or female, he could tell you the basic percentage of them that would fall into any of the 5 political buckets, and he ran that across every variable to come up with the 980,000 buckets, and he had 22 years of voter data to use. And things started going sideways in 2008.
He used voter data over the last 22 years, to track how men, women, and U voted across all repub, mostly repub, half and halfs, mostly dem, all dem.
Basically like taking a 3 number combination lock, and trying every single combination.
So based on total number of registered voters, at any given age group, for male, or female, he could tell you the basic percentage of them that would fall into any of the 5 political buckets, and he ran that across every variable to come up with the 980,000 buckets, and he had 22 years of voter data to use. And things started going sideways in 2008.
At least that is how I understood it.
What voter data exactly? Last I checked this country votes via secret ballot. Did he use polling data? That is very different.
Check Arizona, specifically in this instsnce, and other states if curious about them.
https://www.ncsl.org/research/elections-and-campaigns/access-to-and-use-of-voter-registration-lists.aspx