KEY POINTS FROM THIS ARTICLE - We have no idea how much uncertainty exists in pre-election polls. That leads to very complex poll-based forecast model...
http://centerforpolitics.org/crystalball/articles/poll-based-election-forecasts-will-always-struggle-with-uncertainty/
* gave Hillary Clinton a 98% chance of winning
* polling error is generally larger than the reported margin of error
* when nearly everyone had home telephones
* random sampling error
* weights
* which does result in some biases
* coverage error
* 78% of the population that relies mostly or completely on cell phones
* A handful use “probability” methods
* online “nonprobability” polls
* vary greatly in quality
* likely voters
* can easily result in different estimates, depending on the judgment of the decision-makers
* potential changes in how we vote
* it cannot be accounted for in the data
* the model could be adjusted
* poll variance would have to be set extremely high
* Morris
* acknowledged that the model is a work in progress
* adjusted
* adjustment
* will have some data-based ways to adjust for the pandemic
* forcing a lot of ad-hoc decisions for election modelers.
* Gelman states
* My own experience
* Michigan
* Pennsylvania
* Wisconsin
* Most people don’t have a solid understanding of how probability works
* I tried
* most
* people
* will misunderstand them
* will make incorrect assumptions about what they mean
* will
* filter them through
* their own biases
* as Morris notes
* work hard to communicate model outcomes well
* research shows