Look closer, and you’ll notice the big errors were always there under the hood. It’s just that we never really focused on them.
Polling analysis
Why Theories of Polling Bias Don’t Make Sense
Everyone who works on polling and polling analysis is actively incentivized to forecast election results as accurately as possible; and even if they weren’t, influencing public opinion isn’t as easy as it may seem.
Read more…Read more…
Addendum: Pollster Herding and Sample Bias
If polls herd as much as they did in 2019, weighting by education isn’t necessarily going to save them.
Read more…Read more…
Examining potential causes of the 2019 Australian federal polling error
Using a model of pollster herding and sampling bias, we examine hypotheses for the polling error seen in the 2019 Australian federal election.
State polling has historically skewed to incumbents. Will that continue?
Historically, state voting intention polls have tended to show better results for the party in power than the election-day result. However, recent trends suggest the possibility that this pattern will not continue.
Read more…Read more…