Polling questions answered
November 8, 2012 Leave a comment
I had meant to do a post before the election about all the questions about polling that were going to be answered by this election. Like much in life, ran out of time. That said, we can come to a number of important conclusions from this chart of pollster performance via Ezra:
Well, what can you say about Raleigh’s own PPP. Mighty impressive. Especially since they are a robo-poller that thus cannot call cell phones. And I’ve already mentioned Rasumussen. Okay, then…
1) Having live telephone interviewers and cell phones in the sample is clearly not essential to accurate election polling.
2) Part of that phrasing above refers to the fact that rigorously-conducted on-line polls, notably YouGov and Ipsos/Reuters also fared quite well.
3) Polling is definitely an art as well as a science and it pays to have a good artist. Tom Jensen at PPP is definitely a great artist. Scott Rasmussen not so much.
4) Given the cost advantages and overall good performance in 2012, look for even more on-line polling in the future.
5) I’d love it if someone would take the time to compute the average error by mode of survey to see if there really are systematic differences.