Dem
51
GOP
49
We're not sure if there was ever a golden age of polling, but we are absolutely sure it is not now. As noted above, a quarter of the electorate hates both candidates. No one knows what they will do. Hold their noses and pick the least bad one? Vote third-party? Stay home and listen to soothing music on Nov. 5? With such a large fraction of the electorate angry, pollsters may have a rough year.
In 2016, there were a lot of shy Trump voters—people who were embarrassed to tell the pollsters that they liked Trump and wouldn't talk to them. As a consequence, pollsters oversampled college-educated voters, who were becoming Democrats in droves. The polling results in 2020 also were not good. The error was the worst in 40 years. The predicted "red wave" didn't happen in 2022. Is polling dead?
Gee, we hope not. Pollsters know all these things and are trying to adjust. One big change is that no serious pollster uses phone calls to landlines exclusively anymore. All pollsters call landlines and cell phones at a bare minimum. In any event, random digit dialing is deader than the proverbial dodo. Many pollsters now buy lists of people and randomly select from the lists. The best lists are lists of registered voters, but they are not available in every state.
One key development is the advent of multimodal polling. Some pollsters are calling people, sending them text messages, and contacting them online by various means. They then compare the results of, say, talking to people on the phone with polling via text messages. By comparing the different modes, they are trying to correct for systemic errors in any one mode. And the underrepresentation of noncollege voters has certainly been fixed.
One big remaining problem is trying to determine who is a likely voter. This is important because Donald Trump does better with marginal voters than does Joe Biden. How much correction needs to be applied to deal with this?
Another big problem is "message-sending." Some young voters are angry with Joe Biden over Gaza and are happy to tell pollsters that in no uncertain terms. But in November, many of them are surely going to bite their tongues and vote for Biden anyway because deep down, they know Trump will be much worse on Gaza and given a choice between "bad" and "truly evil," "bad" may win.
So far, the only thing we truly believe we have learned from the polling is that the "northern route" will be easier for Biden than the "southern route." Also, we believe that of the southern swing states, Nevada and Arizona will be easier for Biden than Georgia and North Carolina. But as we keep pointing out, it's only June and, well, stuff happens. (V)