


Opinion polling has had some hard times in recent years — at least, judging by election polls; when issue polls are wrong, there’s no way to prove it — but even for all of their failures in close races, polls can still give a pretty solid read on the range of possibilities in a race, at least if the race has been polled multiple times by competent pollsters. But pollsters have been fighting a constant rearguard action against changes in both technology and poll-respondent habits (as well as demographic shifts within the parties) that have made their jobs harder. Patrick Ruffini warns that the iPhone is about to make that much more challenging:
iOS 26 will move calls and texts outside of a sender’s close contacts to an unknown senders tab. Many will welcome this as a way to fight spam. But a big unintended consequence of this will be to make it harder for pollsters to reach a representative sample of Americans. Polling has been moving online for a while now, but for state and local polls, calling and texting the broadest possible segment of the public is often the only way to get a representative sample. . . .
Online polls are largely opt-in and aren’t always feasible at the local level. The polling industry has adapted to the death of landlines first by moving to cell phone interviewing and more recently to texting. . . . Hiding these calls and texts may render these sorts of polls much more expensive, out of reach for all but the most wealthy and powerful institutions. That’s the best case scenario. The worst case is that they become impossible to do. . . . By design, polls must invite the broadest possible cross-section of the public to participate. Strict opt-in requirements are antithetical to this. . . . Polling will continue to evolve and embrace new technology. Text-to-web polling has been a key part of that over the last few cycles. And it’s important to understanding opinion at the local level, especially with local journalism in decline.
Ruffini, of course, argues that Apple and the law should make it easier to do polls, on the theory of “the vital role polling plays in our democracy.” While I have a vested interest as a political writer in good polling, I’m not entirely convinced that polls really do more good than harm for our system, in which we survived for a long time before Gallup invented modern random-sample polling in 1936. Of course, politicians had used straw polls before that, and at least the concept of a random sample wasn’t new. Consider this July 1858 instruction by Abraham Lincoln, during his Senate contest with Stephen Douglas, to Joseph Gillespie (a Lincoln ally in a key district) on how to random-sample the sentiment of “American” voters (i.e., supporters of the Know-Nothings in the previous election): “Make a little test. Run down one of the poll-books of the Edwardsville precinct, and take the first hundred known American names. Then quietly ascertain how many of them are actually going for Douglas.” Still, the basic idea of American democracy was that public sentiment would be registered in regular elections. Somewhere in the past two decades, we got trapped in a cycle of polls shaping elections by shaping expectations, rather than simply measuring what was happening — or worse, helping stoke conspiracy theories when results diverged from expectations. That is not healthy.