Polls in the U.S. have shown the presidential race between Democratic nominee Vice President Kamala Harris and her Republican rival, former President Donald Trump, in a dead heat for several weeks. Yet over the weekend, a narrative developed that voters were breaking for Harris. A poll out of Iowa by the respected pollster Ann Selzer suggested the shift was being driven by older female voters concerned about reproductive rights. This prompted a bevy of speculation as to whether it is possible pollsters might be wrong, or at least dramatically underestimating support for Harris.
As Dan Drezner summarized here, several commentators over the past few days are beginning to think that is indeed the case. One reason given by Nate Silver for the appearance of a dead heat is that, after several election cycles of embarrassing failures, polling firms may be “herding”—that is, hewing to the general consensus as to where things lie by hiding or distorting poll results that buck those trends.
Additionally, Justin Brown argues that pollsters may not have had enough time after Harris replaced President Joe Biden as the Democratic nominee to figure out how to measure the opinions of her likely voters. On top of that, over the past eight years, pollsters have compensated for failing to predict Trump’s victory in 2016 and underestimating his support in his losing bid in 2020, leaving their tools and models more sensitive to Trump voters. Combined, all of that could mean that Harris’ support is being undermeasured relative to Trump’s.