July 23, 2024, 12:20 p.m.
In case you’ve been under a rock, President Biden ended his candidacy for a second term in office on Sunday and threw his support behind Vice President Kamala Harris as his replacement. Expect a flood of polls in coming days assessing his decision and Harris’s standing as the top-of-ticket candidate, as her status as successor seems like it will go uncontested.
Polls in the next few weeks will not tell us what is going to happen in the fall, and it will be important to remember that the renewed energy—and fundraising—among Democrats may not necessarily result in a suddenly changed race. Here are a few reminders of how to think about polls in the next few weeks.
We’ll start with the key reminder that polls are not precise or predictive enough in July to tell us how a close election will turn out, even under the best of circumstances. It was a misuse of polling data for those who wanted President Biden to step aside to say that the race was “unwinnable.” Data indicated an uphill climb, but 2-4 points nationally and 5-6 points in swing states in July is not “unwinnable.” Remember this if Harris has similar numbers at first.
That’s because even though we are getting to the point where polls are more predictive of outcomes in typical election years (which this is clearly not), polls will still move quite a bit—a historical analysis of polls from 1952 to 2012 shows that polls averaged 3-4 percentage points off the ultimate outcome 100 days out. My own work on 2020 polls shows that average can be wider—July 2020 polls averaged 4-6 points of variance from the eventual outcome.
Given that presidential elections in the last few cycles have turned on 2 percent or fewer of voters in a handful of states, I’ve cautioned for months not to take early election polls too seriously because a lot can happen. I didn’t exactly have a candidate change in mind, but we are living the reality of “a lot can happen.” We truly have no idea what impact the changing of candidates will have on the electorate. It’s never happened this late in the process in the era of modern polling.
What we do know is that Democrats have a divided party to put back together. The infighting has been ugly and counterproductive to the goal of mounting a campaign against former President Trump. It’s unclear how much that turmoil has affected voters, but from the data we have so far, it seems Harris will likely start with a similar deficit to Biden’s in the horse race against Trump.
Some Democrats who didn’t want Biden to step down may not immediately get in line behind a new candidate. These people will most likely come home in the fall, but in the next week or two, they might not respond to polls, or they might say they are voting third-party out of anger. Differential nonresponse—which means that different types of people choose whether to respond to polls depending on their reaction to events in the news—could work in a lot of directions. Conversely, there could be Democrats re-energized by the change who come back after previously saying they would vote third-party. It will take time for these reactions to shake out in the polls.
In the meantime, Trump is still going to command the support he has held for a year—about 46-48 percent of the electorate in a two-way poll question (i.e., Trump vs. Harris), and somewhere around 42-44 percent in a multi-way question that includes third-party and independent candidates. In all the poll wobbling we’ve witnessed over the last few months, it’s not Trump’s numbers that have changed—it’s Biden’s that have moved up and down. Obviously, Democrats hope that Harris will do better than Biden, but again, it may take some time for opinion shifts to show up in the data.
Finally, we can’t lose sight of what we know about polls in this process. Recent reporting has not been great on this dimension. There won’t be any elections to validate the data until November, making careful data collection and transparency extremely important—don’t just accept numbers from any poll. Make sure their methods make sense. Remember that margins of error exist when media outlets tell you that one candidate leads based on a 1-2-percentage-point difference. Remember that it may take time for voters to adjust to the new horse race before you dismiss a race as “unwinnable.”
Remain aware of the uncertainty—in the polls and in the environment. The past month has been pretty unbelievable and illustrates perfectly that no one knows exactly what is going to happen. It’s going to take a minute for voters to catch up.
Contributing editor Natalie Jackson is a vice president at GQR Research.
