How Early Primary Polls Foreshadowed Surprises Like Obama’s Rise And Trump’s Win

The 2020 presidential contest is starting to pick up steam, but we’re still a long way from any votes being cast. To see if we can learn anything at this stage of the election cycle, we’ve been looking at past years’ early primary polls to see how well they predicted the eventual nominee. It turns out that they have a fair bit of value, especially when we adjust for how well known a candidate was.

This is the second part of a three-part series on early presidential primary polls. Last week, we examined polls for competitive nomination processes from 1972 to 1996, and now we’ll look at more recent contests, from 2000 to 2016. To refresh readers on how this works, we took all the polls in the FiveThirtyEight database that were conducted in the calendar year before each election and then split them by whether they were taken in the first half of that year (January through June) or the second half (July through December).

For every person who had been included in at least one of those polls — no matter how likely or unlikely they were to run or win — we calculated two key metrics. The first is an average of all the polls for each candidate (or potential candidate) for each half of the year — candidates were counted as having zero percent support in any poll they did not appear in.¹ The second is an average of each candidate’s standing in the polls adjusted for how well known the person was at the time.²

Because pollsters ask questions in different ways and ask about different lists of candidates, our methods of estimating name recognition had to be treated as rough estimates, so the candidates in each primary cycle have been sorted onto a somewhat subjective five-tier scale to sum up their level of fame.³ These name-recognition scores are represented as five square boxes in the table below, where more black boxes means higher levels of name recognition. (Because these scores can adjust a candidate’s average upward but not downward, the adjusted polling average will add up to more than 100 percent).

This time we’ll start with the 2000 presidential cycle, when both parties had crystal-clear favorites who polled well ahead of the opposition more than a year out. The GOP continued its streak of nominating its early frontrunner, who in this case was George W. Bush. The son of a former president and governor of Texas, Bush began the cycle already pretty well known, and he held on to his lead once the voting started. Notably, Bush’s nomination broke another long-running streak in the Republican primaries: nominating a previous runner-up in the next presidential cycle where there was no Republican incumbent (Bush was running for the first time). Our adjusted polling average picked up on another rising star in the first half of 1999 — Sen. John McCain, who really grabbed the national spotlight in this cycle. McCain emerged as the principal opposition to Bush and even defeated the frontrunner in the New Hampshire primary, setting up an ugly fight in the South Carolina primary, which Bush ultimately won. But like previous GOP runner-ups, McCain would return to the presidential stage.

To face Bush, the Democrats nominated Vice President Al Gore, who ended up having only one opponent — former New Jersey Sen. Bill Bradley. Bradley’s adjusted polling average made him look competitive in the first half of the year, but Gore was still polling above 50 percent in the second half, so Bradley was running out of time to cut into Gore’s lead. Bradley kept it close in New Hampshire, but Gore went on to win every contest and the nomination, making it just the second time in the modern primary era that the Democrat who led the early unadjusted polling average clinched the nomination. Only former Vice President Walter Mondale had earned this distinction previously, in 1984. In the controversial 2000 general election, Gore lost to Bush, but pollsters continued asking about the Democrat in future election cycles.

Because we’re only interested in competitive nomination processes, this means we’ll skip over the 2004 GOP primaries, when Bush ran for re-election without any serious challengers from within his party. As for the Democrats in 2004, their primary field was large and fragmented, and they did not pick the early polling frontrunner as their nominee. Instead, by the second half of 2003, six candidates had an adjusted polling average of 10 percent or more. By this metric, Massachusetts Sen. John Kerry looked somewhat stronger in the first half of the year than in the second, but he still managed to narrowly beat North Carolina Sen. John Edwards in the Iowa caucuses and win the New Hampshire primary as well. Edwards ended up joining Kerry as his vice presidential pick, but the Kerry-Edwards ticket fell to Bush and his running mate, former Secretary of Defense Dick Cheney, in the general election.

In the 2008 cycle, both parties had dynamic primaries in the race to succeed Bush in the White House. Edwards threw his hat in the ring again, but the Democratic race really developed into a titanic duel between Illinois Sen. Barack Obama and New York Sen. Hillary Clinton. Clinton was the better-known candidate thanks to her time as first lady, and she led surveys throughout 2007, but Obama’s adjusted polling average underscored early on that he might have potential as a rising star in the party. And so despite Clinton’s lead in the national polls heading into the first primaries of 2008, she finished third in the Iowa caucuses behind Obama and Edwards. She recovered to win in New Hampshire, but in the end, Obama and Clinton battled all the way until June, the end of the primary season, before Obama clinched the Democratic nomination.

In the GOP primary, McCain renewed the Republican tradition of nominating a previous runner-up. But McCain’s win bucked another trend — he was the first GOP nominee since 1972 who didn’t lead early national polls the year before a primary. In the first half of 2007, former New York Mayor Rudy Giuliani topped national surveys, but former Tennessee Sen. Fred Thompson (also of “Law & Order” fame) matched him in the adjusted polling average in the second half of the year.

McCain was probably helped by the fact that the field was crowded and some of his leading opponents made pretty big mistakes. Giuliani chose to essentially skip the early states to focus on Florida as a springboard to win other big states on Super Tuesday — a strategy that failed. And Thompson waited too long to declare his candidacy, which was then hampered by the rise of former Arkansas Gov. Mike Huckabee, who became the stronger Southern candidate in the field. McCain got off to a slow start in Iowa but edged out former Massachusetts Gov. Mitt Romney in New Hampshire to get on track. He then won narrow pluralities in South Carolina and Florida, fending off Huckabee’s continued challenge in order to win the overall nomination. In the end, McCain lost decisively to Obama in the general election.

In 2012, the GOP nominated Romney — who had been one of the runners-up in the 2008 campaign — to take on Obama. This marked a return to the tradition of the early polling frontrunner eventually winning the party’s nomination. But Romney was probably also helped by Huckabee, who was also a runner-up in 2008, deciding not to run. And then there was Texas Gov. Rick Perry’s decision to defend a Texas policy that allowed immigrants who had entered the country illegally to pay lower in-state rates for college tuition, which hurt him in the polls. Other candidates, such as former House Speaker Newt Gingrich and businessman Herman Cain, had periodic booms in the polls, but Romney was the last one standing, though he did lose to Obama in the general election.

Now for the most recent presidential primary cycle. In at least some respects, the 2016 primary on the Democratic side echoed the party’s primary in 2000, when Gore led the early surveys at more than 50 percent and faced only one significant opponent. Hillary Clinton was the early polling frontrunner in the first half of 2015, when her unadjusted polling average was at 67 percent, the highest mark for any half-year period we looked at from 1972 to 2016. She slipped slightly in the second half of the year, but still led at 61 percent. However, there was evidence in the early polls that despite not being as well known as Clinton, Vermont Sen. Bernie Sanders had potential. His adjusted polling average stood at about 24 percent in the first half of the year, then shot up to 44 percent in the second half. Ultimately, Clinton had to fend off Sanders all the way to the end of primary season, just like in 2008 when she was up against Obama, except this time she won.

But the bigger story in the 2016 cycle was arguably the rise of now-President Donald Trump in the GOP primary field. Prior to his campaign announcement in June 2015, Trump was polling in the low single digits. But the 2016 Republican field was arguably the most crowded one ever, and there was no clear frontrunner. In the first half of the year, seven (!) candidates were polling at more than 10 percent in the adjusted polling average, with Wisconsin Gov. Scott Walker and former Florida Gov. Jeb Bush leading the way. But in the second half of 2015, Trump shot up to take the lead while candidates such as Bush and Walker fell sharply. Trump’s increase from an average of 3 percent in the first half of 2015 to about 29 percent in the second half — using either the regular or adjusted polling average — represents the largest increase from the first half of the year to the second half for any candidate in the 1972 to 2016 period. And once the primary voting started in 2016, Trump consistently won pluralities in most of the early contests, which positioned him to withstand efforts among some in the GOP to stop him from winning the party’s nomination. He then, of course, went on to defeat Clinton in the general election.

So while aspects of Trump’s victory were certainly unprecedented, he did lead in primary polls during the second half of 2015, which means that him winning the GOP nomination maybe shouldn’t have come as such a surprise. In fact, McCain was the only Republican during the 1972-to-2016 period who won the nomination without ever having led in the national polls in the year before the primary. But it’s hard to apply this logic to the 2020 Democratic primary, as Democrats have historically been far less likely to nominate the early poll frontrunner during the modern primary era.

So now we’ve gone through every cycle from 1972 to 2016 and found that the early polling leader often went on to win a party’s nomination, and the early polls, once they were adjusted for name recognition, often foreshadowed the rise of notable candidates. In the final part of our series, we’ll move beyond the descriptive and dive deep into the trends that emerge from the entire 1972 to 2016 period, drawing some statistical conclusions about how meaningful early primary polls really are.