Last week, a new Reuters/Ipsos poll showed that Hillary Clinton and Donald Trump had drawn almost even in a general election match-up. The gap between the two candidates had dropped to just one point.
Headlines blared that the race for the White House was a toss-up. Clinton and Trump were neck-and-neck. Much ink was spilled.
The next day, Reuters/Ipsos was back in the field with their five-day rolling poll. This time, the gap between the two candidates had widened again to four points, a more conventional margin. The poll went mostly unnoticed.
- Unique set of problems dog Donald Trump's campaign
- Donald Trump doubtful he'd have a good working relationship with British PM
Welcome to the fevered coverage of public opinion polling in the U.S. presidential election, which will culminate a mere 173 days from now. Expect polls which show a competitive race to get outsized attention compared to their duller counterparts.
But has the race become more competitive? The landscape has certainly changed since May 3, when Donald Trump became the presumptive Republican presidential nominee after his victory in Indiana, bumping Texas Senator Ted Cruz and Ohio Gov. John Kasich out of the contest.
There is some indication that it has indeed become more competitive — but perhaps not to the extent some recent headlines would suggest.
Clinton still leads tighter race
General election match-up polls have limited value this early in a presidential campaign. But the 2016 race breaks the mould in many ways, including the notoriety of the two candidates likely to be on the ballot. This has the potential to make general election polls somewhat more informative than is normally the case.
(While Trump is the last man standing for the Republican nomination, Clinton is still battling Vermont Senator Bernie Sanders for her party's title. But her lead among delegates puts her virtually out of range of being caught by Sanders.)
Clinton has been leading Trump in the vast majority of polls conducted since the real estate mogul and reality television star launched his unlikely presidential bid last year. That lead yawned to its widest extent (since voters started casting ballots in the presidential primaries) around the end of March and early April, when poll aggregators Real Clear Politics and HuffPost Pollster both put it at 11 points.
That gap has certainly since been reduced.
In polls conducted partially or entirely after May 3, Clinton has averaged 45 per cent support against Trump's 41 per cent. Clinton's lead in these polls has been estimated at between one and six points.
But it is important to compare apples to apples. When these same polling firms were last in the field at Trump's recent nadir, they pegged the gap between Clinton and Trump at eight points. So while the margin between the two has certainly narrowed, it has not dropped as significantly as a comparison to the wider aggregate would suggest.
And Trump's victory in the Republican nomination may have only given him a modest bump. When the polling firms that have been in the field since Indiana were last in the field, before the state all-but-clinched Trump's nod, he was trailing Clinton by an average of a little more than five points.
It is worth noting as well that Clinton's polling lead, while smaller than it was when Trump's numbers had dropped a little over a month ago, is quite typical. From September to early March it was often four points or less.
That four-point lead is wider than the one Barack Obama enjoyed at this stage of the 2008 and 2012 races against Republicans John McCain and Mitt Romney, which he ultimately won by seven and four points respectively. And on average, the last four U.S. presidential elections have been decided by 3.5 points.
Lessons from the primaries
If there is one lesson that many American pundits have been happy to learn from the presidential primaries, it is that all of the prognosticators, predictors and data journalists missed Trump's rise — and so the conclusion must be that the data nerds cannot be trusted.
But this is a misreading of what actually happened.
Trump's unexpected victory happened in part because the data that was most recent was largely dismissed, while the unsuccessful historical record of candidates like Trump was given too much weight.
It was not an unreasonable assumption to make. But that data set only included a handful of U.S. elections. The polls, on the other hand, had given Trump the Republican win almost from the beginning.
- The Pollcast: Subscribe to Éric Grenier's podcast
- Analysis: Partisan interests difficult to avoid in electoral reform debate
To dismiss the polls now is to make the mistake that the data journalists made in early primary season. In those polls, Clinton has beaten Trump in the general election match-up as frequently as Trump was besting his Republican rivals.
That doesn't mean she will win on Nov. 8. Or that because the polls were mostly right during the primaries that we can assume they will be mostly right after the conventions. But the betting odds are still in Clinton's favour. If a case for a Trump win in November is made, the data today can't quite support it.
It does, however, support the race getting more competitive than it was a few weeks ago.
But is it neck-and-neck? Probably not — yet.
Get CBC Politics video, news and analysis in your Facebook newsfeed: "Like" our page here.
A previous version of this story stated that Barack Obama defeated Republicans John McCain and Mitt Romney by four and seven points, respectively, in 2008 and 2012. In fact, McCain was defeated by seven points and Romney by four.May 18, 2016 12:42 PM ET