#TrumpWon the debate? Legitimate polls show otherwise

The hashtag #TrumpWon was trending on Twitter Tuesday. Partially boosted by sarcasm, it was fuelled primarily by Donald Trump, who was quick to claim he won Monday's debate by citing a slew of online polls that pegged him as the winner.

But there's a problem. None of the polls Trump has pointed to are actually legitimate polls. Instead, all of the scientific, real polls published so far have shown him to be the loser of the debate by significant margins.

• Keith Boag: The 'change' candidate can't change his ways in 1st debate
• Presidential Poll Tracker: Check out the latest projections

In the world of public opinion research, the "online polls" that litter news websites are a plague. Whereas real surveys try to assemble representative samples of the population, these online polls are more of a gimmick or a game. Anyone can answer them. Often people can answer them multiple times. They can invite their friends, who likely think just like them, to answer the online polls on social media.

And then they can point to these completely unrepresentative and meaningless results as if they signal something important.

At their very best, these online polls are only a sample of a given website's audience — at least those who see the online poll while it is posted and are interested enough to participate. It is no coincidence that two of the online polls Trump has quoted come from conservative websites like Drudge and Breitbart (the latter run by Trump's campaign CEO).

These online polls are not the same as the legitimate surveys conducted by pollsters that sample from a demographically representative internet panel. And they are certainly not the same as the legitimate surveys that randomly sample from the general population via telephone.

These legitimate polls also differ in their results. Four scientific polls published since the debate using legitimate surveying methods have shown that debate-watchers saw Hillary Clinton as the better performer and the winner of Monday's debate:

• A CNN/ORC poll of 521 debate-watchers taken the night of the debate found that 62 per cent thought Clinton had done the best job, compared to 27 per cent for Trump.
• A PublicPolicyPolling survey of 1,002 debate-watchers also taken the night of the debate showed 51 per cent of viewers thought Clinton had won, while 40 per cent gave the nod to Trump.
• A POLITICO/Morning Consult poll of 1,253 likely voters on Sept. 26 and 27 found that 49 per cent thought Clinton had won, while 26 per cent thought Trump did.
• A YouGov poll of 1,145 Americans conducted on Sept. 26 and 27 found that 57 per cent of those who viewed the debate thought Clinton won, against 30 per cent for Trump.

Nevertheless, some of these legitimate polls still come with some caveats.

The Morning Consult poll, for instance, included voters who did not watch the debate.

The CNN/ORC poll has been criticized for having a slightly larger sample of Democrats than most surveys normally do. But the disproportionate presence of Democrats in the sample was not significant enough to put much of a dent in Clinton's 35-point lead over Trump. And it is impossible to know for certain whether or not the debate audience was disproportionately Democratic in the first place, or that the debate itself pushed more Americans to self-identify as Democrats.

Still, the evidence gleaned from the scientific surveys shows that Clinton won the debate by a wide margin. The last time CNN recorded such a big win in a debate was in 2012, when Mitt Romney bested a tired-looking Barack Obama in that campaign's first tilt. 

Romney used that performance to close the gap on his rival, though his surge did not endure.

Debate impact unknown, for now

We won't know for certain what short-term impact the debate will have on the race for a few more days. A few surveys on voting intentions will trickle out for the rest of the week, but clear trend lines won't be clear until after the weekend. 

The legitimate polls of debate-watchers, despite the clear victory for Clinton, do not necessarily suggest that she is likely to make significant gains. In a two-point race, though, she will gladly take insignificant gains.

On balance, the legitimate polls suggest that the debate may have made Americans slightly more likely to vote for Clinton than for Trump, though most of the enthusiasm for each candidate came from people who were already planning to vote for them. Nevertheless, both the CNN/ORC and PPP polls found that Clinton had done better among Independents than Trump did.

Additionally, the scientific polls found that Republicans were more likely to say Clinton had won than Democrats were to give the debate to Trump.

So Trump may have been scraping the bottom of the barrel in the public opinion world to find these online polls that suggested he had the better night. But that may not be the point. Most voters are unlikely to differentiate much between legitimate and illegitimate polls.

In the end, Trump's tweets and statements about the "online polls" that have shown him as the winner may not be for the consumption of the broader public, but instead to keep his own supporters motivated and optimistic about his campaign. In a post-truth political world, that may be good enough.