In a scientific online snap poll from YouGov, Clinton won a relatively narrow victory, with 47 percent of registered voters who watched the debate saying she did a better job, 42 percent saying Donald Trump did, and 12 percent saying it was a tie.
Undecided voters favored Clinton by a 3-point margin, the pollsters said, with female viewers saying by a 12-point margin that Clinton won and men saying by a 3-point margin that Trump won.
A majority of debate-watchers polled by YouGov said that Clinton was better-prepared, more knowledgeable and more presidential, while Trump won top marks only for being more negative and for interrupting more often.
Clinton scored a bigger win in a snap poll from CNN/ORC, which recontacted people who said in a past poll that they planned to watch the debate. Fifty-seven percent of the debate-watchers polled said that Clinton won, while 34 percent favored Trump.
Clinton’s advantage was in part due to the fact that those tuning into the debate are more likely to be Democrats than the nation as a whole, as CNN’s Jake Tapper noted.
Both snap polls, in fact, suggest that few minds were changed by the debate. The share of those polled by CNN who thought Clinton won is nearly identical to the 58 percent of viewers who said before the event that they supported her.
YouGov, similarly, found their respondents planning to vote for Clinton by a 48-42 margin, neatly mirroring the divide on who won the debate.
The significant difference between the two surveys exemplifies why post-debate snap polls aren’t a reliable way to measure the sentiment of the full electorate. As The Huffington Post wrote after the first presidential debate:
The results of snap polls like CNN’s are a step up from focus groups and online “reader polls,” in that they’re making an effort to be representative of the audience that watched the debate, rather than merely recording the opinions of whomever happens to be around to talk on TV or click on a button. Still, there are plenty of reasons to take them with a grain of salt. First, the universe of people watching the debate isn’t the same as the universe of people who will vote in the election. ... Second, it takes time to reach survey respondents. Polls that rush into the field will, by necessity, make some tradeoffs to do so, such as pre-screening for people who planned to watch the debate and are willing to answer questions afterward. And third, to the extent that debates matter ― which they often don’t ― their impact depends in large part on how they’re covered after the fact.
Polls taken next week will provide a better reflection on how much this debate, as well as Friday’s release of a tape that caught Trump making derogatory comments about women, changed voters’ minds.
But Trump, who lagged Clinton in the polls by 6 points before the latest damaging news cycles and had a less than 14 percent chance of winning per HuffPost’s presidential forecast, needed an exceptionally good night if he hoped to regain ground. The earliest results suggest that’s very unlikely to be the case.
YouGov used an online panel to interview 812 registered voters who watched the debate. CNN recontacted 537 voters, using live interviewers to reach both landlines and cell phones.