Credit Breitbart for conducting a scientific poll whose outcome would be in doubt instead of taking the easy route and posting a Drudge-style online poll widget. The latter are fine if all you want is a broad sense of what your own readers think; if you want to know what the country thinks, scientific is your only reliable option. It amazed me this morning to see how many Trumpers on social media were dismissing the overnight scientific polls by CNN and YouGov that showed Clinton winning by pointing to Drudge-style widgets that have Trump romping 95/5 or whatever. A widget poll posted on a website doesn’t rely on a random sample. A site’s readership is self-selecting and in most cases will be skewed heavily by partisan ideology, race, sometimes age, sometimes gender. On top of that, it’s easy to game those polls by having other sites who are sympathetic to one candidate or the other link to the widget and encourage supporters to vote and run up the score. It’s like trying to gauge a candidate’s support based on crowd size at his rallies. That might tell you something about enthusiasm but it doesn’t tell you much about Election Day. If Breitbart had run a poll like that, given how pro-Trump their readership is, Trump’s margin might have been 98/2 or thereabouts but that wouldn’t tell you anything about national opinion. And Breitbart knows that, even if some of its allies don’t.

So they commissioned their pollster, Gravis, to do a scientific poll. Result: Clinton won narrowly, 48/43, although Trump came out ahead on key questions.

Only five percent of voters said their minds were changed by the debate, and of that number slightly more said they had switched to Trump than to Clinton. If all of that is true there should be virtually no changes in the polls next week due to the debate. On the other hand, YouGov had Clinton winning by a more comfortable margin:

CNN had Clinton winning even more comfortably at 62/27(!), although their sample was D+15. A noteworthy result:

That’s a bad outcome for a guy who needed to show last night that he was qualified, but it may be an artifact of the sample. Democrats split 12/87 on the question while Republicans split 84/14, near mirror images, while independents split 50/49. It’s pretty much a straight partisan divide. On the other hand, Clinton’s own numbers on this question are significantly better: She earns a 67/32 split, with even 38 percent of Republicans admitting that she’d be able to do the job.

What does all of that mean? Nate Silver looked back at post-debate scientific polls and found that the candidate who was perceived to have won usually ended up with an election poll bounce the following week. If that’s true to form, Clinton could be leading again in tight states like Ohio and Colorado next week. On the other hand, Silver’s playing an expectations game that could come back to haunt Clinton: If her numbers fail to rise despite accord among various surveys that she won the debate, the left is going to have a nervous breakdown over the possibility that Trump is now essentially mistake-proof and that his momentum is likely to carry him to victory in November. In that sense a “tie” in next week’s polls is a win for him, leaving Hillary and the left to make a tough decision. Do they want to risk two more debates with Trump despite having gained nothing from a victory in the first one? You won’t see anything from Clinton in the next two that you didn’t see last night, but there’s always a chance you’ll see a disciplined, cogent Trump putting any remaining doubts about him to rest. He’s really got nothing to lose from further debates if he doesn’t take a hit from last night’s performance. Clinton does. Will she bail?