Polls normally do pretty well to get a thousand respondents, but the poll I’m looking at about the second presidential debate has received a total of 907,211 responses, 68% of whom said that Trump “won” the night. Sounds solid, right? Not so fast – there are about 907,211 reasons to dismiss this online poll as garbage.
When polls speak to just a thousand people, they try to make sure those people are a good representation of the entire country. If they get it right, the results shouldn’t change by more than 3 or 4 percentage points whether you’ve polled 1,000 people or 2,000. That wasn’t the case last night.
When I checked in at 10.15pm, the results looked very different from how they look right now. Of the 49,280 votes that had been cast, 93% of them stated that Trump was the winner of the debate. It’s highly unlikely that most people changed their minds in the last 15 minutes of the debate or in the time since.
sure https://t.co/juiFdsFAgx pic.twitter.com/ZEWe1FBarR
— Mona (@MonaChalabi) October 10, 2016
Unlike polls that attempt to get a nationally representative sample of voices, this online poll’s respondents were just any old random people on the internet. Well, not quite any old people. The fact that the poll was created by and appeared on the homepage of drudgereport.com means it was probably more of a representation of Drudge readers than the country as a whole. And, according to surveys conducted by Pew Research Center, 74% of readers of Drudge Report are on the right of the political spectrum.
Lastly, unlike in more reliable polling, people were able to vote more than once in this poll (and still can – at the time of writing, it’s still open). That’s shaky enough but when some people are actively encouraging others to “flip” the poll, the numbers here are totally meaningless.
Why this stuff is meaningless... Also, why does any supporter care if you win an unscientific clicker (aka poll)? pic.twitter.com/IisnQWWMIY
— (((Harry Enten))) (@ForecasterEnten) October 10, 2016
There are more useful numbers available. CNN, together with the market research company ORC, conducted a poll with a more robust methodology, although they only managed to speak to 537 registered voters in total (only 27% of whom identified as Republican).
In the CNN/ORC poll, which was conducted after the debate ended, 57% of respondents said Clinton won the night, compared to 34% who said Trump did. Surprisingly, those numbers are slightly less positive for Clinton than when CNN/ORC asked the same question after the first presidential debate (and before it was revealed that Trump had bragged about grabbing “pussy”). On 26 September, 62% of respondents said Clinton won the presidential debate, while 27% said the same about Trump.
