For weeks since the Democratic National Convention, Donald Trump has trailed in the polls. His supporters have started to sound like conspiracy theorists in their unwillingness to accept the accuracy of the polls they seemed to love so fervently during the primary season.
A reasonable critique of polls would have sounded something like, "This is just the bounce in the polls that Hillary Clinton got from her convention, which, by the way, was a disaster, but you wouldn't know it from the liberal media."
But it's been nearly a month since the Democratic convention ended. Analysis by Nate Silver in 2012 predicted that it would be approximately three weeks from the end of the last convention to the day when polling bounces were finally nullified.
There's been only one national poll conducted completely outside of that three-week window, and it showed Trump down by three points in a two-way race against Clinton. For comparison, he was down 4 points on July 18, right before the Republican National Convention began.
The same poll also asked about a four-way race with Gary Johnson and Jill Stein included. In that case, Clinton's lead was 4 points over Trump, up from 3 points before the GOP convention.
But of course, the election isn't decided in a national poll. It's decided in the states.
In Florida, worth a whopping 29 electoral votes, a Florida Atlantic University poll found Trump with a 2-percentage-point lead.
In Missouri, worth 10 electoral votes, a Monmouth poll showed Trump up by 1.
In North Carolina, worth 15 electoral votes, a Monmouth poll showed Clinton up by 2 percentage points. A CNN poll has her up by one point.
In Ohio, worth an important 18 electoral votes, another Monmouth poll shows Clinton up by 4 percentage points.
All those polls are within the margin of error. Of course, the polls conducted in those states within the three weeks post-convention aren't worthless, just slightly skewed in Clinton's favor.
When you include those polls, you get a decidedly pro-Clinton picture, even after accounting for the skews. For example, a Monmouth poll conducted from Aug. 12-15 is probably skewed 1 or 2 points in Clinton's favor, but it gave her a 9-point lead.
Bettors throw all that information together and put their money on who they think will win the election. And they seem pretty sure Clinton will win.
As of Wednesday afternoon, bettors give Clinton a 76 percent chance of winning the election, roughly tied for her highest point yet in the campaign. It's a high point she reached about a week after the Democratic convention that hasn't showed signs of coming down since. Prior to the conventions, they only gave her less than a 70 percent chance of winning.
In order to discredit some polls less favorable to Trump, his supporters claim there's a secret Trump vote — Trump voters who, for various reasons, won't admit their support to pollsters.
Some also claim polls are skewed to under-represent Trump supporters. In 2012, similar claims even inspired the creation of a website devoted to unskewing the polls back to Republicans' favor. But in the end, the polls were actually "skewed" 2 percentage points in Mitt Romney's favor.
Trump supporters also claim that huge turnouts at his rallies show the polls are wrong. How could he be losing with all those enthusiastic supporters at his rallies?
Ask Walter Mondale, Romney and Bernie Sanders how that worked out for them. Mondale claimed in 1984 that "the pollsters aren't getting it" after drawing huge crowds at his rallies. Then he won only one state. Turns out, Mondale was the one not getting it.
Right before the 2012 election, Romney drew a crowd of 10,000 in the crucial swing state of Colorado that made some observers believe the election was tightening. Instead, Romney lost Colorado by more than 5 percentage points.
Sanders drew huge rallies in the 2016 Democratic primary campaign, frequently out-rallying Clinton. But in the end, Clinton won a majority of states, unpledged delegates and votes.
If this were the 1916 election, rallies might be a good indicator of support. But it's 2016 and we have polls that are better indicators of what's really going on.
Jason Russell is a commentary writer for the Washington Examiner.