(CNN) First things first: The theme song of the week is Greatest American Hero.
Poll of the week: A new Reuters/Ipsos poll puts former Vice President Joe Biden at 47% to President Donald Trump's 40% among registered voters. The poll was taken partially during the Democratic National Convention and partially during the Republican National Convention.
Biden's advantage in the average of all polls has been consistent and most polls (unlike Ipsos) have him with more than 50% of the vote.
What's the point: One of the more interesting phenomenons during this campaign has been watching analysts, pundits and voters grapple with what occurred in 2016. The polling suggested Trump would lose to Hillary Clinton, and he, of course, won in the Electoral College.
One response to the current data and reaction to it comes from Axios' Jim Vandehei. He says the conventional wisdom that Trump can't win is wrong.
The result of the 2016 outcome for this cycle is that the general public doesn't buy the polling showing Biden clearly ahead. They think Trump is going to win.
A Pew Research Center poll published earlier this month demonstrates what's going on quite well. The poll had Biden up by 8 points over Trump, very similar to the average and the Ipsos poll discussed earlier.
Yet, the same poll found that Americans believed by a 51% to 46% margin that Trump would defeat Biden in the election. (Among voters, it was a tighter 50% to 48% spread in favor of Trump.)
The poll indicates that voters either believe the race will shift back to Trump or that the polling is wrong.
Interestingly, the poll was self administered via the internet without live interviewers, so it's not like the voters who said they were voting for Biden had reason to give what they might perceive as the more socially desirable answer (i.e. not voting for Trump).
Despite this, some voters think the polling is off.
Another question getting at the idea of potentially hidden Trump voters shows something similar to the Pew poll. By a 5-point margin, voters in an August Fox News poll said they thought more of their neighbors were voting for Trump over Biden. Biden was ahead in the horserace by 7 points in the poll.
The polling on who voters think is going to win is a marked reversal of what was happening at this point four years ago. Voters believed that Clinton was going to win by a 62% to 28% margin in a mid-August Quinnipiac University poll.
Yet the horserace polling at that time actually had Clinton ahead in the average by less than what Biden is currently up by.
And, as I noted previously, voters overwhelmingly believed Clinton would win at the end of the 2016 campaign.
The fact that the conventional wisdom was wrong in 2016 has clearly had a big effect on people's perceptions and not necessarily in a good way.
A plurality of Americans thought that the Republicans would hold onto the House in 2018, even as polling suggested otherwise. They were blown out.
Today, there seems to be a continued overcorrection of Trump's chances in 2020. Not only do more Americans than not think Trump will win, but the betting markets have Biden only as a nominal favorite.
None of these interpretations of the data are likely correct. Trump may very well defeat Biden, but it's not the most likely outcome.
While caution in interpreting polling data and recognizing that they are capturing only a moment in time is good, downright dismissing it is not the right answer.
Fortunately, most analysts I know are doing no such thing. They recognize that Biden is a favorite, but acknowledge that there is the possibility that Trump can win.
Whether Trump's chance shrinks or grows over the next few weeks will be largely dependent on whether the race shifts following the conventions.
Keep in mind that absentee ballots begin getting sent out in North Carolina later this week and that many people will be voting by mail this year.
If Biden continues to hold a clear advantage in the polls over the next few weeks, Trump's chances will begin to slide significantly.