Dem
49
GOP
51
There have been some national polls taken since the debate. Here are those taken by two serious nonpartisan pollsters as well as some taken by partisan pollsters. That a pollster is partisan doesn't mean that they always make up or fudge the numbers, just that the temptation is there. They also like to be seen as accurate because candidates always want accurate numbers internally, and being seen as incompetent is not a selling point when trying to acquire new clients.
| Harris | Trump | Net | Pollster | Dates |
| 50% | 45% | Harris +5 | Data for Progress (D) | Sept. 12-13 |
| 52% | 46% | Harris +6 | Ipsos | Sept. 11-13 |
| 44% | 42% | Harris +2 | Redfield & Wilton (R) | Sept. 12 |
| 51% | 47% | Harris +4 | RMG Research (R) | Sept. 9-12 |
| 45% | 41% | Harris +4 | TIPP Insights (R) | Sept. 11-13 |
| 50% | 45% | Harris +5 | YouGov | Sept. 11-13 |
The average of the polls is Harris +4.3. The margin of error on these polls is about 4 points, but a bit smaller on the numerical average of the six polls (because the total sample size is bigger) and there are still 7 weeks to go. There could be serious methodological errors (shy racists, overcorrection for the Trump effect, etc.), an October surprise, etc. etc., but it is a safe bet that Jen O'Malley Dillon is sleeping better these days than Chris LaCivita.
If you are interested in the nuts and bolts of modern polling, ABC News has a page explaining how the polls it sponsors with Ipsos work. In short, they don't use random digit dialing anymore. They get a list of street addresses from the USPS and ask some fraction of people on it to fill out the survey online. People who lack Internet access are given a tablet and a free Internet connection. All participants get a small incentive worth $1-2.
National polls are one way of measuring the race, but not the only one. The state polls put Harris above 270 EVs (see map) but just barely, with nine states nominally swing states. Indiana isn't a swing state but there is only one (flaky) poll of it.
Another resource is the betting data. PaddyPower in Ireland allows people to bet on political outcomes. (If you want to follow the link you may need a VPN pointed at Ireland or the U.K.) They have Harris with a 55.6% chance of winning the election and Trump with a 50.0% chance. As to the popular vote, the bettors think that Harris has an 81.8% chance of winning it and Trump has a 25.0% chance. The numbers don't add to 100% in each case due to the vig (the bookie's take—the bookie always wins). But all in all, there is a sea change since early June, when it looked that Joe Biden's goose was cooked and Trump would coast to an easy win. (V)