It’s not just NBC’s tracking poll showing this, either, but it’s as good place to start as any, as you’ll see. The latest iteration of the NBC/Survey Monkey tracking poll shows Hillary Clinton near a majority with an eight-point lead over Donald Trump, 49/41, her largest lead in this series. Two significant changes in demographics undermine key arguments from Team Trump about their prospects:

Hillary Clinton now leads Donald Trump by 8 points, her highest advantage since the general election match-up question was first asked on May 2, according to the latest NBC News|SurveyMonkey Weekly Election Tracking Poll.

This week, Clinton enjoys 49 percent support of registered voters to Trump’s 41 percent. Her 8-point margin over Trump grew from 6 points in last week’s poll.

The lead goes down to six points in a three- or four-way race with Libertarians and Greens, but that’s not exactly good news for Trump or Clinton. He drops to 36% in those matchups, down five while Hillary drops seven to 42%. That split isn’t uneven enough to help Republicans in the presidential race, and suggests that the other two options will pull support from disgusted voters in both parties proportionally.

The good news for Hillary is that young voters appear to be coming home to the Democratic nominee. The NBC survey shows Hillary narrowing the gap among white voters from -12 to -8, a difference outside the ±1.8% margin of error. Another demo shows a much more dramatic change. With Bernie Sanders out of the race and offering support to Clinton, she picked up twelve points among college-age voters:

Bear in mind that this is a difficult demo to turn out in elections. Barack Obama managed to succeed twice in doing so, but he had a superior ground game, and it’s not clear that Team Hillary will come close to matching it — or come close to making the emotional connections required to succeed. However, this does hint that the progressives that gave Bernie their loyalty have begun to rally to the Democratic nominee, as expected, and the head-to-heads will begin to reflect that.

Or perhaps they already do. Take a look at the RCP average for the last three months in the Trump/Clinton H2Hs:

Yikes. This graph does not include this poll, but its 6.8-point spread comes close to mirroring it. That could have some Republicans thinking about a convention revolt, but there isn’t much appetite for it among the rank and file:

Although the Weekly Election Tracking Poll indicates that Clinton is picking up considerable momentum against Trump, Republican and Republican-leaning voters are rallying behind their party’s presumptive nominee and rejecting the “Stop Trump” movement. A large majority of Republican voters would like to see Trump nominated at the Republican convention.

All of these are national polls rather than battleground polls, which show closer races in almost all instances. Yesterday, I argued that national polling doesn’t matter as much as the state-by-state polls do, but with a gap this wide, that might not necessarily be true. Scott Rasmussen, who founded Rasmussen Reports but has personally been out of the polling business since 2013, respectfully disagreed with that take:

The only time that is true is if the national polling is VERY close. If it’s 49-49, then state polls are decisive. But, any change in the national numbers reflect a weighted average of state changes. Bottom line, if Clinton matches Obama and wins by 4, she will have Electoral College margin similar to Obama 2012. If she wins by 7, her Electoral College totals will be similar to Obama 2008.

And, the reverse is true as well. If Trump wins by 4, his margin will be similar to Obama 2012.

I would argue that state polling is an interesting side story at this point. Maybe, for example—maybe—Trump has a shot in Pennsylvania. But if he outperforms there, he will underperform somewhere else relative to the national average. That’s a mathematical certainty.

By the way, I know it’s a whole lot more fun to imagine all the state scenarios, but the simplest tool is probably the best. Take the results from the last election. If Trump does a point or two better nationally, assume he does a point or two better in each state. You can run it all up one side and down the other. And, what you’ll see is that unless the vote is very close, the popular vote winner wins.

Scott has an interesting — if depressing — projection of the Electoral College based on current polling. This in particular bears watching:

Trump is not projected ahead in any states that Barack Obama won in 2012. However, eight Obama states with 109 Electoral College votes are only leaning in Clinton’s direction (Colorado, Florida, Iowa, Michigan, Nevada, Pennsylvania, Virginia, and Wisconsin). If Trump is to pull off an upset, it would most likely come from winning many of these states.

Four of those are covered in my book, Going Red.