Biden starts the general election with a lead in national polls, but it's closer than it looks. The polls show the electorate split along the same demographic lines as 2016, leaving Trump's relative edge in the Electoral College intact. nytimes.com/2020/04/13/ups…
So far, national polls show virtually no change among white voters with or without college degree over the last four years. But Biden does appear to be doing quite a bit better among older voters and weaker among nonwhite voters 
As an aside, a lot of the analysis that you see, suggesting that Biden is doing worse or better among so-and-so group, is based on cherry-picking or an apples-to-oranges comparison of national RDD polls to exit polls.
[clap emoji] it hasn't changed [clap emoji]
[clap emoji] it hasn't changed [clap emoji]
Trump's persistent strength among white voters without a degree means his relative advantage in the white, working class battleground states persists as well. Not much question that Trump continues to fare better in the battlegrounds than he does nationwide
And on top of this, these are polls of registered voters, not likely voters. Turnout's always a question but LVs are almost better for the GOP than RVs, and I'd guess that will hold again.
Over all, Trump's probably doing 4 or 5 points--and perhaps more--among likely voters in the electoral tipping point states than he does among registered voters nationwide
I do think Arizona stands out as a possible bright-spot for Democrats. They would be better off if they had a breakthrough in Florida, instead or as well, which I wouldn't really rule out given Biden's strength among older voters and the sparse polling there to this point
The older voter thing interests me. It's hard to explain and maybe Biden+10 among over 65 will prove to be noise.
It's partially compositional change: Trump led by just 2 pts among voters over age 61 (65 now), with Clinton ahead among voters age 61-65 in 2016
It's partially compositional change: Trump led by just 2 pts among voters over age 61 (65 now), with Clinton ahead among voters age 61-65 in 2016
At the same time, the more conservative "Silent Generation" is getting old enough to start departing the electorate in greater numbers. Together, I suspect Clinton was quite competitive among voters who were over age 61 in '16 and alive today, and maybe even won them
One point that didn't make the cut, due to lack of *recent* evidence: I think most of Biden's weakness among nonwhite voters, compared to Clinton, is among black voters and maybe especially young black voters. If so, harder to envision far higher black turnout, Dem GA/NC breakout
Hard to see it in recent national polls, which often don't breakdown by race. But it was there in our Times/Siena battleground polls in October, which had the same demographic story we see in national polls today
Another, easy to over-look factor in Biden's declining strength among nonwhite voters as a whole is the growing non-black share of nonwhite voters. Non-black nonwhite voters (Hisp/Asian, etc) tend to be more competitive than black voters and are a larger share of new registrants
I also like looking at a longer window for state polling, post-Dem impeachment inquiry on 9/26, for a bigger n. Here's what that looks like
A few recurring questions:
--what about NC/OH. Not enough live interview polls. Over longer term (prior post), NC divided.
--and gender? I don't write about gender in pieces like this, since it doesn't vary by state. But it's possible that the gender gap has slightly widened
--what about NC/OH. Not enough live interview polls. Over longer term (prior post), NC divided.
--and gender? I don't write about gender in pieces like this, since it doesn't vary by state. But it's possible that the gender gap has slightly widened
In these same polls, Biden leads 54-36 among women, v. 51-36 for Clinton; Trump leads 50-42 among men, v. 45-41 in '16 RV. polls.
Difference is small enough that I wouldn't be too confident this is a real shift, though it could be
Difference is small enough that I wouldn't be too confident this is a real shift, though it could be
For me, there's too much variance in the quality of subnational non-live interview polling for me to toss it all together. That's not to say there aren't some ok or even good polls in this category
Next, some wonder about whether turnout will actually help Democrats. There are some demographic shifts that help Dems in this respect, like greater strength among older and well-educated voters. Still, the few polls applying screens/asking LV-type questions suggest GOP will gain
Monmouth recently showed Republicans likeliest to say they were almost certain to vote. CNN has far more GOPers saying they're very enthusiastic about voting (not exactly the same as an LV screen fwiw). In our Oct polling, Trump picked up 1 point in a traditional screen
