USC/Dornsife tracking poll megathread

[jdk]

Wow the legend that is USC/LA Times is back! I cant believe they have Biden+11, have they completely overhauled their polling methodology?

Last time they polled the same people everytime.  So because the first sample was skewed, they were all skewed.  But they gave a clear picture of who had momentum at any given time.

for that reason, I hope they use the same method

[Terry the Fat Shark]

fwiw they have apparently adjusted their poll to make it a bit more accurate demographically. They say their adjusted method would've had Clinton up by 1 in the PV (a 2 point error instead of a 5 point one) albeit still a bit right of the national result. Apparently this time they are going to do releases every 2 weeks instead of daily.

[Landslide Lyndon]

https://election.usc.edu/

Biden     53
Trump    40

[wbrocks67]

Yay it's back. It would appear that there was a little bit of a post-DNC bump.

[It's Perro Sanxe wot won it]

Yay it's back. It would appear that there was a little bit of a post-DNC bump.

Yeah, Biden seems to have had a 3 percentage point convention bump, while so far Trump seems to have had a 1 point bump, without taking into account the final day of the RNC.

[Soaring_Eagle]

Another Trash Poll, also not included in the RCP average

Again Biden being up by 13 is just untrue and anyone can see that

Not a chance he could possibly be up that high

[Penn_Quaker_Girl]

Another Trash Poll, also not included in the RCP average

Again Biden being up by 13 is just untrue and anyone can see that

Not a chance he could possibly be up that high

+13 may be a bit of an outlier status at this point, but time is running out for Trump to turn things around in his favor.  He can't keep relying on one or two major events that will magically change his position. 

[woodley park]

Another Trash Poll, also not included in the RCP average

Again Biden being up by 13 is just untrue and anyone can see that

Not a chance he could possibly be up that high

Are you watching the news? The country has become a complete and total disaster! Especially compared to four years ago when we were almost 7 years into an economic expansion. Now we have rampant death, unrest, and economic catastrophe. It’s a miracle for Trump that it’s only Biden + 13.

[tagimaucia]

Another Trash Poll, also not included in the RCP average
 

Is this a bit? RCP is actually known for having no clear criteria or transparency for what they include in their averages and they include every poll by many pollsters universally known to be crap while completely blacklisting or cherry picking many other ones with good track records.

[wbrocks67]

RCP may include it, its possible they just didnt update yet.

Also is Soaring Eagle a russian bot or

[kph14]

Another Trash Poll, also not included in the RCP average

Again Biden being up by 13 is just untrue and anyone can see that

Not a chance he could possibly be up that high

In 2016, RCP conveniently including them in their average when USC was the only one showing a Trump popular vote lead

[Podgy the Bear]

Good to know that Political Wire has already placed this and the Navigator poll information on their site. 

[BobbieMac]

Just to point out USC/Dornlife are polling on a daily rolling weekly window, so strictly speaking this Biden+13 number is just a snapshot. He's been between +9 and +16 since mid August.

[Gass3268]

Another Trash Poll, also not included in the RCP average

Again Biden being up by 13 is just untrue and anyone can see that

Not a chance he could possibly be up that high

This was RCP’s favorite poll in 2016.

[Sir Mohamed]

Just wait for the #ConventionBump, that will put Trump ahead and give him states like MN.

[Rand]

Another Trash Poll, also not included in the RCP average

Again Biden being up by 13 is just untrue and anyone can see that

Not a chance he could possibly be up that high

Your president is trash. That is why Joe Biden is up so high in the polls.

It is what it is.

[Pollster]

On the day of Biden's inauguration, pundits will still be saying that Trump is "running out of time to turn this around."

[Devout Centrist]

The interesting thing is, this pollster is using mostly the same panel of people as 2016. No idea if it’s actually representative of the rest of the country, but we’ll see.

[It's Perro Sanxe wot won it]

The interesting thing is, this pollster is using mostly the same panel of people as 2016. No idea if it’s actually representative of the rest of the country, but we’ll see.

According to their website, the reason why they overestimated Trump in 2016 was that their sample overrepresented rural voters, something they have now corrected. Without that error, their last poll before the election (which had Trump up by 5) would have shown a one point lead for Clinton.

[jdk]

Another Trash Poll, also not included in the RCP average
 

Is this a bit? RCP is actually known for having no clear criteria or transparency for what they include in their averages and they include every poll by many pollsters universally known to be crap while completely blacklisting or cherry picking many other ones with good track records.

This is their criteria:

If it helps the Republican in the averages include it.

If it helps the Democrat and it's not from a well established pollster, ignore it.

[jdk]

The interesting thing is, this pollster is using mostly the same panel of people as 2016. No idea if it’s actually representative of the rest of the country, but we’ll see.

Where this pollster is particular useful is that they poll the same people every time so even if their samples are skewed you can get a clearer view of where the trends are heading better than any other pollster

[pbrower2a]

Another Trash Poll, also not included in the RCP average

Again Biden being up by 13 is just untrue and anyone can see that

Not a chance he could possibly be up that high

+13 may be a bit of an outlier status at this point, but time is running out for Trump to turn things around in his favor.  He can't keep relying on one or two major events that will magically change his position. 

This is the point. In a basketball game, college or pro, a ten-point lead  with ten minutes to go is not decisive. With one minute to go it is decisive. Maybe a team that is behind by seven points can get into overtime if it is able to make a three-point basket, get fouled on the shot and get the free-throw, and then get the ball back quickly and make a successful three-point shot. Of course that is unlikely. Because the consequences are the same whether one loses by four or ten, a team behind takes its only chance to win.

Joe Biden has done what an effective underdog does in the primaries, seeking strength for winning the more decisive battles and getting it. So far he has been able to negate the usual advantages of incumbency that an incumbent President usually has. Donald Trump has clearly won his base, but he has failed to expand support for him. He has offended too many sensibilities.

No, in the 'contest' between fire-fighters and arsonists, there are not good people on both sides. So it is with violent racists and people who consider ethnic equity, there are not good people on both sides. 

[Amenhotep Bakari-Sellers]

Trump has gained no ground since his RNC speech, it's only a matter how much he loses by.

[BlueSwan]

USC doing the opposite of in 2016.

[TiltsAreUnderrated]

Each data point tracks seven days and about 1000-3000 voters. 538 treats both of the first projections as LVs and ignores the last two types, but there are four: traditional responses, probability-based projections (which incorporates respondents' answers regarding their own beliefs about the probability of them voting for X or Y, etc.), expectations about how others in the state will vote and "social circle voting" (expectations about how one's social circle will vote).

The probability-based one just gives Biden, Trump and someone else, and the traditional question prompts for Biden, Trump, Jorgensen, Hawkins, undecided and would not vote (and then removes "would not vote" responses for the final % projections). These responses are still weighted.