How Strong is Biden's Lead?

Time for another blog update. As always, if you are impatient for one of these updates, the 2020 pages on Election Graphs are updated nearly every day as new polls come in. Or you can follow @ElecCollPolls on Twitter to see all the polls as I add them.

The last update here was 12 days ago on September 5th. Here are the high-level changes since that last post:

Model Metric 5 Sep 17 Sep 𝚫
Probabilities Trump 2σ
Median
Biden 2σ
Biden +38
Biden +152
Biden +254
Biden +44
Biden +152
Biden +256
Biden +6
FLAT
Biden +2
Trump Win
Tie
Biden Win
0.2%
0.1%
99.7%
0.1%
0.0%
99.9%
-0.1%
-0.1%
+0.2%
Categories Trump Best
Expected
Biden Best
TIED
Biden +220
Biden +288
Biden +6
Biden +240
Biden +288
Biden +6
Biden +20
FLAT
Tipping Point Biden +3.3% Biden +5.6% Biden +2.3%

This week's TL;DR: Biden is still significantly ahead. All metrics are equal to or better for Biden than when we did our last update. Most importantly, Biden's tipping point has moved from 3.3% up to 5.6%. This shows that it isn't just a bunch of states that are just barely Biden giving him his electoral college margin, but that instead, he has pretty substantial leads in all of the states he needs to win.

Five states had a net change in or out of "Weak Biden" and "Weak Trump" in these 10 days, so let's look at those first:

At our last update, Biden's lead in Pennsylvania had slipped below 5% into our "weak" category. It didn't last long. The next wave of polls was good for Biden, and his lead is now up to 5.6%, which translates into a 93.2% chance of winning the state.

In our last update, Georgia had slipped just barely to the Trump side of the centerline. Now it is just barely on the Biden side. Right now our average has Biden with a 0.1% lead. But this is in the zone where based on our analysis of Election Graphs results from 2008 to 2016 we actually give Trump slightly better odds of winning the state anyway. At an 0.1% Democratic lead, we have Biden at 46.8% to win, compared to Trump at 53.2%.

But really, that is close enough to 50/50 to call it a coin toss at the moment.

In our last update, Biden had gotten a bunch of really good Arizona polls, and he strengthened into our "Strong Biden" category. More recent polls revert to the more normal pattern, with Biden having a lead in Arizona, but a small one. That lead is currently 3.8%, which corresponds into a 82.9% chance of winning the state if the election was today.

Could we please have some more Iowa polls? For such a close state, it is surprisingly lightly polled. The five poll average currently goes back 1.6 months. A lot happens in 1.6 months.

Anyway, the average at the moment is a dead tie. (Well, looking at the unrounded numbers, Trump leads by 0.033%, but really, that is a tie.) But again, the historical analysis of our poll averages show the polls slightly overestimating Democratic support, so that margin still means a 54.2% chance of a Trump win.

Like Georgia though, Iowa is a complete tossup at the moment. If I really had to guess though, I'd guess that the one poll showing Biden ahead by 6% is an outlier, and when it rolls off of our average, Iowa will be on the Trump side of the fence. (But still not by much.)

Nevada also has had criminally few polls. The average had moved over the 5% Biden lead line for a little bit but is now down to 4.8%. That's still a 90% chance of a Biden win.

Between these five states, you explain the 20 EV movement toward Biden in the expected case, and the 6 EV movement toward Biden in Trump's best case.

In our categorization view, this once again ends up putting Trump's best case as a narrow loss. In other words, he can win all of the states that he leads, plus all of the states where he is behind by less than 5%, and he still loses. To win, Trump needs to win all of those states… plus Pennsylvania, where he is currently down by 5.6%.

In addition to the states above, North Carolina, Minnesota, and Maine's 2nd congressional district also crossed lines between weak categories, but they crossed back again for no net change.

When we start looking at our probabilistic model, any changes in the numbers matter though, not just those that shift categories. So a quick look at the rest of the "close" electoral votes:

Putting all of this together into our probabilistic model, you get this trend:

Our last update was right about the "Early Voting Begins" line. (Yes, people are already voting in Election 2020!) Our probabilistic envelopes show a very slight movement toward Biden, but things have been pretty stable. There are no big swings here.

In the categorization view, we have a few states that are bopping around near our category boundaries, and the all or nothing nature of that model means that the numbers can move a lot when a big state changes categories. But here the fact that Biden leading a state by 0.1% and Trump leading it by 0.1% really is not that much of a difference, so things are smoothed out a bit.

And we see…  stasis. This view of the race has barely moved.

This view looks at the range of electoral college results we should expect though. Our tipping point by contrast essentially looks at how easily that could change.

At our last update, we noted that Biden's tipping point had cratered, going from 6.5% on August 18th to only 3.3% on September 5th. As more polls came in, the 3.3% peak was actually erased. We now have Biden's low point at a 4.0% tipping point, after which he recovered to 5.6%.

3.3% is low enough you are within range of a big news event or systematic polling error erasing that lead. At 5.6%, things are a lot more secure.

But still not completely so. You only have to look at the two week period in June when Biden went from 2.1% to 7.3%. If you had a reversal of the same magnitude, Biden's tipping point lead would only be an extremely narrow 0.4%. But still, things look a lot more solid at 5.6%.

The RCP average of national polls currently shows a 5.9% Biden lead. Comparing to the 5.6% tipping point, this implies that the structure of the electoral college is currently giving Trump only a 0.3% head start, way down from 3.7% in the last update.

The spectrum of states where the margin is less than 10% now looks like this:

There are still LOTS of close states.

The last few updates, I've done a 2016 to 2020 comparison that looks like this:

<48 Days Out> 2016 2020
Expected Case Clinton by 130 Biden by 240
Tipping Point Clinton by 1.6% Biden by 5.6%

This time I can do better, with two new graphs that have been added to the main 2020 page:

This directly compares the evolution of the expected case metric in 2020 vs 2016. Biden has not always been doing better than Clinton. It has gone back and forth.

But at the moment Biden's Expected Case is better than Clinton EVER was able to reach, not just in the time frame shown above, but looking at the entire history of 2016 polling. The best she ever achieved was a 188 electoral vote margin, which she hit twice on the graph above.

Biden is currently at 240. If that holds, it would be the biggest electoral college winning margin since Bush crushed Dukakis with a 315 electoral vote margin in 1988.

But let's look at the tipping point comparison:

Biden is ahead of where Clinton was at this time in 2016. But this time in 2016 was one of Clinton's low points. She bounced back over the next couple of weeks. When we got to 31 days before the election her tipping point had rebounded all the way to 6.0%. That is a stronger lead than Biden has today.

But of course, over those last 30 days, you can clearly see Clinton's lead deteriorate. By the time you got to election day, it was 1.6%, which as we saw, was small enough that a little polling error in the states near the tipping point, plus some movement that happened too late for polls to capture, was enough for Trump to win.

What's different this time?

Aside from the big movement toward Biden in June, Biden's tipping point has been more stable than Clinton's. Clinton's line swung back and forth wildly. Biden is certainly jittering around a bit, but the magnitude of the movements is a lot smaller.

Election Graphs focuses on margins, but looking at the details of the polling and the actual absolute amount of support for each candidate, one big difference in 2020 vs 2016 is that we have a significantly smaller set of undecided voters now than in 2016. So the group who are available to "slosh around" and shift back and forth over time is a lot smaller.

Does that exclude the possibility of a Clinton style collapse?

No. It does not.

But it probably does mean you need a pretty huge news event to cause that kind of movement.  It would not only have to make the undecideds break strongly toward Trump but also keep some current Biden supporters from voting for him, either by actually flipping or just by staying home.

The closer we get to election day, the harder this is. Just because there is less time to change people's minds. Especially since early voting has already started, and will be in full swing for the entire last month.

There were "October Surprises" that made a difference in 2016. We will see soon enough if there are similarly large and impactful events in 2020.

The map below is where things stand today. But it will surely change. Stay tuned!

47.6 days until the first results start coming in for Election 2020.

For more information:

This post is an update based on the data on the Election Graphs Electoral College 2020 page. Election Graphs tracks a poll-based estimate of the Electoral College. The charts, graphs, and maps in the post above are all as of the time of this post. Click through on any image for current interactive versions of the chart, along with additional details.

Follow @ElectionGraphs on Twitter or Election Graphs on Facebook to see announcements of updates. For those interested in individual poll updates, follow @ElecCollPolls on Twitter for all the polls as I add them. If you find the information in these posts informative or useful, please consider visiting the donation page.

Mixed Messages

Time for another blog update. As always, if you are impatient for one of these updates, the 2020 pages on Election Graphs are updated nearly every day as new polls come in.

The last update here was 10 days ago on August 25th. If you didn't notice it at the time, please check that last post for a CORRECTION I added on the 26th. The fixes mentioned on that correction are the baseline used here, rather than the originally posted numbers.

Here are the high-level changes since that last post:

Model Metric 25 Aug 5 Sep 𝚫
Categories Trump Best
Expected
Biden Best
Biden +6
Biden +146
Biden +288
TIED
Biden +220
Biden +288
Trump +6
Biden +74
FLAT
Tipping Point Biden +6.0% Biden +3.3% Trump +2.7%
Probabilities Trump 2σ
Median
Biden 2σ
Biden +50
Biden +148
Biden +258
Biden +38
Biden +152
Biden +254
Trump +12
Biden +4
Trump +4
Trump Win
Tie
Biden Win
0.1%
0.0%
99.9%
0.2%
0.1%
99.7%
+0.1%
+0.1%
-0.2%

This week's TL;DR: Biden is still significantly ahead, and if he wins every state where he leads the polls, he wins by a larger margin than before. However, he is has weakened in a lot of the close states, meaning Trump has more possible paths to a comeback than he did before, and Biden's lead is more precarious than it was.

A bunch of states moved in and out of "Weak Biden" and "Weak Trump" in these 10 days, so let's look at those first:

First up, Texas. The big one. The average in Texas has once again moved from just barely Trump, to just barely Biden. Biden now holds a thin 0.8% lead in our average. In terms of probabilities, we have it 54.2% Biden, 45.8% Trump. If the election was today.

Then Pennsylvania. Biden had been looking pretty strong there, with a lead as high as 7.0% as recently as August 25th. But Trump has gained since then, and Biden's lead here has slipped to only 3.3%, which translates into a 79.4% chance of winning. But Trump winning Pennsylvania is back on the table as a reasonable possibility.


Georgia is perpetually close, and once again crosses the centerline, this time moving from Biden to Trump. Election Graphs now has Trump leading by 1.5%, which is a 71.6% chance of a Trump victory.


Meanwhile, North Carolina goes the other way. After only a very short time on the Trump side, North Carolina is back to "Weak Biden", which is where it has been for most of the last few months. We have Biden up by 1.4%, which is a 63.3% chance of a Biden win.


Biden's lead in Arizona increases, moving it to the "Strong Biden" category. Biden now leads there by 5.7% in our average, which is a 93.5% chance of winning.


There have not been many Nevada polls. We have to go back almost 10 months to get the five polls for our average. But with the latest poll, Biden's lead jumps to 5.4% or a 92.5% chance of winning the state.

Texas, Georgia, and North Carolina crossed the center line. Texas (38 EV) and North Carolina (15 EV) moved toward Biden, while Georgia (16 EV) moved toward Trump. That is a net movement of 37 EV toward Trump, or a 74 EV increase in Biden's expected margin.

Pennsylvania, Arizona, and Nevada on the other hand, impacted what we call Trump's "Best Case". That would be where he wins every single state that he leads our averages, plus every state where he is behind by less than 5%. Pennsylvania (20 EV) moved toward Trump, while Arizona (11 EV) and Nevada (6 EV) moved toward Biden, for a net improvement for Trump of 3 EV, or 6 EV of net margin.

In our categorization view, this ends up putting Trump's best case at an actual 269-269 tie, which would throw the election into the House of Representatives. To get to that Trump would need to win all that states he is ahead in, all five states where Biden is ahead by less than 5%, and also both Maine's 2nd congressional district and Nebraska's 2nd congressional district. But if that happened… tie.

As always though, we want to look beyond the simple categorization of states based on the averages. So a few more states to look at.

In addition to the states above that changed categories and stayed there, there were three additional states that moved in or out of "Weak Biden" and "Weak Trump", but then moved back again, leaving no net category change in the last 10 days, even though the actual average moved around a bit. In some cases, you can't even see the category change in the charts anymore, because the timing of the polls actually erased the change. But here they are anyway:

And finally, the locations with margins under 5% that didn't change categories at all this time around:

Putting all of this together into our probabilistic model, you get this trend:

Looking at the center median line, we can see a move toward Biden as the Democratic Convention happened, with that leveling off, then a move back toward Trump again starting as the Republican convention was going on. As of right now, Biden is a little bit better off in this view than he was right before the conventions started.

Basically though, we been in the same sort of range since the bigger movement toward Biden in June. The median has not left the Biden by 134 to Biden by 180 zone since the end of June.

We HAVE, on the other hand, seen the upper end of these probabilistic bands increase. These are the lines showing the best scenarios for Trump. They bounce around a lot but have generally been creeping upward since the end of June.

This basically indicates a pattern where Biden's electoral college lead has been pretty steady but has been getting more precarious, as leads in a variety of states diminish.

The simulations now have Biden at a 99.7% to win to Trump's 0.2%. The remaining 0.1% is that 269-269 tie possibility.

This is if the election was today of course, and the election is not today. I mentioned that the increases in the upper end of Trump's range indicate that perhaps Biden's lead is a little less solid than the margins would imply. This brings us to the tipping point:

Biden's tipping point has cratered since the conventions began, going from 6.5% on August 18th to only 3.3% now.

If Biden wins every state he leads, he wins by a very impressive 379 electoral votes to Trump's 159. That is a 220 EV margin. That would be a larger electoral college margin victory than any election since Clinton's defeat of Dole in 1996 with exactly that margin.

But Biden's margin in Pennsylvania, the state that currently puts him over the edge, is only 3.3%. You only need that small movement in the "Weak Biden" states to flip the election. Or a polling error of that magnitude.

In June we saw the tipping point move 5% in two weeks. We have just under two months left until the election.

(OK, if you look closely, you'll notice a 3.3% move would only get us to that 269-269 tie, it would currently take a 5.3% move to actually also flip Minnesota and have an outright Trump win, but the point stands…)

The RCP average of national polls currently shows a 7.0% Biden lead. Comparing to the 3.3% tipping point, this implies that the structure of the electoral college is currently giving Trump a 3.7% head start, up significantly from 1.6% in the last update. Like 2016 and 2000, the situation where Trump loses the popular vote but wins the electoral college is a live possibility.

The spectrum of states where the margin is less than 10% now looks like this:


To win, Trump needs all the close states. But all of those states are VERY close. It isn't that big of a stretch from where things are right now.

Now comparing to this time in 2016:

<59 Days Out> 2016 2020
Expected Case Clinton by 5 Biden by 220
Tipping Point Clinton by 0.9% Biden by 3.3%

This point in 2016 was Clinton's late-summer low point, before surging back to a strong lead in October, which then of course completely slipped away in the last month.

The best Clinton saw in the Expected Case in the last 59 days was a 154 electoral vote margin in October. Biden is a lot stronger than that at the moment.

Clinton's best Tipping Point in the last 59 days was 6.0% in early October though. Biden was stronger than that in mid-August, but he isn't anymore.

People argue that the presidential race is much more stable this year than it was in 2016. There are fewer undecided people. And both candidates have less "soft" support that can easily slip away. This does seem to be true. So maybe the chances of a lot of movement are less than there were.

But the amount of movement you need to change the outcome right now isn't all that large.

The map below is where things stand today. Keep watching!

59.0 days until the first results start coming in for Election 2020.

For more information:

This post is an update based on the data on the Election Graphs Electoral College 2020 page. Election Graphs tracks a poll-based estimate of the Electoral College. The charts, graphs, and maps in the post above are all as of the time of this post. Click through on any image for current interactive versions of the chart, along with additional details.

Follow @ElectionGraphs on Twitter or Election Graphs on Facebook to see announcements of updates. For those interested in individual poll updates, follow @ElecCollPolls on Twitter for all the polls as I add them. If you find the information in these posts informative or useful, please consider visiting the donation page.

Summer Doldrums

Time for another update for the blog. As always, if you are impatient for one of these updates, the 2020 pages on Election Graphs are updated nearly every day as new polls come in.

The last blog update was 8 days ago on August 9th.

Here are the high-level changes since then:

Model Metric 9 Aug 17 Aug 𝚫
Categories Trump Best
Expected
Biden Best
Trump +34
Biden +180
Biden +288
Biden +6
Biden +182
Biden +288
Biden +40
Biden +2
FLAT
Tipping Point Biden +4.2 Biden +5.9% Biden +1.7%
Probabilities Trump 2σ
Median
Biden 2σ
Biden +50
Biden +154
Biden +264
Biden +52
Biden +152
Biden +260
Biden +2
Trump +2
Trump +4
Trump Win
Tie
Biden Win
0.1%
0.0%
99.9%
0.1%
0.0%
99.9%
FLAT
FLAT
FLAT

The bottom line here is there is not much net change. If you had to pick a winner, Biden is a bit better off than when we made our last post. But really, things haven't moved all that much.

Let's start with the states that actually moved in or out of the "Weak Biden" and "Weak Trump" categories in our categorization view.

There were three:

After a brief foray into "Weak Biden" territory, Biden's lead in Pennsylvania increased once again to over 5%, moving the state to our "Strong Biden" category, and removing winning Pennsylvania from what we call Trump's "best case".

With this Trump's best case margin drops 40 electoral votes from a 34 electoral vote win to a 6 electoral vote loss.

As of our last update, Georgia had moved from just barely Biden to just barely Trump. It didn't last long. Georgia is now once again just barely Biden. "Just barely" is critical here of course. It is really just too close to call.

North Carolina had been in the "Weak Biden" category since June, but Trump has led the latest batch of polls and thus has pulled North Carolina back to his side of the line. Narrowly. Once again, the truth is the state is too close to call.

With the Georgia move, Biden's margin increased by 32 electoral votes. But the North Carolina move took away 30 of that. So the net change is a 2 electoral vote improvement for Biden in the "expected case" where every candidate wins exactly the list of states they lead in the Election Graphs averages.

The change in the expected case and in Trump's best case can be seen in the chart above.

Before we look at the probabilistic views, here is a quick no-commentary rundown of all of the other close states (and congressional districts that give electoral votes):

Putting all of this together into our probabilistic model, you get this trend:

Looking at the center median line, where half the simulated election results were better for Biden, and half were better for Trump, Biden's most recent peak was a 180 electoral vote margin on July 31st. Since then he has slipped down to a 152 electoral vote margin. So the recent movement has been against Biden.

Maybe this trend will continue. Or maybe things will bounce back in Biden's direction again. In this view, it now looks like things have been staying within a fairly narrow band since the end of June. But we're now at the upper end of that band. Trump may break out.

Conventional wisdom is that there are "convention bounces". And we are about to enter two weeks of political conventions. But this year the conventions are essentially back to back. So if there are such bounces, the Trump bounce may just cancel out the Biden bounce too quickly to even be measured reliably in state-level views such as we have on Election Graphs.

The conventions do however tend to be the point at which both campaigns go into overdrive and non-political folks finally start paying attention to the race. So it would not be surprising to have significant events that move the needle.

As usual, we'll just have to wait and see.

My simulations still have Biden at a 99.9% chance to win. This graph has been quite boring for some time now because Biden has the blue pegged up against the 100% line.

As always, the important caveat that this is if the election was today, which it is not.

There is time for this to change. So how secure is the Biden lead?

Biden's tipping point lead had dropped down to 4.6%, but it has rebounded to 5.9%. This number basically represents how much polls would have to change nationally to flip the winner if the change was uniform across all states.

By comparison, the RCP average of national polls is currently at a 7.5% Biden lead. So it would take a 7.5% move for Trump to win the popular vote, but only a 5.9% move for Trump to win the electoral college. This implies that the structure of the electoral college is currently giving Trump a 1.6% head start.

It is tempting to concentrate on the states that are currently closest. At the moment that would be Iowa, Maine CD2, North Carolina, Ohio, and Texas. But at the moment if Biden won those states it would just be gravy on top of a win that had already been secured. And if Trump won those states, it would not be enough to win the election.

The battle for winning the election is around the tipping point. You want to be as far ahead as possible in the states near the tipping point. That is how you secure the win, and that is generally where most efforts should be concentrated.

Right now, Pennsylvania is the tipping point. Adding in the two states on either side, you bring in Michigan, Wisconsin, Nevada, and Florida.

This is part of why you see tons of polls in Pennsylvania, Michigan, and Wisconsin even though none of those are particularly close at the moment. And of course, Florida, since it is not that far off from the tipping point, AND it is close at the moment.

It does not however explain why there have only been five polls in Nevada in the LAST YEAR. Come on people! Can we get some polls in Nevada?

Anyway… you can see how the tipping point fits in with all the other states in the current spectrum of the states in contention:

So yeah. As with 2016, watching Michigan, Wisconsin, and Pennsylvania makes sense.

Arizona? Georgia? Ohio? Iowa? North Carolina? Texas?

Those are "stretch goals" for Biden. Great as a bonus, but he needs to concentrate on the bread and butter first.

By contrast, Trump NEEDS all those states… plus some. At the moment Trump can once again win ALL of the close states and still fall slightly short. He needs to also bring Pennsylvania back into the fold to win.

How do the metrics compare to where things were at this time in 2016?

<79 Days Out> 2016 2020
Expected Case Clinton by 86 Biden by 182
Tipping Point Clinton by 3.0% Biden by 5.9%

By both metrics, Biden is doing significantly better than Clinton was at the same point in the race four years ago.

The best Clinton saw in the Expected Case in the last 79 days was a 178 electoral vote margin near the end of August. Biden is slightly ahead of that mark as well.

Clinton's best Tipping Point in the last 79 days was 6.0% though, which she didn't hit until October. Biden is below that level, and that 6.0% tipping point lead disappeared in less than a month in 2016.

So that serves as the usual warning of how quickly things can change.

Finally, the current map:

78.7 days until polls start to close on election night, and the conventions are about to begin. The race is about to go into overdrive. Buckle up!

For more information:

This post is an update based on the data on the Election Graphs Electoral College 2020 page. Election Graphs tracks a poll-based estimate of the Electoral College. The charts, graphs, and maps in the post above are all as of the time of this post. Click through on any image for current interactive versions of the chart, along with additional details.

Follow @ElectionGraphs on Twitter or Election Graphs on Facebook to see announcements of updates. For those interested in individual poll updates, follow @ElecCollPolls on Twitter for all the polls as I add them. If you find the information in these posts informative or useful, please consider visiting the donation page.

Trump Narrows The Gap

I am overdue for another update. The 2020 pages on Election Graphs are updated nearly every day as new polls come in. So you can always see the current status there for yourself in between these posts. But here comes another blog post…

Since the last update on July 24th, there have been new state polls in Pennsylvania (x8), Arizona (x7), Florida (x7), Michigan (x10), Ohio (x5), North Carolina (x9), South Carolina (x4), Minnesota (x5), Colorado (x3), Georgia (x5), Texas (x4), Virginia (x3), Wisconsin (x9), New Jersey, Massachusetts, Alaska, Iowa (x5), Maine All (x4), Montana (x2), Maine CD1 (x2), Maine CD2 (x2), Washington, New Hampshire (x3), California, Alabama, Kentucky (x4), Hawaii, Indiana, Oklahoma, Utah, and Kansas.

That's 31 out of the 56 entities that allocate electoral votes in only 16 days. That's more than half. Given that, I'm going to stop calling out the states with polls each update. In short, the answer from now until the election is always going to be "a lot of them", and "multiple polls for all the close states".

It has been 16 days, how much has changed?

First the summary table of changes:

Model Metric 24 Jul 9 Aug 𝚫
Categories Trump Best
Expected
Biden Best
Biden +36
Biden +168
Biden +288
Trump +34
Biden +180
Biden +288
Trump +70
Biden +12
FLAT
Tipping Point Biden +6.5 Biden +4.2% Trump +2.3%
Probabilities Trump 2σ
Median
Biden 2σ
Biden +80
Biden +168
Biden +272
Biden +50
Biden +154
Biden +264
Trump +30
Trump +14
Trump +8
Trump Win
Tie
Biden Win
0.0%
0.0%
100.0%
0.1%
0.0%
99.9%
+0.1%
FLAT
-0.1%

So compared to 16 days ago, what is the TL;DR summary?

Biden still leads by a large margin and actually improves his margin in the "everybody wins every state where they lead the average" expected case metric, but overall Trump has managed to reduce Biden's margin in a number of states, thus increasing the likelihood he might steal those states back and strengthening his position. And if he managed to flip ALL the close states, he could once again pull off a narrow win.

Now some of the charts, starting with the categorization view:

We can see that while Biden improved his expected case, Trump's best case has improved significantly.

So which specific states changed?

First, it was Iowa that moved from barely Trump to barely Biden, improving Biden's margin in the case where each candidate wins every state where they lead the average. But Biden leads Iowa by only 0.2%. And it looks pretty likely that the lead is due to a single outlier poll that showed a 6% Biden lead. So far no other polls indicate a Biden lead of that magnitude. So it would not be surprising at all to see Iowa slip back to the Trump side with the next poll, or at least when that outlier slips out of the average.

But then there is Florida. Biden's substantial lead in Florida has slipped significantly. He still leads, but no longer by huge margins. In our average Biden now only leads Florida by 3.0%. If the election was today, we estimate that translates into a 73.8% chance that Biden would win the state. But we actually have 86 days until the election, and there is plenty of room for this to move further.

The substantial swings here over June and July indicate that as polarized as things tend to be these days, there is still a non-trivial group of voters who have shown they can change their minds over time. In this case, they moved away from Trump for a while, but seem to be coming home again, once again putting Florida within Trump's reach.

Much the same pattern has played out in Pennsylvania. Biden built a substantial lead in June, but over the course of July, it faded quite a bit. Biden still leads Trump by 4.2% in Pennsylvania, translating into an 84.9% chance of winning the state in our model if the election was today. But that is significantly less certain than the 98.8% we had at Biden's peak.

Given Florida's 29 electoral votes and Pennsylvania's 20 electoral votes, together these states can change the margin by a massive 98 electoral votes.

Having these two states close enough to be in play is critical to Trump. At the moment, he is still behind, but close enough that it is easy to imagine scenarios where Trump could come back from behind and win them.

Not all of the movement was bad for Biden.

In Wisconsin, at the time of the last update, Wisconsin had flipped from "Strong Biden" down to "Weak Biden". But that didn't last long, and Wisconsin is once again a strong enough Biden state that we don't include it in Trump's best case. (At an 8.6% lead, our odds estimate is a 99.1% chance of a Biden win if the election was today.)

New Hampshire also moves out of Trump's reach with Biden leading the average by 7.8%, which would be a 98.6% chance of a win if the election was today.

But Wisconsin and New Hampshire are only 14 electoral votes, compared to 49 electoral votes for Florida and Pennsylvania. So the net change is 35 electoral votes in reach for Trump that were not when we did the last blog post, meaning a 70 electoral vote increase in Trump's best case.

So, now to dash through the "weak" states that are in play that didn't switch broad categories this time around, in order by  the number of electoral votes:

When you mash all of these movements together into our probabilistic model that doesn't just say that since one candidate is ahead in a state they are going to win it, but instead simulates many possible election results recognizing that close states can go either way, you get these trends:

Roughly speaking, the darker the color, the more likely the final margin will be in that range.

Looking at the center median line, where half the simulated election results were better for Biden, and half were better for Trump, Biden's most recent peak was a 180 electoral vote margin on July 17th. Since then he has slipped down to a 154 electoral vote margin.

All of the other bands have moved toward Trump as well, each to a differing degree. The moves aren't as dramatic as the categorization view shows, but there has been a steady deterioration for Biden since that peak.

June and the first half of July were a constant stream of bad news for Trump, and his position in election polls reflected that. Since then things have slowly been moving back in his direction.

To be absolutely clear, Biden still has a huge advantage. But less so than he did a few weeks ago.

It is still too small to see directly on the chart, but Biden's chances of winning when we did our last update rounded to 100.0% (it was actually 99.9931% in our simulation), and now they round to 99.9% (it is actually 99.9265% in our current simulation).

So that is still substantial, even if it is less than it was.

But that is if the election was today.

How vulnerable is that lead?

Biden's tipping point lead peaked at 7.3%. It is down to 4.2%. That's a 3.1% decline in a little over a month. If that trend continued, in another month Biden would still have a lead, but it would be quite tenuous.

That is not a prediction. It seems more likely that if Biden's spike in June was just an aberration and things are returning to the mean, that the tipping point will settle out between a 2% and 4% Biden lead.

But that is the level of lead that might just be a systematic polling error. Or that could be erased by a major event in the last few days of the campaign, too late to be measured by the polls.

So how does this look compared to 2016?

<86 Days Out> 2016 2020
Expected Case Clinton by 164 Biden by 180
Tipping Point Clinton by 6.0% Biden by 4.2%

So Biden is ahead of Clinton at this same point in time in terms of expected margins if all the states fell where their polls indicated. But Clinton's lead was actually a bit more secure than Biden's is right now, and of course, it slipped away before the end.

If the election was today, Biden would almost certainly win. But his lead is somewhat precarious. If 2% or 3% of people change their minds from Biden to Trump in a few critical states, Trump takes the lead.

Now the spectrum of battleground states :

Here is what it looked like in our last update:

And here it is today:

Out of the close ("weak") states, Biden only has to hold Arizona and Pennsylvania. He can afford to lose all the rest and still win.

But we're once again in a position where with a sweep of all the close states, Trump could win.

Finally, the current map:

86.0 days until polls start to close on election night. Stay tuned!

For more information:

This post is an update based on the data on the Election Graphs Electoral College 2020 page. Election Graphs tracks a poll-based estimate of the Electoral College. The charts, graphs, and maps in the post above are all as of the time of this post. Click through on any image for current interactive versions of the chart, along with additional details.

Follow @ElectionGraphs on Twitter or Election Graphs on Facebook to see announcements of updates. For those interested in individual poll updates, follow @ElecCollPolls on Twitter for all the polls as I add them. If you find the information in these posts informative or useful, please consider visiting the donation page.

Trump Collapsing?

Since the last update on June 18th, there have been new state polls in Utah, Kentucky, New Hampshire, North Carolina (x5), Minnesota, Alabama, Michigan (x3), Wisconsin (x5), Texas (x2), Ohio, Pennsylvania (x3), Florida (x2), Arizona, Georgia (x2), Missouri, and New York.

In our last update, we noted that while national polls were moving rapidly against Trump, the story with state polls was more complicated.

Not this time. With very few exceptions, these polls have been bad for Trump. State polls have caught up with the national polls, and this time around we see a dramatic move away from Trump and toward Biden across metrics.

As usual, let's start with the summary table of changes, then get into the graphs:

Model Metric 18 Jun 1 Jul 𝚫
Categories Trump Best
Expected
Biden Best
Trump +42
Biden +130
Biden +288
Biden +56
Biden +162
Biden +288
Biden +98
Biden +32
FLAT
Tipping Point Biden +4.8 Biden +7.3% Biden +2.5%
Probabilities Trump 2σ
Median
Biden 2σ
Biden +32
Biden +132
Biden +242
Biden +82
Biden +158
Biden +262
Biden +50
Biden +26
Biden +20
Trump Win
Tie
Biden Win
0.3%
0.1%
99.6%
0.0%
0.0%
100.0%
-0.3%
-0.1%
+0.4%

Not a single one of our national metrics moved toward Trump this time.

Let's start by looking at the states that moved in or out of the "Weak Biden" and "Weak Trump" categories that our categorization model thinks could go either way since the margin is less than 5%.

In order from biggest to smallest:

With a series of really strong polls, Trump seems to be collapsing in Florida at the moment. Now, it would be natural for there to be some reversion to the mean at some point, and maybe the poll showing Biden with a 10%+ lead is an outlier. But for the moment, Florida moves from "Weak Biden" to "Strong Biden", meaning the categorization view no longer sees Florida as being in reach for Trump if the election was today. (Spoiler, it isn't.)

Pennsylvania also moves from "Weak Biden" to "Strong Biden" as a wave of new polls showing him ahead wash the most recent poll showing Trump leading Pennsylvania out of the average.

Now, Florida and Pennsylvania moving out of Trump's reach (for now) is big news, but perhaps the even bigger news is Biden retaking the lead in Georgia. Now, this is just barely. He leads by 0.9% in our average at the moment, which means we give Biden about a 55% chance of winning the state.

Georgia is clearly a battleground state at the moment, which is significant in and of itself, no matter which candidate is slightly ahead. In a world where Trump was doing well, you would not expect him to have to be fighting for Georgia, let alone be losing it.

With these categorization changes, you get this chart for the range between the categorization view's best and worst cases for each candidate:

In terms of the central "expected case" line, we are at a 162 EV win for Biden, which is almost, but not quite back to the 166 EV win we had for him in early May.

But Trump's best case is not only a loss for the first time since last October, but it is also the worst it has ever been in the Biden vs. Trump matchups. At this moment, if Trump were to win EVERY close state, he would still lose to Biden by 56 electoral votes.

Of course, the categorization model is very coarse. Let's look now at the rest of the close states that had new polling since our last update and see how those change the probabilistic model.


Where are Wisconsin and Michigan? Aren't they key swing states? Well, maybe. But at the moment they are not CLOSE states. Biden leads Wisconsin by 7.1% and Michigan by 8.0%. That may not stay that way. In both states, Biden's lead is down a bit from its high. But at the moment, Biden has quite a strong lead in both.

When you mash all of the movements in all of the states with new polls together into our probabilistic model, you get this:

Trump peaked in the middle of April in our simulations, dropped quickly, then plateaued, increasing to a second peak right around the beginning of June before falling again. Then things started to flatten out again, but that most recent bit of the chart is still subject to change as new polls covering that time period come in.

In terms of the median case of our simulations, Biden is now winning by 158 EV, almost at his recent best of 160 EV. His all-time best was 184 EV back in October.

In terms of probability of winning our site now shows Biden at 100.0%. That is rounded though of course. Looking at the unrounded numbers, it is actually 99.9977% at the moment.

This is the time for the usual "if the election was held today" warning. If Biden's polls end up looking like this on election eve, he would almost certainly win. But we have almost 126 days to go. And things can change.

So let's look at the tipping point, which measures just how much things have to change in order to flip the winner.

The tipping point has moved dramatically toward Biden. Between June 9th and June 22nd, it moved from Biden by 2.7% to Biden by 7.3%. That's 4.6% in less than 2 weeks.

On the one hand, that is a dramatic collapse for Trump. But on the other, it shows just how quickly things can move. Something that can go down quickly can potentially go up quickly too.

So for the two metrics we had in 2016 as well as today, how does Biden stack up to Clinton at the same time period?

<126 Days Out> 2016 2020
Expected Case Clinton by 144 Biden by 162
Tipping Point Clinton by 3.2% Biden by 7.3%

So yes, Biden is doing better on both metrics than Clinton was at this same time in 2016.

From this point, Clinton would improve a bit. In August 2016 she got up to a 188 EV lead in the expected case and a 6.1% lead in the tipping point. This, of course, did not last all the way until the election. By the time we got to the election, Election Graphs had Clinton leading, but just barely.

Other analysts looking at the internals of various polls, both at the state and national level are saying that Biden's support is more solid than Clinton's was. That she had weaknesses that Biden does not.

Maybe, maybe not. In any case, he is doing better at this time in the race than she was. A 7.3% tipping point is a SUBSTANTIAL lead.

Another way to look at the change since the last update on June 18th is to look at how the center portion of the spectrum of states changed.

Here is what it looked like in our last update:

And here is how it looks today:

Things are getting bluer.

To close things out, the current map:

And that is where we are.

But it is only July. Time to watch what happens over the summer.

125.7 days until polls start to close.

For more information:

This post is an update based on the data on the Election Graphs Electoral College 2020 page. Election Graphs tracks a poll-based estimate of the Electoral College. The charts, graphs, and maps in the post above are all as of the time of this post. Click through on any image for current interactive versions of the chart, along with additional details.

Follow @ElectionGraphs on Twitter or Election Graphs on Facebook to see announcements of updates. For those interested in individual poll updates, follow @ElecCollPolls on Twitter for all the polls as I add them. If you find the information in these posts informative or useful, please consider visiting the donation page.

Ups and Downs

Since the last update on June 2nd, there have been new state polls in Kansas, Arizona (x4), Florida (x4), Michigan (x6), North Carolina (x4), Pennsylvania (x2), Wisconsin (x3), Texas (x2), Ohio, California, Iowa (x3), Oklahoma, Tennessee, Kentucky, Georgia, Missouri, Arkansas, New Mexico, and Mississippi.

Now, there has been lots of reporting in the last couple of weeks about national polls moving strongly away from Trump and toward Biden.

The view here looking at state polls is a bit more complicated. Some states moved toward Trump while some moved toward Biden. And combining that into an Electoral College view, Trump strengthened a bit at the start, then Biden made up all of that ground and then some by the end. Overall Biden is a bit stronger now than at our last update.

Anyway, let's start with the summary table of changes, then get into the graphs:

Model Metric 2 Jun 18 Jun 𝚫
Categories Trump Best
Expected
Biden Best
Trump +2
Biden +130
Biden +288
Trump +42
Biden +130
Biden +288
Trump +40
FLAT
FLAT
Tipping Point Biden +4.3 Biden +4.8% Biden +0.5%
Probabilities Trump 2σ
Median
Biden 2σ
Biden +22
Biden +122
Biden +234
Biden +32
Biden +132
Biden +242
Biden +10
Biden +10
Biden +8
Trump Win
Tie
Biden Win
0.5%
0.1%
99.4%
0.3%
0.1%
99.6%
-0.2%
FLAT
+0.2%

The most obvious change here is the improvement of Trump's best case in the categorization view by 40 EV.

For the categorization view, a state is considered to be "in play" if the margin is less than 5%. This change was driven by Biden's lead in the Pennsylvania poll average dropping below 5%.

You'll also notice that the "Expected Case" in the categorization view moved 30 EV toward Trump, then reverted to what it was before. This is due to the average in North Carolina moving from just barely Biden, to just barely Trump, and back again within the scope of a few days.

Those are the states that actually shifted categories since the last update. But let's look at the other close states that have new polling this time around too:

Trump is still leading Texas, but it has been moving toward Biden recently. A blue Texas would change the game completely. Biden isn't quite there. But it is close enough it should make the Trump camp nervous.

Biden has had a small lead in the Florida average all along, but it has been trending further in his direction lately. Another really strong poll and Florida might even move over the 5% threshold so we wouldn't even classify it as a close state. (Although it is just as likely that the TIPP poll showing Biden with a greater than 10% lead is an outlier and the average will soon revert to a small Biden lead.)

Ohio is surprisingly sparsely polled, but the average flipped from Biden to Trump in May and the one new poll in the last month didn't move the average at all.

Georgia is barely on the Trump side of the line, and trending toward Biden.

In Arizona Biden has moved from our "strong lead" to our "weak lead" category since May. He still leads, but Trump has been closing the gap.

Iowa is still a Trump state, but the average has been tightening there too.

If you count these up, you'll see five of the eight close states with new polls have been moving toward Biden, and three have been moving toward Trump.

When you mash all this together into our probabilistic model, you get this:

In the last couple of weeks, things moved toward Trump for a little bit, then started moving back toward Biden, and his median position is now 10 EV better than it was when we did our last update.

The switch from things improving for Trump to improving for Biden does seem to be just a few days after the death of George Floyd, just when the aftermath of that event was dominating the news.

In terms of the median case of our simulations, the last time Biden was doing better than he is today was on December 5th.

In terms of probability of winning though, Biden was better off much more recently, on May 19th.

The tipping point is also moving back toward Biden again:

Biden is back where he was in May, but you have to go back all the way to October to find a time he was doing better.

All of this looks very strong for Biden. But remember how fast the tipping point can shift. In 2016 on at least a couple of occasions, it moved 5% or 6% within just a few weeks. And so far in 2020, we have seen rapid swings of nearly 3%. The bigger the movement, the bigger the news event that has to happen to drive the change. But given the last few years, who can doubt the potential for big news events that can change a campaign overnight?

Over the last few days, there have been tons of commentators talking about Biden's national polling being much stronger than Clinton was during the 2016 campaign. Some are saying he is doing better than she EVER did on that metric. That is probably true. But we don't elect people by popular vote.

How does this look in our views?

<139 Days Out> 2016 2020
Expected Case Clinton by 144 Biden by 130
Tipping Point Clinton by 6.1% Biden by 4.8%

So yeah. Biden might well be doing better in terms of national popular vote polling than Clinton was at this stage. But Clinton was doing better when you factor in the state polls and the Electoral College.

Let's be very clear here. Biden is in a very strong position right now in terms of the Electoral College as well as the popular vote. Very strong.

But so was Clinton at the same point in the campaign. People were talking about landslides.

Then there was a bit of a roller coaster. Clinton was a lot weaker by the time we got to mid-September. Then she recovered and was strong again by mid-October. But then she collapsed again in the last few weeks. On election day, she was still favored, but it was clear Trump had a path to win and a Trump victory was very possible.

(That's looking at the Election Graphs analysis of course. Famously, lots of other sites didn't show things to be quite that close at the end. Election Graphs was one of only a handful that did.)

That kind of roller coaster may not happen this year. Biden may stay strong through the rest of the race. We shall see. But nobody should be getting overconfident at this stage.

OK, so to round it out, here is the spectrum of the "weak" states that are actively in play, plus the "strong" states that might be brought into play with some big improvements by one side or the other:

And of course the current map:

And that is where we are.

138.7 days until polls start to close.

For more information:

This post is an update based on the data on the Election Graphs Electoral College 2020 page. Election Graphs tracks a poll-based estimate of the Electoral College. The charts, graphs, and maps in the post above are all as of the time of this post. Click through on any image for current interactive versions of the chart, along with additional details.

Follow @ElectionGraphs on Twitter or Election Graphs on Facebook to see announcements of updates. For those interested in individual poll updates, follow @ElecCollPolls on Twitter for all the polls as I add them. If you find the information in these posts informative or useful, please consider visiting the donation page.

Is Impeachment Hurting the Dems?

Once again, too long between updates. Since the last one on November 20th there have been new polls in Wisconsin (x3), New Hampshire, California (x2), Texas (x2), Arizona, Michigan, Pennsylvania, and Iowa.

If I'd been posting weekly as I had intended, there would have been a post looking at how Election Graphs win odds correlated to the popular vote polling averages at RCP and using that to measure the Democrats' Electoral College disadvantage. (Spoiler: As of right now it looks like Democrats need an approximately 6% popular vote win margin to have a 50/50 chance of winning the Electoral College.)

I also would have done a post talking about how while for us political junkies, it seems inconceivable that people don't know who at least the top five Democratic candidates are, this article in the Washington Post by Robert Griffin lays out a pretty convincing case that a large part of the differences in polling between Democratic candidates vs. Trump is STILL simply that lots of people don't know who some of these people are, something that clearly would change before Election Day if they were to win the nomination. That is obviously very important when interpreting what we see here on Election Graphs, which is 100% driven by state level general election matchup polls.

But I kind of missed talking about both those points when they were fresh. So I'll let the above suffice for now.

The main thing I want to look at today is this:

This shows the median electoral college result from the Election Graphs Monte Carlo simulations for each candidate pair.

For both Biden and Sanders there is a very clear "V" shaped pattern. (OK, if you look closely, you can see a "W", but the large scale pattern is a "V".) In the first half of this V, as new polls came in and the state averages moved from the baseline based on the 2000-2016 elections, to averages based on 2020 polls, almost every poll made Sanders and Biden look better.

Although he dropped out already, you can see a smaller scale version of this "V" pattern with O'Rourke as well.

Buttigieg doesn't show the initial portion of this V. As the initial polls came in, his results basically just stayed at about the same level as the 2000-2016 baseline. Frankly though, there was very little Buttigieg vs. Trump state level polling in this time frame though, which would explain that.

But all four of these candidates share the second half of the "V". Starting at a specific point in time, as new polling results came in, more often than not, the state averages would move away from the Democrat, and so the median electoral college results would also move the same direction.

When you look at the charts, the inflection point seems to be… well… very close to the date Nancy Pelosi announced that she was officially backing the impeachment investigation over the Ukraine issue.

There are some ups and downs, and you could argue that the best numbers for the Democrats were perhaps a little bit earlier or a little bit later, but roughly speaking, leading up to that announcement on September 24th, every week Sanders, Biden, and O'Rourke looked better in the polls than the week before, while Buttigieg held steady. After September 24th though, Sanders, Biden, Buttigieg, and O'Rourke all looked weaker in each subsequent week.

This may be a coincidence. It may just relate to which pollsters put out polls in which states during this time period. It may just be a "reversion to the mean" after a series of earlier polls were more favorable to the Democrats than was really reflected by the ground truth…

But the location of the inflection point compared to the date when impeachment moved from something a few people were talking about, to a real thing that was happening is hard to ignore. It certainly LOOKS like the impeachment efforts are hurting the Democratic candidates in state level head to head polling against Trump. State after state where the polling averages had moved from red to blue, have now slipped back into the red again.

But wait. There has been one clear omission in the discussion above.

Warren's trend line just does not match the patterns followed by the four candidates discussed above. (Neither did Harris's for that matter, but she dropped out, so we won't spend more time there.) There is no clear "V" shape like Sanders, Biden, and O'Rourke. Nor does Warren follow Buttigieg's pattern.

In fact, Warren does not show any change that looks like it coincides with impeachment events. Initially as early polls came in, she looked like she would do worse than the median based on the 2000-2016 averages. Then she started doing a little better than that. And now she's a little worse again.

The fact that the reversal in fortunes that lines up with the impeachment announcement does not seem to apply to Warren (or Harris) seems notable. After all, Warren was one of the first Democratic candidates to come out strongly in favor of impeachment, and she did so strongly. If there was an impact from impeachment, why wouldn't it touch her? Perhaps even more than other candidates? This may in fact be an argument toward this movement NOT being tied directly to impeachment, but rather to something unrelated.

Without information that specifically digs into motivations rather than just candidate preferences, it is hard to say anything definitive. But the alignment certainly is suggestive. We shall see if those trends continue as the impeachment saga continues to play out.

In the mean time, we'll close out by looking at the changes in our main metrics since the last update post, followed by some of the state level charts.

Dem 20 Nov 16 Dec 𝚫
Biden +158 +126 -32
Sanders +88 +56 -32
Warren +50 +24 -26
Buttigieg -6 -44 -38

All four still active Democrats weakened significantly in their "median case" from the Election Graphs Monte Carlo simulations.

This  has not been a good month for the Democrats.

Dem 20 Nov 16 Dec 𝚫
Biden 99.9% 99.4% -0.5%
Sanders 95.0% 83.5% -11.5%
Warren 81.8% 67.2% -14.6%
Buttigieg 44.3% 16.5% -27.8%

Given how far ahead Biden was, his win odds are still 99%+. But the other three took pretty big hits these last few weeks.

Especially Buttigieg. Last time he was already the only candidate with a less than 50% chance of beating Trump. But now that has dropped to a paltry 16.5%.

Dem 20 Nov 16 Dec 𝚫
Biden +210 +178 -32
Sanders +118 +58 -60
Warren +48 +14 -34
Buttigieg +6 -86 -92

Using the older and simpler "expected case" where every candidate simply wins every state where they lead the polls, you see similar across the board drops to what you see in the "median case" from the simulation, but the drops are even more dramatic.

Dem 20 Nov 16 Dec 𝚫
Biden +4.4% +4.3% -0.1%
Sanders +1.8% +1.0% -0.8%
Warren +0.6% +0.6% Flat
Buttigieg +0.2% -1.3% -1.5%

For the tipping point, which measures how much of a gain would be needed to flip the electoral college winner if that gain occurred uniformly across all states, Warren manages to stay flat, but the other three Democrats get weaker.

Finally, presented without additional commentary, the updated state charts in each of the states with new polling. You'll notice the recent swing toward Trump occurs in almost every state.

The news is coming hot and heavy these days. Iowa is now less than 50 days away. Things will develop quickly. Stay tuned.

323.6 days until polls start to close.

For more information:

This post is an update based on the data on the Election Graphs Electoral College 2020 page. Election Graphs tracks a poll-based estimate of the Electoral College. The charts, graphs, and maps in the post above are all as of the time of this post. Click through on any image to go to a page with the current interactive versions of that chart, along with additional details.

Follow @ElectionGraphs on Twitter or Election Graphs on Facebook to see announcements of updates. For those interested in individual poll updates, follow @ElecCollPolls on Twitter for all the polls as I add them. If you find the information in these posts informative or useful, please consider visiting the donation page.

Goodbye Blue Texas

It has once again been longer than I like, but it is time for another post.

Since the last update, there have been new polls in Michigan (x3), Nevada (x2), Texas (x2), Arizona (x2), Pennsylvania (x3), Florida (x2), North Carolina (x3), Wisconsin (x2), Iowa, Georgia (x4), and New Hampshire.

That is a lot. Sorry about that. Oops.

The most striking individual state result is that after peaking in September with three of the top six Democrats leading Trump in the Texas averages, and two more getting close, Texas has been moving back toward Trump.

As of now, none of the six most polled Democrats lead in Texas, and only three of those keep Trump's lead to less than 5%. And one of those (O'Rourke) has already dropped out of the race, leaving only Sanders and Biden still making it close.

Converting this to win odds, Biden has a 30.1% chance of winning the state (if the election was today), and Sanders has a 25.0% chance of winning.

None of the rest (except O'Rourke, who is out) is above 5%.

So Texas is reverting to form. It may be closer than it has been in previous years, but at least for the moment, the Democratic hopes for a blue Texas seem to be fading.

I'll go over other states with new polling at the end of the post, but first, a look at four ways of looking at the changes in the national summary since the last post.

O'Rourke vs. Trump is now in the top six best-polled candidate combinations (replacing Sanders vs. Pence). But since O'Rourke dropped out, we will leave him out and only look at the top five for now.

I haven't done posts showing the update-to-update comparisons for the older "categorization method" before, but since that used to be the bread and butter of Election Graphs, let's start there.

Dem 1 Nov 20 Nov 𝚫
Biden +254 +210 -44
Sanders +190 +118 -72
Warren +38 +48 +10
Harris +20 +20 Flat
Buttigieg +6 +6 Flat

In this "expected case" view, where every candidate wins every state where they lead in the poll average, both Sanders and Biden have lost ground.

Warren improves her position a little.

Harris and Buttigieg are flat.

Dem 1 Nov 20 Nov 𝚫
Biden +5.3% +4.4% -0.9%
Sanders +4.7% +1.8% -2.9%
Harris +1.4% +1.4% Flat
Warren +0.3% +0.6% +0.3%
Buttigieg +0.6% +0.2% -0.4%

Looking at the tipping points, which is analogous to the popular vote, but adjusted for the structure of the electoral college, once again, Warren is the only Democrat who is improving.

Harris is flat.

Biden, Sanders, and Buttigieg all lose ground to different degrees.

Note that while for a short time, Biden had a tipping point greater than 5%, meaning he could win using only states where he led by more than 5% and didn't even need any swing states. That is no longer true.

Now all five of these Democrats have tipping points indicating that they need to win at least some tight states to win.

Now, moving on to the more elaborate probabilistic model I look at a bit more these days…

Dem 1 Nov 20 Nov 𝚫
Biden +184 +158 -26
Sanders +124 +88 -36
Warren +36 +50 +14
Harris +8 +12 +4
Buttigieg -4 -6 -2

This view shows the "Median Case." The median case is the electoral vote margin in the exact middle of the 1,000,001 simulation runs done for each candidate combination when sorted by the margin. About half the time, the Democrat does better than this. About half the time, they do worse.

Warren and Harris both improve a bit. Biden, Sanders, and Buttigieg all lose ground.

Note that while in the classification view, all five Democrats lead Trump, in the probabilistic view, Buttigieg's median case is actually to lose.

But we need to look at probabilities, not the single "median case" estimate. You should not think that since a candidate is ahead or behind on the median case, that maps to winning and losing.

For instance, Buttigieg's median case is a six electoral vote loss to Trump. But if you look at the 2σ range, that is the range of outcomes that you would expect to occur 95.45% of the time; you get a range from Buttigieg winning by 92 electoral votes to Trump winning by 90 electoral votes.

There is a huge range of possibilities. It isn't just "Trump is ahead in the median case, so he wins."

So time to look at the win odds…

Dem 1 Nov 20 Nov 𝚫
Biden 100.0% 99.9% -0.1%
Sanders 98.3% 95.0% -3.3%
Warren 73.1% 81.8% +8.7%
Harris 54.6% 58.2% +3.6%
Buttigieg 46.0% 44.3% -1.7%

The trends above, which cover just under three weeks, show Warren and Harris improving, while Biden, Sanders, and Buttigieg slip back.

But looking back a bit farther, we can see some overall trends going back to September.

Biden continues to be pegged at a 99%+ chance of winning. His lead in some states has slipped, but his overall margins are high enough that this hasn't yet started to impact his chances of winning.

Sanders never had as big a margin lead as Biden, so as some of those states slipped back toward Trump, you see a more significant impact on his odds of winning.

Harris and Buttigieg have never done all that much better than a coin flip against Trump, peaking at around a 70% chance of winning. But since September, they have both dropped significantly, with Buttigieg now only at a 44.3% chance of winning, and Harris only at 58.2%.

The only candidate consistently improving over the last few months has been Warren. She bottomed out at only a 41.7% shot of winning in June, and while there have been ups and downs, the trend is clearly in Warren's direction.

We will, of course, see if that lasts. As can be seen by the spike toward Trump in June, trends can reverse quite quickly.

Now, besides Texas, here are a few additional states where there are trends worth noting. (Since so many places had new polls, I'll skip a few where there is less to comment on.)

No clear trends in Florida except to note that it is an exceptionally close race no matter which Democrat you match up against Trump. As has been usual for the last few presidential races, Florida is right on the line. And it is big. So it makes a huge difference.

The general trend in Pennsylvania, Michigan, and Wisconsin over the past few months has been for Democratic leads to decrease.

In Pennsylvania, there is no Democrat with more than a 5% lead, and Warren (and O'Rourke) are both slightly behind.

At this point, only Biden has a lead higher than 5% in Michigan.

In Wisconsin, all the Democrats still lead, but none by more than 5%.

These are, of course, the three states that gave Trump his victory in 2016. At the moment, they are all looking to be close battlegrounds once again.

The people who say that the Democratic nominee needs to pay close attention to these states are certainly not wrong.

And at the moment, the Democrats seem to be slipping in all three.

Pollsters have not paid as much attention to Georgia as I would like. But there have been a bunch of polls in the last few weeks and they show a competitive state, which is a significant change from the historical average.

The poll average now shows Biden and Sanders ahead, with Warren, Harris, and Buttigieg all bringing Trump's lead under 5%.

(O'Rourke has never been polled in Georgia, and now that he has dropped out, probably will never be. Sniff.)

Although Biden has reversed a bit recently, overall Sanders, Warren, and Biden are making North Carolina a narrowly fought battleground.

Harris and Buttigieg, while they are still keeping Trump's lead under 5%, do not seem to be gaining any additional ground lately.

Arizona has also been moving toward the Democrats. At least for Sanders, Warren, and Biden. Warren and Biden actually are slightly leading. Sanders brings Trump's lead under 5%.

Harris and Buttigieg, on the other hand, are not making things much closer than the historical 7.6% average Republican margin in the state. Just as in North Carolina, they lag behind the stronger Democrats.

So Trump is gaining in Texas, Pennsylvania, Michigan, and Wisconsin.

But the Democrats are gaining in Georgia, North Carolina, and Arizona.

As we get closer, things will heat up; there will be lots more polls and more movement.

Are we having fun yet?

349.7 days until polls start to close on election night.

For more information:

This post is an update based on the data on the Election Graphs Electoral College 2020 page. Election Graphs tracks a poll-based estimate of the Electoral College. The charts, graphs, and maps in the post above are all as of the time of this post. Click through on any image to go to a page with the current interactive versions of that chart, along with additional details.

Follow @ElectionGraphs on Twitter or Election Graphs on Facebook to see announcements of updates. For those interested in individual poll updates, follow @ElecCollPolls on Twitter for all the polls as I add them. If you find the information in these posts informative or useful, please consider visiting the donation page.

Some Dems Up, Some Dems Down

Since last week's update, there have been two new polls in Texas, and one each in New Hampshire, Wisconsin, Michigan, and Pennsylvania.

For the six most polled Democrats against Trump (Biden, Sanders, Warren, Buttigieg, Harris, and O'Rourke) this week saw the following category changes:

  • Biden vs. Trump state category change: WI has moved from Strong Biden to Weak Biden
  • Sanders vs. Trump state category change: MI has moved from Strong Sanders to Weak Sanders
  • Warren vs. Trump state category change: PA has moved from Weak Warren to Weak Trump

Looking at how this impacted the national numbers:

In the "Expected Case" where every candidate wins every state they are ahead in, Pennsylvania moving from slightly Warren to slightly Trump means that Warren now only has a 12 electoral vote margin in this scenario.

Biden's lead dropping below 5% in Wisconsin, and Sander's lead dropping below 5% in Michigan meanwhile, improves Trump's "Best Case" (where he wins not only all of the states where he is ahead but also all the ones where he is behind by less than 5%) against both candidates.

As these five states moved around, several tipping points moved as well.

  • Sanders vs. Trump tipping point change: Sanders by 1.3% in FL -> Sanders by 1.9% in TX
  • O'Rourke vs. Trump tipping point change: O'Rourke by 2.1% in TX -> O'Rourke by 1.8% in TX
  • Biden vs. Trump tipping point change: Biden by 4.9% in VA -> Biden by 4.6% in WI

The tipping point changes already make it clear that this was a mixed week. Some Democrats improved their positions against Trump; some moved in the opposite direction. This mix in the situation exists despite all of the actual category changes being in Trump's direction.

The movement within categories that starts to be shown by the tipping point is something we can see even better if we switch to the probabilistic simulation view:

Dem 8 Sep 15 Sep 𝚫
Biden 99.9% 99.8% -0.1%
Sanders 95.8% 96.9% +1.1%
O'Rourke 80.0% 80.8% +0.8%
Warren 71.1% 69.0% -2.1%
Buttigieg 67.1% 67.0% -0.1%
Harris 64.3% 65.7% +1.4%

There were a lot of mixed results in a lot of states this week. But when you balance it all out in terms of the odds of winning in the Electoral College:

  • Harris, Sanders, and O'Rourke strengthened against Trump
  • Warren, Biden, and Buttigieg weakened against Trump

Another way of looking at this is the median margins in the Monte Carlo simulations for each of the candidate combinations:

Unlike the categorization "Expected Case" view, where Harris narrowly loses to Trump, all six Democrats lead Trump in the probabilistic "Median Case."

Warren, Buttigieg, and Harris all lead by 30 or less electoral votes, though. In electoral vote terms, that is a very tight race. Any state with more than 15 electoral votes slipping to the other side of the centerline would switch the outcome.

O'Rourke, Sanders, and Biden, each, in turn, have more of a margin than the candidate before, so additional buffer. They can afford for more to go wrong before they lose.

Now, let's look at each of the five states that got polled this week.

Of the two polls in Texas, one matched other recent surveys and was mostly favorable to the Democrats. The other showed clear Republican leads.

The net impact of the two polls on the averages was that Biden, Sanders, Warren, and Harris continued improving in Texas, while Buttigieg and O'Rourke got a little weaker.

You do get one piece of oddness, though. I use the methodology I described in my Polling Error vs. Final Margin post to translate margins into odds of winning. When I use the straight-up unsmoothed numbers from the 2008 to 2016 elections, there are a few places where the relationship between the polling average and the win odds are not monotonic.

This behavior means that even though O'Rourke's polling average dropped from a 2.1% lead to a 1.8% lead, my computed "odds of winning Texas" for O'Rourke INCREASED from 66.4% to 68.2%.

Now, I could have smoothed the handful of places where this non-monotonic behavior occurs out of existence. But I chose when I first set up the odds based view to use the numbers that came straight out of my analysis of the previous election results without any further manipulation.

Given the historical data and the methodology I used, a 1.8% Democratic lead is indeed slightly more likely to win than a 2.1% Democratic lead. Of course, this does seem a little crazy and is quite possibly just an example of overfitting.

But it isn't a huge difference and occurs at only a few specific margin zones, so I'll leave it be.

We still end up with a situation where three of the six Democrats are leading in Texas though, and another two make it close.

Texas continues to be one of the most important states to watch.

Sanders improved slightly in Pennsylvania.

Biden's lead deteriorated significantly.

Warren lost her lead to Trump.

The poll did not include the others.

Biden, Sanders, and Warren all weaken against Trump in Michigan with this week's update.

The poll did not include the others.

Warren got stronger in Wisconsin.

Biden and Sanders got weaker.

The poll did not include the others.

Biden, Sanders, Warren, and Harris all got weaker in New Hampshire with the latest polling.

The poll did not include Buttigieg or O'Rourke.

And that's it for this week's update.

415.6 days until polls start to close.

For more information:

This post is an update based on the data on the Election Graphs Electoral College 2020 page. Election Graphs tracks a poll-based estimate of the Electoral College. The charts, graphs, and maps in the post above are all as of the time of this post. Click through on any image to go to a page with the current interactive versions of that chart, along with additional details.

Follow @ElectionGraphs on Twitter or Election Graphs on Facebook to see announcements of updates. For those interested in individual poll updates, follow @ElecCollPolls on Twitter for all the polls as I add them. If you find the information in these posts informative or useful, please consider visiting the donation page.

Most Dems improve against Trump while Buttigieg weakens

Since the last update, there have been new polls in Kentucky, Wisconsin, Michigan, Pennsylvania, Florida (x2), and North Carolina.

With this update, Buttigieg vs. Trump has now replaced Harris vs. Trump in the "five best-polled candidate combinations" that I'll spend time on in these posts. The new list is Biden, Warren, Buttigieg, Sanders, and O'Rourke all paired against Trump.

With this latest batch of polls, we have the following changes:

National:

  • Trump best case vs. Warren has changed: Warren 214 to Trump 324 -> Warren 204 to Trump 334
  • Warren vs. Trump tipping point change: Trump by 0.1% in VA -> Warren by 0.5% in FL
  • Warren vs. Trump expected case change: Warren 266 to Trump 272 -> Warren 295 to Trump 243
  • Buttigieg vs. Trump tipping point change: Buttigieg by 1.1% in IA -> Buttigieg by 0.6% in PA
  • Sanders vs. Trump tipping point change: Sanders by 1% in IA -> Sanders by 1.3% in FL
  • Sanders vs. Trump expected case change: Sanders 272 to Trump 266 -> Sanders 301 to Trump 237
  • O'Rourke vs. Trump tipping point change: Trump by 0.5% in FL -> Trump by 0.3% in FL

Wisconsin:

  • Warren vs. Trump state category change: WI has moved from Strong Warren to Weak Warren

Florida:

  • Sanders vs. Trump state category change: FL has moved from Weak Trump to Weak Sanders
  • Warren vs. Trump state category change: FL has moved from Weak Trump to Weak Warren

OK, now on to the graphs… national first:

In the expected case, Warren and Sanders improve.

Notably, Warren moves from losing to winning with this update.

Biden continues to do the best of the five Democrats, while O'Rourke does worst.

In the tipping point Warren, Sanders, and O'Rourke improve, while Buttigieg weakens.

Once again, you can see Warren crossing the center line to the winning side of the field.

And once again Biden does best, while O'Rourke does worst.

Now let's look at all of the states that had new polls in this batch:

In Florida, O'Rourke and Buttigieg are losing to Trump. Biden, Warren, and Sanders are beating him. As usual, Biden is doing better than the other Democrats. But every single matchup here is close. As usual, Florida is a battleground state.

Biden has a healthy lead in Pennsylvania. O'Rourke is behind Trump. All the rest of these five Democrats are in the lead over Trump, but they have narrow edges.

In Michigan, Biden is doing best. Warren is doing worst. Strong leads for Biden, Sanders, and Buttigieg. Warren and O'Rourke ahead by narrower margins. But all five leading Trump.

Biden is the only Democrat leading Trump in North Carolina. But all five would make it a close race. North Carolina is definitely in play at the moment.

In Wisconsin, Biden is doing best, O'Rourke is doing worst. Biden and Sanders have strong leads. Buttigieg, Warren, and O'Rourke are only weakly ahead. But all five lead Trump.

Kentucky is very very solidly red. With the very first Kentucky poll of the cycle, it looks like it may be even redder this time around than the average of the last five elections.

OK, finally, the odds based view of the national race. I haven't managed to make this view live on the site with graphs and all yet, but I'll at least provide updates here.

Democrat Dem Odds Trump Odds Tie Odds
Biden 99.4% 0.4% 0.1%
Sanders 86.0% 13.2% 0.8%
Buttigieg 65.5% 33.2% 1.2%
Warren 53.2% 44.9% 1.9%
O'Rourke 50.5% 47.8% 1.7%

For those missing Harris, since she dropped off my top five, she's now at a 62.3% chance of winning.

It remains striking what a vast range there is between Biden on the one end of this spectrum and O'Rourke on the other.

As always, the caveat that things can and will change, but if the election was today, Biden looks about as close to a sure thing as you can get (much stronger than the median odds based view Clinton had, which was at about 86% on election day), while meanwhile, O'Rourke looks like a coin toss.

If this is primarily due to name recognition and the main deciding factor for voters is simply Trump vs. non-Trump, you should see the divergence between various Democrats reduce over the next few months as people get to know some of the lesser-known Democrats.

If on the other hand, these kinds of differences persist as we get closer to the Iowa caucuses, then there are lots of people out there where the choice of Democrat does indeed influence them on their final vote. In this case, "electability" becomes a valid criterion for Democrats to consider when making their choices in the primaries and caucuses.

3.1 days until the first Democratic debate.

499.1 days until polls start to close on Election Day 2020.

For more information:

This post is an update based on the data on the Election Graphs Electoral College 2020 page. Election Graphs tracks a poll-based estimate of the Electoral College. The charts, graphs, and maps in the post above are all as of the time of this post. Click through on any image to go to a page with the current interactive versions of that chart, along with additional details.

Follow @ElectionGraphs on Twitter or Election Graphs on Facebook to see announcements of updates. For those interested in individual poll updates, follow @ElecCollPolls on Twitter for all the polls as I add them. If you find the information in these posts informative or useful, please consider visiting the donation page.