Early Voting Tracker- Elections Lab

[texasdago 7378]

1 hour ago, Js1 said:

That’s because Lupe was bad. Awful. Dreadful. She didn’t even try 

Funny, I voted for her and still wanted to type, "who!?"

[Nice Guy Eddie 13276]

4 hours ago, bolverk said:

 

Mid-sized cities away from the border that rate poorly:

• Corpus Christi
• Amarillo proper (not the suburbs to its south)
• Temple/Killeen/Copperas Cove
• Wichita Falls
• Abilene
• Odessa (Midland's doing okay)
• Victoria

 

 

are you still talking about voting or is it general commentary about those cities?

[Js1 82428]

Just now, texasdago said:

Funny, I voted for her and still wanted to type, "who!?"

We all voted for her out of necessity.  She was fucking awful.  I literally don't recall her even running a campaign

Andrew White would have at least tried. 

[bolverk 28892]

6 minutes ago, Huckleberry said:

I was actually going to look into doing something like this so I appreciate what you're doing. What did you use as a factor for the trend continuation? For example, if a county went 40% Clinton then 42% Biden are you saying 44% this year or are you continuing but dampening the trend? I would want to look at historical info but without enough time to devote to it, my first instinct would be to cut the last 4 year difference in half so call the estimate for the above hypothetical county 45% Harris this year instead of 46%.

I'm not a stats wizard like yourself, so I just cheated and used the FORECAST formula option in Excel (using 2012, 2016, and 2020) to guestimate what 2024 should look like. I honestly have no idea if that was a defensible option, but that's what I went with. 

Most of the rural red counties showed a huge drop in support for the Dem candidate from 2012 to 2016. Some continued that decline in 2020, and so the trend continued to show them getting redder in 2024. Some dropped in 2016, but support inched up in 2020 toward Biden and the 2024 output kind of split the difference between 2016 and 2020. The ones that were steadily getting bluer continued to do so.

[Bateshorn 9489]

21 minutes ago, Huckleberry said:

I was actually going to look into doing something like this so I appreciate what you're doing. What did you use as a factor for the trend continuation? For example, if a county went 40% Clinton then 42% Biden are you saying 44% this year or are you continuing but dampening the trend? I would want to look at historical info but without enough time to devote to it, my first instinct would be to cut the last 4 year difference in half so call the estimate for the above hypothetical county 45% Harris this year instead of 46%.

 

6 minutes ago, bolverk said:

I'm not a stats wizard like yourself, so I just cheated and used the FORECAST formula option in Excel (using 2012, 2016, and 2020) to guestimate what 2024 should look like. I honestly have no idea if that was a defensible option, but that's what I went with. 

Most of the rural red counties showed a huge drop in support for the Dem candidate from 2012 to 2016. Some continued that decline in 2020, and so the trend continued to show them getting redder in 2024. Some dropped in 2016, but support inched up in 2020 toward Biden and the 2024 output kind of split the difference between 2016 and 2020. The ones that were steadily getting bluer continued to do so.

[Huckleberry 32153]

Okay, that's a linear trend then. I think it will overstate the vote trends for this election but it's not a bad calculation to look at. With another step it might be better. If you used FORECAST on the difference in vote shares instead of on the vote shares themselves and then added that estimated 2020-2024 difference to the 2020 vote share it would be a better reflection, I would think off the top of my head.

As an example, if a county went 40% Obama, 42% Clinton, 43.5% Biden then FORECAST on the vote shares themselves will give you 45.3% Harris. But if you run FORECAST on the deltas (2%, 1.5%) then it will of course estimate a 1% delta this election, and you will have a prediction of 44.5% Harris in this election.

Of course this also has risks for counties that reversed trends. So probably not worth the effort for message board posting.

Edited Wednesday at 06:08 PM by Huckleberry

[Js1 82428]

[heso 2557]

On 10/29/2024 at 11:26 AM, Js1 said:

Nate Cohn touched on this - it is people picking up the phone for a survey, saying "I am voting for Trump!" and hanging up the phone and not completing the survey. 

Why are those types of responses even counted? 

[Huckleberry 32153]

Just now, heso said:

Why are those types of responses even counted? 

Because pollsters believe that completely discarding them is what made their product suck in 2016. So as is human nature they are overcorrecting and counting them all as Trump votes.

[WBT 20133]

Huck's being a real 3rd derivative right now

[Js1 82428]

1 minute ago, Huckleberry said:

Because pollsters believe that completely discarding them is what made their product suck in 2016. So as is human nature they are overcorrecting and counting them all as Trump votes.

And in 2020 too.  

[Bateshorn 9489]

That moment when the Surl makes me feel like a paste eating child is always special. 

Edited Wednesday at 06:13 PM by Bateshorn

[hookem2010 2510]

I'm still working on subtracting 3-digit numbers in my head, with varying degrees of success.

[bolverk 28892]

16 minutes ago, Huckleberry said:

Okay, that's a linear trend then. I think it will overstate the vote trends for this election but it's not a bad calculation to look at. With another step it might be better. If you used FORECAST on the difference in vote shares instead of on the vote shares themselves and then added that estimated 2020-2024 difference to the 2020 vote share it would be a better reflection, I would think off the top of my head.

Just shared a link to a copy of my spreadsheet in your DMs if you want to play with the numbers.

[Js1 82428]

28 minutes ago, bolverk said:

Just shared a link to a copy of my spreadsheet in your DMs if you want to play with the numbers.

Is this how nerds sext?

[Huckleberry 32153]

Sort of. I sent him back a more conservative version that dampens the trends. It has Harris at 46.3% of the vote.

Of course his county percentages are based on early voting turnout but using full election vote shares from 2020. I don't think there's a better way to do it, though, considering that reasonable people expect the early versus election day voting split to be 2022ish, not 2020ish. Just saw that poll that said 50% plan to vote on Election Day versus 37% in 2020.

[dieucla98 2767]

9 minutes ago, Js1 said:

Is this how nerds sext?

"Democrat firewalls of 390k in PA and cross tabs" do it for you better? 😁

For the record, I support everyone's efforts here to make sense of what is going on.

[Js1 82428]

5 minutes ago, Huckleberry said:

Of course his county percentages are based on early voting turnout but using full election vote shares from 2020. I don't think there's a better way to do it, though, considering that reasonable people expect the early versus election day voting split to be 2022ish, not 2020ish. Just saw that poll that said 50% plan to vote on Election Day versus 37% in 2020.

Yep.  We are moving to a totally different model than 2020, so really hard to just compare things to 2020. 

[bolverk 28892]

5 minutes ago, Huckleberry said:

Sort of. I sent him back a more conservative version that dampens the trends. It has Harris at 46.3% of the vote.

Of course his county percentages are based on early voting turnout but using full election vote shares from 2020. I don't think there's a better way to do it, though, considering that reasonable people expect the early versus election day voting split to be 2022ish, not 2020ish. Just saw that poll that said 50% plan to vote on Election Day versus 37% in 2020.

Yeah, it doesn't consider at all which way the Early Voters might swing, but like you said, I don't know how you'd calculate that in Texas. That said, 36% of registered Texas voters were already in the bad as of yesterday, but I haven't looked at EV turnouts at the local level in past elections.

[bolverk 28892]

2 minutes ago, Js1 said:

Yep.  We are moving to a totally different model than 2020, so really hard to just compare things to 2020. 

The spreadsheet also has county results for 2012 and 2016, for whatever that's worth.

[heso 2557]

18 hours ago, Aqua Buddha said:

I keep saying this.  Ever since Dobbs, Dems have outperformed all of the polling in actual elections by about 5-8 points.

Because women are afraid to speak openly to a pollster but vote alone. 

[sushihorn 3215]

1 hour ago, Huckleberry said:

Because pollsters believe that completely discarding them is what made their product suck in 2016. So as is human nature they are overcorrecting and counting them all as Trump votes.

When response rates are <5%, I bet they’re glad of any response at all. 

[heso 2557]

4 hours ago, Aqua Buddha said:

Georgia's early vote and mail in total is already at 75% of the 2020 total.  Granted, the state has grown more so there's that.  Both parties are definitely cannibalizing election day vote.

Georgia almost 57% women, BTW.  Harris ain't losing GA.

GA was 56% women in 2020. Depending on how Election Day shakes out it would be a stretch for that to slip back low enough to flip the state back to red. 

[sushihorn 3215]

3 hours ago, heso said:

GA was 56% women in 2020. Depending on how Election Day shakes out it would be a stretch for that to slip back low enough to flip the state back to red. 

With EV being so huge, the ballots cast so far should look a lot like the final electorate. That’s bad news for Harris. Young adults are 10.5% of the 2024 EV vs 16.2% of the 2020 total vote. Black vote share is down a point (28.9 v 29.9). Urban share is 16.7% v 19.0% 4 years ago and rural is taking share from the suburbs as well as the cities.

https://targetearly.targetsmart.com/g2024?count_prefix=final_eday_voted_count_&demo_filters=[{"key"%3A"age"%2C"value"%3A"All"}]&state=GA&view_type=state

You can’t replicate the 2020 electorate, much less the Obama Coalition when young voters and urban voters stay home. How do you move the numbers when 70-75% of the votes are already cast? The shortfall of black support is smaller a can be made up by the 25-30% yet to vote but that is not enough by itself. Like NV, the remaining votes have to look radically different for Kamala to win. 

Edited Wednesday at 11:21 PM by sushihorn

[Bozo_Casanova 25529]

4 minutes ago, sushihorn said:

With EV being so huge, the ballots cast so far should look a lot like the final electorate. That’s bad news for Harris. Young adults are 10.5% of the 2024 EV vs 16.2% of the 2020 total vote. Black vote share is down a point (28.9 v 29.9). Urban share is 16.7% v 19.0% 4 years ago and rural is taking share from the suburbs as well as the cities.

https://targetearly.targetsmart.com/g2024?count_prefix=final_eday_voted_count_&demo_filters=[{"key"%3A"age"%2C"value"%3A"All"}]&state=GA&view_type=state

You can’t replicate the 2020 electorate, much less the Obama Coalition when young voters and urban voters stay home. How do you move the numbers when 70-75% of the votes are already cast? The shortfall of black support is smaller a can be made up by the 25-30% yet to vote but that is not enough by itself. Like NV, the remaining votes have to look radically different for Kamala to win. 

[Firemans4Horn 16872]

Nate Silver so take it with a tablespoon of salt

 

 

[heso 2557]

3 hours ago, heso said:

GA was 56% women in 2020. Depending on how Election Day shakes out it would be a stretch for that to slip back low enough to flip the state back to red. 

I meant to say that it *wouldnt* be a stretch for that gender split for fall back a couple of points and flip the state red. 

[sushihorn 3215]

Early voting was generally a small part of total turnout in 2016. FL was a major exception: half the vote was early. Notice that the error there was 2.6% which is better than most traditional polls did that year. If 50% is reasonably predictive, I’m going to say 70-75% should be also. 

[Dahobbs 17097]

13 minutes ago, sushihorn said:

Early voting was generally a small part of total turnout in 2016. FL was a major exception: half the vote was early. Notice that the error there was 2.6% which is better than most traditional polls did that year. If 50% is reasonably predictive, I’m going to say 70-75% should be also. 

The fact that it randomly correlated doesn't mean it should be thought of us predictive. We know that there are often differences in early voting and election day voting. This is true across states and across elections. You're assumption is simply not warranted. 

[heso 2557]

The advantage didn't get us much.
The Republicans have mostly controlled the State House, the Senate, and won many statewide elections in that time. Not to mention the 2016 Presidential election in Pennsylvania. 2022 was the first time that Senate seat was won by a Democrat since...1962. 
The PA D registration advantage didn't seem to actually mean advantages in votes. If it has been eroding away for 20 years than why were results in 2020 and 2022 better than the results in 2014 and 2016? That doesn't seem like something eroding away.
But I don't know all the dynamics of Pennsylvania politics or what it really means to be registered D.
 

That advantage did show up in election results. At its peak it gave Obama the largest MOV in an election in PA since Nixon-McGovern.

Before hypothesizing about why the change happened, some numbers:

Since 2008, 2,127,861 Pennsylvanians changed their party affiliation thusly:

D to R: 633,000
I to D : 372,000
R to D : 366,000
I to R: 289,000
D to I: 272,000
R to I: 196,000

Giving Republicans a gain of 360,000.

Unsurprisingly these changes have peaks in election years, the biggest being 2016.

Both on a total and percentage basis this has happened in southwestern PA more than any other part of the state. A Democratic stronghold for a century, it’s now a super trumpy area. It was the land of steel mills and coal mines and unions. I wouldn’t be surprised if southwestern PA had the biggest electoral shift in the country over the last 30 years. Politically, PA used to be referred to as a T because southeast and southwest were solid blue. Now Allegheny County is an island in the west.

I think this is the single biggest factor in that change:



You have a bunch of rural communities that never recovered. Plus a lot of latent racism, and FOX news constantly telling them that big city coastal elites are working against them. Add in that anyone smart enough to go to college gets the fuck out.

Then here comes Trump promising to put everything back to the way it used to be. When you have dozens of communities that peaked in the 1930s-60s and Trump is telling them in simple words that he can fix it.

Which is borne out in the annual data. 2016 was the largest shift of voter registration. 113k D to R in 2016. The rise of MAGA. It’s been declining since then: 81k in 2020, 71k in 2024. As have shifts in other directions as well.

Rural PA, lead by SW PA is making the same shift that SE Ohio and West Virginia are/have made. Dying small towns that are looking for someone to blame. Trump gives them someone to blame.

[heso 2557]

A silver lining on the reddening of PA: the largest age group of republicans in PA is 55-64 followed by 65-74. The largest age group for democrats is 25-34. Every age group except 55-65 and 75+ is larger for Dems over Reps. 

[sushihorn 3215]

I’ve been looking for this data and finally found it. Women are not a monolithic group and they are a large group so it makes sense to drill down. Pew Research looked at political preferences among women by relationship status and there are huge group differences. 

 

Married, widowed and divorced women are narrowly divided by party preference: married R+5, widowed D+1, divorced D+7. Cohabitators are D+31 and never married are D+48. Lesbian and Bi are D+71.

It’s pretty clear where the gender gap comes from and why it looks like the Grand Canyon in socially liberal cities. I think cultural attitudes towards marriage are a major reason for the rural-urban contrast. 

[TwiceHorn 30733]

7 hours ago, WBT said:

Huck's being a real 3rd derivative right now

Third derivative of what with respect to what?  Otherwise, I got that reference.

[Dahobbs 17097]

17 minutes ago, sushihorn said:

I’ve been looking for this data and finally found it. Women are not a monolithic group and they are a large group so it makes sense to drill down. Pew Research looked at political preferences among women by relationship status and there are huge group differences. 

 

Married, widowed and divorced women are narrowly divided by party preference: married R+5, widowed D+1, divorced D+7. Cohabitators are D+31 and never married are D+48. Lesbian and Bi are D+71.

It’s pretty clear where the gender gap comes from and why it looks like the Grand Canyon in socially liberal cities. I think cultural attitudes towards marriage are a major reason for the rural-urban contrast. 

You're conflating cause and effect here. 

Edited yesterday at 02:08 AM by Dahobbs

[Bateshorn 9489]

2 hours ago, Firemans4Horn said:

Nate Silver so take it with a tablespoon of salt

 

 

Nate Silver is a 1.5 hit wonder who has devolved into a degenerate gambler. He is right about one thing: we don’t know shit about fuck.

[Pato del Muerto 51917]

42 minutes ago, sushihorn said:

I’ve been looking for this data and finally found it. Women are not a monolithic group and they are a large group so it makes sense to drill down. Pew Research looked at political preferences among women by relationship status and there are huge group differences. 

 

Married, widowed and divorced women are narrowly divided by party preference: married R+5, widowed D+1, divorced D+7. Cohabitators are D+31 and never married are D+48. Lesbian and Bi are D+71.

It’s pretty clear where the gender gap comes from and why it looks like the Grand Canyon in socially liberal cities. I think cultural attitudes towards marriage are a major reason for the rural-urban contrast. 

This is why the libs are using the schools to make as many gays as possible!!

[sushihorn 3215]

2 minutes ago, Pato del Muerto said:

This is why the libs are using the schools to make as many gays as possible!!

Hollywood says “hold my Chardonnay.”

[StassneyHorn 6411]

[Huckleberry 32153]

3 minutes ago, StassneyHorn said:

So their big takeaway is that in consecutive midterm elections each party did better when they weren't in the White House. How insightful. 

[StassneyHorn 6411]

I don’t know how you are going to find reliable relationship stats on registered women voters, so I give you race breakdowns.

[Valmy77 12527]

Obviously if we have a 2018 type result, we will have a landslide win.

But we don't need a landslide win, just a win.

[sushihorn 3215]

8 minutes ago, StassneyHorn said:

I don’t know how you are going to find reliable relationship stats on registered women voters, so I give you race breakdowns.

Target Smart actually has marital status as a demographic slice. It’s binary though, married v non-married. It’s not very detailed but it provides another directional indicator. 

[Valmy77 12527]

1 hour ago, sushihorn said:

I’ve been looking for this data and finally found it. Women are not a monolithic group and they are a large group so it makes sense to drill down. Pew Research looked at political preferences among women by relationship status and there are huge group differences. 

 

Married, widowed and divorced women are narrowly divided by party preference: married R+5, widowed D+1, divorced D+7. Cohabitators are D+31 and never married are D+48. Lesbian and Bi are D+71.

It’s pretty clear where the gender gap comes from and why it looks like the Grand Canyon in socially liberal cities. I think cultural attitudes towards marriage are a major reason for the rural-urban contrast. 

That doesn't make any sense. Why would it look like the Grand Canyon in Liberal cities? The trends seem pretty consistent for both genders.

 

 

Edited yesterday at 03:18 AM by Valmy77

[StassneyHorn 6411]

11 minutes ago, sushihorn said:

Target Smart actually has marital status as a demographic slice. It’s binary though, married v non-married. It’s not very detailed but it provides another directional indicator. 

What states ask for relationship status? What states update this? Are they cross referencing marriage and divorce records?

Edited yesterday at 03:25 AM by StassneyHorn

[berlinerbaer 663]

 

Top line for PA: Even

For reference, the actual VBM partisan split in PA as of this Monday is 58-31-10 D-R-I. Giving the GOP all the indies, and assuming no other partisan switching makes it 58-41. This poll clearly seems to have oversampled Republicans.

MOE of 9% though, so small sample size of like 100.

[BamaATL 15750]

29 minutes ago, berlinerbaer said:

 

Top line for PA: Even

For reference, the actual VBM partisan split in PA as of this Monday is 58-31-10 D-R-I. Giving the GOP all the indies, and assuming no other partisan switching makes it 58-41. This poll clearly seems to have oversampled Republicans.

MOE of 9% though, so small sample size of like 100.

We talked to 10 people at Taco Bell....

MOE of 9%

[atomheartbevo 88753]

9 hours ago, Js1 said:

Yep.  We are moving to a totally different model than 2020, so really hard to just compare things to 2020. 

How are the COVID deaths factoring in?

[elfenix 34001]

2 hours ago, Bateshorn said:

Nate Silver is a 1.5 hit wonder who was always a degenerate gambler. He is right about one thing: we don’t know shit about fuck.

ftfy

[scottsins 1921]

With EV being so huge, the ballots cast so far should look a lot like the final electorate. That’s bad news for Harris. Young adults are 10.5% of the 2024 EV vs 16.2% of the 2020 total vote. Black vote share is down a point (28.9 v 29.9). Urban share is 16.7% v 19.0% 4 years ago and rural is taking share from the suburbs as well as the cities.
https://targetearly.targetsmart.com/g2024?count_prefix=final_eday_voted_count_&demo_filters=[{"key"%3A"age"%2C"value"%3A"All"}]&state=GA&view_type=state
You can’t replicate the 2020 electorate, much less the Obama Coalition when young voters and urban voters stay home. How do you move the numbers when 70-75% of the votes are already cast? The shortfall of black support is smaller a can be made up by the 25-30% yet to vote but that is not enough by itself. Like NV, the remaining votes have to look radically different for Kamala to win. 

Has this been addressed yet by anyone on the “other side” of this analysis? I read it when I woke up to take a piss, and it’s making it difficult for me to go back to sleep now.

[Valmy77 12527]

7 hours ago, heso said:

That advantage did show up in election results. At its peak it gave Obama the largest MOV in an election in PA since Nixon-McGovern.

In one election Obama did well. Good for him.

But otherwise the Republicans have done well there. Especially in 2010. So if this advantage has been eroding away why would the results in 2022 be so much better for the Democrats than in 2010? Pretty worthless advantage...except for Obama in 2008.

Edited 20 hours ago by Valmy77