Slate/Votecastr real time election projections

[muon2]

I think that this model is inaccurate. They know only turnout and project results based on pre-election polling. In real exit polls they ask voters after they vote, and here they don't aske anyone about anything. This makes no sense to me.

This is the model that big campaigns have used for years, and one of the consultants on this project is Obama's former director of microtargeting. That was a very successful electoral forecast model. It's also the model used by networks to make state calls. Exit polls are less accurate in general since their sample size is smaller than this method.

[muon2]

Isn't Colorado a state where GOP wins early vote and Democrats election day vote? So it makes sense?

Yes. More Dems turn in ballots on election day

CO is all vote by mail now.

So, this is projection on 1,53m votes in Colorado. Anyone can tell me for who is this good? Who votes on election day more in CO, D or R?

This is similar to the RCP polling average (43.3-40.4-6.1-3.1) with undecided votes spread proportionally to the candidates (46.6-43.5-6.6-3.3). I'd say that's good for Clinton in CO.

[muon2]

I think that this model is inaccurate. They know only turnout and project results based on pre-election polling. In real exit polls they ask voters after they vote, and here they don't aske anyone about anything. This makes no sense to me.

This is the model that big campaigns have used for years, and one of the consultants on this project is Obama's former director of microtargeting. That was a very successful electoral forecast model. It's also the model used by networks to make state calls. Exit polls are less accurate in general since their sample size is smaller than this method.

Exit polls cannot be less accurate because they interview people after voting, and only real voters, not likely one.

This model is also based on real voters. The CO Secretary of State identifies everyone who voted by name and location in their database, just not who they voted for. The statistics are huge. It's better than a small sample of exit voters where you know who they voted for.

[muon2]

I think that this model is inaccurate. They know only turnout and project results based on pre-election polling. In real exit polls they ask voters after they vote, and here they don't aske anyone about anything. This makes no sense to me.

This is the model that big campaigns have used for years, and one of the consultants on this project is Obama's former director of microtargeting. That was a very successful electoral forecast model. It's also the model used by networks to make state calls. Exit polls are less accurate in general since their sample size is smaller than this method.

Exit polls cannot be less accurate because they interview people after voting, and only real voters, not likely one.

This model is also based on real voters. The CO Secretary of State identifies everyone who voted by name and location in their database, just not who they voted for. The statistics are huge. It's better than a small sample of exit voters where you know who they voted for.

I agree. But I have a question, is there some sort of margin of error like in the polls?

Of course there's an MoE like any other sampling technique. There's a component based on the model analytics and a component based on the sample size. Polls that weight their sample based on demographics are relying on the same sort of analytics as this model so that factor is going to be comparable, and unfortunately its never reported with the poll results. The sample size in most polls is around 1000 which gives the 3-4% MoE that is reported. Here the sample size is more than a hundred times larger and 1000 times larger n the case of CO. That puts the MoE from sample size at a few tenths of a percent.

[muon2]

One thing to remember about this model is that it is like the play-by-play for a sports game. It will move over the day based on when certain voters cast their votes. However if there are large leads that build up early, it will show just like a team that scores a lot early in the game.

[muon2]

They are missing votes both there and in Florida. 

I know they are. But they are using voter file for EV in all states, right?

If it was a good model it should.  Considering how bad it is so far I doubt it.

What, specifically, has been so awful about it so far? Do we know there are systematic issues in the algorithms they are running or the data they are pulling? We know some of the data is a bit off perhaps, but do you have tonights results to compare the current projections too? Because if you don't I really don't think its wise to label it as a 'good' or 'bad' model at this point.

It's almost impossible to talk about how good the model is until the polls close in a specific state. At that point the "game" is over and the model should forecast the result well before the networks do. The exception is CO where the votes are cast by mail so they should be coming in as an unbiased sample so that is the best test of the model before polls close.

[muon2]

Just read their "How It Works" article again.

Only now do I see that this has ZERO exit poll component. This is just yet another pre-election day poll masquerading as actual voting data.

So much for all their hype about breaking the exit poll embargo.

You didn't already realize that? Only Edison Research conducts those.

They never said they were breaking the exit poll embargo. They are breaking the embargo on analytical data based on turnout that the networks and campaigns watch all day as the polls are open.

[muon2]

I'm off now for a while. I have to go collect some midday turnout counts to send in to party HQ to feed into our models. Just like votecastr is doing.

[muon2]

I'm off now for a while. I have to go collect some midday turnout counts to send in to party HQ to feed into our models. Just like votecastr is doing.

I'm back. My polling place had about 25% vote early and 30% at the door by 1:00. That seems like a fairly normal turnout.

[muon2]

Clinton now up almost 300k in Florida. I am starting to doubt this one.

FL is shot. It's over if they are right.

The key disclaimer was the comment on the livestream that Cuban vote is driving that. If their estimate of second generation Cubans isn't as Dem as they think it is, then the model will shift the prediction a bit.

[muon2]

And the bad news for Trump keeps coming:

Update: 3:06 p.m.: The state maps previously displayed on this page represented only Election Day data but not early vote estimates. They have been removed and will be reposted when they are updated.

You freaking mean they just allotted the damn early vote based on today's vote.  God damn they are amateur hour.

Actually, they did have maps based just on early vote earlier today (Wisconsin/Iowa/Ohio were much more blue, FL about the same) and then those maps turned quite a bit more Trump friendly. I think they are saying they forgot to merge the data from Map 1 into Map 2, and that Map 2 was just ED vote. I think.

That's the way I see it too. Their biggest problem seems to be integrating the actual data model with the multi-platform displays.

[muon2]

Pennsylvania: C+3
Clinton 48%
Trump 45%

Florida: C+3
Clinton 48%
Trump 45%

Colorado: C+2
Clinton 46%
Trump 44%

Iowa: T+1
Clinton 45%
Trump 46%

Nevada: C+1
Clinton 46%
Trump 45%

Ohio: T+1
Clinton 45%
Trump 46%

Wisconsin: C+5
Clinton 48%
Trump 43%

New Hampshire: C+4
Clinton 47%
Trump 43%

Which would mean:
http://www.270towin.com/maps/wOZZQ

Where are these numbers from? Are they projections for end-of-day, or the current state?

This is the current state based on votes cast, not a projection.