538 Model Megathread

[Likely Voter]

Other than AZ, it looks ok ...

I'd give AZ to Trump in a close race, because the state's laws that will block Latinos from voting.

well if you look at the polls plus version, the model gives AZ and NC to Trump. So the 'projection' is that 2016 would be exactly the same as 2012 from an EV point of view. However, currently the model has Clinton ahead in AZ and NC with polls only and 'now-cast'

[Likely Voter]

If you sort the states by Trump's %chance to win, then NH is actually the tipping point at 32.8% vs 27.6% for VA. 

So Trump's narrow path ( sorted by % chance to win) is Romney + OH + CO + IA + FL + NH

[Likely Voter]

Perhaps Clinton going after NE-2 is their backup plan in case Trump wins ME-2. That would be pretty crazy outcome withe the race coming down to NE and ME and their weird rules.  You can bet the NE legislature would be kicking themselves for not going through with the idea of switching to WTA.

[Likely Voter]

bear in mind that these models are only as good as the available data. The biggest input is state polling. Half the battlegrounds have had little to no polling in last few weeks, so that leads to anomalies like NC looking more D than IA.

[Likely Voter]

I continue to be skeptical of the IA, NH, NV, ME2 bank shot, but that is where the polls say is Trump's best shot.  The biggest question would be, are the NV polls right this time and has it really moved to be more R than VA, CO and PA?  Also is NH really more R than NC? Really? probably not.

There is a reason why Nate Silver's book was titled "The Signal and the Noise"

[Likely Voter]

Here is how Polls-Plus sorts the battlegrounds...

MI   D+7.3     
WI   D+7.1       
MN   D+5.3     
VA   D+5.3       
CO   D+4.6       
PA   D+3.7

NATL AVG   D+3.6   
   
NV   D+3.0       
NH   D+2.3       
ME2   D+1.7     
IA   D+1.6   
FL   D+1.5     
OH   D+1.2       
NC   R+0.0   
AZ   R+3.7       
NE2   R+3.8       
GA   R+4.4       
MO   R+5.6 
 

[Likely Voter]

IIRC the NYC model doesn't factor in national polls and they aren't using these 50 state online polls.  538 model basically uses pretty much every poll there is (albeit with weighting) and those national polls are used to create momentum at the state level. So there is a level of garbage in/garbage out with 538, especially in periods where there are fewer live voter polls coming out, yet lots of data from these tracking polls and '50 state' polls.

[Likely Voter]

I think the issue is that Silver's model most wants lots of live caller state polls. If every day it had Fox, Monmouth, Marquette, Marist, CNN, etc state (and national) polls feeding it, it would work best. It is essentially an engine tuned for those. But this cycle the vast majority of data being put into the model are the online polls and notably the trackers and more recently, the '50 state polls'.  If they knew what data was coming this cycle, I bet they would have built the model differently, but they have what they have. 

I suspect that after the debates there will be more live-call polls and they will increase in frequency both at the state and national level. Then perhaps you will see less of a disparity between it and the Upshot model. 

[Likely Voter]

RNC Nate has somehow 'forgot to add in the Monmouth poll that has Clinton up 5 in Florida or the St Leo  poll. Of course Nate did remember to add in the Monmouth poll that has Trump up 8 in Iowa. Nate is looking much less credible this cycle.

they are there. The polls are sorted by weight not just age. Although the weighting is always a bit hard to figure out. There are five polls weighted higher than the Monmouth poll (which is the most recent and from a A+ rated pollster)

[Likely Voter]

It appears the debate about 538 model vs other models has got the notice of Nate Silver. Here is a series of tweets he sent out early this morning...

You must be logged in to read this quote.

He also retweeted this earlier they came up with comparing models....

[Likely Voter]

If you compare the 538 model to the other models, they all agree on almost every state. All agree on Clinton winning her 'firewall' states (MI, WI, PA, MN, NH, CO, VA). All agree Trump holds AZ and GA and all agree he picks up IA and OH. Almost all also agree on Trump winning ME2. So if those picks are off, 538 team can just point to polling errors and say how everyone got it wrong. 

However, where they are out on a limb is NV, NC and FL. They are the only ones saying Trump is ahead in those states. And this is where he Huffpo article comes into play because it is the 538 "trendline adjustment" that gives Trump the lead in those states. Even 538s own adjusted polling averages have Clinton ahead in those 3 states. So if Clinton wins, they can't say "it was the polling, everyone else got those wrong." If Trump wins all 3, Nate Silver will be deemed an uber genius who saw what no others could see.

But if Trump loses all 3, the 'secret sauce' of the 538 (aka the 'trendline adjustment') is going to get a lot of flack. You can see some walkback  in the article about Nevada, but again the other models are saying Trump will lose so they are alone, and can't just blame the polls (even though NV polls have sucked in the past).   

Of course by Tuesday morning the 538 model may magically see the roughly 2% trendline adjustment (in Trump's favor) go down or even reverse, and then their model will probably align with Upshot and others.

[Likely Voter]

As I  noted before, the sole reason the model had Trump up in NC, FL and NV was the "trend line adjustment" which was pushing  every state a bit more than 2 points in Trump's direction.  Today that has been cut in half and that flips those three states back to Clinton.  Now 538 aligns with all other models in terms of state wins, but still is more bullish on Trump in terms of margins and odds .

Was it purely data that changed the trend line in the last few hours?