State of the race, mid September

[AUDub 11,148]

Probably the best example:

 

[McLoofus 35,182]

1 minute ago, AUDub said:

Probably the best example:

 

What does this mean?

Also, how many beers are y'all gonna drink tonight? I think I'm going to drink 7 beers. And maybe some bourbon.

[AUDub 11,148]

1 minute ago, McLoofus said:

What does this mean?

Also, how many beers are y'all gonna drink tonight? I think I'm going to drink 7 beers. And maybe some bourbon.

RCP's average has an arbitrary cutoff that results in the average not really reflecting the polls. The following post provides a good example. Here it is again. 

So those three partisan polls, Trafalgar, InsiderAdvantage and Susquehanna are artificially given more weight in the average than reputable firms like Monmouth, Reuters, NYT, TheHill or WaPo. 

[AUDub 11,148]

I expect to be drunk and lose count by 9pm lol. 

[McLoofus 35,182]

5 minutes ago, AUDub said:

RCP's average has an arbitrary cutoff that results in the average not really reflecting the polls. The following post provides a good example. Here it is again. 

So those three partisan polls, Trafalgar, InsiderAdvantage and Susquehanna are artificially given more weight in the average than reputable firms like Monmouth, Reuters, NYT, TheHill or WaPo. 

Ahhh, gotcha. I wasn't paying attention to the cutoff date. 

[TitanTiger 20,482]

1 hour ago, AUDub said:

Nate Cohn makes some valid points about the RCP average this year.  Thread:

 

I was noticing this early today.  I couldn't find any rhyme or reason as to why certain polls were part of the RCP average and others taken over the same time period weren't.  Seems very dishonest.

[AUDub 11,148]

14 minutes ago, TitanTiger said:

I was noticing this early today.  I couldn't find any rhyme or reason as to why certain polls were part of the RCP average and others taken over the same time period weren't.  Seems very dishonest.

They set some strange arbitrary cutoff. I understand including troll polls in the average, but having them bump more reputable firms from the average artificially alters the result and is not a good thing.

[TitanTiger 20,482]

2 hours ago, AUDub said:

They set some strange arbitrary cutoff. I understand including troll polls in the average, but having them bump more reputable firms from the average artificially alters the result and is not a good thing.

So I did some quick math, looking at the six battlegrounds they list (FL, NC, PA, MI, WI, and AZ).  All of them use odd and arbitrary inclusions or cutoffs to short Biden an average of 1.2 points per contest.  Here are some specifics:

Florida
RCP average:  Biden +0.9
If RCP included recent polls from Reuters, NYT/Siena, and Emerson, all of which fall within the same poll ending dates and have similar sampling sizes as the ones they did include, Biden's lead goes to +1.3

North Carolina
RCP average:  Trump +0.2
If RCP included CNN and NBC/Marist, it flips to Biden +1.


Pennsylvania 
RCP average:  Biden +1.2
If RCP included Monmouth, Emerson, Reuters and NYT/Siena, it goes to Biden +3

Michigan
RCP average:  Biden +4.2
If RCP included Reuters and CNN, it goes to Biden +6.1

Wisconsin
RCP average:  Biden +6.7
There's no poll listed that warrants being added, but interestingly here, they included two polls several days old (ended on 10/25) compared to the rest which all ended between 10/30 and 11/1.  And what do you know?  Both of the older polls are among the lower numbers for Biden.  Take those old polls out and stick with the most recent ones and it goes to Biden +8.

Arizona
RCP average:  Biden +0.9
Did a little combo thing here.  They included a poll that ended 10/28, which conveniently shows Trump up by three but left out CNN that is more recent (ended 10/30).  Swap those out so all the polls are 10/29 or newer and it's Biden +1.9.

 

There's simply no rhyme or reason to it.  But at the same time, it cannot be an accident either.  Not when all of them help Trump's numbers look better.  And that's before you even question whether Trafalgar, Rasmussen and Susquehana are accurate to begin with.

 

[AUDub 11,148]

Oh good you did the legwork.

I'm not entirely sure why they're doing it this way. RCP does lean right, but seemed be a fairly neutral arbiter on things like their average. I can only assume bad faith.