I should just link

my post from the other thread but here goes.

What is the purpose of a March 2012 senate poll for Florida?  Frankly its mostly for amuesment because its not clear who the GOP nominee will be and the political climate for Nov 2012 is a little hard to determine right now.  So I don't get too excited about it.

My second point is what exactly is the proper way to poll for a Nov 2012 election in march 2012?  There are no set rules.  PPP is showing a pool of registered but active voters that are more democratic, less independent and more liberal in Nov 2008.  If you do believe that will occur you can buy the PPP results.  If you think that will occur or its pointless to guess about Nov 2012 turnout you can ignore the PPP polls.

Where I do have a problem with PPP is this.  In all these redo election polls, when ask if you were to vote today who would you vote, they are still using a projected Nov 2012 turnout model.  The same pool of voters for the Nelson poll was asked about Scott-Sink. I contend the buyers remorse polls for WI-OH-FL should not be using a 2012 turnout model but rather a 2010 or a march 2011 turnout model.  

That being said kudos to PPP for showing  their work.  I might add that other polls show that your mix of voters can have a huge impact on results.  

---

by: rdelbov @ Thu Mar 31, 2011 at 01:45:04 AM UTC

Why are you still talking about turnout models?

PPP isn't "using a projected Nov 2012 turnout model". They are polling registered voters who have voted in at least one of the three elections. That's it. And that is probably the right way to do things, because we don't know what the 2012 turnout will look like so applying a 2010 turnout model or a 2008 turnout model or a "march 2011 turnout model" (whatever that means) could easily skewer your results.  

Male, VA-08

---

by: drobertson @ Thu Mar 31, 2011 at 01:59:28 AM UTC

[ Parent ]

The partisan breakdown

of any poll determines or influences the results.  That's polling 101.

If I polled Florida, called 500 active registered voters like PPP did, and found 400 republicans my poll would be inaccurate.  There is no way I can randomly call 500 active registered voters and poll 80% republicans.

Yes PPP is just call active registered voters but what was the mix of his democrats, republicans and indies.  That's what determines his results.  I am comparing the blend of his poll respondents to exit polls to determine if his polling reflects what goes on in Florida's elections.  Thats polling 102.

I hope this helps.  

---

by: rdelbov @ Thu Mar 31, 2011 at 02:41:57 AM UTC

[ Parent ]

There are a couple things you don't seem to understand......

First, as andyroo has demonstrated in comments here over the years, and as I've found on my own by doing my own back-of-the-envelope math, changing a turnout model doesn't change toplines as much as you might think.  When you actually sit down and do the math, applying crosstabs to a different turnout model, the outcome rarely changes, and the margins, too, change very little.

Second, OK, yeah, the turnout model matters, but PPP isn't claiming to know what 2012 turnout will be.  They're simply coming up with whatever turnout model a random sample produces, and as it turns out their polls turn out to be by-and-large right.

Now, it's March 2011, so yes it's too soon to take any polls now to the bank.  PPP's polls looked better for Dems in March 2009 than later in that year.  So things can change.  But whatever polls say now tell us quite a bit about the state of public opinion today.

And regarding the turnout models being very different from 2009 which you've constantly harped on elsewhere, it's a given the turnout model will be far more favorable to Dems in 2012 than it was in 2010.  Just about every Republican campaign operative in America will concede that point.  The 2012 turnout will be a lot closer to 2008 than 2010, and maybe about the same as in 2008 given the fact that Republican-favoring demographics continue a linear decline, and Dem-favoring demographics, mainly nonwhites and young whites, continue linear growth as a share of the electorate.

43, male, Indian-American, Democrat, VA-10

---

by: DCCyclone @ Thu Mar 31, 2011 at 02:53:20 AM UTC

[ Parent ]

I would further clarify that

polls are not meant to predict the outcome except maybe in the last month of an election. Their main purpose is to show a snapshot of public opinion i.e what would happen if the election was today? PPP isn't making any prediction on the composition of the Nov. 2012 elctorate. What they're finding is the March 2011 electorate if the election was happening right now. The party id crosstabs allow us to track which types of voters are enthusiastic and engaged throughout the election cycle and the breakdown is by no means constant nor should it be.

21, Progressive Democrat, MN-08 (home), MN-05 (college)

---

by: JonathanMN @ Thu Mar 31, 2011 at 04:52:46 AM UTC

[ Parent ]

Exactly right

Excellent post.

---

by: conspiracy @ Thu Mar 31, 2011 at 11:17:09 AM UTC

[ Parent ]

Thanks! N/t

21, Progressive Democrat, MN-08 (home), MN-05 (college)

---

by: JonathanMN @ Thu Mar 31, 2011 at 20:12:41 PM UTC

[ Parent ]

mn-8 and mn-5.

Where exactly in the 8th are you from and where in the 5th do you live now? Didn't realize we were from the same districts.

26 White Male. Born and raised in MN-8, currently living in MN-5.

"A man who stands for nothing will fall for anything."

---

by: OGGoldy @ Thu Mar 31, 2011 at 18:01:38 PM UTC

[ Parent ]

I was born and raised in

Duluth but I go to college at the U of M in the Twin Cities. I could have gone to UMD but that was a little bit too close to home (only a ten minute drive max).

What about you?

21, Progressive Democrat, MN-08 (home), MN-05 (college)

---

by: JonathanMN @ Thu Mar 31, 2011 at 20:19:17 PM UTC

[ Parent ]

Me

I was born in Hibbing, and lived outside of Bovey as a boy. Family moved to the cities when I was in middle school. I also went to the U (and actually am a graduate student there now, in addition to working full time in the private sector). I live in Columbia Heights now, which is just north of campus about 6 miles.

26 White Male. Born and raised in MN-8, currently living in MN-5.

"A man who stands for nothing will fall for anything."

---

by: OGGoldy @ Thu Mar 31, 2011 at 22:48:44 PM UTC

[ Parent ]

I used to think partisan weighting should be used

Now I'm not so sure. SurveyUSA don't do it and neither do PPP. It can't be a coincidence these two firms have the best record over recent cycles.

---

by: conspiracy @ Thu Mar 31, 2011 at 11:03:12 AM UTC

[ Parent ]

How many times do you have to be told

That they weight for physical demographics? They have a track record for accuracy using the same methods and their current polling is right on target with everybody else. I seriously don't know what more you want.

---

by: conspiracy @ Thu Mar 31, 2011 at 10:55:27 AM UTC

[ Parent ]

To be right.

About everything.

19, Self Appointed Chair of the SSP Gay Caucus (I claimed it first :p), male, Dem, IN-09 (College IN-09) (Raised IL-03, IL-09)

---

by: ndrwmls10 @ Thu Mar 31, 2011 at 10:58:52 AM UTC

[ Parent ]

And to ignore polls

One does not like. We really should just shut up shop since it is pointless to discuss this stuff until much closer to election day!

---

by: conspiracy @ Thu Mar 31, 2011 at 11:06:53 AM UTC

[ Parent ]

Polling

You make it sound like PPP just randomly calls a set number of people, records their answers, and they're done with it.  That's NOT how polling works.  PPP sets demographic targets for each poll before conducting the poll and then weights the responses based on those demographic targets.  So when PPP says a poll is 48% Sink voters, 41% Scott voters, and 11% non-voters, those are the targets (or near the targets) that they set, which may or may not have anything at all to do with reality.

Why is PPP spending all this money and throwing all these polls out 20+ months before the election?  Easy answer...they are trying to set the narrative and drive up base enthusiasm.  There's no other reason.

All this buyers' remorse stuff is nonsense, really.  You can't look at the Florida poll and tell me it's not nonsense.  All the Florida poll tells me is that if you get a sample of majority Sink voters and people who didn't vote, they prefer Sink.  Well duh.  Buyers' remorse would be if you had a poll of majority Scott voters and it says they would prefer Sink.  Let's see a poll of 55% Scott voters and 45% Sink voters and see who comes out on top.

---

by: elucas730 @ Thu Mar 31, 2011 at 10:46:01 AM UTC

[ Parent ]

The buyers remorse stuff is interesting

But largely irrelevant. Again, the compare and contrast shows they are producing polls comparable to other firms.

---

by: conspiracy @ Thu Mar 31, 2011 at 10:51:08 AM UTC

[ Parent ]

To your last point

"16% of those who say they voted for him (Scott) would switch sides, as would 21% of all Republicans."

---

by: conspiracy @ Thu Mar 31, 2011 at 10:58:20 AM UTC

[ Parent ]

You basically ignored the diary

It clearly shows that their polling is comparable to other firms where data is available. If anything, I'd argue they are actually more often than not more favorable to Republicans than Democrats.

---

by: conspiracy @ Thu Mar 31, 2011 at 10:48:54 AM UTC

[ Parent ]

"Oversampling liberals" is silly, they should know better......

It's obvious that when you offer more ideological choices, you're going to get more people self-identifying in varying degrees of "liberal" and "conservative" than with just 3 total choices.  The RRH guys should be smart enough to recognize that, and also to recognize if liberals are oversampled, so are conservatives

There's nothing wrong with PPP except maybe their continued low favorability/job approval ratings, but those are across-the-board.  But at least they're not as bad as last cycle.

I gotta say, as you and others have noted I really like PPP dominating polling early this cycle, a stark contrast to Rasmussen last cycle.  Rasmussen lost almost all its remaining credibility with its 2010 flop...which is amazing since it's awfully tough to have an inaccurate GOP skew in a GOP wave year.

43, male, Indian-American, Democrat, VA-10

---

by: DCCyclone @ Thu Mar 31, 2011 at 03:01:19 AM UTC

Personally

I want to see Rasmussen return to their ways of 2002-2008. They really were the cream of the crop before 2009, and they have the potential to be back there. But their glaring inaccuracy in 2010 means they are on probation in my boom for the 2012 cycle. If they turn their stuff around, and get back on the ball, It would be good for the political community.

26 White Male. Born and raised in MN-8, currently living in MN-5.

"A man who stands for nothing will fall for anything."

---

by: OGGoldy @ Thu Mar 31, 2011 at 04:20:18 AM UTC

[ Parent ]

They weren't very good in 2008 either

Certainly not in swing states.

---

by: conspiracy @ Thu Mar 31, 2011 at 10:51:40 AM UTC

[ Parent ]

They are just placing too much emphasis

On unreliable crosstabs. And ignoring the fact PPP have a history of producing accurate toplines even though the inards of the polling looks nothing like the exit polls. Take NJ-Gov 2009 for example. The final PPP poll had Christie winning by 6 points. The partisan breakdown of the sample was D44-R36-I20. On election day Christie won by 4 points and the exit poll reported turnout as being D41-R31-I28.

---

by: conspiracy @ Thu Mar 31, 2011 at 11:15:27 AM UTC

[ Parent ]

Great comments but I am confused?

 No one here is shocked but that comment.  Please clarify for me what PPP has does?

1st do they weight each poll in each state?  Do they set targets for women, men, young people, old people, democrats, indies or republicans?

2nd or do they randomly call 800 people and just count the 1st 500 who response?

If they weight their polling pool its fair game to question their results because of how they decide to set their numbers. This is a judgement call on their part and its far to look at the basis of their decisions.  

If they do not have set goals for age, sex, party or ideology then its fair to judge what they came up with based on looking at exit data and other polls.

That has been the gist of my comments on PPP this week. I think David thought I was a smart*** for talking about the 7% percent solution for disappearing indies.  I regret that as I prefer to keep discussions on a higher plane.  My basic points are listed above and they are what they.  

---

by: rdelbov @ Thu Mar 31, 2011 at 12:26:33 PM UTC

Sigh

I'm not even going to reply because you obviously can't or won't read.

---

by: conspiracy @ Thu Mar 31, 2011 at 12:33:40 PM UTC

[ Parent ]

They do both 1 and 2.

They call say 1000 people. If 750 answer and take the poll then they count all of their responses. Next, they weight all the date for demographics but NOT PARTY ID. So if the sample was 60% women and 40% men, they would multiply each of those so that it matches some other number like exit polls from 2006, 2008, or 2010 or an average of some sort (maybe the average during recent presidential election years). Weighting by demographics is definitely justified because people don't usually change them. Party ID is much more fluid and can change often so they don't try and weight for that. For example, Democrat that is mad at the party for some reason may say they're independent when they really should be counted as a Democrat.

21, Progressive Democrat, MN-08 (home), MN-05 (college)

---

by: JonathanMN @ Thu Mar 31, 2011 at 14:08:17 PM UTC

[ Parent ]

Okay

I buy that.  I can read.  

Its fair game, however, to judge the results of polls based on how the weighting comes out. I have no problem with PPP not weighting based on partisan ID.  That's certainly their privilege.  

I reserve the right, however, to put where the partisan ID or ideological breakdowns differ either with other polls or with exit polls.  My point about 80% GOP for a New York stands up as its totally apporiate to judge polls by the party ID & Ideological breakdowns.  

---

by: rdelbov @ Thu Mar 31, 2011 at 15:15:19 PM UTC

[ Parent ]

If you can read

Why did you need it saying yet again? Everything JonathanMN told you has been pointed out to you ad nauseum for over a week.

---

by: conspiracy @ Thu Mar 31, 2011 at 15:20:04 PM UTC

[ Parent ]

I did not want to disrespect someone

who posted a response to my comment. I just assume that if someone addresses a point of mine its impolite not to response.  If someone says "good morning" to me I respond.

Just as I response to your comment to me.  I think I am clear on what PPP does.  I was not before I posted on David's thread the other day.  It is clear in my mind now.  

Thanks for starting this discussion and I appreciate your efforts in this posting all this poll information.  I know its just does not happen.  

---

by: rdelbov @ Thu Mar 31, 2011 at 15:45:14 PM UTC

[ Parent ]

You are welcome

So do you accept that PPP is right in line with all this other polling?

---

by: conspiracy @ Thu Mar 31, 2011 at 16:38:31 PM UTC

[ Parent ]

I would say that PPP's polls

have to be examined one by one and every one should be examined that way. I would say the very thing about every other poll that is listed. Where PPP & another pollster have similar results there you have it. Where PPP and another pollster have different results check to see where the difference is.

I say Kudos to PPP because they show their work and have lots of detail.

Here's two polls that I have looked.  Quinnepac national polls and PPP national poll of 03-16-2011. Q had the President at 42% unfavorable while PPP had the President at 47%. That's a five % difference and both polls are -+4%. You can say both polls are right.  

I will say that Q has the breakdown 35R-35D while PPP is 42D-38R.  Maybe that accounts for the 5% difference in the Presidents approval numbers?  The polls seem close so probably both are right as far as polls in march 2011 can be.  

---

by: rdelbov @ Thu Mar 31, 2011 at 18:54:56 PM UTC

[ Parent ]

Fair enough

That is a decent example. Though it could be that Quinnipiac are wrong. I would note the latest Gallup is 49-43 and the AP chimed in at 53-45 yesterday. Anyway, I don't want to tempt you into breaking your promise to David so I will finish with a quote from Republican pollster Whit Ayers talking about PPP.

"A lot of the numbers I see from them make sense to me."

http://www.politico.com/news/s...

---

by: conspiracy @ Thu Mar 31, 2011 at 20:41:00 PM UTC

[ Parent ]

You are getting really close to a banning

Because what you're doing is borderline trolling - but even worse, because an obvious troll would just get ignored. You, instead, are taking advantage of people's good nature and patience here. No one, and I mean no one, needs things explained to them as often as they have been here to you. And the explanations you've gotten have been crystal clear, repeated many times, and come from some of the smartest polling junkies you're going to find anywhere on the Internets.

You also promised you'd shut up about PPP the other day, but you aren't keeping that promise. So, this is your last chance.

---

by: DavidNYC @ Thu Mar 31, 2011 at 19:05:19 PM UTC

[ Parent ]

no more PPP

from me.  

---

by: rdelbov @ Thu Mar 31, 2011 at 20:05:25 PM UTC

[ Parent ]

I see not this PPP versus the rest

Some of the examples are not clear, and the Magellan poll for New Hampshire is in the same line than the resutls of PPP for close (in PVI) states.

---

by: abgin @ Fri Apr 01, 2011 at 21:22:19 PM UTC

I think they are quite clear

They all show basically the same thing whoever did the poll.

---

by: conspiracy @ Sun Apr 03, 2011 at 13:57:01 PM UTC

[ Parent ]