About that PPP poll
Thursday, Jul 30, 2015 - Posted by Rich Miller
* Dave Fako wrote a very insightful comment yesterday which explains why the new PPP poll isn’t all that helpful to gauge the general election contest…
PPP has a good track record, and I have no comment on accuracy or methodology, but high quality standards and reliability and disclosure are and always have been vital to me and our polling firm (Fako Research & Strategies). Below is the methodology statement from the poll which discloses the breakdown of partisanship by primary voting history and the mode - in this case 80% of interviews are via phone (presumably landline) and 20% via Internet among households without landlines.
My only comment about the poll’s methodology is this: their methodology statement says the poll was conducted among 931 registered voters, which was made up of 409 Democratic Primary voters and 369 Republican Primary voters, leaving 153 non-primary voters in their sample.
Is PPP saying only 16% of Illinois general election electorate is made up of non-primary voters (frequently IDed as “Independent”) on most Illinois voter files?
In 2012 ~ 69% of all general election voters were NOT Democratic or GOP Primary voters.
He’s right.
So, we probably shouldn’t give this particular poll too much weight regarding general election issues.
* But it does have some use for primary contests, so back to PPP…
Scott Walker leads the Republican field in the state with 23% to 18% for Donald Trump, 11% for Jeb Bush, 8% for Chris Christie, 7% for Ben Carson, 6% for Marco Rubio, 5% each for Mike Huckabee and Rand Paul, 4% for Ted Cruz, 3% for Carly Fiorina, 2% each for Bobby Jindal and Rick Santorum, 1% each for Lindsey Graham, John Kasich, and Rick Perry, and the dreaded less than 1% rounding to 0 for Jim Gilmore and George Pataki.
Walker is very popular in his neighboring state, with a 64/15 favorability rating. To put those numbers into perspective, no other Republican hopeful does better than a 51% favorability in the state. Walker leads both with voters describing themselves as ‘very conservative’ and ’somewhat conservative,’ while Trump is the choice of moderate Republicans.
On the Democratic side Clinton is dominant. She gets 60% to 23% for Sanders, with Martin O’Malley at 4%, Jim Webb at 3%, and Lincoln Chafee at 1% rounding out the field. Clinton gets at least 54% within every group we track by ideology, gender, race, and age with her support peaking among African Americans with whom she gets 76% and Hispanics with whom she gets 72%.
- Mama - Thursday, Jul 30, 15 @ 11:18 am:
Scott Walker leads. Friends in Wisconsin (un-union jobs) said Walker ruined their state.
- Chupacabra - Thursday, Jul 30, 15 @ 11:33 am:
Who are the nine people who voted for Lincoln Chafee? What are their names and phone numbers? I need to talk with them, they need help. Lincoln Chafee is the former Gov. of Rhode Island who had worse public approval ratings than Dr Walter Palmer. Chafee’s economic platform consists of introducing the metric system.
- A guy - Thursday, Jul 30, 15 @ 11:41 am:
Thank you for this additional analysis. It puts things into much greater context.
- Tournaround Agenda - Thursday, Jul 30, 15 @ 11:43 am:
McLean County is larger than Rhode Island. Just saying.
- Robert the Bruce - Thursday, Jul 30, 15 @ 11:47 am:
==Chafee’s economic platform consists of introducing the metric system.==
I far prefer the metric system to trickle-down economics.
- Shore - Thursday, Jul 30, 15 @ 11:58 am:
those are great numbers for walker, who I don’t like , because I see trump eventually collapsing and his supporters fleeing to the strongest conservative bush challenger.
I love MSK, but this should be a wake up call to him and the squad to clean up his game.
- Anon - Thursday, Jul 30, 15 @ 12:29 pm:
Interesting. I usually have a lot of faith in PPP
- Keyser Soze - Thursday, Jul 30, 15 @ 2:01 pm:
I am a math-science person so polling has always been a source of fascination. Fako’s commentary is insightful and illustrates the ever increasing difficulty in pinning down reliable results. Consider landlines vs cell phones, the emergence of a great many limited issue candidates (e.g., 16 republican primary candidates, holy cow), the growth of partisan talk radio and television, etc.
Add to that the surprising factoid that 69% of voters in a general don’t vote in a primary. This calls into question a poll’s assumption that it is relying on “likely voters.” Apparently, the likelihood of anyone actually voting is very questionable, guesswork at best. In the instant case, the heavy reliance on primary voters, as representative of general voters, most likely yields results that are skewed to the partisans but which greatly discount 69% of the voters. So, are the results reliable? I mean no disrespect to PPP because all pollsters face the same issues and most use similar models. I don’t have an answer to the inherent polling problems. I wish I did. Nonetheless, the process of forecasting elections continues to fascinate.
- Dave Fako - Thursday, Jul 30, 15 @ 2:07 pm:
I want to emphasize my comments were not directed at PPP in terms a reliability or accuracy, and in fact they have a good track record, and should be commended for disclosing enough details to make some assessments of the poll. My comments are more intended to emphasize the need to evaluate every poll publicly released before jumping to conclusions, etc. The fact is based on the methodology disclosure, this particular poll is simply composed of too many partisan primary voters - and if you look at the results of partisan cued questions, they closely track the partisan split (by vote history, not self ID). I would be very interested to see the results weighted by a more normalized ratio, although that would be excessive weighting that would not fit my firm’s quality standards.