Here’s Why Quinnipiac’s Polls Showing Trump Leading Clinton Are Totally Wrong (GD 16)

[Her Sister]

By Jason Easley on Wed, Jul 13th, 2016 at 11:30 am
The media has hopped all over three new Quinnipiac University polls showing Trump leading Clinton in Pennsylvania and Florida, but the problem is that Quinnipiac is rigging their polls for the best possible Trump outcome.


The media has hopped all over three new Quinnipiac University polls showing Trump leading Clinton in Pennsylvania and Florida, but the problem is that Quinnipiac is rigging their polls for the best possible Trump outcome.

The new Quinnipiac polls show Trump leading Clinton 43%-41% in Pennsylvania, 42%-39% in Florida, and tied with Clinton 41%-41% in Ohio. These numbers do not match national polling, which has shown no movement towards Trump, so what’s going on here?

To understand why Quinnipiac’s polls are always more favorable to Trump, it is important to look at the assumptions that Quinnipiac makes about the composition of the 2016 electorate. Each Quinnipiac University poll assumes that white voter turnout will go up, African-American turnout will stay the same, and Hispanic turnout will drop.

The chart below shows white voter turnout in Florida, Ohio, and Pennsylvania in 2012 versus Quinnipiac’s 2016 projections:


If Quinnipiac’s white voter turnout was adjusted to actual voter turnout levels, Clinton would lead by 3 points in Pennsylvania, 5 points in Ohio, and be tied in Florida.

However, the oversampling of white voters is only half of the problem. Quinnipiac’s polling also underrepresents minorities.

http://www.politicususa.com/2016/07/13/quinnipiac-universitys-poll.html

[BlueNoMatterWho]

1. Correct. The Q poll is under sampling Dems by a few points as well.

If the turnout from 2012 remains unchanged in 2012 these swing states would either be a toss up or go Clinton by a couple of points. Only if turnout is as Q is sampling would you get these results. Possible, but not probable. We will know in November.

[rjsquirrel]

2. This doesn't make them wrong

Polls are not right or wrong. It is good to describe the assumptions they make about the electorate when comparing them but no poll can prospectively be guaranteed to describe the actual future electorate. Every poll has some bias and some margin of error even within its bias. Even if the assumptions are wrong successive polls can accurately capture movement over time and it makes sense Clinton would be back on her heels a bit after the Comey silliness.

It's important not to ever leap to despair or celebration over any single poll, and it's particularly speculative this far out, before conventions and VP picks and in a volatile and unprecedented election situation. The campaigns need to make assumptions and pivot with respect to different possible models of the electorate too.

[Her Sister]

3. Yep if I only poll white people results are "right"

for them!

[rjsquirrel]

4. Technically true

I'm not sure why it's funny but Q poll didn't only poll white people.

I'm glad you're sure everything is fine.

[Her Sister]

5. A- All right !

Technically!

[rjsquirrel]

7. In the sense of

being obvious, a truism, uninteresting, meaningless.

The question is whether the Q poll assumes lower Hispanic turnout especially incorrectly. There are a lot of reasons to think it does. But it's interesting to model if it doesn't and if white turnout is really as motivated as Q also assumes.

The point is no one knows. It's all speculation and inference and probability and model-dependent. The polls you like and the ones that scare you.

Maybe it's all a horserace for some folks. But polls also drive voter behavior in well understood ways. I'd rather have some scary polls to remind us that the only poll that matters is on Election Day. Results showing Clinton has it in the bag create a false sense of complacency. We need urgency. We damn well need to be scared of fucking this one up.

[Lord Magus]

9. They just assume that white people will be a larger share of the electorate anybody else predicts.

Why anybody would think that's the case is beyond me.

[rjsquirrel]

10. Because Trump appears to appeal to white grievance

The theory is a lot of disaffected white voters (their class profile is complicated but it's not just working class whites) who may not normally vote may be excited by Trump's overt appeal to their racism.

It's only a theory of the crime, as it were. But it's not implausible at all given our knowledge of the history of racist populism in the US and elsewhere.

[LanternWaste]

12. Yet the corollary seems just as plausible. Trump's candidacy may turn out minority voters

Because Trump appears to appeal to white grievance..."

Yet the corollary seems just as plausible. Trump's candidacy may in and of itself, turn out minority voters who may not normally vote because he scares the bejesus out of rational people who dislike racism.

[rjsquirrel]

18. I agree entirely

That's one of the interesting questions here.

I would suggest it's far too early to be sure of either scenario and that as in most elections GOTV will be crucial.

[Lord Magus]

20. That would assume there is any such vast untapped group of white racist voters.

The reality is that the racists were already showing up to vote Republican. Trump's overt racism makes them more enthusiastic, but that doesn't give them any extra votes.

[rjsquirrel]

22. There is no doubt about it really

plenty of white middle class people and even more white working class folks don't vote and describe themselves as politically alienated. This is pretty well documented. potental voters are not registered voters are not likely voters.

Same is also true for other groups. 40% of eligible voters never show up at all.

Can they be won? Are they susceptible to Trump's pitch? Are Latinos who wouldn't normally vote going to be motivated to oppose Trump? These are the questions of the day. He definitely turned out an unusually large number of new voters for a republican primary winner.

[Her Sister]

30. 20. That would assume there is any such vast untapped group of white racist voters.

Especially against a black candidate!

[Lord Magus]

35. Exactly.

The racists already showed up in droves to vote against President Obama. The people the media likes to euphemistically call "white working class" who are supporting Trump? They're the same racists who showed up in 08 and 12 to vote Republican because the Democratic candidate was black.

[rjsquirrel]

37. Trust me there's more of them out theye

plenty of racists left who didn't vote in 08 and 12.

Never underestimate the number of racists in America.

[aggiesal]

25. Here's their logo ...

[Her Sister]

6. I like Statistics and you're right!

It's important not to ever leap to despair or celebration over any single poll, and it's particularly speculative this far out, before conventions and VP picks and in a volatile and unprecedented election situation. The campaigns need to make assumptions and pivot with respect to different possible models of the electorate too.

[rjsquirrel]

8. Alright! Let's win this thing for team blue!

[Her Sister]

13. Yes! Let's stay Hungry!

and Thirsty!

[Bill USA]

17. the proper term is 'unreliable' due to sample not being an accurate representation of the likely

turnouts of various demographic groups.

[Hortensis]

36. Of course polls are right or wrong and usually wrong.

But some are far better than others, and some far worse. We're also in a recognized "summer doldrums" period where voters become relatively grouchy and disintersted and "standard deviations" bear no relation to the wild numbers showing up in polls.

In this highly charged election, it’s no surprise that the news media see every poll like an addict sees a new fix. That is especially true of polls that show large and unexpected changes. Those polls get intense coverage and analysis, adding to their presumed validity.

The problem is that the polls that make the news are also the ones most likely to be wrong. And to folks like us, who know the polling game and can sort out real trends from normal perturbations, too many of this year’s polls, and their coverage, have been cringeworthy.

Stop the Polling Insanity
http://www.nytimes.com/2016/05/20/opinion/stop-the-polling-insanity.html

[Gothmog]

11. The Q polls are becoming the favorite polls of the GOP

[toddwv]

14. I'd hate to see us "unskewing" polls like the right has done in the past.

Mostly because I use that as a point to browbeat conservatives whenever I'm given the chance...

[Lord Magus]

21. It's not unskewing the polls, it's discarding ones that make racist assumptions about the electorate

And that's exactly what Quinnipiac is doing by projecting that blacks and Hispanics will stay home if there isn't a POC on the top of the ticket.

[Cicada]

33. Quinnipiac just thinks the 2012 Exit Polls undercounted white voters

The exit polls had 72% white voters in 2012. But the Census polls showed it was 74% and a well-respected Democratic-linked database using actual voting records had it 76%. So Quinnipiac guesses it will be 73% in 2016. They may be wrong but they ma be correct.

Exit polls are not always correct.

A great discussion of this was in the New York Times blog called Upshot. Google Upshot 2012 exit poll to see why Quinnipiac may be right.

[andym]

15. Unskewing polls is not a great idea

Better to use meta methods like 538 does...

[Bill USA]

16. Quinnipiac had to come up with a sampling aberrration that would boost GOP candidates numbers

"Hispanic turnout will drop." ...LOL, with the things Trump has said????


Quinnipiac is churning out fake poll results at the behest of the GOP. They're no longer a polster. They are a disinformation mill for the GOP. (are they dancing to win conditional endowment money?)

[Her Sister]

31. Many Hispanics are becoming citizens and registering just so they can vote this year against Trump!

[NewJeffCT]

19. Didn't Quinnipiac also under-sample women?

Women typically make up 53-54% of the voting public in presidential elections, yet I think they only had 51% women in their poll. That would give Clinton another 2 point bump from the 4-5% change in actual voting.

I might be thinking of a previous Q poll?

[ailsagirl]

23. I saw that on PoliticusUSA's website

I'll remember that Q's polling is skewed (to say the least!) when I see future polls

[Cryptoad]

24. Quinnipiac is the Polling Whore of the GOP

and it seems they are even worst this cycle that ever. Big money must have changed hands here. Rumor Control is Nate Silver is contemplating banning them from use in 538 analysis .

[caquillo]

28. Are you sure?

He posted this yesterday evening:

1. Check the pollster’s track record. Has the pollster been around for a while? If so, has it tended to produce fairly accurate results? FiveThirtyEight’s pollster ratings are based on data from hundreds of elections since 1998. Quinnipiac, for example, earns a good grade; its polls have usually been more accurate than average. Thus, Quinnipiac polls are worth taking seriously — and they get more weight in the FiveThirtyEight forecast than an unknown pollster would.

http://fivethirtyeight.com/features/election-update-when-to-freak-out-about-shocking-new-polls/

[Cryptoad]

29. Im never sure of anything rumor control puts out!

[Doodley]

26. If this bias is true, does this mean that Quinnipiac University is racist?

Or at least those at Quinnipiac University who are responsible for their polling?

[Her Sister]

27. Yeah 2% here ~ 2 % there kinda racist!

Joking!

[Lord Magus]

34. Oversampling whites by almost 10% on the basis that there's no black candidate this time...

...seems fairly racist.

[krawhitham]

32. Great, now we are starting with "unskewed" polls

Are we really that fucked?