Pollster.com

Articles and Analysis

 

The Demographics of the Indiana Surveys

Topics: 2008 , Barack Obama , Divergent Polls , LA Times/Bloomberg , Pollster , Pollsters , Zogby

And following-up on my post this morning on the demographics for the North Carolina polls, here is the same set of statistics, when available, for the recent surveys of Indiana. Since African-American's are a much smaller share of the population in Indiana, the Clinton-Obama results are not quite as sensitive to their percentage of the Democratic electorate as in North Carolina. However, should the Indiana result be close, the size of the African-American population will be important. Also, the tables show that the polls vary on age as much as in other states.

The following table shows demographic composition statistics for those pollsters that have released them. Click on the table to display a larger version that also includes the vote preference results for reach poll.

05-06_INDemos2.png

The table excludes statistics from pollsters that have not publicly released demographic information for their North Carolina surveys (or perhaps more accurately, have not published anywhere I could find it): LA Times/Bloomberg, Indianapolis Star/WTHR/Selzer, Howey-Gauge (and thanks again to Pollster reader jac13 for sharing the demographic profile data that Zogby makes available to paid subscribers).

In Indiana we see the same wide variation in the age distribution among pollsters seen elsewhere: Even The percentage of 18-to-29-year-olds varies from 8% to 22%, the percentage 18-to-44 varies from 26% to 51%.

With the exception of one pollster, the variation in racial composition is smaller. Most show an African-American percentage of somewhere between 9% and 12%, with Research2000 (13%) and Suffolk University (15%). The most extreme value is the Howey-Gauge survey, which reported a much higher percentage of African-Americans (20%) among likely primary voters.

Brian Schaffner noticed last week that larger percentage of African-Americans in the Howey-Gauge poll explained how they showed Obama with a two-point advantage while other firms showed Obama trailing by seven or more percentage points. He has some interesting speculation about the composition of the Indiana electorate, but ultimately I have to agree with his bottom line conclusion: Given the lack of an Indiana benchmark for past Democratic presidential primaries, "we don't really know what to expect in terms of African American turnout."

[Updated table to include new surveys from PPP and InsiderAdvantage]

 

Comments
jac13:

How much of a variation in the AA turnout would it take for Obama to eke out a win in Indiana?

____________________

RS:

Mark:
Thanks! I was about to ask for this post :-)
[Of course, the one thing that would make this better is inclusion of the candidates' vote shares along side the demographic splits!]

@jac13: In SUSA's poll, if the African-American turn-out matches the Howey poll's 20%, things could get interesting.

Here's something I just posted on the SUSA thread:

IN regions:
SUSA says Senator Clinton wins Northern IN (38% of total vote) 53-42. I think northern IN includes Gary.
The Suffolk poll has northcentral/northeast/northwest IN at just over 30%, and evenly split 47-46. Of course, this assumes their regional demarcations are similar...

The difference in turn-out is made up in south IN; the vote splits are similar for the two polls.

This regional breakdown reminds me of PA - if turnout is high in Gary and Indy, that could make the race much closer.

____________________

JS:

Given the small sample size for AA's (for samples of ~500 it's a little over 50) I'm surprised there are wilder swings.

Or, do they oversample AA's?

____________________

RS:

@JS:
By "wilder swings," do you mean the differences in the African-American fraction (9-20%)? That's likely because some pollsters force an estimated voter composition on their turnout estimates (unlike SUSA).

SUSA weights its respondents to fit the demographics of the state, then applies a likely-voter filter. PPP, on the other hand, has been known to bump up certain demographics, e.g. youth and African-Americans.

____________________

JS:

RS

Thanks for the comment.

I wasn't clear though. I actually meant I'm surprised there isn't wider variation in the estimates for vote preferences among AA's. The total sample size of AA's nets down to a very small size, making the statistical analysis of HRC-BHO split unreliable -- large MOE if you are just looking at the AA subgroup.

Unless they oversample.

____________________

RS:

@JS:
Oh OK. Yes, you'd think that, wouldn't you? I think some of them oversample, though neither SUSA nor PPP say as such.
But surprisingly, the greater variation is seen in the larger sub-groups like 30-44/35-49 and 46-65/50-64 (not exactly comparable, but close).

e.g.
age-group/pollster/split/share of overall vote:

18-29/18-34 SUSA 53-42 (22%); PPP 53-41 (14%); IA 64-32 (18%) (Obama-Clinton)

46-65/50-64 SUSA 59-39 (29%); PPP 48-45 (40%); IA 46-38 (33%) (Clinton-Obama)

Weird! Or maybe those are the *swing* voters ;-)

____________________

JS:

RS:

Thanks for that analysis. Interesting.

Suprised that Obama is not doing better among younger voters in the SUSA and PPP poll, and HRC not doing better among older voters in PPP (though she seems over stated in SUSA).

Could be a difference between polls. And could also be a problem analyzing crosstabs:

(1) You can get the vote preference right, and still get the underlying demographics wrong (unless they have a correlation of 1.0 with vote preference, which they don't). But if the underlying stuff is a mess, it is an indicator, just one that people on this page may take too seriously. (2) Relatedly, and more importantly, when you are looking at vote splits among groups, if it's a small subset (in your examples they were ~ 100-150 age group samples sizes), it's even harder to make too much of it. You have to effectively recalculate your MOE for the sub-groups. Though the overall trends should generally hold up.

I assume Obama over-samples AA's, and Clinton white working class, etc. for just that reason. Same for regions. For example, I heard that Clintons internal tracking picked up a trend to her in NW IN (in Chicago media market where they are running alot of Wright stories still) ... would be hard to estimate if they weren't designing a sample for that area.

Well, those my thoughts anyway.

Would be happy to hear you reactions

JS

____________________

RS:

@JS:
I often wonder how these polls extrapolate from a sample of 600-1200 to a voting population of almost a million... Oversampling definitely helps with the smaller groups.
But some of the problem could just be random noise - does Senator Clinton lead 46-64 by 20%, or by just 3%? Or by something in-between?
Also, does PPP really expect this age group to be 40%? [I find it hard to imagine the 46-49 group could be 11%!]

I guess what I am saying is that broadly I agree with you!
It sure would be interesting to see the campaigns' internal polls and compare that to other polls/the final result :-)

Just 24 hours more!

____________________

JS:

RS

Yes. Sounds like we agree. A first for this page!

As to your first point, about extrapolating to large populations, is all basic probability theory, with a little bit of pollster art form, since they can't adhere exactly to all the assumptions of that theory. If you take a barrel of white and black balls evenly divided, and draw random samples from it, they will converge on half white and half black, and be w/in the MOE (which is based on the sample size) 95% of the time. Doesn't make any difference (virtually) how big the barrel of balls is. A million or a 100 million.

It's magic. Boggles the mind, but works.

The reason these guys are all over the board is that (a) there is some random error, so you expect that; (b) they deviate for theoretical randomness in different ways; (c) they ask the question in different ways -- if you don't ask the right question it doesn't make any difference how good your sample is!; (d) they make different assumptions about whether you are actually going to vote or not (a source of huge variation)

It's amazing they get as close as they do!

We'll see tomorrow.

JS

____________________

RS:

@JS:
You are right, it's amazing they get it close at all.
I feel more comfortable the larger the sample size - e.g. the recent NYT poll had under 300 Democrats/leaners! Even AP/Ipsos had just 400-500. Give me something with 2000 respondents, and I feel happier (guess that gives SUSA and PPP an advantage.)
When the polls are so close and with really small sub-groups, it's best to remember the MOE and let them pass on by :-)

____________________

Ciccina:

Its funny how they all seem to be predicting a less than average proportion of female voters.

The average from the previous 27 primaries (excluding MI - screwy ballot - and DC - no exits) is 58%, at least according to the data at CNN.com and my primitive math skills.

Howey Gage, Rassmussen and SurveyUSA seem particularly "out there" with 51-52% - I don't believe we've seen numbers that low in any of the past primaries. Only OK (53%), CA (54%), NM and MO (each 56%) are sort of in the ballpark.

After that, 6 states had 57%, 5 states had 58%, 5 states had 59%, and 4 states had 60% and 3 had 62-63%.

On the one hand, I know the IN samples are specific to data from IN. On the other hand, it seems strange that these folks are predicting an f/m mix that's nearly unprecedented this season.

Or maybe I'm reading this completely wrong.

RS, this seems like the kind of thing you could crunch with your eyes closed, but my skills can carry me no further.

____________________

Ciccina:

Correction - in my previous post, first paragraph "they" should have referred to Howey Gage, Insider Advantage and SUSA.

My cutting, pasting and editing skills are also primitive.

Obviously TeleResearch is predicting better than average and ARG and Downs are close to average.

____________________



Post a comment




Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR