Worst Poll Ever. Great Job CNU Polling Dudes!

By: Lowell
Published On: 9/17/2008 7:56:41 PM

Wow, I've seen some bad polls in my day, but they don't get more FUBAR than this one by the Center for Public Policy at Christopher Newport University.

The survey, taken last week for the Center for Public Policy at Christopher Newport University, found McCain with the support of 48 percent of state voters, compared with 39 percent for Obama. Just under 13 percent were undecided.

Now, before you get even the slightest bit concerned about this "poll" (in quotes for reasons I'll now discuss), there are just a few minor, wee-bitty problems here.

1. According to Real Clear Politics, every single poll - 15 of 'em - since Barack Obama clinched the Democratic nomination show the race in Virginia within 4 points. Most show it within 1 or 2 points. Except for this poll, which is a wild, crazy, total outlier. The emphasis in that last word, by the way, should be on "lier," also spelled "liar."

2. As Tim Craig points out, this poll grossly underestimates African American and young voter turnout.

The poll assumes African Americans will make up a little under 10 percent of the electorate. In the 2004 presidential race, African-Americans accounted for 21 percent of all voters in Virginia, according to exit polls.

The poll is also based on an assumption that voters age 18 to 29, a group Obama expects to win handily, will be 3 percent of the electorate. Those voters made up 17 percent of all voters in 2004, according to exit polls.

Again, insane.  I mean, this isn't even defensible, not even semi-plausible, just completely laughable.  

3. Comparing to a serious poll, SurveyUSA assumes 18-34 year olds will make up 25% of the electorate on November 4. CNU assumes that 18-29 year olds will make up 3%, so maybe 18-34 would be 5% or 6%?  What. The. Hell?!?!?

3a. Another data point is the 2004 Virginia exit polls, which indicated that 18-44 year olds made up 49% of the electorate. Again, and not to be too delicate about this, but CNU has its head so far up its ass that a search and rescue party would take weeks to locate it, but by then the head would be dead and decomposing.

4. SurveyUSA assumes African American turnout of 19%, which I actually think is very conservative. Again, CNU assumes LESS THAN 10% African American turnout.  That's beyond idiotic, bordering on complete brain death.  

In sum, I don't even know how these people can call themselves pollsters and why anyone would ever listen to them again.  In-sane.

P.S. I just thought I'd add that this is not an indictment of CNU in general, since I'm sure it's a fine school, but it's DEFINITELY an indictment of the pollster and the poll - possibly the Center for Public Policy as well.


Comments



Yet another reason I chose to go 300 miles away (hallcr3 - 9/17/2008 8:16:07 PM)
to Virginia Tech instead of stay at home in Newport News and go to CNU.


Former U.S. Senator Trible (R) is the president of CNU. (martin lomasney - 9/17/2008 8:17:47 PM)
What would you expect from an institution that graduates a grand total of 12% of its students and is led by that hack?

I'm surprised the students can still count after attending even one year at such a place.  



You must be confused (notwaltertejada - 9/18/2008 12:19:12 AM)
First of all I am a CNU student, and CNU graduated over 800 students last year out of a population of a little over 4,000. How is that 12%? I think you may want to check YOUR numbers.

Second, You are the kind of person who will attack any Republican no matter what their views are. Paul Trible supported John Miller in the 1st senate district last year and helped him get elected in a conservative district. Miller is the main reason the Senate is in the hands of Democrats. By the way, Miller was a PROFESSOR AT CNU. Guess he can't count either and must be an idiot!

Third, like him or not Paul Trible has done a lot for CNU, and in turn the economy of Newport News...which I can tell you is struggling. CNU was not even on the map 15 years ago and would not be today had Trible not been President.

Yes, Trible may seem like a typical politician and be a bit cheesy at times but I doubt he is a bigger hack than you are!

Oh and by the way, CNU is now just as hard to get into as Tech for you and hallcr.  



Um... (DanG - 9/18/2008 2:22:35 AM)
As a Tech student... no, it isn't.  Don't go there, man, or I will whip out rankings and shit the likes you have never seen.  You could've made your point without going after the Hokies.  If you want to roll out against the Hokie Nation, by all means, good luck.  But we make "Pit Bull" Sarah Palin look like a frickin' Chihuahua.


By the way (DanG - 9/18/2008 2:27:40 AM)
Quention Kidd's response to this horrible poll doesn't look good for you guys.  

"Polls are a snapshot in time."

Yeah, but that doesn't matter if you're taking picture with a cracked lens.  The model for the poll is wrong, therefore the poll is pretty much useless.  For a smart guy, Dr. Kidd sure didn't look in control there.  Though it pains me to compliment the Hoos, where's Larry Sabato when you need him?



talking to the new york times (bcat - 9/18/2008 9:47:29 AM)

Where he puts the whole election on Hampton Roads.

http://www.nytimes.com/2008/09...

And it's okay for you to compliment UVA. We have a few things going for us, too.



Are you guys going to start (Catzmaw - 9/18/2008 7:46:42 AM)
singing "Be True to Your School" now?  BTW, I think it's a little unfair to condemn an entire school for one poll out of one center at the school.  I know nothing about CNU, but it seems like a stretch, to say the least.


As far as I know... (Lowell - 9/18/2008 8:03:03 AM)
this polling debacle was strictly the responsibility of the pollster and the Center for Public Policy.  They should retract the poll, or correct with reasonable assumptions about African American and young voter turnout.


Agreed on that, but I'm seeing statements here (Catzmaw - 9/18/2008 8:27:31 AM)
which appear to be attacking the whole school.  The poll is amateur hour, but for all I know CNU has a fantastic English lit program or accounting program or something else unrelated to political polling.  


Wait, that wasn't an attack on Tech (Julie Crum - 9/18/2008 9:40:59 AM)
Dan, hold on.  Not Walter said that CNU was just as hard to get into as Tech.  That's not an attack!  And from what I can see, it looks like it's pretty much a true statement--the College Board website give "percent admitted" numbers that show that CNU admits 54% of applicants, and Tech 67%.  I perceive Not Walter's comment to be a defense of CNU rather than a put-down of Tech.

As a parent of CNU students (one graduate, one current student), I have been pretty impressed by what I've seen there.  They do a nice job with their physics department.



cnu and vt (bcat - 9/18/2008 10:10:37 AM)
Not to quibble too much, but acceptance rate isn't the only measure of how hard it is to get into a school. Students will tend not to apply for schools that they don't think they can get into. You also have to look at things like the average GPA of incoming freshmen (VT: 3.86, CNU: 3.65) and the average SAT (1238 and 1182).

But all that is beside the point, because those numbers are pretty good by most standards (UNCW, for example, which is a fine school, has an average incoming GPA of 3.74 and an average incoming SAT of 1160; Alabama has an average incoming GPA of 3.4 and an average incoming SAT of 1140). With the possible exception of California, there's not another state in the country with so many rock-solid public universities. There's W&M, JMU, VT, UVA, VCU; CNU is just another example. As Virginians, these are all our schools, and we should be proud of all of them.



i almost forgot (bcat - 9/18/2008 10:13:58 AM)
George Mason, which is very quickly establishing itself as the UC-Irvine of the East Coast.


Just a friendly suggestion from a proud 'Hoo (UVAHoo - 9/18/2008 10:51:30 AM)
You may not want to tout your pride in VaTech by talking about pitbulls....

Did you want to talk about rankings? :)



I'll admit (DanG - 9/18/2008 10:54:10 AM)
You guys have a better ranking than us.  Like 35 places ahead.  Still, we get to feel a little better when we kick your ass in football every year. ; )


I'll admit (DanG - 9/18/2008 10:54:11 AM)
You guys have a better ranking than us.  Like 35 places ahead.  Still, we get to feel a little better when we kick your ass in football every year. ; )


I knew I tanked... (UVAHoo - 9/18/2008 10:56:22 AM)
...my burgeoning pro football career by not going to VaTech.  Geez! ;)


Well, not this year (DanG - 9/18/2008 11:11:56 AM)
The one thing we say on Campus is, "Well, at least we can still beat the Hoos."


Okay. (notwaltertejada - 9/18/2008 12:46:51 PM)
I worked in the admissions office last year so I know all the stats.
I was not attacking Tech, I was just defending my school which is just as good as Tech.  


Wally, Beav, chill out... (FredFred - 9/18/2008 11:18:48 PM)
because see, I went to UVA, and if you guys don't stop it I'll whip out "rankings and stuff" that will embarass the two of you :-P

We're on the same team.  Chill.  

-Fred



12% grad rate is the number Trible reported to the U.S. Dept of Education as required (martin lomasney - 9/18/2008 4:11:09 PM)
by Federal law and reported in the U.S. New and World Reports Guide to U.S. Colleges.  Go argue with him.

The only reason the CNU is hard to get into is that there are so many qualified college students from No Va that they've filled up the regional allocations at all of the other, more highly regarded, Va. public colleges and universities and settle for CNU out of financial necessity.  Their grades and SAT scores demonstrate they deserve better.  God knows their parents are paying enough of the State's income taxes to justify getting better.



12% grad rate (Quizzical - 9/18/2008 5:12:11 PM)
Just curious -- is it the rate that graduates in 4 years?  Or the total percentage of incoming students who ultimately graduate?  The former sounds more plausible.  


It is the 4 year rate which is the rate that parents are concerned about unless they plan on paying 5-6 years of tuition (martin lomasney - 9/19/2008 3:40:23 PM)
which I'm sure Pres. Trible wouldn't mind.  BTW the 5 and 6 year grad rates are also abysmal.


Isn't it more complicated than that? (Quizzical - 9/20/2008 7:36:10 AM)
I'm not addressing CNU directly here, as I don't know one way or the other.  However, I don't think the 4 year graduate rate is necessarily a measure of academic value -- I think it is more a measure of the financial resources to finish in 4 years, both in terms of the financial aid available and the financial resources of the parents. Isn't there a direct correlation between the 4 year graduate rate and the endowment per student ratio?  


What's truly amazing . . . (JPTERP - 9/17/2008 8:30:54 PM)
EVEN with those screwed up cross tabs, McCain still can't crack 50 percent support.  


whoa there... (West Ailsworth - 9/17/2008 8:40:11 PM)
I'm a CNU student so I guess it's my duty to speak up...  

It is my understanding that CNU simply pays for these polls and sticks their name on it.  I didn't think CNU had anything to do with the sampling methodology.  Am I wrong? So is it really CNU's fault, or the crappy polling outfit they had conduct it?  

I suppose, come to think of it, that hiring a crappy outfit is CNU's fault.

(BTW, I'm an accounting major who can count just fine)



I don't know (Lowell - 9/17/2008 8:46:11 PM)
but this is the worst poll in human history. If I were in any way, shape or form associated with these pollsters, I'd be totally embarrassed.


well... (West Ailsworth - 9/17/2008 8:49:40 PM)
at least Jim Gilmore didn't go to CNU.


Ha. (Lowell - 9/17/2008 8:52:50 PM)
Good point! :)


Think of it this way (DanG - 9/17/2008 8:55:55 PM)
Mark Warner is up 24 points with these absurd numbers.  So he might be up by 35 points with something that remotely resembles reality.


Well. Here are some good polls. (Pain - 9/17/2008 8:56:20 PM)
48-43 Obama

http://www.msnbc.msn.com/id/26...

But the Times/CBS News poll suggested that Ms. Palin's selection has, to date, helped Mr. McCain only among Republican base voters; there was no evidence of significantly increased support for him among female voters in general. White women are evenly divided between Mr. McCain and Mr. Obama; before the conventions, Mr. McCain led Mr. Obama among white women by a margin of 44 percent to 37 percent.

By contrast, at this point in the 2004 campaign, President Bush was leading Senator John Kerry of Massachusetts, the Democratic challenger, by 56 percent to 37 percent among white women.

From CNN
http://www.cnn.com/2008/POLITI...

Florida: McCain 48%, Obama 48%
Ohio: Obama 49%, McCain 47%
North Carolina: McCain 48%, Obama 47%
Indiana: McCain 51%, Obama 46%
Wisconsin: Obama 50%, McCain 47%



Yes, those look reasonable (Lowell - 9/17/2008 8:59:49 PM)
Actually, they look somewhat optimistic in North Carolina, Florida, and Ohio if you ask me, but they're certainly within the realm of plausibility, at least in Ohio and Florida (not sure about NC, any thoughts?).


I'm not a math wiz (Pain - 9/17/2008 9:06:30 PM)

But, I like the trend over the past few days, and just hope it holds.  If we go into the election up, even by a hair, then I feel we've got a good shot given our ground game.  I'd like 5% which would offset the closet racist factor, but as long as we're on top then we're still in this thing.


Oh, and if you were asking about NC specifically, (Pain - 9/17/2008 9:09:14 PM)

I think that is a tall order.


Not really (DanG - 9/17/2008 9:33:29 PM)
Higher African-American percentage than Virginia, lots of colleges and universities.  It would be a stretch, but it is within the realm of possibility, especially if the economy remains so important.


Two Words: Jessie Helms (norman swingvoter - 9/17/2008 9:43:08 PM)
I think that any state that sends a racist scumbag like Jessie Helms repeatedly to Washington is probably not going to go for Obama.  Of course I could be proven wrong and, if so, I will not be complaining.


We voted for George Allen and Jim Gilmore... (FredFred - 9/18/2008 11:22:21 PM)
... and lots of more liberal folks have moved into the Charlotte area and the research triangle.  Still, I buy Nate at fivethirtyeight.com's argument that because of demographics, if Obama carries NC, he has already carried VA and won the election.  NC is unlikely to be a 'tipping point' state, it's more likely part of a landslide.

Which would suit me just fine, of course.  But I'm still very very nervous.  The election turned on a dime 3 weeks ago and went McCain's way, then this week it has gone Obama's.  If the election were tomorrow I would like our chances.  But this fed bailout of every single bank may make people think the economy is safe long enough for people to focus on more "important" things like lipstick, flag lapel pins, gay marriage, and brown scary people.

-Fred



I'm nervous too, but (Pain - 9/19/2008 6:30:03 AM)

The bounce McCain got was mainly over his shinny new penny.  It took 2 weeks for all but the republican base  to figure out is was a diversion and now we're back to pre-convention numbers.

I know it's close and we're not out of the woods, but the only thing McCain got from Palin was a solified base and quick look by the independants....who, it appears, don't really like what they see.



The poll may be a deliberate effort (Teddy - 9/17/2008 9:27:48 PM)
by the Republicans to begin laying the groundwork to "explain" a sudden flip on election day; I'll think more like this if other polls, interviews, studies, etc. begin popping up hinting at the same pro-Mac results around the country. Yes, this sounds like conspiracy theory, but I ask you, how many times have the Republicans said outright they will do "anything to win," (hence Sarah Palin and endless lies), that they "can't afford" to "let" Obama win. Remember the tortured explanations of why exit polls were wrong in battleground states? I wonder if the lesson the Republicans learned from that little problem has taught them that you have to plan in advance to explain outlier results from electronic machines (which have been hacked) or why votes have been lost or destroyed. Laugh now and cry later.  


I was thinking the exact same thing (linlu - 9/17/2008 10:04:40 PM)
Our city has those abominations called Diebold Voting Machines.  I suspect that this same sort of 'poll' results before the election will occur in every single swing state in order to hide the stealing of a THIRD election.  Unfortunately our officials haven't done squat to get rid of them while all the other states are busy decertifying them and hopefully replacing them.


Can a voter request (Teddy - 9/17/2008 10:26:05 PM)
a paper ballot or a paper receipt? Is there something about that in Virginia law somewhere?  Also, watch out for Republicans trying to supress the vote by showing up with lists of foreclosed homes and denying the former occupants/owners a vote because "they don't live at that address anymore."


Did you see the Virginian Pilot online? (chspkheel - 9/17/2008 10:58:33 PM)
They had it posted earlier on their front page like this poll is the clincher for McCain.  When I saw the sample size, I had a feeling that this was a suspect poll.  Why would you only poll 500 in a statewide race?  The Virginian Pilot ought to be embarrassed!  Well, at least they reported the other polls that show it closer.  

http://hamptonroads.com/2008/0...  



Write an LTE (Teddy - 9/17/2008 11:42:47 PM)
to the Pilot and point out the small sample, the ridiculousness of the poll, slam the Pilot for bias, and dare them to print it. Worth a try.


Earlier the Pilot only reported the CNU poll (VA Breeze - 9/18/2008 12:09:13 AM)
I know I was one who wrote in the comments section and posted in the comments the PPP poll. The comments have now been disabled and they beefed up the story with the PPP poll.

To be fair-the disabling of the comments doesn't necessarily follow from the Pilot's poor reporting of only the CNU poll at first - people have been posting some really nasty comments.



The Pilot choked (DanG - 9/18/2008 2:25:33 AM)
They do often.  Printed the story without taking a look at the sample.  The numbers would've revealed that this poll is total horse-crap.


Why would you NOT poll 500 in a statewide race? (UVAHoo - 9/18/2008 10:54:47 AM)
The whole point of polls are that they're a representative sample of the electorate.  They could have polled 750 voters and only dropped the margin of error slightly and still had the same percentage of African-Americans in the poll.  Its the make-up of the electorate that matters, not the size of it.  A 500 person poll is pretty standard.


Wow. (Silence Dogood - 9/18/2008 9:49:53 AM)
Only 10% of the sample is African American?  10%??  I'm not sure if it's been that low since the Voting Rights Act of 1965 was enacted.


Well, McCain probably wants to reverse that (Lowell - 9/18/2008 11:08:17 AM)
Maybe he's jumping ahead? ***snark***


Sorry I started the CNU vs. VT crap (hallcr3 - 9/18/2008 11:55:53 AM)
I was born and raised in Newport News. I actually have a lot of pride in CNU and if they offered a more diverse selection of courses, I probably would've gone there.

LOL. Just thought I'd mention that.



Problems with Quentin Kidd's argument that... (DCCyclone - 9/18/2008 10:48:51 PM)
...the poll somehow still supports that McCain is winning......

Kidd argues that McCain is winning because his flawed poll shows McCain winning independents and running even with women.

Um, Quentin, no, that's not correct.

If the poll very badly undersamples black voters and young voters, then it very badly undersamples black women and young women.  So the subsample of women is not representative and is skewed toward McCain.

And if the poll undersamples Democrats overall and its constituent groups such as black voters and young voters, then it oversamples independents and perhaps Republicans to make them a larger share of the vote than they really will be.

Quentin seems very amateurish not to recognize that these subsamples are interlocking and interdependent, not mutually exclusive groups.



Oh, and one more problem...... (DCCyclone - 9/18/2008 10:52:50 PM)
I forgot to mention in my comment immediately above that in addition to the subsample problems I pointed out there, the fact is these subsamples of women and independents are going to have much higher margins of error than the sample as a whole, reducing the values of the subsample numbers.  Independents in particular will have a VERY high error margin since they're one of 3 groups and in size far below one-half the total sample.

At some point the margin of error is so large that the numbers aren't worth much.

That's why you can see 2 polls in a state give the same result with the same error margin and confidence interval, but the subsamples can have wildly different results.  I see this all the time particularly between Rasmussen and SurveyUSA, who both use the same basic polling methodology (with notable differences in analyzing data).