Analysis of Real Clear Politics Poll Averages: How accurate have they been in the past?

11/5/2006 | Me

[Western Civ 4ever]

It occurred to me as my fellow Freepers are likely as gut-wrenched as I am over the next poll release, to look at comprehensive averages of all non-partisan public polls in the last two election cycles (the two in which Karl Rove's GOTV apparatus has been in place). We're counting on this wonderful, well-oiled machine to pull us accross the line in the close ones. Did we really defy the polls to an extent that this is likely this year? There are a couple of ways to measure this. We can examine how the poll averages compare in terms of the margin, or spread, if you prefer. In this, we find that in 2002 that in all of the Senate races deemed "Competitive" by Real Clear Politics, the spreads were less Republican than actual results by an average of 2.68%, and actually suggested incorrect winners in four races out of the thirteen it followed: CO, GA, MN, and SD. The first three were in our favor, the heartbreaking Thune/Johnson race was not. (Note: I left LA out of this analysis because of the lack of consistent polling in the Landrieu/Terrell runoff.)

In 2004, eight Senate races were deemed "Competitive." Polling projected none incorrectly, but did underproject average GOP strength in the final spreads by .75%. Also, in the seventeen "Battleground" states in Bush/Kerry, average margins were more GOP by .58%.

So, in theory, the Rove machine has a shot at saving us, but here's where I'm worried. Most poll junkies know that even the most 11th hour of polls will never capture all undecideds. And oftentimes, especially in years where one party is swimming upstream, the final support levels for that parties candidates don't rise much further than their already committed voters reflected in the last poll. For example, in the poor GOP year of 1998 in my home state of Georgia, the GOP candidate for governor had enjoyed outside MOE leads for the entire year. The final poll by the Atlanta Journal gave him a 45-39 edge over the Dem, Roy Barnes. The final result? Dem 52, GOP 44. Barnes picked up nearly all the undecided vote.

With that perspective, how well have polling averages done at accurately predicting candidate support levels? In 2002, GOP support levels were understated 3.88%, Dem levels understated 1.19%. This totals 5.07% average undecided two-party votes, of which the GOP picked up 76.5%.

In 2004, GOP support levels were understated 3.58%, Dem leves understated 2.83%, meaning a GOP pickup of 55.8% of two-party undecideds.

How will this measure pan out this year, in such a horrible climate for the party. We had President Bush at 68% approval in 2002, and barely under 50% in 2004. At a minimum, he was not a drag on us 2004, but even the sunniest optimist among us can't say the same this year. Rove, here's hoping you really the Magnificent Bastard, 'cuz we need you!

[msnimje]

Sorry, but none of this matters.
Polls are for the media spin machines, not voters.
The only thing to do now is vote and get others to do so as well.

[IllumiNaughtyByNature]

How accurate have they been in the past?

[Man50D]

Four points to consider regarding polls:

1. Only a very small fraction of the population is polled.
2. Pollsters can target a group of people they know will give the pollsters the answers they desire.
3. A question can be structured in such a way as to elicit a certain answer.
4. A pollster can record a different answer than was given by the respondent to satisfy pollsters achieving the result they desire.

The most accurate poll and the only one that counts will occur November 7th.

[Strategerist]

The reality is that poll averages, overall, haven't been nearly as horrifically inaccurate as a lot of people here would like to believe.

There's a lot of picking and choosing of one particular poll from one time for one candidate in previous elections that was seriously in error and then people concluding that all polls are completely worthless (well, actually, that all polls that aren't telling you what you want to hear are completely worthless.)

[speedy]

Thanks for crunching the actual numbers. It does give some hope, in that the GOP has taken most of the undecideds the last two elections. Whether or not that is enought to put us over the top this time remains to be seen, but it's good to see some real-life comparisons instead of just hunches.

[JLS]

There are several schools of thought on the late deciders:

1. Some think the break against the incumbent.

2. They could break away from the media supported party, if some people don't want to tell strangers they are voting for the party the press is carping at. Given the press climate this could easily happen this year.

3. They could break a against black dims and for black GOP candidates if some white voters won't tell strangers on the phone they are voting against the black candidate and some black voters won't tell strangers on the phone they are voting for GOP candidate.

So there can be lots of things going on late in this election. I particularly will be interested in the Swann, Steele and Blackwell races to see if 3 is true. Point 1 is the beltway view.

[kingu]

A typical conservative views voting as a duty, a typical liberal views it as chore. With poll checking of ID, increased use of absentee voting, and a lot of right to life and protection of marriage laws at the polling place, a conservative is far more likely to go in and vote.

I've a personal belief in that a Dem has to be three points ahead to even have a chance if there's anything to bring out the conservative vote, four points if they're running against a real conservative.

I think we'll be keeping the Senate - the House, just isn't enough data on enough races to really tell if we'll have a majority or not, and beyond, how those votes will fall if it is really close.

I would feel a whole lot more comfortable if there were more conservatives running.

[catholicfreeper]

Good post. I am a big fan of polls myself. Pollls help us focus resources where we did too and alert us to problems. It also alerts us to where there are opportunities. IN the end the polls are pretty much in the margin of error.

I do have concerns about the undecided. I think the Senate races are becoming more nationalized but the House I am not so sure. I think often we fool ourselves and forget there are like 100 different local issues that affect these races. SOmetimes a canidiate loses because their staff sucks and they have a bad record of getting back to Maw maw when there is a gripe about the social security check.

The key I think is Dem turnout. There is an assumption that they are energized. Are they? I am not so sure.

[RasterMaster]

The DUmbasses in the lame-stream, has-been, drive-by dinosaur DNC media have to cover Saddam's sentencing today and tomorrow.

He's already been sentenced to DEATH, any further trials are moot and we are again reminded of why our troops are in Iraq - depose Saddam (or REGIME CHANGE as it was referred to when Clinton adopted it as policy, but did nothing more than lip service). DUmocrats wanted the election to be about Iraq, and so it is.

1. Jean-Fraud Kerry infers that soldiers in Iraq are dumb.

2. Soldiers respond

3. NYT admits Saddam had nuke program.

4. Families of fallen visit Iraq (no quagmire)

5. Saddam sentenced (death by hanging)

News cycle seems to be working for GOP advantage thus far. DUmocrats yesterday counter with "Rusmsfeld must resign" and "recruiters caugh lying" stories. Wow. That's new....NOT!

In addition to Saddam sentencing, Fox will be running OBSESSION - THE MOVIE, and we get to see the face of the enemy again. Yesterday, Fox is also showed "waterboarding" (Steve Harrigan, a Fox reporter volunteered to show how it is done).

DUmocrats come out of hiding, but have NOTHING to give anyone a reason to trust them with power. They've even invited BJ and Al to speak for them - kiss of death.

I don't know what Monday brings, but I've got a good idea about the day after.

[SolomoninSouthDakota]

"The alleged power to charm down insanity, or ferocity in beasts, is a power behind the eye.---Emerson."

We are facing an insane media allied with a democrat party ferocious for THEIR power back. We need character and confidence, the power behind the eye, to come out ahead.

"Florence Nightingale tells of soldiers suffering with dysentery, who, scorning to report themselves sick lest they force more labor on their overworked comrades, would go down to the trenches and make them their deathbeds. Say what you will, there is in the man who gives his time, his strength, his life, if need be, for something not himself,--whether he call it his queen, his country, his colors, of his fellow man,--something more truly Christian than in all the ascetic fasts, humiliations, and confessions that have ever been made."---"Pushing to the Front", Orison Swett Marden. The Riverside Press, Cambridge, Mass., 1894.

We need men and women like these now. Now is the time for all good men to come to the aid of their country. I am sure all of us here on FR are doing what we can for the good of our country. I think it is clear enough that it's not going to be easy to hold the House and Senate. But do it we must. Let's roll.

[mware]

Actually I think the key is republican turnout. I heard Morris the other day say that the problem the Republicans have is not with the independent voters. He thinks we will get enough of them, he said that the Republican turn out is what is going to determine the results.

[YaYa123]

Yes, for sure the key is turn-out. I mean, we have to "turn out" more legitimate voters than inner-city, democrat controlled, corrupt democrat precinct workers, AFL-CIO, ACORN affiliated mobsters can invent, or dig up.

I'm glad Feds are already in St. Louis, Kansas City, and Memphis, investigating voter fraud, but they can never stop all of it.

[bpjam]

And, what could be the biggest shortfall of polling, is that they are not reporting the number of people who did vote in the last election who are NOT planning on voting now and vice versa. THAT is where the turnout might get noticed prior to the actual election. Who are the people who DIDN'T vote in 2004 but ARE planning on voting today voting for?

[mikeybaby]

I think voter I.D. laws could be a sleeper issue.
I know here in Indiana, our resident liberal nutjob Julia Carson (who needs a full time assistant to wipe the drool off her chin) tried to use some bogus I.D. to vote and was initially refused-was later allowed to vote.
Made the news, though.
I would be interested in what other Freepers think.

[11th_VA]

I totally agree - the new Voter ID laws AND the elimination of 'Straight Ticket Voting' will hold down the Democratic Vote - especially in down ticket races.

Rain in the East will help the GOP there.

[YaYa123]

I'm not familiar with Voter ID ballot issues. Here in Georgia, the courts over-turn photo-required ID progress anyway.

I think one man one woman marriage amendment initiatives will help Corker and Allen. The EMBRYONIC stem cell issue will help in Missouri rural areas.

[beezdotcom]

The most accurate poll and the only one that counts will occur November 7th.

Yeah, except that, to my knowledge, they haven't been polling dead people - but a large number of them WILL be voting Democrat...

[thile44]

There is also a as yet unquantified factor in this election. I don't know if it will matter or not, but voters today have far more accurate information available than ever before, i.e. the Internet and blogs to be swayed only by what the MSM tells them. It will be interesting to see if the MSM is correct on this one. I have a feeling that if they are correct it will be by the thinest of margins.

[PUGACHEV]

Polls can't really say who is actually going to vote, and that's where I think all this media spin about a Democratic tsunami bound to sweep the the Republicans out of office is going to help us. A likely Democratic voter may well have become complacent, thinking that there is little reason to go to the trouble of voting, since the results are a apparently foregone conclusion, just as we've been told, over and over. The Republican voter, however, alarmed at the possibility of Pelosi & etc. moving into a position of leadership is certain to be more motivated, and thus more likely to vote.