Nevada pre-caucus polling off by over 30 points! Why is that? (VANITY)

1/20/08 | vanity

[1curiousmind]

According to RealClearPolitics.com the three final polls leading up to the Nevada caucus had Romney up an avg of 5 points over McCain. Yet he beat him by 38 points. Yet I have yet to hear anyone raise a question about that (unlike the NH polling).

Please explain.

[1curiousmind]

Did these polls keep Romney from focusing more on SC the days preceding the primary?

[JaneNC]

Polls are skewed. I think they are all pretty much fake. No one should take them seriously.

When have they been right?

MSM is pimping Mckennedy. The is one of them...a libtard.

[peyton randolph]

Relatively low turnout. Disproportionately Mormon. Thus, polling samples were skewed. No different than undercounting evangelicals when determining the Huckster’s early support in Iowa.

[Moral Hazard]

“No different than undercounting evangelicals when determining the Huckster’s early support in Iowa.”

Except that Mormon were 25% of the vote in Nevada while evangelicals were 60% of the vote in Iowa.

[CharlesWayneCT]

Actually it was extremely HIGH turnout, not low turnout. But that itself throws things off.

The really funny poll though was the last ARG poll in South Carolina. Usually they adjust their last poll to be closer to accurate, to hide that they seem to use their previous polls to push their candidate.

But this time, it was just wrong, showing wild movement in a 2-day period that if correct should have resulted in a much different result.

[Russ]

Both pre-election and exit polling have gotten way out of hand. They aren’t being used to report trends, they are used in an attempt to sway opinions and votes. They have replaced issues as a focus of the campaign and are reported as news. The real problem, though, is that it is only going to get worse.

[Old Retired Army Guy]

Caucus Polls are always a guess. Never know who is going to show. If you look at total votes cast in NV Republican Caucus, it was really low. Romney got 94 % of the Mormon vote which was 25% of the total. I don’t think that was polled. I think we are through with Caucus states now (I hope).

[Strategerist]

It’s really hard to poll for caucuses.

[Dante3]

Several times the polls have been off. In some cases, it might be difficult to devise proper sampling design, in other cases the design might be “unscientific,” or there is an attempt to manipulate the vote.

[Strategerist]

When have they been right?

The average of the last set of South Carolina polls was almost exactly right.

Poll Average: McCain 27%, Huckabee 26%, Romney 15%, Thompson 15%, Paul 4%, Giuliani 3%.

Actual Results: 33%, Huckabee 30%, Thompson 16%, Romney 15%, Paul 4%, Giuliani 2%.

Differences can be attributed to there being no "undecideds" in the actual election as opposed to polls - the undecides broke for either McCain or Huckabee.

Of note is that Thompson got basically exactly what the polls were predicting (other than the anomalous ARG poll.)

[1curiousmind]

“the undecides broke for either McCain or Huckabee.”

______________

I guess that is what bothered me about the polls being so far off. I (of course as a Romney supporter) feel that he could have beat Thompson, and possibly finished in the low 20’s had he not basically “pulled out” of SC in the final week.

It’s unclear if he did that just to save face (as apparently everyone else did in WY and NV) or if he was really worried he might lose NV.

[COBOL2Java]

"There is nothing wrong with your election cycle. Do not attempt to adjust the polls. We are controlling the ballot boxes. We will control the Democrats. We will control the Republicans. We can roll the candidate; make him flutter. We can change my focus to a soft blur or sharpen it to crystal clarity. For the next 11 months, sit quietly and we will control all that you see and hear. We repeat: there is nothing wrong with your election cycle. You are about to participate in a great adventure. You are about to experience the awe and mystery which reaches from the illusion of liberty to the outer limits of government control."

[LS]

I don't know what to think about the polls. In 1996, a statistics prof from Indiana did an article on why the polls were all wrong (I have the citation somewhere). He concluded that every major poll was off in the 1996 presidential race, but more important, they were all off in the direction of Clinton, and most were off outside the margin of error. He said that statistically, if this were just an "error," the chances that they would ALL be off in Clinton's direction would be 240,000 to one.

The pollsters did apparently try to "clean up their act," especially since Zogby came out of 1996 looking like a pro (he was the closest to the real outcome). They were close in 2000 and 2004 on the presidential side, but missed as many races as they got right on the key senate races (I, on the other hand, correctly picked every competitive senate race in 2002 and 2004 except for Thune in 2002, where he lost by 500 Indian votes, and Salazar in 2004, who won a close race).

Given the pollster's record, I figured they were WAY off in 2006 when they all predicted a GOP debacle. Well, they got it right in 2006, and I completely missed all my predictions.

However, both in NH and now in NV, it is clear that they are again having troubles. Since they are missing on the Dem side as well, perhaps it suggests that it's not so much their bias as it is once again a flawed methodology that is failing to get people to tell the truth about their intentions.

[Strategerist]

In Nevada there was very little polling effort at all, really. Many organizations didn’t even bother to poll there.

That also reduced accuracy.

[Fresh Wind]

LOL!! That’s magnificent!

[gruna]

Yes.

[CPOSharky]

People are so sick of pollsters that they are lying to them.

[Bernard]

This early preponderance of caucuses provides too much importance to caucasians. That should be spread around more.

[NRG1973]

“”MSM is pimping Mckennedy.””

Correct!!! I watched the post primary news coverage and the MSM (including Fox News) was implying that McCain had it all wrapped up. But...he only has about 1/3 the number of delegates that Romney has. Also, McCain hasn’t run well in a state where only Republicans can vote in the primary.

I am hopeful that the rest of the primaries go against McCain and that the Republican party rejects the MSM’s pimping.

[diotima]

A poll is only as good as its sample. For caucuses, it is difficult to figure out what population you are sampling.

When designing a sample you choose how to mix some combination of the following: reliable voters, people who have voted in 4 out 4 of the last elections, likely voters 3 out of 4 of the last election; voters from the last 2 out of 4 elections; voters who have never voted but are registered. In addition you have to make sure the proportion of your voter sample is representative of the populations in each zip code and represents the proportion of male/female in the general population.

WHEN YOU READ THE POLLS, REMOVE THE UNDECIDED AND UNKNOWNS AND ‘REFUSED TO ANSWERS’ and then recalculate the proper percentage. If there are enough of these it will change the margin of error. Many times the public polls do not give you this information, so they are misleading. You do not see the raw count and you do not see how many people fall into the categories I mentioned above. If you look at the exit poll cross tabs, a good number of people decide who to vote for pretty close to the last minute: those people will either show up in a poll as ‘unknown’ or will show up too late to impact a tracking poll. Most polls are conducted over the course of three days+ and actually complete around 100 calls a day. One hundred is too small of a sample to scientifically pick up a trend, which may have been the case in NH where Hillary boomeranged at the last minute.

In a state like New Hampshire where there is no length of residency requirement or prior registration for the primary process, there is no way to even get these people into the polling sample. With a serious get out the vote effort, polling prediction is seriously damaged.

In Nevada, Republican turn out was LOW and Democrat turn out was HIGH.

In addition to the polls that are public, each candidate has his own pollster for internal consumption and these polls are not released to the public. The polls we all see are purchased buy the media and as much as everyone here thinks there is some conspiracy, if the polls are wrong or really off, polling companies will lose business.