Blue Crab Boulevard

About Interesting Polls (Verily)

One of the regular commenters here asked what I thought of the Newsweek poll that has diametrically opposed numbers than the fast Washington Post poll. My first impression is the WaPo poll was a very fast reaction poll and admitted it was done that way - the margin of error is likely higher, but it may actually be quite accurate. I have not seen the raw data, but I'm guessing they simply did cold calls and did not worry about how many Democrats and how many Republicans, etc. The Newsweek poll was done by an outfit I am not real familiar with, and there is no raw data, so it's impossible to say how they set the sample up.

Frankly, if it's up to the standard of many polls I have commented on, it's likely got a badly - and I mean badly of biblical proportions - skewed sample. Most of the latest polls have had absolutely absurd samples. Republicans outnumbered by over 10% by Democrats and and an enormous independent count. Far, far outside the rather consistent over time Harris Poll numbers.

So, until I see raw data, I won't be able to judge. Interestingly, the Mystery Pollster thought the WaPo survey was pretty well done.

UPDATE: Actually, this is probably enough to do a new post on, but I'll post it here. First, Dan Riehl of Riehl World View was actually called by Gallup and is one of their poll sample. Read his take on the poll, because it is important. Assuming his description is accurate (and I am willing to bet it is) the form of the questions provide a "push". Look - I was a psych major once upon a time and had to take design of experiments. That's a push poll. There is a vaguely threatening air about the questions - be intellectually honest. (I know this is a different poll than the Newsweek, but I think it is relevant).

Second, Curt over at Flopping Aces has got the sample data for the Newsweek poll. It is heavily biased. There are 9% more Democrats and 32% ind pendant? 32%? Are you serious?

Now I'm calling it a biased poll, and a serious problem.

I have to ask a question and ask that you answer honestly. Is gaining political power so important that you are willing to do so by using fabricated data to do it? Is it so important that you are willing to lie, cheat and steal to gain it?

If the answer is yes, then explain exactly how are you one iota better than the people you charge with using those tactics?

I'll wait.

Media | Gaius Saturday, 13 May , 2006

By Curt, Saturday, 13 May , 2006 @ 6:58 pm

Got the raw data here, and your right, oversampled Democrats and Independents once again.
By Tano, Saturday, 13 May , 2006 @ 7:47 pm

I think this is a complete canard.

If you do a random sample, then you get what you get. If the polls are finding what you consider to be more democrats than what you believe there to be, the obvious answer is that people are changing their party affiliations. After all, who wants to admit being a Republican these days?

Other polls simply ask for party affiliation. Recent ones show dems with about a 3 pt. advantage. Now, tell me - why do you think that those polls are accurate, while another poll, which asks for party ID as a secondary question, is inaccurate (besides your preference for a certain answer)? There isn’t anything magic about a straight party ID poll that makes it inherently more accurate than, say, a presidential approval poll.

A posteriori poll weighting, to reflect what you think the partisan breakdown “really” is, is contrary to good polling methodology (although many firms do it). If your sampling protocol is valid, then your results will be valid (within the margin of error at 95% certainty). If your sampling protocol is not valid, then you should fix it so that it is, rather than fudge the numbers afterward to reflect what they “should” show.

I’ve recently seen an example of this (usually republican) spin carried to an extreme. One clown claimed that since Bush got 51% and Kerry 48%, therefore all polls should be weighted to reflect these numbers (including eliminating most all who claim to be independents). Sorry I dont remember where I saw that - and yes, it is an extreme example of idiotic spin, but following this logic only a little bit is equally bad.

Interesting that you recommend the Harris poll. Thats the one that just came out showing Bush at 29%!
By Gaius, Saturday, 13 May , 2006 @ 7:51 pm

Read the update.
By Tano, Saturday, 13 May , 2006 @ 8:17 pm

Thank you Gaius! Yup, that was the idiot I was referring to!

Lets eliminate a third of the respondants, because they claim to be independents, when we know from the last elections nearly everyone voted for a Dem or a Repub.! (not counting of course the 40+% who didnt vote).

Can you believe such idiocy?

As for you. What on earth are YOU talking about? First you recommend to us the Harris poll of party identification to make your point about skewed data. Then you moan (along with the moron) that the Newsweek poll shows 32% independent.

So, remind me please. What does your Harris poll show as the number of independents?
By Tano, Saturday, 13 May , 2006 @ 8:30 pm

Gaius,

As to the “push” poll concept.

You really gotta stop taking on faith the testimony of horribly biased spinners. Here is the actual question from the Newsweek survey.

14. Now on another subject. . . As you may know there are reports that the NSA, a government intelligence agency, has been collecting the phone call records of Americans. The agency doesnâ€™t actually listen to the calls but logs in nearly every phone number to create a database of calls made within the United States. Which of the following comes CLOSER to your own view of this domestic surveillance program?

It is a necessary tool to combat terrorism.
It goes too far in invading peopleâ€™s privacy.
Donâ€™t know.

Seems about as straightforward as you could hope for.
By Tano, Saturday, 13 May , 2006 @ 8:49 pm

If you want to actually think about what a stat guy has to say about this whole party id business, I would recommend the following post:

http://politicalarithmetik.blogspot.com/2006/03/partisanship-moves.html

Its a bit long, lots of data, but much better for the mind and soul than running around playing partisan-spin games.
By Gaius, Saturday, 13 May , 2006 @ 8:50 pm

Did you bother to read the Harris poll I linked or just assume it was one you thought it was?

24% Independant. 3% differance between Dem/Repub.

Tell me again how that sample is fair as can be.

Your other point - moot. I said the Newsweek poll was not Gallup. You assume bias on Dan Riehl’s part - I don’t. You’re using one question. What are the rest of the questions and the phrasing? Push polling is more than one question taken out of context from the whole.
By Gaius, Saturday, 13 May , 2006 @ 8:53 pm

I have no idea who that is. Do you?
By Tano, Saturday, 13 May , 2006 @ 9:51 pm

< ?xml version="1.0" encoding="utf-8"?> < !DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
Gaius, Yes, I saw the Harris Poll. 24% is a lot closer to 32% than what your friend ***********was claiming. And that Harris Poll number was from 2004. Go to the article I recommended above and see the range, over time, of about a dozen different firms on this question. Range for independents is 20-40%. As for Dan, you know perfectly well that he is a partisan (I assume you read him regularly), so I dont know what you are basing your statements on. As for the "one question". Well hell, thats the question we are talking about isnt it? Bottom line here: you are making a serious charge - that the polling firm is deeply unethical, on no basis WHATSOEVER. THere are good words to describe that behavior, but I'll be nice. Dont you have any regard for your own reputation? Dont undrstand your last post. Who is who? You asking me?

Edited By Gaius - Personal attacks will be automatically deleted in the future - not edited. Final warning.
By FrauBudgie, Sunday, 14 May , 2006 @ 4:39 am

Popularity blows with the wind; polls are a reflection of what’s popular in current thinking. Are they important — yes, to a certain extent.

Should they be used to influence policy? Maybe not.

There’s a difference between a prom queen and a statesman.

The above is not a defense of the Bush administration, although I am a conservative.

What’s interesting to me is that this is not the first poll that’s come out with a bias in sampling. Good article.
By Juggler, Sunday, 14 May , 2006 @ 10:44 am

“n NSA intervention thatâ€™s not new, not illegal, and not intrusive into private livesâ€¦”

It’s a program begun under this administration, and it’s been unknown, so its existance is certainly new. It remains to be seen whether it’s illegal, but it doesn’t seem to meet Fourth Amendment criteria. We’ll probably never know how intrusive into private lives it’s been.

Conservatives used to say the scariest words in the English language are “I’m from the government, and I’m here to help you.” Now they say the most comforting words in the English language are “I’m from the government, and I’m watching you in order to protect you.”