Matthew Yglesias

First one must be cutting taxes rather than cutting spending, or else the leadin makes no sense.

Also, haven't you heard that Missouri is the archetypal American state now? It was in the Atlantic last month, I think. I can't find a link.

I'm generally wary of Rasmussen because:

1. He runs a "machine" poll, where the questions are asked by a recording, not a human being.

2. His past history of inaccuracy. He ran the "Portrait of America" poll during the 2000 race, which predicted a Bush victory margin of 7-9 percentage points on election eve.

If you look at the table to the left of the story at the link, it becomes clear that "spending", at the end of paragraph 2, should be "taxes".

Which is odd. How is cutting spending versus balancing the budget an either/or? And to the degree it IS an either/or, in that you could have a balanced budget with the same tax rate and higher spending versus a deficit budget with the same tax rate and less spending, how do you come up with option 2?

And if you say, well maybe they mean cutting taxes.....but then how do you reconcile that with the first part of the poll. If lower taxes/lower spending is better than higher taxes/higher spending, regardless of the deficit, then why wouldn't cutting taxes by itself be more popular?

Odd, odd, odd.

hehe...that should be in The Onion.

Of course the polling results are incoherent. Anyone who isn't a policy wonk wanst four mutually incompatible things:

1. Lower taxes, expecially on people like me;

2. Higher spending on what benefits people like me;

3. A balanced budget, and;

4. Lower spending on people not like me.

So what else is new?

Can we now add Rasmussen to the list of polls to ignore?

Here's my list:

1) American Research Group
2) Survey USA
3) Rasmussen

Zogby flirts every now and again, but stays off for now...

When Zogby does a targeted poll of something specific, it's usually paid for by somebody who wants a very specific answer. He's nonpartisan about that, too, so sometimes you get the "even the liberal Zogby" effect on polls that were, say, paid for by AEI.

His presidential polls, though, are pretty good as long as you account for their systematic biases. For instance, his job rating for the president is often reported along with other "approve/disapprove" polls, even though his is actually something like a "very good/pretty good/fair/poor" rating, and "fair" and "poor" are grouped under disapproval. That produces a significant disapproval bias relative to the others; it's not a big deal except that you have to take it into account when trying to compare apples and oranges.

Ill never understand why people have such weirdly irrational hatred of the macine polls. Every analysis of the data shows them to be at least as accurate as traditional methods....

i think what the poll shows is:

never mind

Also -- maybe I just don't understand polling, but... +/- 3% margin of error, with 95% level of confidence? Wouldn't that mean 98%?

Who are these bozos?

I believe it means they're 95% confident that the results are within 3 percentage points of the "actual" ones?

It's standard statistically terminology. You'll find it at the end of most survey results.

Funny thing is, polls often show logical contradictions in voters...but this has to be a typo.

samian writes:
Ill never understand why people have such weirdly irrational hatred of the macine polls. Every analysis of the data shows them to be at least as accurate as traditional methods....

------

My answer is... they're not accurate and they have not been shown to be statistically valid. Take SurveyUSA for example. If you look at their polling for the recent presidential primaries, their results are outside the MOE (often waaaaay outside) almost half the time. The second point is that, to the best of my knowledge, the auto-dialers like SurveyUSA, etc. simply plow through lists until they have the correct sample size. This type of sample is very far from random and again cannot be considered statisitically valid. I don't know where samian gets "Every analysis of the data shows them to be at least as accurate as traditional methods...." - it seems easy to verify that, in fact, mnachine polls are often much less accurate tahn a reputable "traditional" pollster.

My beef with Zogby is that he adds in his little weights here and there without explaining how or why. In my book, that is doctoring data. Take a look at Zogby's NH primary poll and then his WI primary poll. On the eve of the NH primary, he wildly adjusted his numbers and magically brought them in line with everyone else (and the outcome). In WI, he didn't and came out wildly wrong. In his defense, the rest of the primary polls were decent with a continuing problem of underestimating Edwards and misestimating Kerry, which may simply reflect late swings in voters choices rather than methodology problems.

"Can we now add Rasmussen to the list of polls to ignore?

Here's my list:

1) American Research Group
2) Survey USA
3) Rasmussen"

Agree with 2 and 3, but not 1. ARG is a legitimate poll that has proven to have a fairly decent track record. Personally, when it comes to state polls, I'll take Mason-Dixon. Zogby's national poll has been right on the last two elections. The computer generated poll used by 2 and 3 is suspect for the basic reason that you don't know who's responding. It could be some 14-year-old.

Also, haven't you heard that Missouri is the archetypal American state now? It was in the Atlantic last month, I think.

Oh dear. As a Missourian: we're all screwed.

If a pollster asked the questions the right way and in the right order, he could probably engineer a circular result, in which the poll would show that:

(a) Balancing the budget is more important that cutting taxes;

(b) Cutting taxes is more important than cutting spending;

(c) Cutting spending is more important than balancing the budget;

(d) and so on forever ad nauseum.

Also -- maybe I just don't understand polling, but... +/- 3% margin of error, with 95% level of confidence? Wouldn't that mean 98%?

Who are these bozos?

No.. actually, the statement makes a sort of statistical sense. However, you need to take a semester of college statistics, or experimental statistics to understand it. I wouldn't try to explain on a blog. There is a problem with the use of the term "error".. but setting that aside, what they mean is roughly this:

Say they repeated the poll over and over, using the exact techniques they used previously, (same question same day, same number of people.), and calculated the result.

Then in 95 % of the "polls", the results would be within +- 3% of those stated here. In 5% of the "polls" the results might differ by more than +-3%.

The problem with the word "error" is that most people reading the poll are trying to learn what people *think* and not necessarily how they answered the peculiar question posed by the poll.

hmm.. I guess I said I wouldn't try to explain it.. and then I did! LOL!

Paleo-

Quick note on ARG - they suck. Again, look at their data for the recent primaries (or 2000 for that matter where they picked Bush over McCain by about 5 pts). With the exception of NH, every single poll they did was well outside the alowable reported margin of error. Moreover, the have a reputation for not checking their data and their methodology for picking numbers is one step above autodialing.

Take a look here for an analysis of Zogby, ARG and SUSA for the first few primaries. I didn't bother to the the rest because it all dropped off too quickly...

It seems that it was a typo and has been corrected.

Rasmussen now has
* By a 52% to 39% margin, American voters say that balancing the budget is more important than cutting **taxes**.

not

"By a 52% to 39% margin, American voters say that balancing the budget is more important than cutting spending."

A Rasmussen slip not a14% reverse preferences based on word order (that would be embarassing).
Oddly hard core applied general equilibrium fresh water chicago economists agree with the majority that what matters is govt spending not the deficit.

internatinoal data and
I disagree with the 55% of national missourians and, I think, 93% of the dept of economics of U Chicago.
The data on economic growth in different countries

Robert,
I think you are assuming that those polled *assume* that these questions are related to the issue "revving up the economy" and answer based on what they believe will rev up the economy.

National missourian's may want limited government spending and low taxes for other reasons.