A Point On The Statistics Of The Newsweek Survey

This might get moved to GD:2004, but i hope it doesn't, because it's not about the campaign. It's intended to illuminate the polls that have recently been trumpeted by Newsweek, (the originator), and the rest of press.

Supposedly, the change in poll numbers, which by the way, are NOT reflected by Zogby or Pew, was caused by a convention bounce. So, let's look at the numbers, shall we?

Given the limited coverage of the RNC festival, the ratings indicate that about 17 million people watched the speeches. (You can look those up, too! They were in LOTS of newspapers.) Based upon the breakdown of the Dem convention, it's known that about 70% of the people watching these conventions are partisan, positively. About 20% are truly independent/undecided, and about 10% are partisan, negatively.

So, that means that about 5.1 million people (give or take a few hundred thousand) who weren't going to vote for Li'l Georgie anyway, saw the speeches.

Now, the poll touted by Newsweek would have us believe that a dead heat at 44:44, two weeks previous, went to an 11 point lead! Hmmmm?!?!? That means that 11 million likely voters had their minds changed, when only half that number saw the convention. So, how can a convention bounce be attributed for a value twice that of the number who actually saw the speech? Answer: It Can't. No amount of positive coverage by Faux, or CNN, or anybody else would have that level of impact.

The only shift in trends that profound occur when the person LOSING the shift has done something incredible stupid or illegal. (Think Gary Hart.)

A EWMA analysis of the last 20 weeks of polling data, from the same organization as the current poll in Newsweek, indicates that given the state of the electorate, and the values over those 20 weeks, that a shift of that magnitude is one of 17 standard deviations!

For the folks here for whom statistics is torture, let me tell you what a shift of 17 standard deviations means. It means that the chances of someone predicting that would be roughly 1 in 17 trillion trillion! Yep, i in 17, followed by 24 zeroes! And yet, some pundits were predicting a post convention bounce. Hmmmm, again!

Let me assuage any fears about this poll. There is NO WAY POSSIBLE that these numbers are valid. This was obviously a push poll in which the pollers were trolling for an answer.

When Kerry didn't see a big post-convention bounce, the "conventional wisdom" suggested that since the vast preponderance of the electorate had already made up its mind, we shouldn't expect a big bounce. Then, when Bush gets one, the same people suggest that it's a post-convention bounce.

So a president, about whom we know just about everything, gets a bounce because people didn't know who they were going to vote for, but the challenger doesn't get a bounce when the american people learn more about him? No negatives, just no bounce? But the guy we already know gets a bounce? This fails the logic test.

So, add these two things together. A near statistical impossibility compounded by an abject lack of logic in explaining a bounce of ANY size would seem enough to indict these poll results as invalid and meaningless.

Hope you feel better now.
The Professor

However, the damage has been done. The reason is that the average American has the critical thinking skills of a houseplant. The average person reads that Bush got a bounce, so that becomes the truth. They read that Kerry got no bounce and that becomes the truth. So, in the mind of the average American, Kerry mut have done something wrong, and Bush something right.

If I were King of the World, I would mandate that no poll could be posted without a) posting who funded the poll and b) posting the exact questions asked of the respondents.

Spelling edit.

I don't think it matters who funded it as much, if you know the way the questions were asked. A push poll is pretty obvious in its questions.

Let's say that Newsweek funded this poll. That would tell us nothing. But, if the question were to read "Now that you've heard all the Swift Boat Veterans information, and since Kerry didn't get a post-convention bounce, and the RNC just had a very successful convention, who are you more likely to vote for, Bush or Kerry", how do you think the question would be answered?

If this were the case, and we knew the EXACT wording of the question, we'd know Newsweek's agenda. But, merely knowing that they funded it tells us nothing. So, i only think you need to know the latter.
The Professor

another-I usually ignore them because they generally have 0% face validity--otoh I answer the call to DU polls with relish.

You have pointed out some good things here to the being counters, Professor.

I took the polls and DID a scientific analysis of them, and they turn out to be so improbable as to be impossible.

So, while you suggest that the polls themselves have no validity, i'm reinforcing that by trying to show why. Just tryin' ta help!

Just thought DU'ers would like to know.
The Professor

and media control. The quadratic equation isn't political at all!
:hi:

It illustrates pretty definitively that Rove (or his surrogates) is somehow able to directly manipulate Time and Newsweek to release flawed polls and coordinate it w/ the end of the convention and a major media push of "Kerry is collapsing/in dissarray!"...It's illuminating as to how much power they really do wield over the mainstream media and it's frightening...

Link to a thread I started regarding this:

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=132x768727

Hey, i know i'm preaching to the choir here! No surprises!

But i did write a LTTE to both Newsweek and the Trib to point these things out. Can't hurt. Might not help, either.
The Professor

quite scandalous; that the media and polls are being manipulated like this...*sigh*

Thanks for all your great work as usual Professor!

Apparently Zogby has found the flaw in the in their poll.

Two new polls came out immediately after mine (as of this writing) by the nation's leading weekly news magazines. Both Time's 52% to 41% lead among likely voters and Newsweek's 54% to 43% lead among registered voters give the President a healthy 11 point lead. I have not yet been able to get the details of Time's methodology but I have checked out Newsweek's poll. Their sample of registered voters includes 38% Republican, 31% Democrat and 31% Independent voters. If we look at the three last Presidential elections, the spread was 34% Democrats, 34% Republicans and 33% Independents (in 1992 with Ross Perot in the race); 39% Democrats, 34% Republicans, and 27% Independents in 1996; and 39% Democrats, 35% Republicans and 26% Independents in 2000. While party identification can indeed change within the electorate, there is no evidence anywhere to suggest that Democrats will only represent 31% of the total vote this year. In fact, other competitors have gone in the opposite direction. The Los Angeles Times released a poll in June of this year with 38% Democrats and only 25% Republicans. And Gallup's party identification figures have been all over the place.

This is no small consideration. Given the fact that each candidate receives anywhere between eight in ten and nine in ten support from voters in his own party, any change in party identification trades point for point in the candidate's total support. My polls use a party weight of 39% Democrat, 35% Republican and 26% Independent. Thus in examining the Newsweek poll, add three points for Mr. Bush because of the percentage of Republicans in their poll, then add another 8% for Mr. Bush for the reduction in Democrats. It is not hard to see how we move from my two-point lead to their eleven-point lead for the President.



http://www.zogby.com/news/ReadNews.dbm?ID=859

Good article. I also like Zogby. I trust their methodology quite a lot. I also like Pew, but i'm biased, since i know statisticians who work there. I don't know their methodology in detail, but i trust those people to do scientifically valid work.

The Professor