Uechi-Ryu.com

A little bump here, a little bump there...


USA Today/CNN/Gallup Poll Results

I have to admit... The overall trends say what they say, but the details baffle me.

- Bill

Watch what happens if Osama gets captured a week before the election.

A sneeze will do...

This is what chaologists refer to as "the butterfly effect."

- Bill

Zogby International is a poll I trust. Very up to date.

On the eve of the third and final presidential debate, President George W. Bush remains in a toss-up race with Senator John Kerry at 45% apiece, according to a new Reuters/Zogby daily tracking poll. The telephone poll of 1232 likely voters was conducted from Saturday through Monday (October 10- 12, 2004). The margin of error is +/- 2.9 percentage points.

Zogby poll details

- Bill

Here's another result after debate number three. IMO, it is only useful in the respect that the conditions for asking the questions were the same as for the first two debates. But these immediate response polls don't necessarily tell you alot. We in my field call these - at best - proximal outcomes.

USA TODAY/CNN/Gallup post-debate poll-results

Do these results from last night pass your "sniff test?" Most of these CNN/USAToday/Gallup polls have left me scatching my head. The overall poll says one thing; the details say something else. Sometimes that means you have either a flawed sampling methodology or a faulty survey instrument. For example, it doesn't help to compare if you feel more or less favorably about two people when you know one pretty well (already) and the other is a stranger to you. And it still doesn't tell you how you feel about them (only how those feelings have changed).

Stay tuned for my next Zogby result, when it comes. This is a little more reliable IMO, and a better reflection of the "distal outcome." Their results tend to be real close to actual election results, and their survey people know what they are doing.

- Bill

We won't know for sure until November 2nd. However this difference is outside the statistical margin of error (not likely due to chance or "noise").

A day after the final debate, President Bush has opened a four-point lead over Democratic rival Senator John Kerry, according to a new Reuters/Zogby daily tracking poll. The telephone poll of 1220 likely voters was conducted from Tuesday through Thursday (October 12-14, 2004). The margin of error is +/- 2.9 percentage points.

See here.

ZOGBY INTERNATIONAL




- Bill

This most recent difference is within the statistical margin of error.



Zogby International

- Bill

The USA Today/CNN/Gallup poll though is not as close (8% difference).

USA Today/CNN/Gallup

It just goes to show how sensitive these polls are to sampling technique and survey instrument design.

- Bill

The biggest problem I see with all these polls is that none of them take into concideration the Electoral College. As we saw in the 2000 election, you can win the majority vote and still lose the election.

As someone who lives in Massachusetts, it is pretty well assumed that Kerry will win by a very large margin and collect all of our EC votes. But go to Ohio and the winner will probably only win by a small margin and will collect all of thier EC votes. The survey I have not seen very often is who will be winning the "battleground" states...

Cheers,
ljr

You get an A+ in Civics, sir. You are absolutely correct.

Zogby does that very poll. I get to look at it from time to time on my Wall Street Journal Online subscription. There's a beautiful, interactive map that allows you to go from state to state with your cursor and see the details. Only problem is that there is no way I can put that thing online. WSJ knows that. Zogby knows that. They give us these teaser polls with some nice bar and/or line graphs. Then a smart person like you comes up and asks the $50,000 question. Want to know the answer to that? Buy WSJ online, or get a subscription to Zogby online.

Pretty cool marketing technique, no?

Indeed with most states (like my own state of Virginia and your state of Massachusettes) sewn up, the election comes down to a handful of states where the election is very, very close. In fact, it has been said that the person who wins 2 out of 3 of Ohio, Pennsylvania, and Florida will win the election.

Check out the travel plans of Kerry and Bush these days. You will see that their behavior reflects this reality.

However...

I am an engineer by training. We engineers understand the value of approximation. I was also a geochemist in a previous life. I also knew the value of approximation there.

For example, I was part of an effort that won my previous employer a massive contract with the DOE to take about 100,000 soil samples from around the United States and measure the level of 27 different elements. What helped us get the contract was pointing out that relative and not absolute accuracy was most important. After all, they wanted to know what minerals existed hundreds of feet below the earth's surface by measuring the elemental composition just a few feet below the surface. We already knew the latter would only show traces of the minerals of interest. All you really want to know is hot spots and cold spots so you know where to dig first. And it is a LOT easier (and cheaper) to do assays that have relative as opposed to absolute accuracy.

What's important about these polls is to look at the trends. For example the conventional wisdom is that the political conventions give a candidate a bit of a "bounce" in the polls. However one comedian commented that after the DNC, he saw a bigger bounce in Laura Flynn Boyle's bra. The point was made; somehow Kerry goofed.

Similarly the conventional wisdom would be that the winner of the 3 debates would probably win the election While Kerry got a great bounce from the first debate (probably because undecideds got to know him a bit better), he was not able to translate good performances in the next two into numbers in the opinion polls. In fact just the opposite is happening. Why? Good question... I think I know why, but it's too early to tell yet and I'm probably perceived as not being objective here.

The bottom line here is that the trends - which went in Kerry's way for a bit - are starting to turn around. Why they are isn't as important as the fact that they are. Will it translate into enough electoral college votes? Who knows? But if I were a betting man...

And smarter minds than mine on the matter have commented that - just as in 2000 - either one of these candidates could win the popular vote and lose the election. And that's just the way our electoral process works. Study it enough and you'll see the advantage of such a system.

- Bill

And just when you think you have it all figured out...



And that flies in the face of the latest Gallup poll which spots Bush by 8%.

It ain't over 'till it's over.

- Yogi Berra.

The electorate (like your truly) is quite fickle. Is there a better sign that they are not happy with the available choices?

- Bill

I really like what I see.

Also, a recent poll of military members, and their families, prefer Bush by about 2 to 1. Then there are the troops overseas that will vote by absentee ballot. Those votes will go about 3 to 1 for Bush. In 2000 there were members of the DNC who were plotting to have those votes thown out! One DNC leader (cannot remember who) went so far as to call those deployed overseas as 'tax dodgers'!

We will see the the margin grow in Bush's favor in the remaining days before the election.

Rich

Aren't some of the polls biased, as in registered voters vs. those likely to vote?

Good question.

By all means, look at the details. I have provided the links for you to do that.

The polls I am quoting, are likely voters. They are specifically designed to reflect the popular vote - in total.

Conventional wisdom has it that good weather favors Democrats, and bad weather Republicans (due to transportation issues amongst a small core of lower socioeconomic class Democrats). So there are all these little factors that can come in to play on election day.

- Bill

here is another page with electoral college predictions...
http://www.electoral-vote.com/

and here is an interactive map like you mentioned...

http://www.latimes.com/news/politics/el ... 1284.flash

cheers,
ljr