Polls apart: Why survey results don't agree

[BridgeTheGap]

WASHINGTON — Barack Obama is galloping away with the presidential race. Or maybe he has a modest lead. Or maybe he and John McCain are neck and neck.

Confusing? Sure, thanks to the dueling results of recent major polls.

In the past week, most surveys have shown Democrat Obama with a significant national lead over Republican McCain. Focusing on "likely voters" — as many polling organizations prefer this close to Election Day — an ABC News-Washington Post survey showed Obama leading by 11 percentage points. A Wall Street Journal-NBC News poll had the same margin, while the non-partisan Pew Research Center gave Obama a 14-point edge.

But others had the race much closer. CNN-Opinion Research detected an Obama lead of 5 points. The George Washington University Battleground Poll had Obama up by 4 points. And an Associated Press-GfK poll showed Obama at 44 percent and McCain at 43 percent — in effect, a tie.

http://www.kentucky.com/329/story/566496.html

[ShortnFiery]

1. To cut to the chase: Polling validity is based on sound methodology.

The Sample must be large and reflective of the composition of the voting populace.

Methodology is all important. :patriot:

[volatileblob]

2. Right the polls that have the race nearly tied think....

that evangelical/born again Christians are 44% of the population (AP) OR that 18-24 year olds are going about 75% for McCain (IBD).

[bemildred]

3. Well, the thing is, it's like religions.

They disagree, and only one can be right. But if that is so, why single one out? Why have the special case? It is simpler and more reasonable that they are all, similarly, wrong; all fallible products of fallible humans.

In the same way, the polls do not agree, so why assume any are right? It's simpler and more consistent to assume all are meaningless, than it is to assume some subset of the constant bombardment of them are "right", and then try to figure out which they are, and how "right" they are. It is perfectly clear that they are USED TO MANIPULATE PERCEPTIONS, so why pay attention at all? Why the obsession with knowing how the election will come out before it is held? Isn't the whole point of elections to find out what "the people" REALLY think?

[groovedaddy]

4. Polls are approximations. And they're sure doing a roaring business

for a methodology that is "useless."
As approximations, there is no "wrong or right." That's why they have margins of error. They certainly can be "used to manipulate perceptions," but that isn't their only use.
FDR was the first President to use polling, in an effort to determine the American people's opinion on going to war to stem the rising tide of fascism. The public was overwhelmingly against the notion all the way up until Japan attached Pearl Harbor.
Polling can also be used to gauge the effectiveness of messaging and the impact of events on people's perceptions (i.e. the collapse of wall street and the bail out). Elections alone don't reveal this aspect of polling.
Polls aren't going to go away any time soon.

[bemildred]

5. I understand the math quite well, Sir.

So if there is a lot of money spent on it, it must be good?
Sorry if I touched a nerve.

[groovedaddy]

6. No, you didn't touch a nerve. As much as I don't like the over all effect of polling

I simply don't believe it's going away.

[bemildred]

7. Not likely, true. nt