naiive question about media campaign polls (Gallup, Rasmussen, etc)

[steve2470]

I, like most of you, have been astonished to see Obama and McSame so close in the polls. I want to believe that the polls are all BS, for one reason or another. What is the evidence that they are BS ? Yes, I know the state polls and the Electoral College are the real bottom line. Thank you for your time, as always.

[NYC_SKP]

1. Check this post earlier today.

Read the OP, and replies 5 and 17 for some good points:

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=132x7075953

:patriot:

[PermanentRevolution]

2. My (long-winded and largely instinctive) opinion

Let me preface this by saying that I don't actually have any specific knowledge of statistics or political polling beyond what I've read on the internet. So this shouldn't be mistaken for any kind of "insider" perspective. It's not. But I think I'm relatively bright and decently shrewd when it comes to politics in general.

So here goes:

As I see it, you can't just say "all polls are BS" any more than you can say "all polls are gospel truth." You just have to recognize exactly what a poll is, and what it says.

It's not a representation of what the country is thinking. That would be impossible. There's far too many variables for such an assessment to be possible, short of polling every single person in the country while they're strapped to a lie detector. Polls are an educated guess, based on taking a small sample and extrapolating that to a representation of the whole, based on a series of criteria such as race, age, sex, location, party affiliation, and voting history.

So, each pollster has their own methodology on the best way to reach that estimation, which will produce some degree of variety. Two polling companies could poll the exact same sample of people, receive the exact same answers, and, based solely on their particular method of weighting, arrive at two different figures.

But they're NOT using the exact same sample, so the disparity will increase depending on random chance. One pollster may, by chance, reach a sizable number of people whose opinions are actually not in line with the norm. That poll's results would be skewed by these opinions to a degree where it would not reflect the actual situation - not through intentional manipulation, but through an aberrant sample.

So it's not that they're necessarily bullshit, it's that there's almost no chance that two polls taken by two different companies will reach the exact same result, given the number of variables involved - REGARDLESS of any intentional skewing - and therefore highly unlikely that any one poll will represent with 100% accuracy the actual situation.

As far as deliberate poll manipulation, there's certainly potential - after all, they're the ones who decide what questions they'll ask, who they'll call, and how they'll weight specific demographics to reach their final numbers. So it's certainly possible, which is why pollsters that release the specifics of their weighting methods and their samples are generally considered "better" than those that don't.

It all boils down to who you trust and how much. Personally, like it or not, I'm inclined to think that the race is pretty much tied right now. It's easy for us to say "How can it possibly be tied after what Bush has done? Have people not been watching?" but we're predisposed towards that viewpoint since we're here at DU. But the fact that we can't understand a differing viewpoint doesn't mean that it doesn't actually exist.

In my opinion, the evidence points to the contrary - that there is a sizable portion of the country for whom the Republican brand is not an automatic negative, who feel that McCain represents a different and better path for the future than Bush OR Obama - and ignoring, rationalizing, or marginalizing them won't make them go away.

That being said, I'm also convinced of an Obama victory in November, but only if we get our side to the voting booths and in massive numbers. Turnout will swing this for us - it depends on getting all those newly-registered voters to show up in the end, regardless of what the polls are showing.

So, to sum up: Polls - take 'em or leave 'em, but the ground game is what really matters.

[woolldog]

3. They're not BS

Some people here are just in denial.

[Bonn1997]

4. I agree

I find the polls surprising but I also think the people claiming the polls are fixed here have low standards of proof. Many of them often post the poll results by these same organizations in excitement when they show a good result for Sen Obama. These are the same organizations that had Sen Obama up by about 8 points 2 weeks ago. A much simpler possibility might just be that about 45% of the public votes Republican and 45% Democrat because of many moral issues (ranging from economic to social to national defense) regardless of who the candidate is, leaving only about 10% undecided. I'm not sure that this idea is correct but it's an idea that I've been considering. I'm not saying the polling results are valid, though. I just want some stronger evidence than I've seen to indicate that they aren't.