I'm not big on polls. Despite what anyone might say to defend them, I find them to be largely a big waste of time. Between how the questions are asked, how many people are generally asked, and who those people are, there's not a lot to support giving them the time of day. The best polls are elections. The amount of people "polled" and the fact that something definite occurs as a result makes them more worthy of notice. But any poll with direct questions and a large pool of respondents is better than ambiguous questions to groups of people 5000 or less (and usually MUCH less).
So I was interested in an AOL poll today with two questions. At the time I answered the question "What grade do you give Obama's presidency so far?", the total votes was 234,124. It played out as follows:
F-40% 92,926 votes
D-19% 45,139 votes
A-19% 44,652 votes
B-11% 26,183 votes
C-11% 24,224 votes
The second question was "What grade do you give Obama's economic policy so far?" and the total votes was 236,298. The results at the time were:
F-49% 114,630 votes
D-15% 35,598 votes
A-13% 31,253 votes
B-13% 30,169 votes
C-10% 23,648 votes
Now, I don't know about any of you, but my momma was never pleased with C's on my report cards, tests or any other school work. But if we say that a C means one isn't failing, then 59% of those responding (234,124 did) think Barry's failing. (To Momma, a D meant you were failing, even if you got promoted to the next grade level.) And it gets worse when looking at Obama's economic policies. 64% think he sucks out loud. I wonder how many of that almost 1/4 million respondents voted for the guy?
Friday, March 13, 2009
17 comments:
Your First Amendment right to free speech is a privilege and comes with a measure of responsibility. You have the right to exercise that responsibility here but we reserve the right to inform you when you've used that right irresponsibly.
We are benevolent dictators in this regard. Enjoy.
Subscribe to:
Post Comments (Atom)
You understand the difference, I suppose, between scientific polls and junk questionnaires?
ReplyDeleteYeah, we understand your understanding, Danielsan, whic is:
ReplyDeleteIf its conducted by liberals...it's scientific.
If by conservatives...not so much.
Did I get it right?
What I'd like to know is how many times respondents were allowed to vote. The answer would go a great distance toward either credibility or complete trash.
ReplyDeleteOn the other hand, I have to agree with Al's assessment.
What I understand, Dan, is pretty much what I said in the post. Most surveys are poorly worded and give results based on the response of maybe 2000 people, if you're lucky. Yet we are to take such as some sort of reliable indicator of national sentiment. Nonsense. The poll I highlighted included questions fairly stated and offered five options for answers. And as I just checked the AOL archive before posting this comment, the total responses for each question is around 360,000 people. Scientific or not, I would put more confidence in the results of this poll than the average polls often put forth as the be-all and end-all of national opinion. (BTW, the percentages stayed pretty much the same with the extra 130K responses. Another BTW, one can only vote once from a given computer. To try again only gives results and not the chance to take the poll again. "Cheating" would require finding extra computers and voting several times. Not likely anyone would for an AOL poll.)
ReplyDeleteScientific or not, I would put more confidence in the results of this poll than the average polls often put forth as the be-all and end-all of national opinion.
ReplyDeleteScientific or not. Is it your opinion, then, that polls that agree with your hunches are more likely to be right, but those that disagree with your opinions are more likely not to be valid?
And you base that on, what?
Dan pseudo-righteously asks, "You understand the difference, I suppose, between scientific polls and junk questionnaires?"
ReplyDeleteI do. Anything Dan agrees with, which is usually anti-Christian and anti-Obama, is scientific, and anything representing truth, justice, and the American way is junk questioning.
Eric,
ReplyDeleteTo answer your question, once I responded to the poll, my computer would not allow another vote. So, if I wanted to pad the results, I would have to find another computer, perhaps each of those at the library, if they are all separate from each other, and vote again that way. Seems like a lot of trouble just to pad an AOL poll. I don't know of any other way to cheat, so I'd say the results are at least as credible as a Minnesota senatorial election.
Dan asked,
ReplyDelete"Scientific or not. Is it your opinion, then, that polls that agree with your hunches are more likely to be right, but those that disagree with your opinions are more likely not to be valid?"
I thought I was pretty clear in that in my opinion, polls, by and large, are crap and a big waste of time. The reasons for this, once again, is due to the way the questions are often phrased together with both the limited number of respondents (usually around 2000 people or less), and the inability of someone like either of us knowing who was asked the question. How do we know that the pollers really sampled the diversity of our nation and not just, say, Berkley, California?
This AOL poll has the advantage of being put to anyone who happens to have AOL and the desire to respond. Of all their users, over 360,000 responded. The sheer numbers alone makes the poll more credible than the usual polls from the usual suspects. And each question allowed for five answer choices to further focus in on the true pulse of the nation.
So, wise guy, the results don't mean a freakin' thing as to whether or not I think a given poll is credible. But the criteria under which the poll or survey was taken means that one would have more credibility than another.
Even more to the point, if a poll was taken that asked, for example, who's position on homosexual marriage was closer to God's Will, and the choices were Dan Trabue and Marshall Art, the whole world could agree with you and I'd still be right. I don't look to polls to tell me what's right, only what others MIGHT be thinking according to the limitations of the polling questions.
Does that answer your question?
"Scientific or not. Is it your opinion, then, that polls that agree with your hunches are more likely to be right, but those that disagree with your opinions are more likely not to be valid?" - Trabue
ReplyDeleteSpeaking for myself, Dan...no. It is my opinion that YOU believe "that polls that agree with your hunches are more likely to be right, but those that disagree with your opinions are more likely not to be valid."
Didn't you already get this from other comments on the thread?
Didn't you know, D? Turnabout is fair play.
ReplyDelete"The sheer numbers alone makes the poll more credible than the usual polls from the usual suspects."
ReplyDeleteActually no. Online polls inherently have the failures of the medium and the bias of the audience.
1. You can easily have more lots of multiple voting. All someone has to do is clear the cookies on their browser to be able to vote again. Or people like me who only keep cookies for a day could vote every day. This of course also doesn't weed out auto voting scripts which are really easy to implement.
2. Online audiences tend to be younger and more affluent. You also need a base education to get and surf online. The AOL poll undoubtedly undersamples seniors, the economically disadvantaged and also those with dialup connections.
3. AOL has no reputation for journalistic truthfulness. I would not be surprised at all to find they added about 10,000 votes to each category to get the poll started.
From an uneducated laymans point of view it is suspicious that you can diagnos national trends from samples of 1,000 - 2,000 people. However if you take a course or read an introductory text to statistics you can understand how they do it. This is what we call science. And the amazing thing is that with science, what works for one person works for everyone else too. And you can test out what works for other people.
Cool eh?
"AOL has no reputation for journalistic truthfulness." - BenT
ReplyDeleteNeither does Gallup, NYT, LAT, WaPo, Zogby, etc.
Unless you're a lberal...then you'll believe anything these organizations feed you.
"AOL has no reputation for journalistic truthfulness. I would not be surprised at all to find they added about 10,000 votes to each category to get the poll started."
ReplyDeleteThis is true, however, AOL would skew the numbers in the other direction if indeed they truly add votes, because they are left leaning.
BenT,
ReplyDeleteShame on you for knowing how to cheat! But really, do honestly think too many are that intent of discrediting Obama to take the time? I can't stand the guy, and I wouldn't dream of it.
I understand that there are "scientific" explanations for the use of small sample polling. I just don't buy 'em. 1000-2000 people represent a country of 300mil? I don't think so. It's a cheap and lazy way to influence opinion by claiming it can represent the feelings of so many more. I don't waste my time with them. But I will pay a little more attention to those that survey over 100,000 or more.
Three other problems with the AOL poll is that they don't correct the results for gender and location.
ReplyDeleteThe overall average ratio is 1.03 women to every man. Online the ratio is almost twice as many men to women. But this poll doesn't correct for those glaring differences.
Second online activity is much more rare in rural areas. And population is greater on the east and west coasts.
There are too many problems with this poll to say that it reflects national opinion.
For an "uneducated layman" Ben is awfully authoritative, and confident of his "facts".
ReplyDeleteI think, usually, the results of a survey can be skewed by the way in which a question is formed. That is why I put little credence to them, in general. If the question is "Do you approve of Obama?", the answer would be either yes or no. Kind of hard to read anything else in the answers to that. Either you approve of him or you don't.
ReplyDelete