@Ticomaya,
Ticomaya wrote:
The poll result I posted was from the cable news broadcast last night ("
Text 3945 and push "A" for Biden, "B" for Palin ..."). Any evidence the viewers of Fox News are not "uncommitted voters"? (As opposed to the regular viewers of CNN, whom we all know to all be pinko leftists, and die-hard Obamaniacs.
)
OK, back to polling 101.
Professional pollsters select respondents by random digit dial or other ways of collecting a random, nationally representative sample.
If one or the other demographic group nevertheless ends up being underrepresented (for example because the response rate from that group was lower), many pollsters then weight the results to correct for that.
It's an approximate science for sure, but the end goal is always to get a sample that is representative of the population (or rather, usually, of registered voters, sometimes likely voters) overall.
This is qualitatively different, of course, from the online and text message polls you see on websites or that you saw on TV. Why? Because the respondents are not randomly sampled from the population, but are self-selected. The respondents are all people who a) watched that particular TV station or read that website (and will therefore tend to skew politically to the extent in which the station or site has a political colour) and b) were motivated enough to spontaneously volunteer their opinion.
In short, these are samples that by definition will be far from representative. That's why on those online polls it says stuff like, "not a scientific survey". And why such online and text messaging polls aren't taken seriously by any pollster, researcher, journalist, or politician. Not to mention the ample opportunities to vote more than once. Or to appeal to your partisan, activist website's visitors to all go to the fox or msnbc site and help swing the vote there.
Now the insta-polls during or directly after the debate by professional pollsters are a subgenre with its own imperfections. Public opinion about the debate tends to shift in the course of the hours and days after it happened once a common wisdom emerges in the punditry and commentary, so the results of the insta-polls are very preliminary. Moreover, there's questions of methodology: samples are by necessity smaller in polls that are done within an hour or two, and will be less representative because the pollster could only reach those who were at home right at that time.
But they're still miles removed from those website polls. They're not based on a self-selected sample consisting of those most eager to proclaim their opinions; they're based on a random sample of people who watched the debate. That alone voids your equation of the Fox online poll with the CNN survey or the CBS poll.
That's not all though. The CBS poll was one specifically of voters who had identified themselves as uncommitted. That's different from a regular poll, and it's even more different from an online/text messaging poll on, say, the msnbc or foxnews website, which are likely to be dominated by partisans. The same goes for the focus group that CNN had take part in their running evaluation of the debate. Those were Ohioan voters who identified themselves as being uncommitted.
Now there may be some politicking going on, I dunno - Democrats or Republicans pretending to be uncommitted in order to get into the focus group so they can influence the results or something. But none of that will amount to much more than tinkering (I assume they even do some filtering). The contrast remains one between a poll that is at least roughly based on a representative group, in this case of uncommitted voters specifically, and an online or SMS poll which never even aims to be representative in any way.
What's probably important to emphasise here: This is not a Fox News vs CNN thing. Fox does regular, standard professional polling which is as reliable as your average polling, and CNN has its own online polls that are worth nothing. This is about being able to make the basic distinction between representative professional polls and bogus online polls.
)
