You hear it every day. The national media quotes “the latest polls”. But, who are the ones doing the polling? And, who are the ones polled? One would think they’d find pollsters around every corner in every town…right next to McDonald’s. So, for all the hub-bub over their stats, what’s the SD on national political polls?
Have you heard of Pew or Gallup? The media almost always says what poll or pollster they are talking about when they announce the results.
And you don’t encounter them all over the place because they only need to interview a very small proportion of the population to get very good figures.
Here’s alist of recent polls, with the organization that conducted them.
Here’s alist of recent polls, with the organization that conducted them.
Here’s the details on the sampleused by one of the polls.
There is another category of polls known as “push polls”. They are generally paid for by candidates or interest groups, and their purpose is to manipulate the results in the direction they want by the questions and the way they are worded. I ran into one of these once that I later learned was sponsored by The Club for Growth. There were question like “Do you support Scott Walker’s plan to save the state billions of dollars by decreasing the the unfair powers of public employee unions?”
And yes, the questions were really blatantly that partisan.
Virtually all news stories will say specifically who conducted the poll. The headline or lead-in may summarize the poll as a “national poll” or “poll of New Hampshire voters” or whatnot, but read on and it will virtually always say what company conducted it.
Mostly people with telephones. Traditionally landlines, but more cell phone users are being polled because younger people are cutting the cord – or never had a cord to begin with.
I ‘subscribe’ to a polling organisation in the UK called you.gov. At least once a week I get a questionnaire to complete online; usually commercial stuff like “have you ever heard of these companies; do you buy these products; and, of course, who will you vote for?” Naturally they know my age, political views and economic status.
I do get rewarded for my efforts and it adds up to around £100 per year and the chance of more in prize draws. Importantly, it does not generate any advertising spam. The results are frequently quoted in the media and seem generally accurate.
Agree. …
But only for legitimate news stories put out by legitimate news organizations. For outlets like Fox, which is not a legit news organization, push polls are about all they quote. Each one carefully anonymous as to which entity paid for the doctrinally correct results.
Not knowing which kinds of media outlets the OP listens to / reads, we can’t say what version of reality he’s getting.
It pains me to defend Fox News, but that is simply not true.
First of all, the polls conducted by Fox News are fairly well regarded – not the gold standard, but not obviously tampered with. Cite. Second, Fox News reporting, so far as I am aware, consistently cites the polls they report on. I simply do not know of an instance where Fox News has not reported who conducted the poll. Do you have an example of this?
Yup. Survey sampling (be it for political polling or market research) is based on the well-proven theory that a sufficiently large sample (in the absolute) will yield results that are likely to be statistically very close to the result you would get if you conducted a “census” (i.e., surveyed everyone).
In most cases, this means that a sample of somewhere between a few hundred to (at the most) a few thousand respondents is more than sufficient. (Pollsters who interview larger numbers of respondents tend to do so because they want to be able to have sufficiently large numbers of respondents in various subgroups: younger vs. older, male vs. female, etc.)
When poll / research results wind up being “wrong”, it’s nearly never because “they didn’t ask enough people”. More likely, the reasons for the results being off are because the sample population excluded an important subgroup (as Ravenman mentioned, traditional polling is done using respondents’ landline phones, but another big issue with political polling is identifying “likely voters”), or the wording of the question was confusing or misleading.
It is kind of amusing that cellphones are the problem today, since the great Literary Digest poll disaster of 1936 was due in part because many people did not have phones at all.
For those unaware, Literary Digest magazine predicted that Alf Landon would win with 57% of the vote. Not hardly. The magazine died two years later partly thanks to this.
I don’t think that cite is a sterling endorsement for Fox polls. It gives them an overall rating of C+. In comparison, ABC, NBC, and CNN got A-'s and CBS got a B.
Fox News also got a separate grade of B. There are two grades on the list. It appears that most of the recent polls conducted by Fox is with the two companies that received the “B” grade. (ETA: Apparently Fox News stopped using the company with the C+ grade in 2011.)
Y’all remember that Dewey was famously predicted to beat Truman in 1948. That was the result of a poll which went very wrong.
I read an explanation once, somewhere: It was an early attempt at doing a telephone poll. In those days, telephones were far from universal, and it was mostly the better-off who had them. So just by limiting the sample to people with telephones, the results were grossly biased.
If I learned (and remember) one thing from my Intro Statistics class, it’s that selecting a good sample is an elaborate and exquisite art and science.
ETA: Just noticed this:
Am either I or you or everyone conflating two stories here?
The famous “Dewey Defeats Truman” headline was in the 50s not the 30s, so there’s either two separate stories or one of them is badly wrong.
on edit – the headline was in '48, IIRC.
Okay, I missed that Fox News was listed twice. And you’re right; one of their polling services (Fox News/Opinion Dynamics Corp.) is rated C+ and the other (Fox News/Anderson Robbins Research/Shaw & Company Research) is rated B.
I note however that the survey analyzed thirty-four polls conducted by the first group and only four polls conducted by the second. Does that reflect the relative amount which they were used? If so, then Fox was using the C+ service more than eight times as much as the B service.
It seems Fox News stopped using the C+ service in 2011, and started using the other one.
Push polls are not done to manipulate news organizations for the most part. Mostly they are done to try to influence the people being polled and identify supporters. In your case the Club for Growth was trying to get you to vote for Walker and/or see if you were someone who might be willing to donate money or time to Walker.
Generally speaking, they can’t be used to identify supporters, because they don’t get your name. I been polled a number of times, including push polls, and they don’t ask, nor do I volunteer, my name or address.
No, they are made to influence media headlines by issuing press releases claiming their polling data supports their candidate’s position on X. And lazy reporters (there are many) don’t dig deep enough to learn exactly what questions were asked or what methodology was used to justify this claim.