Back in the day when we all used land lines we would get calls from various groups or politicians wanting our opinions or wanting us to take part of polls. However in todays world with cell phones we dont have those.
Online. I subscribe to a poll taker who pays (peanuts) for my opinion on all kinds of things. Not just political but TV, advertising and consumer stuff as well.
A lot of people in my circle of acquaintances get at least occasional polling calls on their cell phones.
I used to work for a polling firm, and–while every survey was a little bit different–it was fairly common to ask the respondent, “Did I reach you on a cell phone or a land line?”
I still get called at home. I’ve gotten at least three calls (that I answered) this political season. I bet the polls are less accurate than they used to be, though.
And how do they make sure they’re calling someone in the right locality? My area code is Montana, but my voter registration is Ohio, and I’m far from the only one in that situation.
FiveThirtyEight.com addresses this topic regularly, as well as assessing the predictive efficacy of various polling organizations. Polls have never been particularly accurate, especially at the local level or on quick reaction to contemporary events where there isn’t time to get a true statistical sampling; at best, you are getting a sense of whether an issue or candidate has broad or narrow demographic support, and whether it is overwhelmingly favored for one side or up in the air, hence why Nate Silver uses weighted aggregate models along with other historical factors to try to correct for pollster bias and methodological error.
It can be a bit of a challenge since polling organizations cannot be assured that a particular phone number corresponds to a known geographic location; on the other hand, the amount of demographic information available today from internet marketing companies means that pollsters can actually target their polls, and presumably refine a credible result based on those factors rather than relying on “random” selection and self-reporting. This can also lead to bias, conscious or otherwise, by polling organizations with a partison bent or a deliberate agenda.
But even in the early days of telephone polling back in the late ‘Seventies and early ‘Eighties, polls missed certain demographics; in particular, the working poor who may not own a phone, and others who did not respond to poll requests. It is virtually impossible to set a statistically significant subset of every key demographic even on a national level or to get an accurate representation of public opinion in near real time in response to an event. A good public policy reporter with an experienced intuition about public opnion is probably about as good as the best poll, at least on local and statewide issues; unfortunately, with the demise of city newspapers, that is a dying breed.
I got a push poll on my cell 2 weeks ago. I was sick at home, so was a little slow in asking who they were with. Next time I’ll make sure it’s reputable.
Well, frankly, as Stranger On A Train notes, the polling companies have really struggled with the decline in land-line phones, and with ensuring that newer methodologies (cell phones, online surveys) are representative. (Non-political market research, which is my field, has suffered from the same challenges.) Getting a representative, accurate sample was certainly an issue even back when 95+% of households had a land line (and were willing to answer their phones), but it’s a much, much bigger issue now.
Gallup, a research and consulting company whose name has been synonymous with political polls for decades, used their political polls as a way to build recognition of their company’s name, so that they could sell other sorts of polling and surveying to clients. But, they’ve been stepping back from political polling for several years, because of the rapid increase in the difficulty of being able to run representative, accurate polls.
And, as 538 will frequently note, there are a lot of political polls out there, and quite a few of them suffer from poor quality (and some of them are likely to be intentionally biased).
Years ago in the malls they would have these people out doing polling or asking them to try various products. With the demise of malls that has gone away.
“Mall Intercept” testing was largely (if not entirely) market research studies (i.e., research among consumers for products and services), not political polling. It’s not that malls have entirely died (though certainly some have), but the bigger issue for researchers is that the cross-section of people who regularly go to shopping malls has become even less representative of the general population than it was before.
Even when shopping malls were more popular among consumers, most political pollsters didn’t use them for their polls, as telephone studies were still generally held to be the gold standard for representative sampling.
As mall intercept testing declined, the market research industry largely moved to online tests. For the relatively small proportion of market research studies which require face-to-face interaction between respondents and researchers (such as testing food products), there are “central location” facilities which conduct these studies (in my previous job, our company ran such a facility).
Almost every household had exactly one home phone and
People there would answer it and talk to you if you called it!
So getting a random sample by calling phone numbers was possible.
Both of those things have changed. Lots of people don’t have a landline any more, and lots of people won’t bother talking to people who call them. The polling companies can use models of which sorts of people have which phones, or respond to internet polls, or whatever, but none of that is as good as getting a real random sample.
An important component in understanding the accuracy of a poll when you know you’re not starting from a simple random sample is being able to compare the characteristics of the people answering the poll against the general population. We need Census data to be able to do that, so Americans, please do your part and complete your Census form in 2020.