This was in today's Washington Post. I thought that calls to cellphone users were limited to those who had agreed to be called, This would suggest that polls involving cellphone users are likely to be biased by the views of persons choosing to opt in and be polled, perhaps to promote their views, while landlike users are much closer to a random sample.
Wes Leatherock snipped-for-privacy@aol.com snipped-for-privacy@yahoo.com (mailto: snipped-for-privacy@yahoo.com) The politics of cell phones
By E.J. Dionne
The Pew Research Center has performed _an important piece of analysis_
formatting link
that ought to shape the way we look at polls over the next two years. And its findings have real political impact, so please read on even if you're not obsessed with public opinion surveys.
There has been much debate in recent years over whether telephone surveys conducted only on landlines have produced distorted findings because so many Americans, particularly the young, now use only cell phones only. Pew has shown that this fear is justified. And it notes that this problem is growing. In essence, landline-only polls give the Republicans a measurable advantage.
Here's the key finding: "Across three Pew Research polls conducted in fall
2010 -- conducted among 5,216 likely voters, including 1,712 interviewed on cell phones -- the GOP held a lead that was on average 5.1 percentage points larger in the landline sample than in the combined landline and cell phone sample."
By E.J. Dionne | November 23, 2010; 8:28 AM ET Categories: _Dionne_
*Robocalls* to cellphones are limited in that way. If you have an actual human dialing the randomly-chosen numbers -- even if they work in some telemarketing boiler-room -- then it's OK.
There have always been issues with getting a representative sample by any means of polling -- whether automated telephone or in-person interviews or anything in between -- because many people (most people?) will refuse to participate. The polling industry has historically assumed that (non-)participation is effectively uncorrelated with the variables the survey is investigating; this is usually handled by weighting the survey results to match an objective population model (usually taken from census data and, for political polling, historical election results). Pew's data show conclusively that cell-phone-only households are not representative of the population as a whole[1] and therefore cannot be modeled in this way. Of course, there is still a large nonparticipation rate even when cell numbers are included in the sample, so we still don't know whether the nonparticipating cell-phone-only households have the same bias as participating cell-phone-only households.
-GAWollman
[1] Specifically, cell-phone-only households are younger, more Democratic, and less affluent than landline households, if I remember the Pew results correctly. You can go to Pew's Web site and find the complete survey protocol.
No, they're not subject to the TCPA, but one problem is that since no cell numbers are listed in the (fast disappearing) white pages, and people often keep their cell number when they move, it is hard to find a reasonable sample of cell numbers.
The other problem, of course, is that people are not particularly inclined to use up their airtime minuts being polled.
But what about predictive dialers? The dialer targets numbers in a sequence or geography, dials every line number attempting to make a connection (hence all those hangups), then assigns them to outbound calling positions. Do paid pollsters actually dial numbers themselves without use of predictive dialers? I wonder. Volunteer pollsters using temporary phone banks or even donated telephone lines probably would dial themselves.
. . . assumptions that are not safe to make, unless you have a very well conceived poll on nonparticipation and actually survey those who wouldn't otherwise participate, which is a contradiction.
We've changed the definition of household, well, not for common sense purposes but for combinations of people who are readily surveyed. Because it was painful to get Ma Bell to combine residential bills for multiple households, this billing concept was useful for polling purposes. With cell phones, what is a household? If pollsters have access to cell phones, then they have access to something derived from billing records. Not every user on a cell phone family plan resides in the same house. I know a number of small businesses that provide cell phones for employees using family plans.
I'm not sure it matters, for no concept of a household means that polling one individual is representative of other members of the same household. Furthermore, there's some relationship between an adult child's opinion and the demographics of his parents but it's not necessarily the strongest influence. Depending on what age you reach the young adult, he'll have completed a different amount of college than his parents and at a young age won't actually know if he'll complete graduate school. Polls trying to establish basic demography ask about level of college completed.
A cell phone user, knowing he's paying for air time, may be less willing to participate in the poll than a land line user. I'm leery of any conclusions of cell phone only users, because we still don't know what a representative sample is or who those reached represent with those ubiquitous quick and dirty polls reported in the newspapers.
In my state, one pollster was consistently reporting that voters favored statewide Republican candidates early in the cycle. I asked a friend who runs various campaigns who knew who the guy was. He explained that he's well known for surveying in rural areas and smaller cities and not known for surveying in major metropolitan areas. He said he's a good pollster, however, his results can be regarded as good only for the areas surveyed and cannot be extrapolated to the entire state.
Candidates know these things. They attempt to publicize somewhat weighted polls to improve their own fundraising ability. Newspapers and television nightly news often do a bad job on reporting the known strengths and weaknesses of pollsters.
Suppose you surveyed a representative group of people. You compare in person results to telephone results. In person, you ask people extensive questions about their telephone services and usage and use this data to weight the results of those reached by various forms of telephone. It's still likely that "corrected" telephone survey results are different from in person results. Why? Survey takers themselves may prefer one method over the other.
Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.