January 7, 2015

Pew Research will call more cellphones in 2015

The number of Americans who rely only on a cellphone for their telephone service continues to grow. Fully 43% of U.S. adults live in a household with a cellphone and no landline phone, according to new government data for the first half of 2014. That’s up four percentage points from just six months earlier. According to an extrapolation by Pew Research Center, an estimated 46.5% of adults are cell-only today.

To keep pace with this rapid trend, the Pew Research Center will increase the percentage of respondents interviewed on cellphones in its typical national telephone surveys to 65%; 35% of interviews will be conducted by landline. Last year, we increased the ratio to 60% cellphone, with 40% conducted on landline. Back in 2008, when we first started routinely including cellphones in our phone surveys, just one-fourth (25%) of all interviews were done by cellphone.

Cell-Only Households Increase in NumberOur goal in making this change is to ensure that all adults are adequately represented in Pew Research Center surveys. Although cellphone-only households are very common today, there are sizeable demographic differences between people living in cell-only households and those with landlines.

For example, young adults, Hispanics, renters and the poor (as defined by the U.S. Census Bureau’s poverty thresholds) are all far more likely to be cell-only. To the extent that cell-only households are underrepresented in our samples, these groups are also underrepresented. In a typical Pew Research Center national telephone survey, a little more than half of respondents interviewed on a cellphone report that they have no landline telephone; consequently, the share of all respondents who are cell-only depends heavily on how many total cellphone interviews are conducted. By raising the share of all interviews conducted on a cell phone to 65%, we expect that about 37% of our total sample will be cell-only – still short of the target of 46.5%, but closer.

The question naturally arises: Why not interview everyone on a cellphone? In fact, at least one major national survey is going to do just that. The Surveys of Consumers, conducted by the University of Michigan, will begin calling only cellphones this month.

But we are not ready to make that change just yet, for at least two reasons. One is that there remains a small share of the public that is not reachable by cellphone. In the newly released data from the National Center for Health Statistics, 7% of adults live in households with a landline phone but no wireless phone. In addition, some people with landlines and cellphones may turn their cellphones on only to make calls or when they are expecting to be called. If these kinds of people are demographically different from those who are more easily reached on a cellphone, then the resulting sample will be less representative of the full population.

The other reason to continue calling landlines is that it remains more expensive and takes more time to call cellphones. Cellphone interviews cost roughly 50% more than landline interviews. Federal law prohibits the use of automatic dialing devices to call cellphones. Interviewers must manually dial cellphones, which takes more time than having the number automatically dialed.

Patterns of telephone use have been changing for more than a decade in the U.S., and the trend in the rise of cell-only households shows no signs of abating. Pew Research will continue to monitor the trend and modify our samples as circumstances warrant.

Topics: Research Methods, Technology Adoption

  1. Photo of Scott Keeter

    is senior survey advisor at Pew Research Center.

  2. Photo of Kyley McGeeney

    is a research methodologist at Pew Research Center.


  1. Donna Rosenberg2 years ago

    Which area code do the researchers call from? I don’t normally answer calls from unknown numbers but I would certainly want to answer this one. I often wonder who ARE these respondents because the results are just not what I hear on the grapevine … call me please.

  2. Wayne Bauer2 years ago

    When you call me, what will appear on my caller ID?

    1. Scott Keeter2 years ago

      Unfortunately there is no single answer to this question because caller ID systems differ throughout the country. Usually it is the business name of the survey organization we partner with to do our calling. Often it is just the phone number. Experiments have shown that particular messages on caller ID can help, but there is no way to ensure that the same message is seen by all households.

  3. Walter Johnson2 years ago

    This problem is actually much more complicated than most people realize because one and line phone number can represent an entire household or only one person, depending on whether or not the land line subscriber pays for distinctive ring service, while with few exceptions cell phones represent a just one person. Each in other words poses different technical challenges.

    Also I have only a cell phone and I was not willing to answer any surveys when I had limited minutes per month, so for such a call to be even answered would require knowing the organization ID of the caller and before answering that the caller would pay the per minute charge. I now have unlimited minutes so that is less of an issue but I know people with very limited call minutes during the day and into the evening. All this demands enough oversampling that the final set of respondents represents the population sampled and not some recognizable sub group only.

    My own expertise is surveying was way before cell phone service was affordable for individual personal use of the middle class but I don’t envy any of the researchers who have to deal with this issue today, but doing so is absolutely imperative. The article is right land line users and cell phone users represent different populations and opinions.

  4. Johanna e MacPherson2 years ago

    How many people called on landlines actually answer? I helped on a journalism school survey 21 years ago . And the only ones then who answered were retired old men and young mothers
    Of our total attempts less than 10% answered, and the complicated demographics we were trying to follow never once could be met
    I have never trusted surveys since. Except that I like the Pew. But phones? Who answers?

    1. Walter Johnson2 years ago

      You are right many people will not answer calls form unfamiliar numbers. I am like that for out of area code calls, although I allow a message to be left. For those calls with no caller id transmitted though they cannot even leave a message.

      However, in defense of surveys the same thing happens with Land Line Service when the customer gets caller ID service. Very few like to be hounded for donations or political contributions let alone telemarketers.

      I ran a government survey that had mandatory response (one of only 6 at the time), but we still could never get a 100% response rate. After 3 years I finally got a 99.9% response rate but that was after people were familiar with the survey and we followed up with three mailings to non-respondents before sending them a certified return receipt form mailing at considerable expense. Today just the postage for that last mailing costs something like $7 per survey and is done with all of he mandatory surveys.

      You will notice that many government monthly statistics are revised for two months following their initial published data. That is due primarily to late reports. A single major employer with a significant change in a month can be directly responsible for a significant revision made the following month for the prior month simply by being a day late in reporting. Running surveys is not fun and games but hard work and being cursed by people who don’t want to be bothered is an occupational hazard. With the survey I did we usually only got a 65% to 68% response to the first mailing and improvement with each subsequent mailing. If you call the same phone numbers at different times of day though you might still not get an answer. All phone done surveys have specific limitations, but they are still valuable and worth doing.

    2. Scott Keeter2 years ago

      Our response rates average between 8-10% for both landline and cellphone numbers. The percentage of people who answer the phone, however, is much higher. We actually reach someone at more than half of the numbers we call that we think are households. But many people decline to be interviewed. The combination of refusals and non-contacts results in our final response rates. And, unfortunately, the trend is downward, with fewer people cooperating than in past years.

      The good news is that we are still obtaining good cross-sections of the public. It’s true that older people are easier than younger people to reach. That’s partly a function of lifestyle. And certain kinds of people are more likely than others to cooperate. But, on the whole, the samples look a lot like the general public, and statistical weighting helps to make them match up even more precisely.

      You can see an overview of the trend in nonresponse, and more details about it, in this report we did in 2012: people-press.org/2012/05/15/asse…

  5. LEONARD MCCARTHY2 years ago

    Since I have my land line and cell phone number listed on the Governments “DO NOT CALL LIST”, does that mean that because you are conducting a survey, you are exempt from complying with the list?

    Also, since any use of a cell tel costs the user money for the air time whether they are the one making or receiving the call, has this been taken into consideration when you make your calls to cell tel phones?

    1. Scott Keeter2 years ago

      You are correct that legitimate survey research is exempt from the federal government’s Do Not Call List. But we tell our callers to always respect respondents’ wishes to be added to our internal Do Not Call list.
      When we first started calling cell phones, it was more common for cell phone users to be paying by the minute. For this reason, we decided to offer all cell phone respondents a $5 reimbursement to help offset any cost that might incur in talking with us. Only a minority of our cellphone respondents actually say they want the money, but we continue to think it’s important to offer it. One reason is that some people have pre-paid cell phones that don’t have contracts but are instead are loaded with a limited number of minutes. Because this type of phone may be used by very low-income people, we think it’s especially important to make sure those individuals don’t turn down our interview request because of concerns about the cost. It’s important that we represent people from all income levels.

      1. Walter Johnson2 years ago

        Scott Keeter, You made a very wise decision to offer the $5, but unless that information is included in your caller ID you may find you still don’t get a phone answer. All cell phone services I am aware of include caller ID and some cell phone apps automatically reject by no answer any unknown calls (no caller ID sent). If you want that policy change to matter you need to find a way to add it to your caller id transmission by your phone carrier.

  6. Cathy2 years ago

    Do you ever call to make an appointment for a follow-up call? I agree with Ray that unanticipated survey calls are usually not welcome.

    1. Scott Keeter2 years ago

      Our interviewers are instructed to offer respondents the opportunity to talk later at a more convenient time, by appointment, if they tell us that they are unable to talk with us when we call. We are always happy to do that.

  7. Richard2 years ago

    Call me!

    I have a lot to say and want to share!!!!!!!

    1. Walter Johnson2 years ago

      Since you were not asked your phone number and did not give it I would not expect a call. The survey respondents are randomly selected and everyone in the USA has a statistically small chance of being selected. If you want to raise administrative or other questions I suggest you call them during business hours.

  8. David Garrett2 years ago

    I often wonder who actually answers cold calls from an unknown number, regardless of which device – land line or cellphone – they own, and how representative their answers might be. The advent of caller ID was the end of my answering any call that just happened to come in; and probably the end of effective telephone interviews, too. Perhaps I am an outlier, but if you are not in my directory the odds of my answering your call are pretty close to 0.

  9. John Witmer2 years ago

    As a citizen without a cellphone, I’m glad you are continuing to poll our group.
    At the same time, I recognize that the vast number of people with only a cellphone need to be represented in getting a fair sample of opinion. Just don’t forget about us troglodites!

  10. rich bauter2 years ago

    I’m already signed up and consider it a major goal to read PEW each week. It is refreshing, as unbiased as you can find and always revealing some surprises. I really appreciate the reports.

    A question; how do you select your call numbers? Is it random, does it constantly change, how do you collect the demographics, etc.

    1. Scott Keeter2 years ago

      Hi — I took a shot at answering your first question in response to a comment from Laura below. Our cell phone numbers are “fresh” for each survey — that is, they represent a new random sample from the universe of available numbers. We collect demographic information by asking. At the end of a typical survey, we have a set of questions that asks about race, education, age, income, religion and the like.

  11. Ray Sullivan2 years ago

    I object to receiving ANY phone surveys by any group for any purpose. It’s an invasion of my privacy. The only way any survey is acceptable is via USPS mail or a general inquiry via email allowing the respondent to opt in or out. Email or snail mail responses are not representative of the population you say? That’s the price for assuring that one’s daily routine and demands on one’s time is not violated. Interrupting someone with a cellphone or landline call with a survey is an invasion that should not be tolerated. I rather enjoy doing surveys online at MY leisure but I would NOT respond to such a call if I were indisposed, at my supper table, in bed, out in my vehicle,etc etc. it is virtually impossible to “capture” your audience when it is convenient for the caller AND responder.

  12. Laura2 years ago

    I am curious how you get your cell phone number listings. Thanks!

    1. Scott Keeter2 years ago

      There is no master list of cell phone numbers out there, but there is a list of the chunks or “banks” of numbers assigned to wireless service. A bank is a set of contiguous numbers with a given area code and exchange — e.g., 202-419-XXXX. Even though we don’t know which numbers within a bank of, say, 100 contiguous numbers are actually assigned to people (or even that all numbers in a bank are assigned), we can randomly sample from those and call them. Using what is known about the assignment of numbers across the country, we can construct a valid national sample of cell phone numbers in which each assigned number has a known chance of being included. This meets the definition of a “random sample.” More details about the process can be found on our website at people-press.org/methodology/sam…