Washington Post Staff Writer
Tuesday, January 13, 2009 11:00 AM
Washington Post staff writer David Brown was online Tuesday, Jan. 13 at 11 a.m. ET to discuss the challenges public health researchers are facing because many homes, that they use to survey, no longer have land lines.
The transcript follows.
David Brown: Good morning chatters! Welcome to the Live Online about the story in yesterday's Post about some of the difficulties in getting representative and reproducible data from population-based surveys, such as those run by the Centers for Disease Control and Prevention (CDC) that provide a picture of the behaviors and health status of Americans.
We are very fortunate today to have on the chat Stephen J. Blumberg, a senior staff scientist at the National Center for Health Statistics, which is part of CDC. He was the principal author of a report in December that described the great increase in existence of cellphone-only households, and the finding that people who use cellphones exclusively are a distinct population from the people who use them some of the time or not at all, and therefor must be sampled directly.
So let's go.
Cincinnati, Ohio: Is it not true that all cell phones can be equipped with a GPS-like device to give location of the phone? If not, is it possible to do so?
Sonny Boy in Cincy
Stephen Blumberg: Many cell phones do have the ability to broadcast GPS coordinates, and some specialized survey researchers have been thinking about how to use such information. For example, the Bureau of Transportation Statistics could use this information for surveys regarding people's movemments around metropolitan areas. I am not aware of any surveys, however, currently using GPS.
David Brown: I believe that every cellphone can be traced to a physical location by triangulation of signal from cellphone towers. That is different, however, from having a GPS capability built into the phone.
Portland, Ore.: Great topic. I sure you cover what research firms are doing to address this issue and how it impacts the validity of their research. Are they moving away from phone surveys or do they have a plan on how to get cell phone numbers?
Stephen Blumberg: Yes, most of the reputable polling firms have been experimenting with the inclusion of cell-phone-only households in their surveys. Most are currently using randomly generated cell phone numbers and conducting interviews on cell phones. Some have added address-based samples to their surveys, drawing addresses from the postal services delivery lists and mailing questionnaires or requests for phone numbers on which potential respondents can be reached.
David Brown: The polling firms' stock in trade is accuracy, so it does not pay for them to ignore this now unignorably large population of cellphone-only users.
Washington, DC 20024: It would seem that very delicate information is being transmitted over the cells with health records, etc. Can you transmit this private information on totally secure cell phones?
Sincerely, Steven Garrett, Chairman www.Genesis-Key.com
Stephen Blumberg: Unlike earlier technology, today's cell phones are digital and fairly difficult (though not impossible) to tap. Your home phone, with its lines extending from your home, is probably easier.
That said, survey researchers need to remain concerned about collecting sensitive information on cell phone surveys when people are in public locations. We all have had experience where we have overhead someone on a cell phone saying things that we didn't want to hear. Survey researchers need to be aware that their respondents often don't self-censor. Therefore, before sensitive questions are asked on cell-phone surveys, the interviewer should ask if the respondent is in a location where they can answer the questions without being overhead. Or, the interviewer should only ask YES/NO types of questions.
David Brown: Another issue that arises with interviews in which the respondent is on a cellphone is that in many cases the person is doing something else that requires attention. They are not more likely to shorten their answers or give "Do not know" respones, and they don't appear to be in a greater rush to get off the phone. While they are concerned with attention issues they don't know yet whether it is a problem. It may be that people who are willing to answer a survey on a cellphone while doing something else are accomplished multitaskers, and that people who can't divide their attention as easily or don't want to simply refuse to participate.
Princeton, N.J.: This is a problem that affects not only public health research, but all polling. Since Princeton is a hotbed of polling organizations, it is a frequent topic of conversation here. Most pollsters claim there is little difference between those who have landlines and those who don't and claim to have studies where they sent people to the home of those without landlines to check. When looked at closely, I find these studies unconvincing. I believe the Pew Foundation has worked on this problem.
As for the public health case, why not send survey forms to randomly selected physicians and ask them to give them out, or better yet, to have patients fill them out while waiting?
Finally, I can't resist saying this is yet another difficulty that could be more easily handled with a government run single payer system, as survey forms could be included in monthly or quarterly summaries of one's account.
Stephen Blumberg: In fact, there are significant differences between adults living with landlines and adults living in cell-phone-only households. Adults living in cell-phone-only households are 4 times as likely to be renting their home. They are also more likely to be 18-29 years of age, to be Hispanic, and to be living in or near poverty. Also, after controlling statistically for these differences, cell-phone-only adults are more likely to binge drink, to smoke, to engage in regular physical activity and to be uninsured (among others). On health characteristics, these two populations are different.
However, on political attitudes, after taking age into account, cell-only adults have been similar to landline adults. This was true in 2004, though there is some evidence that in 2008, cell-phone-only adults were slighty more favorable toward Obama than were landline adults of the same age.
David Brown: I think there would be huge methodological criticism of a health survey that relied on doctors handing out questionnaires to patients. First of all, one couldn't be assured of complete or random distribution (whichever was the instruction), and second one would only get people who go to the doctor, which is far from being representative of the whole population.
Pompano Beach, Fla.: Are there any studies linking cell phones to brain cancer, especially in young people?
Stephen Blumberg: This is a concern that is being investigated by other non-CDC researchers, but it is not something that the CDC has any information about at the current time.
Washington, D.C.: Are there methodologies we can adapt from developing countries where many people skipped landlines altogether and went straight for cell phones?
David Brown: I am not aware of any methodologies adapted from developing countries that have been imported here. However, I know that some software developers are writing programs and creating services that can be used on cellphones in developing countries, where many people have them and access to computers and landlines is limited. A Washington-based non-profit, DataDyne, is doing this. Some of their work is described here:
Vienna, Va.: 1. Couldn't a system be devised for general and survey use where the caller pays ALL charges for calls to cell phones? Of course the system would have to alert the CALLED party in advance that they will incur no charges for answering the call.
The other obvious question is: Why don't surveys that are about a certain communication medium use a different medium? If your survey is about telephone use, send out the survey by e-mail or even regular mail, broadcast or even print advertising?
Stephen Blumberg: In most European countries, the caller pays for all charges to cell phones. Perhaps as a result, the cell-phone-only rate in several European countries exceeds 50% and people there do not object to including their cell phone numbers in telephone directories.
Yes, if a survey was collecting information about people's use of communications technology, a landline telephone survey would not be ideal. Even a cell-phone survey would be problemmatic, and combining both a cell surveys and landline survey into a single estimate would be difficult.
Hamden, Conn.: As long as I am charged for my minutes, I do not want pollsters calling on my cell phone. It is the same issue as merchandisers sending me unsolicited faxes and using my fax toner and paper. We could use a system of "pre-paid" calls where the caller is charged for used minutes at both ends of the call.
David Brown: This point of view is the reason that most, although not all, surveys that include cellphone numbers offer some sort of incentive for the cellphone user to defray the user's expense. These can be in the form of checks sent to the respondent or, to avoid asking the respondent to provide his or her address, is in the form of a number one can put into a merchant's website to get a gift credit of, say, $10 at Amazon.
Austin, Tex.: Given all these limitations to survey research esp the BRFSS -- what are other alternatives that are (were) tested and prove(d) useful!!!
Stephen Blumberg: There are several alternatives that are available, though research is needed to determine which is most useful for each particular situation. The difficulty in any survey research endeavor is to reach a representative population. Randomly generated telephone numbers (when they include cell phones) and randomly selected addresses help do that. Alternatives such as surveys conducted in doctor's offices are more problemmatic. Approximately 15% of adults in the US do not have a usual place where they go for medical care, and nearly 10% have financial barriers to getting care. So surveys in doctor's offices and medical record check studies may not reach a representative sample of persons.
I think the future will see more multi-mode surveys, where survey researchers are willing to collect information by phone, by mail, by web, or by any other way that the respondent is willing to provide it. We are seeing a resurgence in such surveys now, where people are first contacted by mail, then by phone, and then in-person.
Of course, if money were no object, all surveys would be fielded in-person, door-to-door, but that is very expensive. Nevertheless, highly regarded surveys such as the National Health Interview Survey and the National Health Examination Survey are conducted that way.
Oakland, Calif.: I would like to know what information you have on immigrants use of cellphones compared to other populations. In many other countries, cellphones are an excellent substitute for unreliable or expensive landlines. I am interested in what information you might have on cells vs landlines for these populations who, while making their transition to this country, might defer use of important prevention services and thus appear in the system when they have developed more serious and preventable conditions. Deeper knowledge about greater penetration of cell users among Immigrants could help us to better predict population health status. There are also intervention strategies that directly reach out to cell users that might lend themselves to some health status polling strategies.
Stephen Blumberg: We do know that Hispanic adults are more likely than non-Hispanic Black adults and non-Hispanic White adults to be living in cell-only households. We also know that cell phone surveys reach a higher number of non-English and non-Spanish speakers than do landline surveys. But, to my knowledge, no surveys have looked at the use of cell phones by immigrants.
Lithuania: Hello, It would be interesting to hear your opinion about the impact of emissions from antennas, which GSM operators put on the roofs of apartment blocks, for the health of people living there. What is the minimum safe distance to live from such a antenna? Thank you. Kestutis
David Brown: A number of chatters have asked about the personal safety of cellphones (effect on the brain) and of transmissions from cellphone towers. Neither Dr. Blumberg nor I know the evidence for or against various theories well enough to address these issues. They are good questions, we just don't have the answers. Sorry.
Newark, Del.: Are you worried about stationary demographical information or health issues related to cell phone use? I think the latter will prove to be a greater issue as time goes by, and anyway, many people with cell phones still have land lines. I myself plan to give up the land line if I move and don't need it for Internet connection. It's all about the $$$$$.
Stephen Blumberg: Many people with cell phones do retain landlines. In fact, 14% of adults have both and tell us that they receive all or almost all calls on their cell phones. These cell-mostly adults may be retaining a landline because they need one for their computer, fax machine, alarm system, or for buzzing someone into their apartment building. However, I do expect that the current economic environment will result in more people considering a switch away from landline telephone service.
David Brown: I'm sure there is a huge generational thing going on in this regard. Many young people have a cellphone in their hand about half the day, and within easy reach the rest of the time. Older people tend to leave them on the kitchen counter or in a pocketbook.
Longs, S.C.: My comment is that I receive about three to four telemarketer calls a day on my cell phone. So there are ways to get cell phone numbers and conduct the health survey. Now maybe you can explain how cell phones are impacting public heath because I do not really understand what cell phones have to due with it. Is it federal regulated that you can not call cell phones, I read that you can not speed dial. I need more clarification please.
Stephen Blumberg: The Telephone Consumer Protection Act prohibits any call made without consent to a cellular telephone number IF an automatic dialing device is being used. Telemarketers rely on such predictive dialing devices to call many telephone numbers at a time for little cost. Therefore, in the past, they have generally avoided calling cell phones. There are no exclusions in the law for survey researchers. Therefore, to legally call cell phones, survey researchers (and telemarketers) must hand-dial each cell phone number, unless they already have a relationship with the called party.
David Brown: It is the hand-dialing, and the fact that about half the people reached on cellphones are children and ineligible to participate in most surveys, that makes them so much more expensive than landline surveys. It takes about 9 completed calls (a call that rings and is answered) to get one completed survey when the number is a cellphone number, compared to 5 for landline numbers.
Freising, Germany: I've read that many people in Third World or developing countries have cell phones but no land lines. I'm not sure if Cell phone use is expanding in Europe at the expense of land lines, but your article indicates that this is the case in the U.S.
I'd like to ask if you think that this trend is likely to continue, but perhaps the more pertinent question is, why would someone give up their land line for a cell phone, even though a cell phone is more expensive?
Stephen Blumberg: In the United States, landline telephone service can be more expensive than cell phone service. In addition, the installation of a landline telephone comes with expensive charges and often comes with a credit check. Cell phones provide more features (such voice mail, caller-ID, free long distance calls) and, with pay-as-you-go plans, allow for better budgeting of monthly expenses and allow purchasers to avoid credit checks.
Rockville, Md.: What plans does the National Center for Health Statistics have for including cell-phone users into their land-line only surveys, and at what point would land-line only surveys become non-representative of the general population?
Stephen Blumberg: Landline-only surveys remain representative of the population of older adults for nearly all topics. And estimates from landline-only surveys (with proper statistical weighting adjustments) have been shown to be comparable to estimates from in-person surveys for many topics (e.g., health status) even when looking at younger populations. However, for younger populations of adults, it is probably true that the landline-only samples are no longer representative.
The National Center for Health Statistics conducts very few telephone surveys. Most of its major surveys are conducted in-person because this mode produces much more reliable and valid estimates.
However, for the telephone surveys that it does conduct (such as the National Survey of Children with Special Health Care Needs, the National Survey of Children's Health, and the National Immunization Survey), NCHS has been experimenting with using address-based samples to supplement random-digit-dial landline samples. NCHS has also been experimenting with conducting surveys on cell phones to supplement the RDD landline samples. In 2010, I expect that one of these "new" sampling frames will be included as a regular part of all NCHS telephone surveys.
Albany, Ore.: I do understand the importance of surveys and how they can be used to help the population. There was one statement that was a little concerning. My apologies for sounding alarmist. The comment was 'For example, the Bureau of Transportation Statistics could use this (GPS) information for surveys regarding people's movements around metropolitan areas' There needs to be two questions. 1)How can this information be used to help. 2)How could this information be used to hurt. Asking the second question makes sure the help from the information is significant enough and that all measures have been put in place to ensure that information does not fall into the wrong hands. Again, sorry for sounding alarmist.
Stephen Blumberg: Of course, permission of the cell phone user, assurances of confidentiality, and strong Federal protections of those data would be necessary. The Confidential Information Protection and Statistical Efficiency Act (CIPSEA) provides such protections against disclosure (even in response to subpoena) if the data were collected by a federal statistical agency (such as the Bureau of Transportation Statistics).
Research Triangle Park, N.C.: Is there any technology or billing operations that will allow the research firm to take on all the call charges and have the cell-phone respondent not use their minutes for the survey (i.e. automatic reimbursement by research to mobile carrier)?
Stephen Blumberg: Not currently. Unfortunately, if such a technology were created, it could also be used by telemarketers, and they would no longer be bound by the federal Telephone Consumer Protection Act (TCPA). The TCPA's restrictions on the use of autodialers do not apply if the called party is not charged for the call.
Pittsburgh, Pa.: This is an interesting topic. I remember wondering, after seeing so many polls showing Obama and McCain in a dead heat but with election results skewing so differently, if a similar trend may have been affecting the polling results.
If +/- accurate data are needed, how can this be worked around? Is it back to the days of censusing on foot? Do cell-only households correspond with any other trends, such as frequent internet/email use?
Stephen Blumberg: Most major polling firms did include cell phone numbers in their pre-election polls last year. When data from those cell-phone numbers were included in their estimates, there was generally a 1-2 percentage point increase in support for Obama. Because Obama had a lead greater than that in most pre-election polls, the inclusion or exclusion of cell phone numbers did not strongly influence conclusions drawn from these polls. However, if the race had been closer, I am sure this would have been a bigger issue.
Remember that pre-election polls need to be conducted frequently in a short period of time. It is highly unlikely they will ever be conducted using door-to-door interviewing.
David Brown: The Pew Research Center included cellphone users in all its election surveys from August onward. Exit polls on election day found that 20 percent of voters were cellphone-only people, which is a higher percentage that what was found in the recent NCHS survey, which found 16 percent of adults used cellphones only. The Pew Research Center describes some of its methodology at:
David Brown: We're out of time, but I think we covered quite a bit of ground, thanks to the presence of Stephen Blumberg. I am very appreciative he was able to join us. Thanks to the chatters, too.
Editor's Note: washingtonpost.com moderators retain editorial control over Discussions and choose the most relevant questions for guests and hosts; guests and hosts can decline to answer questions. washingtonpost.com is not responsible for any content posted by third parties.