The widespread use of cellular phones has impacted both the SC BRFSS and CHAS. Previous studies have found differences in "cell phone only" compared to landline telephone populations in demographic, economic and health characteristics.
Cell phone only samples are more likely to be male, African American, Hispanic, under the age of 34, employed, of lower income and/or unmarried compared to landline only samples.1 Significant differences in health care access and behaviors also have been found. For example, "cell phone only" adults are more likely to be binge drinkers, smokers, people who engage in regular physical activities, have an unmet need for medical care due to cost and have used preventive health care services.2 However, demographic weighting adjustments greatly reduce these biases such that when data from landline telephone surveys are weighted to match population demographic characteristics, bias is similar to the margin of sampling error on the landline sample (less than 2 percentage points) for the majority of health indicators.3Although greater bias (1-5%) has been found for some estimates of health care and behavioral health indicators (e.g. binge drinking, smoking, financial barriers to medical care) to specific populations (i.e. young or low-income adults), bias can be attenuated to some extent through weighting adjustments.4BRFSS post-stratification adjustments include age, race, sex and ethnicity. The CDC is currently developing weights for the BRFSS data using raking methodology that will also allow for marital status and education to be included in the post-stratification weighting process. With the addition of education and marital status, these raked weights will yield a weighted sample that is more representative of the state's less educated or lower income population. Thus, the raked weights may further account for these differences between cell phone only and landline populations and aid in producing less biased estimates.
As more South Carolinians abandon landline phones and become "cell phone only," the State Center for Health Statistics recognizes the importance of incorporating these households into the survey framing scheme. In 2009, the SC BRFSS added a cell phone component to its landline survey. However, fiscal constraints make expanding surveys to include "cell-phone only" households difficult. Including a cell phone component substantially increases program expenditures given that a cell phone interview costs roughly two and a half times the cost of a landline interview. This is because it takes significantly more time (e.g. log on hours) and effort (e.g. more dialings) per interview to complete a cell phone interview compared to a landline interview.5 Therefore, additional funding dedicated to collecting a cell phone sample will be necessary for the SC BRFSS and SC CHAS to survey households that are serviced only by cell phones.
1Link MW, Battaglia MP, Frankel MR, Osborn L, Mokdad AH. Reaching the US cell phone generation: Comparison of cell phone survey results with an ongoing landline telephone survey. Public Opinion Quarterly 2007; 71(5): 814-839.
2Hu SS, Balluz L, Battaglia MP, Frankel MR. Expanding the BRFSS to a dual frame telephone survey: Comparison of risk
factors and health conditions. Presentation at the 27th Annual Conference of the Behavioral Risk Factor Surveillance
System (BRFSS), San Diego, CA, March 2010.
3Hembroff LA. Cellphone Supplementation to the MiBRFS: 2008 & 2009. Presentation at the 6th Annual BRFSS Training Workshop, Centers for Disease Control and Prevention, Atlanta, Georgia, October 2009.
4Blumberg SJ, Luke JV. Reevaluating the Need for Concern Regarding Noncoverage Bias in Landline Surveys, Am J Public
Health 2009; 99(10): 1806-1810.
5Cassell, J. South Carolina Cell Phone Pilot Study: Moving Beyond the Core Questionnaire. Presentation at the 6th Annual BRFSS Training Workshop, Centers for Disease Control and Prevention, Atlanta, Georgia, October 2009.