This presentation reviews how the addition of surveys with the cell-only population can impact the results of traditional landline surveys, and considers the costs and benefits of adding this hard to reach, but growing, population of telephone users. The Kentucky Health Issues Poll (KHIP) is a random digit dialing telephone health opinion survey of more than 1500 adults that has been conducted annually since 2008. A sample of more than 200 cell-only households was included on the 2009 and 2010 KHIP. The demographic characteristics of landline-only samples have been deficient when compared to Census estimates for Kentucky in representation of both younger and non-white respondents. In both 2009 and 2010, as compared to the unweighted landline sample, the unweighted cell-only sample was significantly more likely to be non-white and to be less than 35 years of age. Combining landline-only and cell-only surveys resulted in improved sample coverage, but did this improved coverage impact results? For example, the cell sample permitted the inclusion of considerably more uninsured Kentuckians than would have otherwise been possible with a landline-only sample.
1. Reflecting the Community: Demographic
Variations in Landline and Cell-Only
Households for a Statewide Telephone Survey
Sarah Walsh, MPH, CHES, Foundation for a Healthy Kentucky
Jennifer Chubinski, MS, The Health Foundation of Greater Cincinnati
Michelle Vargas, Institute for Policy Research, University of Cincinnati
Eric Rademacher, PhD, Institute for Policy Research, University of Cincinnati
2. Presenter Disclosures
The following personal financial
relationships with commercial interests
relevant to this presentation existed
during the past 12 months:
NO RELATIONSHIPS TO DISCLOSE.
3. Kentucky Health Issues Poll (KHIP)
• Questions vary by year, but include:
– Health status
– Socioeconomic variables
– Public opinion on health related topics
• KHIP gives policymakers and advocates
valuable data for keeping health on the public
agenda.
5. KHIP Methodology
• Conducted annually, since 2008
• Random Digit Dial Telephone survey (RDD)
– Cell phone sampling started 2009
• Data collection done by the Institute for Policy
Research (IPR) at the University of Cincinnati
• Analysis done by Foundation staff
6. KHIP sample over time
1,632 landline interviews
2008 January 16-February 11, 2008
1,464 landline; 205 cell-only interviews
2009 October 8-November 6, 2009
1,469 landline; 208 cell-only interviews
2010 December 3-22 and 27-28, 2010
≈ 1,200 landline; ≈ 300 cell any
2011 interviews
September 27-present
7. Cell-only Households, 2009-2010
Blumberg SJ, Luke JV, Ganesh N, et al. Wireless substitution: State-level estimates from the National Health Interview
Survey, January 2007-June 2010. National health statistics reports; no 39. Hyattsville, MD: National Center for Health
Statistics. 2011. Available at http://www.cdc.gov/nchs/nhis.html
8. Why Cell?
• KY high cell-only state
• Vulnerable populations and more likely cell
phone-only users important to our mission
– Low-income
– Minority
– Young adults
• More reliable sample of population
14. 2009 Results
Weighted Weighted
Landline Only Combined Sample
Currently Uninsured 28.0% 27.5%
Uninsured in Past Year 31.5% 31.7%
Below 100% FPL 27.1% 26.4%
Below 200% FPL 47.4% 47.3%
15. 2010 Results
Weighted Weighted
Landline Only Combined Sample
Currently Uninsured 21.3% 22.5%
Uninsured in Past Year 29.8% 31.6%
Below 100% FPL 23.6% 24.7%
Below 200% FPL 48.1% 49.1%
16. Results (weighted)
• Most differences lost in weighting procedure
– Weights based on 2008 American Community
Survey (ACS)
– Sample weighted on age, race, sex, education,
regional distribution, number of adults and phone
lines in the household.
• HOWEVER, the cell phone sample reduced the
magnitude of the weights
17. Lessons Learned- KHIP 2011
• Cell sampling changed from cell only to cell
mostly
• Larger percentage of cell phones
• Larger cost with cell sampling, but cost is
going down
– Previously 5x a landline response
– Currently 3x a landline response
18. Draw your own conclusions
• Raw data available in free statistical portal
– Online Analysis and Statistical Information System
(OASIS) www.oasisdataarchive.org
19. Draw your own conclusions
• Data available for mapping in free mapping
portal- HealthLandscape
www.healthlandscape.org
20. Questions?
Sarah Walsh Jennifer Chubinski
Senior Program Officer, Local Director of Community
Data for Local Action Initiative Research
(502) 326-2583 (513) 458-6608
swalsh@healthy-ky.org jchubinski@healthfoundation.org
Notas do Editor
APHA-mandated slide
Health status = BMI, current smoker, etc.Socioeconomic = age, race, sex, education, income, etc.Public opinion = support for smoke free laws, support for the ACA, identification of legislative priorities.
Something to the effect of this is the history of KHIP – I’m hoping we will have 2011 out of the field by then and can report on the number of completes.Point out that 2009 and 2010 are the only years we are discussing in this paper.
When we were first contemplating adding a cell sample for KHIP, studies estimated the number of cell-only households about 21%. I remember thinking that “more than 1 in 5” was such a huge proportion. More current data suggests that we’re now talking about “nearly one in three.” This map shows state-level estimates of cell-only households for our study period.The authors estimate that 31.5% of Kentucky adults, ages 18 and older, lived in “wireless-only” households between July 2009 and June 2010. This puts us in the top ten for the nation. (An additional 13.8% of adults lived in “wireless-mostly” households). Over the summer, early results from the National Health Interview Survey were released for the second half of 2010. While state-level figures are not available, Blumberg and Luke estimate that in the South, 31.1% of adults live in wireless-only households, suggesting that the rate of cell-only households continues to climb.
The sample is weighted to the 2008 American Community Survey population estimates based on gender, race, age, educational attainment and region of the state.Both 2009 and 2010 were weighted to the 2008 ACS. A similar pattern emerges for both years – the cell-only sample tends to be younger, is more likely to be non-white, is more likely to be male.While the overall impact on the pooled sample is modest, inclusion of the cell-only respondents moves the total sample slightly closer to the weighting target, thereby limiting the impact of the weights and their associated biases.
The sample is weighted to the 2008 American Community Survey population estimates based on gender, race, age, educational attainment and region of the state.Both 2009 and 2010 were weighted to the 2008 ACS. A similar pattern emerges for both years – the cell-only sample tends to be younger, is more likely to be non-white, is more likely to be male.While the overall impact on the pooled sample is modest, inclusion of the cell-only respondents moves the total sample slightly closer to the weighting target, thereby limiting the impact of the weights and their associated biases.
DO WE NEED TO PRESENT THE DEMOGRAPHICS OF KY – MOSTLY WHITE NOT BIG HISPANIC POPULATION?
Ultimately, the findings did not change significantly for other variables based on the inclusion of the cell sample in the weighted total.
Ultimately, the findings did not change significantly for other variables based on the inclusion of the cell sample in the weighted total.
DO WE NEED TO PRESENT THE DEMOGRAPHICS OF KY – MOSTLY WHITE NOT BIG HISPANIC POPULATION?