Do cell phone users differ by frame and should we care?

Do cell phone users differ by
frame and should we care?
Presenter
Michel Boudreaux
AAPOR, Phoenix, Arizona
May 13, 2011
Funded by a grant from the Robert Wood Johnson Foundation
Acknowledgments
• Funding support:
– MN Department of Human Services
– MN Department of Health
• Coauthors:
–
–
–
–
–
www.shadac.org
Kathleen Thiede Call
Donna McAlpine
Tim Beebe
Brett Fried
David Dutwin
2
National/Midwest Cell Phone Use
Adults Living in Wireless Only Households
30
26.6
Percent
25
24.9
20
15
10
14
US
Midwest
12.6
5
0
Jan-Jun 2007
www.shadac.org
Jan-Jun 2010
3
To screen or not to screen
• Lack of consensus in field on whether to:
1) Screen for cell phone only (CO) households
2) Include all but landline mostly
3) Interview the entire cell phone frame
• Excluding some cell users could introduce
error
• Including all users complicates weighting
due to overlapping probabilities of
selection
www.shadac.org
4
Research questions
• Are cases omitted when screening for CO
households different depending on which
frame they are captured in (LL vs. Cell)?
• Do these differences disappear when
controlling for post-stratification variables?
www.shadac.org
5
Data Sources
• 2009 MN Health Access Survey (MNHA)
– Ongoing survey of health insurance/access
– Accepted all cell phone completes
• 2010 MN Survey on Adult Substance Use
– Focus on prevalence of drug/alcohol use and
treatment patterns
– Accepted only CO and CM cases
www.shadac.org
6
Analysis
• T-test/pairwise comparisons of:
– Cell mostly sampled in LL versus Cell frame
• Regression models with telephone usage
as predictor, controlling for weighting
variables
• Full sample weights applied
www.shadac.org
7
Unadjusted Demographics, MNHA
Cell Mostly Users by Sample Frame
LL
90.4% 86.2%
2.3%
65+
www.shadac.org
Cell
87.1%
*
76.1%
26.3% 23.3%
*
5.3%
White
Home Owner
College
8
Unadjusted Health Insurance, MNHA
Cell Mostly Users by Sample Frame
LL
70.5%
Cell
*
57.0%
*
22.6%
15.%
5.6%
Public
www.shadac.org
Group
*
9.6%
Individual
8.8%
10.9%
Uninsured
9
Adjusted Health Insurance, MNHA
Adjusted Difference, LL vs Cell
0.15
0.00
0.04
-0.04
Uninsured
Public
Group
www.shadac.org
Individual
10
Unadjusted Demographics, MNSASU
Cell Mostly Users by Sample Frame
LL
Cell
89.6% 86.8%
45.5%
4.1%
4.1%
65+
www.shadac.org
*
37.7%
White
College
11
Unadjusted Health-Related Outcomes,
MNSASU
Cell Mostly Users by Sample Frame
LL
Cell
*
14.2%
9.6%
10.6%
9.0%
2.5%
Drug Use
www.shadac.org
Heavy Alcohol Use
2.4%
SMI
12
Adjusted Health-Related Outcomes,
MNSASU
Adjusted Mean Difference, LL vs. Cell
www.shadac.org
0.02
0.01
0.00
Drug Use
Heavy Alcohol Use
SMI
13
Discussion
• CM users from landline vs. cell frames do
differ on health outcomes
• However, post-stratification weighting
appears to correct for these differences
• This suggests that we can safely screen
for CO
• However, cost and variance remains an
issue
www.shadac.org
14
Contact information
• Michel Boudreaux
• State Health Access Data Assistance
Center (SHADAC)
– [email protected]
www.shadac.org
©2002-2009 Regents of the University of Minnesota. All rights reserved.
The University of Minnesota is an Equal Opportunity Employer
15