SlideShare a Scribd company logo
1 of 71
Download to read offline
Random digit dialing cell phone surveys and 
  surveillance systems: data quality, data 
       ill        t     d t      lit d t
      weighting strategies, and bias


             Cristine D. Delnevo, PhD, MPH & Daniel A. Gundersen, MA,
                           UMDNJ‐ School of Public Health
                 Randal S. ZuWallack, MS & Frederica R. Conrey, PhD
                 Randal S ZuWallack MS & Frederica R Conrey PhD
                               ICF Macro International

                      Presented at 137th Annual Meeting & Exposition
                      P     t d t 137th A      l M ti & E      iti
                                      Philadelphia, PA
                                    November 7‐11, 2009


Work supported in part by the National Cancer Institute (R21CA129474 ) and a contract from the New 
      Jersey Department of Health and Senior Services, through funding from the Cigarette Tax
Wireless substitution (US), 2005 2008
  Wireless substitution (US), 2005–2008




Source: Blumberg & Luke  Wireless substitution: Early release of estimates from the National Health Interview Survey, July‐
December 2008. National Center for Health Statistics. May 2009.  
Demographic differences
                      Demographic differences
    •    Gender: men more likely than 
                                                               Wireless substitution by age (and time)
                                                               Wireless substitution by age (and time)
         women to be wireless only
                    b i l           l
    •    SES: Adults living in (30.9%) and 
         near poverty (23.8%) more likely 
         than higher income adults 
          h hi h i              d l
         (16.0%) to be wireless only
    •    Region: Wireless substitution 
         highest in South (21.3%) and 
         hi h t i S th (21 3%) d
         Midwest (20.8%) vs. Northeast 
         (11.4%) or West (17.2 %)
    •    Race/Ethnicity: Wireless 
         R /Eth i it Wi l
         substitution highest  among 
         black (21.4%) and Hispanic 
         (25.0%) adults vs. Non‐Hispanic 
         (25 0%) adults vs Non Hispanic
         white adults (16.6%)  
Source: Blumberg & Luke  Wireless substitution: Early release of estimates from the National Health Interview Survey, July‐
December 2008. National Center for Health Statistics. May 2009.  
Biased health estimates?
        Biased health estimates?
• Potential for biased health estimates due to sample
  Potential for biased health estimates due to sample 
  under‐coverage remains a real, growing threat to 
  RDD health surveys
• Cell‐phone only also differs with respect to health 
  behaviors and the validity of some health estimates 
  based on traditional RDD surveys are increasingly 
  questionable
Health estimates by phone status
   Health estimates by phone status
                                          Has a landline Wireless-only
                                                                     y            No telephone
                                                                                         p
                                           telephone

    NHIS July – December 2007
5+ alcoholic drinks in 1 day                    17.7                37.3               27.1
Current smoker                                  18.0                30.6               38.6
Uninsured                                       13.7
                                                13 7                28.7
                                                                    28 7               44.1
                                                                                       44 1
Has a usual place for care                      87.5                68.0               61.8
Flu vaccination                                 32.7                16.6               20.9
Ever tested for HIV                             34.7                47.6               45.8


Source: Blumberg SJ, Luke JV. Coverage bias in traditional telephone surveys of low‐income and young 
adults. Public Opin Q. 2007;71:734–749
Cigarette smoking among young adults, 
       2003‐2005, NHIS & BRFSS
       2003 2005 NHIS & BRFSS




Source: Delnevo, Gundersen & Hagman (2008) Declining prevalence of alcohol and smoking estimates 
among young adults nationally: artifacts of sample under‐coverage?  Am J of Epidemiology
New challenges: Wireless Mostly?
                                                                      Telephone Status, NHIS July‐
                                                                            December 2008

      • The percentage of adults 
        living in wireless‐mostly 
        households has been 
        increasing
      • Who are they?
        Who are they? 
        (demographic & health 
        behaviors?)
      • Will they respond to 
        landline surveys? 


Source: Blumberg & Luke  Wireless substitution: Early release of estimates from the National Health Interview Survey, July‐
December 2008. National Center for Health Statistics. May 2009.  
Reactions?
• Starting in 2009, the Center 
  for Disease Control and 
  f Di         C      l d
  Prevention (CDC) is 
  requiring states to 
  incorporate cell phone 
  incorporate cell phone
  interviews in their regular 
  BRFSS sample
• Yet there is no widely 
                        y
  accepted methods of 
  evaluating data quality or 
  data weighting, particularly 
  for state and local area 
  for state and local area
  surveys. 
• AAPOR Cell Phone Task 
  Force report
  Force report
This special session:
         This special session:
• Analysis of data quality of cell phone
  Analysis of data quality of cell phone 
  surveys
• Demonstrate weighting procedures for 
  merging cell phone samples with landline 
  samples
• An assessment of bias in landline only
  An assessment of bias in landline only 
  surveys, and 
Assessing Data Quality of
Cell Phone Random Digit
                      g
      Dial Surveys
      Frederica R Conrey
       Randy Zuwallack
Data Quality

• Q lit of Responses
  Quality f R
• Quality of Sample
2008 New Jersey Adult Tobacco
              Survey
• R d
  Random Di it Di l L d (RR 19%) and
           Digit Dial Land (RR=19%) d
  Cell Phone (RR=16%)
• Short version
  – 49 Questions (min=37; max=70)
  – 534 Cell completes
  – 468 Landline completes
Survival Analysis

• Diff
  Different people get diff
          t     l    t different surveys
                               t
  because of skip patterns
• Survival analysis
  – Measures the impact of survey mode on non-
    response
  – Controls for differences in survey length
Quality of Responses
Item Non-Response
               Failure Rates by Phone Mode
                     Mean
                     M              Median
                                    M di     Std
  Cell               3.3%            2.2%    7.4%
Landline             3.4%            2.0%    5.5%




         Survival predicted by p
                  p          y phone mode:
         Hazard=1.00, p>.95
Open Ended Response

                           Total Responses p Open
                                    p      per p
Open end reports of the
O       d       t f th     End by Phone Mode
  events of 2 recalled                  Responses
  commercials                             / OE    DK / OE

  – 251 respondents were       Cell        .52      .49
    asked at least one       Landline      .48      .55
    open ended question         P          .56      .33
Quality of Sample
Unit Non-Response

Data
D t quality i th t
       lit is threatened if
                       d
  – Response rates are low AND
  – The people who DO NOT respond are
    different from those who DO.
If cell phone respondents are less likely to
   respond, then there is non-response bias.
Survival Analysis

• C ll and l dli respondents may get
  Cell d landline          d t           t
  different surveys
• Response rates alone don’t tell the story
• Survival analysis tells whether cell
                y
  respondents are more likely to break off
  g
  given the same survey length
                         y   g
Survival Model

                      25%
Sample Breaking Off




                      20%
              g




                      15%

                      10%                                         Cell
     e




                                                                  Landline
                      5%

                      0%
                            0   20         40           60   80
                                     Survey Questions
                                                              p<.001
                                                                 001
What does a difference in survey
          survival mean?
• C ll respondents quit sooner th l dli
  Cell       d t      it       than landline
  respondents.
• The sample under-represents cell phones
• The longer the survey, the worse the
          g            y
  nonresponse bias
• The solution?
  – Careful weighting
  – Short surveys
In a population study of tobacco use
              behavior…
Minimal difference
Mi i l diff                  Substantial difference
                             S b t ti l diff
  between cell and             between cell and
  landline in response         landline in response
  quality                      rate
• No difference emerged in   • Sample quality may be
  item-nonresponse             threatened if cell phone
• No difference emerged in     surveys are too long or
  richness of open end
   i h      f        d         weighted incorrectly
  responses.
Weighting Cell Phone Surveys 
Weighting Cell Phone Surveys



          Randal Zuwallack
         Frederica R Conrey
         Frederica R Conrey
Thanks

• Cris Delnevo
  Cris Delnevo
• Dan Gundersen
• NCI (R21CA129474), New Jersey Dept of 
   C ( 2 C 29       )                 f
  Health and Senior Services
Dual Frame



                          D




A.  Adults in landline households with no cell phone,
A Adults in landline households with no cell phone
B.  Adults in landline households with a cell phone, and
C. Adults in non‐landline households with a cell phone (cell only).
Dual Frame



                         D




Common designs:
Common designs:
  Dual frame w/ no overlap: Landline (A+B) + Cell (C)
  Dual frame w/ overlap: Landline (A+B) + Cell (B+C)

Uncommon design:
   Dual frame w/ no overlap: Landline (A) + Cell (B+C)
   Dual frame w/ no overlap: Landline (A) + Cell (B+C)
Weighting Challenges

• Challenge 1: How do we put the dual frames
  Challenge 1: How do we put the dual frames 
  together?

• Challenge 2:  Differential Nonresponse
Challenge 1

• How do we put the dual frames together?
  How do we put the dual frames together?
• No overlap
  – Estimate of cell‐only population size?
    Estimate of cell only population size?
     • Internal estimate
     • External estimate: NHIS (Blumberg et al.)
• With overlap
  – Must determine group membership
  – Adjust for multiple selection probabilities
  – Estimate of phone group population sizes?
Cell Survey

• “In addition to your cell phone is there at
   In addition to your cell phone, is there at 
  least one telephone inside your home that is 
  currently working and is not a cell phone?  Do 
  currently working and is not a cell phone? Do
  not include telephones only used for business 
  or telephones only used for computers or fax 
  or telephones only used for computers or fax
  machines.” 
  – ‘yes’ = dual user, while those who responded 
     yes = dual user while those who responded
  – ‘no’ = cell‐only
Landline Survey

• “In addition to your residential landline
   In addition to your residential landline 
  telephone, do you also use one or more cell 
  phone numbers?
  phone numbers?” 
  – ‘yes’ = dual user
  – ‘no’ = landline only.
     no = landline only
Example 1‐‐Colorado

• Combine with BRFSS
  Combine with BRFSS
• Group membership 
  –KKnown for cell
          f     ll
  – Unknown for landline
• Limited to dual frame w/ no overlap
  – Used 15% (NHIS state estimates) for merging 
    landline and cell
  – Poststratified dual sample to age and sex.
Example 2

• You are midway through a landline survey and
  You are midway through a landline survey and 
  want to add cell phones.  You don’t know who 
  has a cell phone and who doesn t. What are 
  has a cell phone and who doesn’t What are
  your options?
  1) Add cell only
  1) Add cell only
  2) Add cell and dual‐users 
Challenge 2

• Differential Nonresponse
  Differential Nonresponse
• Cell‐only overrepresented when conducting 
  cell phone surveys. 
  cell phone surveys
  – Contact rate
  –CCooperation rate
             i
• Those who rely more on their cell phone will 
  be easier to reach.
Telephone Reliance
                         C
L
                         e
a
                         l
n
                         l
d
l
i                        P
n                        h
e                        o
                         n
                         e
Landline Frame
                               C
L
                               e
a
                               l
n
                               l
d
l
i
         Landline households   P
n                              h
e                              o
                               n
                               e




    Landline sample
Cell Phone Frame
                                      C
L
                                      e
a
                                      l
n
                                      l
d
l
i
           Cell Phone Users           P
n                                     h
e                                     o
                                      n
                                      e




                        Cell sample
                        Cell sample
Dual Frames

 Landline sample             Landline sample




    Cell sample                 Cell sample



Dual Frame sample           Dual Frame sample


      Ideal                      Realistic
Our Goal

               Rebalance 
               R b l
                 on cell 
                reliance



Rebalance 
on landline 
 reliance
Measuring Telephone Reliance

• Cell only landline only
  Cell only, landline only
• Classify Dual users
  – “Of ll th t l h
    “Of all the telephone calls that you receive, are…”
                            ll th t          i        ”
     • All or almost all calls received on a cell phone? (cell‐
       mostly)
     • Some received on a cell phone and some on a regular 
       landline phone? (true‐dual)
     • Very few or none received on a cell phone? (landline‐
       mostly)
Telephone Reliance
                                                 C
L
                                                 e
a
                                                 l
n
d   Landline   Landline   True    Cell    Cell   l
l
i
      Only      Mostly    Dual   Mostly   Only   P
n      (0)       (1)       (2)    (3)      (4)   h
e                                                o
                                                 n
                                                 e
Response Propensity

• Adjust for differential nonresponse by
  Adjust for differential nonresponse by 
  benchmarking against NHIS
• Logistic regression model
  Logistic regression model
  – Dependent: Survey type
     •1 b
       1 = observe cell user in national cell phone survey
                     ll      i    ti   l ll h
     • 0 = observe cell user in NHIS
  – Independent: Cell phone reliance (1 4) age race
    Independent: Cell phone reliance (1‐4), age, race
Data sources

                  National cell sample
                  National cell sample   NHIS
                        n=500

Landline mostly           8%             23%

True dual                27%             42%

Cell mostly              23%             17%

Cell only                42%             18%

Cell users               100%            100%
Applying the model

• Applied to same data—poststatification
  Applied to same data poststatification
• Applied to independent data
  –A
   Assumption: national cell sample measures the 
            ti     ti    l ll       l          th
   odds ratio for observing a cell‐only respondent in 
   a cell sample relative to a dual user.
   a cell sample relative to a dual‐user


• S
  State, local surveys 
         l l
Applying the model

                                  Colorado
                                Cell Sample
                       w/o NR adj             w/ NR adj


Landline mostly (1)       9%                    26%
True dual (2)             26%                   37%
Cell mostly (3)
C ll    tl (3)            19%                   15%
Cell only (4)             46%                   21%
Total cell sample        100%                  100%
Applying the model

• Assume landline only is 20% (we don’t know)
  Assume landline only is 20% (we don t know)
                         CO

     Landline only      20%

     Landline mostly    21%

     True dual          30%

     Cell mostly        12%
                                         NHIS 
                                         NHIS
     Cell only          17%              state 
                                       estimate 
     Total population
           p p          100%            = 15%
City Sample

                   Landline     Cell Sample         Combined 
                    Sample                           Samples

                              w/o NR 
                              w/o NR    w/ NR 
                                        w/ NR    w/o NR 
                                                 w/o NR    w/ NR 
                                                           w/ NR
                               adj       adj      adj       adj

Cell‐only             ‐        43.5     18.9      35.5     13.4
Cell‐mostly         12.4       25.1     19.8      13.8     12.2
True Dual           30.1       23.6     35.5      18.7     25.1
Landline mostly
Landline‐mostly     19.1       7.9      25.8      9.9      17.2
Landline‐only       38.4        ‐         ‐       23.9     32.1
Conclusions

• Dual‐frame
  Dual frame
  – There are ways to combine the data, even when 
    we don t have a full picture of group membership.
    we don’t have a full picture of group membership
• Differential nonresponse
  – Response propensity model rebalances the cell
    Response propensity model rebalances the cell 
    sample based on cell reliance. 
  – Can be applied at state and local levels when no
    Can be applied at state and local levels when no 
    benchmarks exist.
  – Next steps: explore a response propensity model
    Next steps: explore a response propensity model 
    for landline.
Thank you

Randal.Zuwallack@macrointernational.com
Randal Zuwallack@macrointernational com
Frederica.Conrey@macrointernational.com
Examining the bias in landline only 
            g                        y
 surveys: How does the cell phone only 
   population differ from the landline 
   population differ from the landline
population on health indicators, and are 
estimates from landline surveys biased?

       Daniel A. Gundersen, MA, UMDNJ‐SPH
    Cristine D. Delnevo, PhD, MPH, UMDNJ‐SPH
         Randy S. ZuWallack, MS, ICF Macro
Cell Phone Substitution and RDD surveys
 Cell Phone Substitution and RDD surveys
• RDD surveys (e.g. BRFSS) have traditionally only sampled 
  household telephones (i.e. landlines)
• Up until early 2000s, rate of cell phone only households 
  was small
• From mid 2000s, rate of substitution has grown 
  substantially
   – 6 7% of adults in 2005 to 18 4% in 2008 nationally1
     6.7% of adults in 2005 to 18.4% in 2008 nationally
• Higher among certain demographic groups1
   – Young adults
   – Hispanics and Blacks
   – Poor and near poor
What is bias due to coverage error in the 
              sampling frame?
                   li f      ?
• Non‐covered population is different from
  Non covered population is different from 
  covered population on some variable of 
  interest


  – If proportion of non‐covered (    ) is small, bias will 
    be small
  – If difference between the covered and 
                                      ,
    noncovered                is small, bias will be small
Previous Research
                Previous Research
• Data from Jan 2004‐June 2005 NHIS found2
   – Greater than 1 percentage point bias in binge drinking, smoking 
     prevalence, usual place for medical care, receiving influenza 
     vaccine
• Data from 2007 NHIS found3
  Data from 2007 NHIS found
   – Bias increased slightly for past year binge drinking and receiving 
     influenza vaccine
   – These biases were larger among young adults and low income
     These biases were larger among young adults and low income 
     persons
• Data from 2001‐2005 BRFSS on 18‐24 year olds found4
   – Prevalence of binge drinking, heavy drinking, and cigarette 
     smoking declined during 2003‐2005; coincided with large 
     increase in wireless substitution among young adults
   – NHIS and NSDUH did not observe similar declines during this 
     period
Our Study:
                   Our Study:
• Objective:
  Objective: 
  – Assess the presence of bias in landline RDD due to 
    exclusion of cell phone only  on select health 
    exclusion of cell phone only on select health
    indicators
• Data Source and Instrument:
  Data Source and Instrument:
  – Cell phone RDD of adults in Colorado (n=501)
     • May to September 2008
       May to September 2008
     • Instrument was shortened version of BRFSS
  – BRFSS from same data collection period (n=4,527)
    BRFSS from same data collection period (n 4,527)
Methodology
• Cell Phone sample:
  Ce     o e sa p e:
  – Design weights account for probability of selection
• BRFSS
  – Standard BRFSS design weight accounts for strata, 
    number of landlines and adults in the household
  – Postratified b
            f d by age(7)*sex*race
                      ( )* *
• Merged data
  –D i
    Design weights scaled to represent share of 
               i ht    l dt          t h      f
    population by phone status
  – Postratified by age(7)*sex*race
                  y g ( )
Statistical Analyses
               Statistical Analyses
• Comparisons of BRFSS landline and Cell Only based on design 
  weights
• Comparison of BRFSS landline and merged data are 
  postratified to demographic makeup of CO




• We assume the merged data to be unbiased (i.e. no coverage 
  error due to cell only exclusion)
  error due to cell only exclusion)
• All analyses conducted in STATA v.10.1 to account for complex 
  sampling design
Table. CO BRFSS vs. CO Cell Only, May‐September 2008 (n=5,028)
                               y,   y p              (   ,   )


                               BRFSS landline    Cell only
                                                         y     Difference
Health Indicator                 % (95%CI)      % (95%CI)      % (95%CI)

Smoking*
       *                        15.29 (±1.14) 28.14 (±4.46) ‐12.85 (±4.60)
                                      (     )       (     )        (     )

Ever had HIV test*              36.64 (±1.85) 52.51 (±5.09) ‐15.87 (±5.41)

Having health insurance*        88.36 (±1.11) 72.46 (±4.38)   15.9 (±4.51)

Having primary care provider* 85.91 (±1.16) 60.39 (±4.85) 25.52 (±5.00)

Not affording care due to cost 12.27 (±1.08) 20.42 (±3.94)
Not affording care due to cost* 12 27 (±1 08) 20 42 (±3 94)   ‐8.15 (±4.08)
                                                              ‐8 15 (±4 08)
*p<.05; data weighted to correct for sampling design
Figure 1. Bias of Smoking Prevalence, CO BRFSS (n=4,527) and CO Cell Only 
 RDD (n=501) May‐September 2008
 RDD (n 501) May September 2008
 10


7.5
75


  5


2.5


  0


‐2.5


 ‐5


‐7.5


‐10
Figure 2. Relative Bias of Smoking Prevalence, CO BRFSS (n=4,527) and CO Cell 
Only RDD (n=501) May‐September 2008
Only RDD (n 501) May September 2008

     50%

     40%

     30%

     20%

     10%

      0%

    ‐10%

    ‐20%

    ‐30%

    ‐40%

    ‐50%
     50%
Figure 3. Bias of Ever had an HIV test, CO BRFSS (n=4,527) and CO Cell Only 
 RDD (n=501) May‐September 2008
 RDD (n 501) May September 2008

 10

7.5

  5

2.5

  0

‐2.5

 ‐5

‐7.5

‐10
Figure 4. Relative Bias of Ever had an HIV test, CO BRFSS (n=4,527) and CO 
Cell Only RDD (n=501) May‐September 2008
Cell Only RDD (n 501) May September 2008


 50%

 40%

 30%

 20%

 10%

  0%

‐10%

‐20%

‐30%

‐40%

‐50%
 50%
Figure 5. Bias ‐ Having health insurance, CO BRFSS (n=4,527) and CO Cell Only 
   RDD (n=501) May‐September 2008
   RDD (n 501) May September 2008


 10

7.5

  5

2.5

  0

‐2.5

 ‐5

‐7.5
‐7 5

‐10
Figure 6. Relative Bias Having Health Insurance, CO BRFSS (n=4,527) and CO 
Cell Only RDD (n=501) May‐September 2008
Cell Only RDD (n 501) May September 2008


 50%

 40%

 30%

 20%

 10%

  0%

‐10%

‐20%

‐30%

‐40%

‐50%
 50%
Figure 7. Bias Has primary care provider, CO BRFSS (n=4,527) and CO Cell Only 
  RDD (n=501) May‐September 2008
  RDD (n 501) May September 2008


 10

7.5

  5

2.5

  0

‐2.5

 ‐5

 75
‐7.5

‐10
Figure 8. Relative Bias Has Primary Care Provider, CO BRFSS (n=4,527) and CO 
Cell Only RDD (n=501) May‐September 2008
Cell Only RDD (n 501) May September 2008


 50%

 40%

 30%

 20%

 10%

  0%

‐10%

‐20%

‐30%

‐40%

‐50%
 50%
Figure 9. Bias ‐ could not afford health care due to cost, CO BRFSS (n=4,527) 
  and CO Cell Only RDD (n=501) May‐September 2008
  and CO Cell Only RDD (n 501) May September 2008

 10


7.5


  5


2.5


  0


‐2.5


 ‐5


‐7.5
 7.5


‐10
Figure 10. Relative Bias ‐ could not afford health care due to cost, CO 
 BRFSS (n=4,527) and CO Cell Only RDD (n=501) May‐September 2008
 BRFSS (n 4 527) and CO Cell Only RDD (n 501) May September 2008


50%
40%
30%
20%
10%
 0%
‐10%
‐20%
‐30%
‐40%
‐50%
Summary of Findings
             Summary of Findings
• Bias is present not only among those with high wireless 
  substitution rates
• Smoking prevalence underestimated among those with 
  higher wireless substitution rates
  higher wireless substitution rates
• Ever had an HIV test substantially underestimated 
  among all groups
   – R l ti bi l
     Relative bias large among those with high wireless 
                               th       ith hi h i l
     substitution rates (young adults, non‐whites, low SES)
• Bias for health care insurance and having primary care 
  provider is underestimated among non‐whites, but 
       id i     d      i     d              hi   b
  overestimated among other groups
   – Relative bias is small
Implications for study design and 
                 analysis
                     l
• When possible, include an RDD of cell phone only 
  population (BRFSS now does this)
   – If you can’t, be aware of the potential for bias and interpret 
     findings accordingly
• If you’re analyzing landline RDD data from past years
   – Interpret findings with potential bias in mind
   – Historical trend may observe artificial changes due to coverage 
     error
   – Wireless substitution rates differ by geographic region so 
     problem may be less in certain areas
• A bi
  A bias present today may not be the same historically  or in 
                   d           b h         hi    i ll       i
  the future
   – Characteristics of the early adopters may not be the same as the 
     current cell only population today or laggards
     current cell only population today or laggards
Limitations
• Unable to assess bias in some subpopulations
  Unable to assess bias in some subpopulations 
  due to small sample size
• Study does not account for cell phone mostly
  Study does not account for cell phone mostly 
  population
References
1.   Blumberg SJ & Julian V. Luke. (2009). Wireless Substitution: Early 
     release of estimates from the National Health Interview Survey, 
       l      f ti t f          th N ti      l H lth I t i S
     July‐December 2008.
2.   Blumberg SJ, Luke JV & Marcie L. Cynamon. (2006). Telephone 
     coverage and health survey estimates: evaluating concern about 
     coverage and health survey estimates: evaluating concern about
     wireless substitution. American Journal of Public Health. 96(5): 
     926‐931.
3.   Blumberg SJ & Julian V. Luke. (2009). Reevaluating the need for 
     concern regarding noncoverage bias in landline surveys. American 
     Journal of Public Health. 99(10): 1806‐1810.
4.   Delnevo CD, Gundersen DA & Brett T. Hagman. (2008). Declining 
     estimated prevalence of alcohol drinking and smoking among 
     estimated prevalence of alcohol drinking and smoking among
     young adults nationally: artifacts of sample undercoverage? 
     American Journal of Epidemiology. 167(1): 15‐19.
Contact Info
                    Contact Info
      Cristine Delnevo, PhD, MPH delnevo@umdnj.edu
      Cristine Delnevo PhD MPH delnevo@umdnj edu
      Daniel A. Gundersen, MA gunderda@umdnj.edu
Randal ZuWallack Randal.Zuwallack@macrointernational.com
   Riki Conrey Frederica.Conrey@macrointernational.com

More Related Content

Similar to Random digit dialing cell phone surveys and surveillance systems: Assessing data quality and bias

Cell Phone Sampling in Kentucky Health Survey
Cell Phone Sampling in Kentucky Health SurveyCell Phone Sampling in Kentucky Health Survey
Cell Phone Sampling in Kentucky Health Surveyjchubinski
 
Coverage Nonresponse Trade-Off
Coverage Nonresponse Trade-OffCoverage Nonresponse Trade-Off
Coverage Nonresponse Trade-OffStephanie Eckman
 
Identifying Subpopulation Least Likely to Use Mosquito Nets after Mass Distri...
Identifying Subpopulation Least Likely to Use Mosquito Nets after Mass Distri...Identifying Subpopulation Least Likely to Use Mosquito Nets after Mass Distri...
Identifying Subpopulation Least Likely to Use Mosquito Nets after Mass Distri...MEASURE Evaluation
 
AAPOR 2013 Langer Research: Bloomberg CCI
AAPOR 2013 Langer Research: Bloomberg CCIAAPOR 2013 Langer Research: Bloomberg CCI
AAPOR 2013 Langer Research: Bloomberg CCILangerResearch
 
The Cell Phone Challenge To Survey Research
The  Cell  Phone  Challenge To  Survey  ResearchThe  Cell  Phone  Challenge To  Survey  Research
The Cell Phone Challenge To Survey Researchmjomane
 
Interphone and beyond evidence for harm or safety armstrong
Interphone and beyond evidence for harm or safety  armstrongInterphone and beyond evidence for harm or safety  armstrong
Interphone and beyond evidence for harm or safety armstrongLeishman Associates
 
giornate nefrologiche pisane: Cancarini La dialisi peritoneale oggi: indicazi...
giornate nefrologiche pisane: Cancarini La dialisi peritoneale oggi: indicazi...giornate nefrologiche pisane: Cancarini La dialisi peritoneale oggi: indicazi...
giornate nefrologiche pisane: Cancarini La dialisi peritoneale oggi: indicazi...Giuseppe Quintaliani
 
Workshop session 4 - Optimal sample designs for general community telephone s...
Workshop session 4 - Optimal sample designs for general community telephone s...Workshop session 4 - Optimal sample designs for general community telephone s...
Workshop session 4 - Optimal sample designs for general community telephone s...The Social Research Centre
 
Initial Lessons From Implementing a Telecolposcopy Program on a High Risk Pop...
Initial Lessons From Implementing a Telecolposcopy Program on a High Risk Pop...Initial Lessons From Implementing a Telecolposcopy Program on a High Risk Pop...
Initial Lessons From Implementing a Telecolposcopy Program on a High Risk Pop...MobileODT
 
Mobile healthforthemasses.2015
Mobile healthforthemasses.2015Mobile healthforthemasses.2015
Mobile healthforthemasses.2015Eric Larson
 
RESULTS ON UTILIZATION OF INSECTICIDE TREATED MOSQUITO NETS IN KISAUNI MOMBASA
RESULTS ON UTILIZATION OF INSECTICIDE TREATED MOSQUITO NETS IN KISAUNI MOMBASARESULTS ON UTILIZATION OF INSECTICIDE TREATED MOSQUITO NETS IN KISAUNI MOMBASA
RESULTS ON UTILIZATION OF INSECTICIDE TREATED MOSQUITO NETS IN KISAUNI MOMBASAStanford Kapere
 
Psychiatric disorders in HIV Positive individuals in urban Uganda by Mugerwa ...
Psychiatric disorders in HIV Positive individuals in urban Uganda by Mugerwa ...Psychiatric disorders in HIV Positive individuals in urban Uganda by Mugerwa ...
Psychiatric disorders in HIV Positive individuals in urban Uganda by Mugerwa ...Hidzuan Hashim
 
Measuring ICT Access and Use by Households and Individuals
Measuring ICT Access and Use by Households and IndividualsMeasuring ICT Access and Use by Households and Individuals
Measuring ICT Access and Use by Households and IndividualsWael Kassem
 

Similar to Random digit dialing cell phone surveys and surveillance systems: Assessing data quality and bias (20)

Cell Phone Sampling in Kentucky Health Survey
Cell Phone Sampling in Kentucky Health SurveyCell Phone Sampling in Kentucky Health Survey
Cell Phone Sampling in Kentucky Health Survey
 
Coverage Nonresponse Trade-Off
Coverage Nonresponse Trade-OffCoverage Nonresponse Trade-Off
Coverage Nonresponse Trade-Off
 
Identifying Subpopulation Least Likely to Use Mosquito Nets after Mass Distri...
Identifying Subpopulation Least Likely to Use Mosquito Nets after Mass Distri...Identifying Subpopulation Least Likely to Use Mosquito Nets after Mass Distri...
Identifying Subpopulation Least Likely to Use Mosquito Nets after Mass Distri...
 
AAPOR 2013 Langer Research: Bloomberg CCI
AAPOR 2013 Langer Research: Bloomberg CCIAAPOR 2013 Langer Research: Bloomberg CCI
AAPOR 2013 Langer Research: Bloomberg CCI
 
Impact of COVID-19 on rural women and men in Cross River and Kaduna states, N...
Impact of COVID-19 on rural women and men in Cross River and Kaduna states, N...Impact of COVID-19 on rural women and men in Cross River and Kaduna states, N...
Impact of COVID-19 on rural women and men in Cross River and Kaduna states, N...
 
The Cell Phone Challenge To Survey Research
The  Cell  Phone  Challenge To  Survey  ResearchThe  Cell  Phone  Challenge To  Survey  Research
The Cell Phone Challenge To Survey Research
 
Technology taking yourmissionmobile
Technology taking yourmissionmobileTechnology taking yourmissionmobile
Technology taking yourmissionmobile
 
Taking Your Mission Mobile
Taking Your Mission MobileTaking Your Mission Mobile
Taking Your Mission Mobile
 
hikCpw 5-brick
hikCpw 5-brickhikCpw 5-brick
hikCpw 5-brick
 
Interphone and beyond evidence for harm or safety armstrong
Interphone and beyond evidence for harm or safety  armstrongInterphone and beyond evidence for harm or safety  armstrong
Interphone and beyond evidence for harm or safety armstrong
 
giornate nefrologiche pisane: Cancarini La dialisi peritoneale oggi: indicazi...
giornate nefrologiche pisane: Cancarini La dialisi peritoneale oggi: indicazi...giornate nefrologiche pisane: Cancarini La dialisi peritoneale oggi: indicazi...
giornate nefrologiche pisane: Cancarini La dialisi peritoneale oggi: indicazi...
 
Workshop session 4 - Optimal sample designs for general community telephone s...
Workshop session 4 - Optimal sample designs for general community telephone s...Workshop session 4 - Optimal sample designs for general community telephone s...
Workshop session 4 - Optimal sample designs for general community telephone s...
 
Initial Lessons From Implementing a Telecolposcopy Program on a High Risk Pop...
Initial Lessons From Implementing a Telecolposcopy Program on a High Risk Pop...Initial Lessons From Implementing a Telecolposcopy Program on a High Risk Pop...
Initial Lessons From Implementing a Telecolposcopy Program on a High Risk Pop...
 
Mobile healthforthemasses.2015
Mobile healthforthemasses.2015Mobile healthforthemasses.2015
Mobile healthforthemasses.2015
 
RESULTS ON UTILIZATION OF INSECTICIDE TREATED MOSQUITO NETS IN KISAUNI MOMBASA
RESULTS ON UTILIZATION OF INSECTICIDE TREATED MOSQUITO NETS IN KISAUNI MOMBASARESULTS ON UTILIZATION OF INSECTICIDE TREATED MOSQUITO NETS IN KISAUNI MOMBASA
RESULTS ON UTILIZATION OF INSECTICIDE TREATED MOSQUITO NETS IN KISAUNI MOMBASA
 
Psychiatric disorders in HIV Positive individuals in urban Uganda by Mugerwa ...
Psychiatric disorders in HIV Positive individuals in urban Uganda by Mugerwa ...Psychiatric disorders in HIV Positive individuals in urban Uganda by Mugerwa ...
Psychiatric disorders in HIV Positive individuals in urban Uganda by Mugerwa ...
 
Measuring ICT Access and Use by Households and Individuals
Measuring ICT Access and Use by Households and IndividualsMeasuring ICT Access and Use by Households and Individuals
Measuring ICT Access and Use by Households and Individuals
 
Internet use and health information seeking behavior of adolescents in Mbarar...
Internet use and health information seeking behavior of adolescents in Mbarar...Internet use and health information seeking behavior of adolescents in Mbarar...
Internet use and health information seeking behavior of adolescents in Mbarar...
 
Media Insights & Engagement 2016
Media Insights & Engagement 2016Media Insights & Engagement 2016
Media Insights & Engagement 2016
 
Growing up with Media pilot study: Examining exposures to violence
Growing up with Media pilot study: Examining exposures to violenceGrowing up with Media pilot study: Examining exposures to violence
Growing up with Media pilot study: Examining exposures to violence
 

Recently uploaded

Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024The Digital Insurer
 
Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Allon Mureinik
 
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonData Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonAnna Loughnan Colquhoun
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreternaman860154
 
The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024Rafal Los
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonetsnaman860154
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersThousandEyes
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationRidwan Fadjar
 
Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slidevu2urc
 
🐬 The future of MySQL is Postgres 🐘
🐬  The future of MySQL is Postgres   🐘🐬  The future of MySQL is Postgres   🐘
🐬 The future of MySQL is Postgres 🐘RTylerCroy
 
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationGenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationMichael W. Hawkins
 
Google AI Hackathon: LLM based Evaluator for RAG
Google AI Hackathon: LLM based Evaluator for RAGGoogle AI Hackathon: LLM based Evaluator for RAG
Google AI Hackathon: LLM based Evaluator for RAGSujit Pal
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationRadu Cotescu
 
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfThe Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfEnterprise Knowledge
 
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking MenDelhi Call girls
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Igalia
 
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 3652toLead Limited
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsMaria Levchenko
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationSafe Software
 
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhi
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | DelhiFULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhi
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhisoniya singh
 

Recently uploaded (20)

Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024
 
Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)
 
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonData Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt Robison
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreter
 
The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonets
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 Presentation
 
Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slide
 
🐬 The future of MySQL is Postgres 🐘
🐬  The future of MySQL is Postgres   🐘🐬  The future of MySQL is Postgres   🐘
🐬 The future of MySQL is Postgres 🐘
 
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationGenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day Presentation
 
Google AI Hackathon: LLM based Evaluator for RAG
Google AI Hackathon: LLM based Evaluator for RAGGoogle AI Hackathon: LLM based Evaluator for RAG
Google AI Hackathon: LLM based Evaluator for RAG
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organization
 
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfThe Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
 
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
 
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed texts
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
 
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhi
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | DelhiFULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhi
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhi
 

Random digit dialing cell phone surveys and surveillance systems: Assessing data quality and bias

  • 1. Random digit dialing cell phone surveys and  surveillance systems: data quality, data  ill t d t lit d t weighting strategies, and bias Cristine D. Delnevo, PhD, MPH & Daniel A. Gundersen, MA, UMDNJ‐ School of Public Health Randal S. ZuWallack, MS & Frederica R. Conrey, PhD Randal S ZuWallack MS & Frederica R Conrey PhD ICF Macro International Presented at 137th Annual Meeting & Exposition P t d t 137th A l M ti & E iti Philadelphia, PA November 7‐11, 2009 Work supported in part by the National Cancer Institute (R21CA129474 ) and a contract from the New  Jersey Department of Health and Senior Services, through funding from the Cigarette Tax
  • 2. Wireless substitution (US), 2005 2008 Wireless substitution (US), 2005–2008 Source: Blumberg & Luke  Wireless substitution: Early release of estimates from the National Health Interview Survey, July‐ December 2008. National Center for Health Statistics. May 2009.  
  • 3. Demographic differences Demographic differences • Gender: men more likely than  Wireless substitution by age (and time) Wireless substitution by age (and time) women to be wireless only b i l l • SES: Adults living in (30.9%) and  near poverty (23.8%) more likely  than higher income adults  h hi h i d l (16.0%) to be wireless only • Region: Wireless substitution  highest in South (21.3%) and  hi h t i S th (21 3%) d Midwest (20.8%) vs. Northeast  (11.4%) or West (17.2 %) • Race/Ethnicity: Wireless  R /Eth i it Wi l substitution highest  among  black (21.4%) and Hispanic  (25.0%) adults vs. Non‐Hispanic  (25 0%) adults vs Non Hispanic white adults (16.6%)   Source: Blumberg & Luke  Wireless substitution: Early release of estimates from the National Health Interview Survey, July‐ December 2008. National Center for Health Statistics. May 2009.  
  • 4. Biased health estimates? Biased health estimates? • Potential for biased health estimates due to sample Potential for biased health estimates due to sample  under‐coverage remains a real, growing threat to  RDD health surveys • Cell‐phone only also differs with respect to health  behaviors and the validity of some health estimates  based on traditional RDD surveys are increasingly  questionable
  • 5. Health estimates by phone status Health estimates by phone status Has a landline Wireless-only y No telephone p telephone NHIS July – December 2007 5+ alcoholic drinks in 1 day 17.7 37.3 27.1 Current smoker 18.0 30.6 38.6 Uninsured 13.7 13 7 28.7 28 7 44.1 44 1 Has a usual place for care 87.5 68.0 61.8 Flu vaccination 32.7 16.6 20.9 Ever tested for HIV 34.7 47.6 45.8 Source: Blumberg SJ, Luke JV. Coverage bias in traditional telephone surveys of low‐income and young  adults. Public Opin Q. 2007;71:734–749
  • 6. Cigarette smoking among young adults,  2003‐2005, NHIS & BRFSS 2003 2005 NHIS & BRFSS Source: Delnevo, Gundersen & Hagman (2008) Declining prevalence of alcohol and smoking estimates  among young adults nationally: artifacts of sample under‐coverage?  Am J of Epidemiology
  • 7. New challenges: Wireless Mostly? Telephone Status, NHIS July‐ December 2008 • The percentage of adults  living in wireless‐mostly  households has been  increasing • Who are they? Who are they?  (demographic & health  behaviors?) • Will they respond to  landline surveys?  Source: Blumberg & Luke  Wireless substitution: Early release of estimates from the National Health Interview Survey, July‐ December 2008. National Center for Health Statistics. May 2009.  
  • 8. Reactions? • Starting in 2009, the Center  for Disease Control and  f Di C l d Prevention (CDC) is  requiring states to  incorporate cell phone  incorporate cell phone interviews in their regular  BRFSS sample • Yet there is no widely  y accepted methods of  evaluating data quality or  data weighting, particularly  for state and local area  for state and local area surveys.  • AAPOR Cell Phone Task  Force report Force report
  • 9. This special session: This special session: • Analysis of data quality of cell phone Analysis of data quality of cell phone  surveys • Demonstrate weighting procedures for  merging cell phone samples with landline  samples • An assessment of bias in landline only An assessment of bias in landline only  surveys, and 
  • 10. Assessing Data Quality of Cell Phone Random Digit g Dial Surveys Frederica R Conrey Randy Zuwallack
  • 11. Data Quality • Q lit of Responses Quality f R • Quality of Sample
  • 12. 2008 New Jersey Adult Tobacco Survey • R d Random Di it Di l L d (RR 19%) and Digit Dial Land (RR=19%) d Cell Phone (RR=16%) • Short version – 49 Questions (min=37; max=70) – 534 Cell completes – 468 Landline completes
  • 13. Survival Analysis • Diff Different people get diff t l t different surveys t because of skip patterns • Survival analysis – Measures the impact of survey mode on non- response – Controls for differences in survey length
  • 15. Item Non-Response Failure Rates by Phone Mode Mean M Median M di Std Cell 3.3% 2.2% 7.4% Landline 3.4% 2.0% 5.5% Survival predicted by p p y phone mode: Hazard=1.00, p>.95
  • 16. Open Ended Response Total Responses p Open p per p Open end reports of the O d t f th End by Phone Mode events of 2 recalled Responses commercials / OE DK / OE – 251 respondents were Cell .52 .49 asked at least one Landline .48 .55 open ended question P .56 .33
  • 18. Unit Non-Response Data D t quality i th t lit is threatened if d – Response rates are low AND – The people who DO NOT respond are different from those who DO. If cell phone respondents are less likely to respond, then there is non-response bias.
  • 19. Survival Analysis • C ll and l dli respondents may get Cell d landline d t t different surveys • Response rates alone don’t tell the story • Survival analysis tells whether cell y respondents are more likely to break off g given the same survey length y g
  • 20. Survival Model 25% Sample Breaking Off 20% g 15% 10% Cell e Landline 5% 0% 0 20 40 60 80 Survey Questions p<.001 001
  • 21. What does a difference in survey survival mean? • C ll respondents quit sooner th l dli Cell d t it than landline respondents. • The sample under-represents cell phones • The longer the survey, the worse the g y nonresponse bias • The solution? – Careful weighting – Short surveys
  • 22. In a population study of tobacco use behavior… Minimal difference Mi i l diff Substantial difference S b t ti l diff between cell and between cell and landline in response landline in response quality rate • No difference emerged in • Sample quality may be item-nonresponse threatened if cell phone • No difference emerged in surveys are too long or richness of open end i h f d weighted incorrectly responses.
  • 23. Weighting Cell Phone Surveys  Weighting Cell Phone Surveys Randal Zuwallack Frederica R Conrey Frederica R Conrey
  • 24. Thanks • Cris Delnevo Cris Delnevo • Dan Gundersen • NCI (R21CA129474), New Jersey Dept of  C ( 2 C 29 ) f Health and Senior Services
  • 25. Dual Frame D A.  Adults in landline households with no cell phone, A Adults in landline households with no cell phone B.  Adults in landline households with a cell phone, and C. Adults in non‐landline households with a cell phone (cell only).
  • 26. Dual Frame D Common designs: Common designs: Dual frame w/ no overlap: Landline (A+B) + Cell (C) Dual frame w/ overlap: Landline (A+B) + Cell (B+C) Uncommon design: Dual frame w/ no overlap: Landline (A) + Cell (B+C) Dual frame w/ no overlap: Landline (A) + Cell (B+C)
  • 27. Weighting Challenges • Challenge 1: How do we put the dual frames Challenge 1: How do we put the dual frames  together? • Challenge 2:  Differential Nonresponse
  • 28. Challenge 1 • How do we put the dual frames together? How do we put the dual frames together? • No overlap – Estimate of cell‐only population size? Estimate of cell only population size? • Internal estimate • External estimate: NHIS (Blumberg et al.) • With overlap – Must determine group membership – Adjust for multiple selection probabilities – Estimate of phone group population sizes?
  • 29. Cell Survey • “In addition to your cell phone is there at In addition to your cell phone, is there at  least one telephone inside your home that is  currently working and is not a cell phone?  Do  currently working and is not a cell phone? Do not include telephones only used for business  or telephones only used for computers or fax  or telephones only used for computers or fax machines.”  – ‘yes’ = dual user, while those who responded  yes = dual user while those who responded – ‘no’ = cell‐only
  • 30. Landline Survey • “In addition to your residential landline In addition to your residential landline  telephone, do you also use one or more cell  phone numbers? phone numbers?”  – ‘yes’ = dual user – ‘no’ = landline only. no = landline only
  • 31. Example 1‐‐Colorado • Combine with BRFSS Combine with BRFSS • Group membership  –KKnown for cell f ll – Unknown for landline • Limited to dual frame w/ no overlap – Used 15% (NHIS state estimates) for merging  landline and cell – Poststratified dual sample to age and sex.
  • 32. Example 2 • You are midway through a landline survey and You are midway through a landline survey and  want to add cell phones.  You don’t know who  has a cell phone and who doesn t. What are  has a cell phone and who doesn’t What are your options? 1) Add cell only 1) Add cell only 2) Add cell and dual‐users 
  • 33. Challenge 2 • Differential Nonresponse Differential Nonresponse • Cell‐only overrepresented when conducting  cell phone surveys.  cell phone surveys – Contact rate –CCooperation rate i • Those who rely more on their cell phone will  be easier to reach.
  • 34. Telephone Reliance C L e a l n l d l i P n h e o n e
  • 35. Landline Frame C L e a l n l d l i Landline households P n h e o n e Landline sample
  • 36. Cell Phone Frame C L e a l n l d l i Cell Phone Users P n h e o n e Cell sample Cell sample
  • 37. Dual Frames Landline sample Landline sample Cell sample Cell sample Dual Frame sample Dual Frame sample Ideal Realistic
  • 38. Our Goal Rebalance  R b l on cell  reliance Rebalance  on landline  reliance
  • 39. Measuring Telephone Reliance • Cell only landline only Cell only, landline only • Classify Dual users – “Of ll th t l h “Of all the telephone calls that you receive, are…” ll th t i ” • All or almost all calls received on a cell phone? (cell‐ mostly) • Some received on a cell phone and some on a regular  landline phone? (true‐dual) • Very few or none received on a cell phone? (landline‐ mostly)
  • 40. Telephone Reliance C L e a l n d Landline Landline True Cell Cell l l i Only Mostly Dual Mostly Only P n (0) (1) (2) (3) (4) h e o n e
  • 41. Response Propensity • Adjust for differential nonresponse by Adjust for differential nonresponse by  benchmarking against NHIS • Logistic regression model Logistic regression model – Dependent: Survey type •1 b 1 = observe cell user in national cell phone survey ll i ti l ll h • 0 = observe cell user in NHIS – Independent: Cell phone reliance (1 4) age race Independent: Cell phone reliance (1‐4), age, race
  • 42. Data sources National cell sample National cell sample NHIS n=500 Landline mostly 8% 23% True dual 27% 42% Cell mostly 23% 17% Cell only 42% 18% Cell users 100% 100%
  • 43. Applying the model • Applied to same data—poststatification Applied to same data poststatification • Applied to independent data –A Assumption: national cell sample measures the  ti ti l ll l th odds ratio for observing a cell‐only respondent in  a cell sample relative to a dual user. a cell sample relative to a dual‐user • S State, local surveys  l l
  • 44. Applying the model Colorado Cell Sample w/o NR adj w/ NR adj Landline mostly (1) 9% 26% True dual (2) 26% 37% Cell mostly (3) C ll tl (3) 19% 15% Cell only (4) 46% 21% Total cell sample 100% 100%
  • 45. Applying the model • Assume landline only is 20% (we don’t know) Assume landline only is 20% (we don t know) CO Landline only 20% Landline mostly 21% True dual 30% Cell mostly 12% NHIS  NHIS Cell only 17% state  estimate  Total population p p 100% = 15%
  • 46. City Sample Landline Cell Sample Combined  Sample Samples w/o NR  w/o NR w/ NR  w/ NR w/o NR  w/o NR w/ NR  w/ NR adj adj adj adj Cell‐only ‐ 43.5 18.9 35.5 13.4 Cell‐mostly 12.4 25.1 19.8 13.8 12.2 True Dual 30.1 23.6 35.5 18.7 25.1 Landline mostly Landline‐mostly 19.1 7.9 25.8 9.9 17.2 Landline‐only 38.4 ‐ ‐ 23.9 32.1
  • 47. Conclusions • Dual‐frame Dual frame – There are ways to combine the data, even when  we don t have a full picture of group membership. we don’t have a full picture of group membership • Differential nonresponse – Response propensity model rebalances the cell Response propensity model rebalances the cell  sample based on cell reliance.  – Can be applied at state and local levels when no Can be applied at state and local levels when no  benchmarks exist. – Next steps: explore a response propensity model Next steps: explore a response propensity model  for landline.
  • 49. Examining the bias in landline only  g y surveys: How does the cell phone only  population differ from the landline  population differ from the landline population on health indicators, and are  estimates from landline surveys biased? Daniel A. Gundersen, MA, UMDNJ‐SPH Cristine D. Delnevo, PhD, MPH, UMDNJ‐SPH Randy S. ZuWallack, MS, ICF Macro
  • 50. Cell Phone Substitution and RDD surveys Cell Phone Substitution and RDD surveys • RDD surveys (e.g. BRFSS) have traditionally only sampled  household telephones (i.e. landlines) • Up until early 2000s, rate of cell phone only households  was small • From mid 2000s, rate of substitution has grown  substantially – 6 7% of adults in 2005 to 18 4% in 2008 nationally1 6.7% of adults in 2005 to 18.4% in 2008 nationally • Higher among certain demographic groups1 – Young adults – Hispanics and Blacks – Poor and near poor
  • 51. What is bias due to coverage error in the  sampling frame? li f ? • Non‐covered population is different from Non covered population is different from  covered population on some variable of  interest – If proportion of non‐covered (    ) is small, bias will  be small – If difference between the covered and  , noncovered                is small, bias will be small
  • 52. Previous Research Previous Research • Data from Jan 2004‐June 2005 NHIS found2 – Greater than 1 percentage point bias in binge drinking, smoking  prevalence, usual place for medical care, receiving influenza  vaccine • Data from 2007 NHIS found3 Data from 2007 NHIS found – Bias increased slightly for past year binge drinking and receiving  influenza vaccine – These biases were larger among young adults and low income These biases were larger among young adults and low income  persons • Data from 2001‐2005 BRFSS on 18‐24 year olds found4 – Prevalence of binge drinking, heavy drinking, and cigarette  smoking declined during 2003‐2005; coincided with large  increase in wireless substitution among young adults – NHIS and NSDUH did not observe similar declines during this  period
  • 53. Our Study: Our Study: • Objective: Objective:  – Assess the presence of bias in landline RDD due to  exclusion of cell phone only  on select health  exclusion of cell phone only on select health indicators • Data Source and Instrument: Data Source and Instrument: – Cell phone RDD of adults in Colorado (n=501) • May to September 2008 May to September 2008 • Instrument was shortened version of BRFSS – BRFSS from same data collection period (n=4,527) BRFSS from same data collection period (n 4,527)
  • 54. Methodology • Cell Phone sample: Ce o e sa p e: – Design weights account for probability of selection • BRFSS – Standard BRFSS design weight accounts for strata,  number of landlines and adults in the household – Postratified b f d by age(7)*sex*race ( )* * • Merged data –D i Design weights scaled to represent share of  i ht l dt t h f population by phone status – Postratified by age(7)*sex*race y g ( )
  • 55. Statistical Analyses Statistical Analyses • Comparisons of BRFSS landline and Cell Only based on design  weights • Comparison of BRFSS landline and merged data are  postratified to demographic makeup of CO • We assume the merged data to be unbiased (i.e. no coverage  error due to cell only exclusion) error due to cell only exclusion) • All analyses conducted in STATA v.10.1 to account for complex  sampling design
  • 56. Table. CO BRFSS vs. CO Cell Only, May‐September 2008 (n=5,028) y, y p ( , ) BRFSS landline Cell only y Difference Health Indicator % (95%CI) % (95%CI) % (95%CI) Smoking* * 15.29 (±1.14) 28.14 (±4.46) ‐12.85 (±4.60) ( ) ( ) ( ) Ever had HIV test* 36.64 (±1.85) 52.51 (±5.09) ‐15.87 (±5.41) Having health insurance* 88.36 (±1.11) 72.46 (±4.38) 15.9 (±4.51) Having primary care provider* 85.91 (±1.16) 60.39 (±4.85) 25.52 (±5.00) Not affording care due to cost 12.27 (±1.08) 20.42 (±3.94) Not affording care due to cost* 12 27 (±1 08) 20 42 (±3 94) ‐8.15 (±4.08) ‐8 15 (±4 08) *p<.05; data weighted to correct for sampling design
  • 61. Figure 5. Bias ‐ Having health insurance, CO BRFSS (n=4,527) and CO Cell Only  RDD (n=501) May‐September 2008 RDD (n 501) May September 2008 10 7.5 5 2.5 0 ‐2.5 ‐5 ‐7.5 ‐7 5 ‐10
  • 65. Figure 9. Bias ‐ could not afford health care due to cost, CO BRFSS (n=4,527)  and CO Cell Only RDD (n=501) May‐September 2008 and CO Cell Only RDD (n 501) May September 2008 10 7.5 5 2.5 0 ‐2.5 ‐5 ‐7.5 7.5 ‐10
  • 66. Figure 10. Relative Bias ‐ could not afford health care due to cost, CO  BRFSS (n=4,527) and CO Cell Only RDD (n=501) May‐September 2008 BRFSS (n 4 527) and CO Cell Only RDD (n 501) May September 2008 50% 40% 30% 20% 10% 0% ‐10% ‐20% ‐30% ‐40% ‐50%
  • 67. Summary of Findings Summary of Findings • Bias is present not only among those with high wireless  substitution rates • Smoking prevalence underestimated among those with  higher wireless substitution rates higher wireless substitution rates • Ever had an HIV test substantially underestimated  among all groups – R l ti bi l Relative bias large among those with high wireless  th ith hi h i l substitution rates (young adults, non‐whites, low SES) • Bias for health care insurance and having primary care  provider is underestimated among non‐whites, but  id i d i d hi b overestimated among other groups – Relative bias is small
  • 68. Implications for study design and  analysis l • When possible, include an RDD of cell phone only  population (BRFSS now does this) – If you can’t, be aware of the potential for bias and interpret  findings accordingly • If you’re analyzing landline RDD data from past years – Interpret findings with potential bias in mind – Historical trend may observe artificial changes due to coverage  error – Wireless substitution rates differ by geographic region so  problem may be less in certain areas • A bi A bias present today may not be the same historically  or in  d b h hi i ll i the future – Characteristics of the early adopters may not be the same as the  current cell only population today or laggards current cell only population today or laggards
  • 69. Limitations • Unable to assess bias in some subpopulations Unable to assess bias in some subpopulations  due to small sample size • Study does not account for cell phone mostly Study does not account for cell phone mostly  population
  • 70. References 1. Blumberg SJ & Julian V. Luke. (2009). Wireless Substitution: Early  release of estimates from the National Health Interview Survey,  l f ti t f th N ti l H lth I t i S July‐December 2008. 2. Blumberg SJ, Luke JV & Marcie L. Cynamon. (2006). Telephone  coverage and health survey estimates: evaluating concern about  coverage and health survey estimates: evaluating concern about wireless substitution. American Journal of Public Health. 96(5):  926‐931. 3. Blumberg SJ & Julian V. Luke. (2009). Reevaluating the need for  concern regarding noncoverage bias in landline surveys. American  Journal of Public Health. 99(10): 1806‐1810. 4. Delnevo CD, Gundersen DA & Brett T. Hagman. (2008). Declining  estimated prevalence of alcohol drinking and smoking among  estimated prevalence of alcohol drinking and smoking among young adults nationally: artifacts of sample undercoverage?  American Journal of Epidemiology. 167(1): 15‐19.
  • 71. Contact Info Contact Info Cristine Delnevo, PhD, MPH delnevo@umdnj.edu Cristine Delnevo PhD MPH delnevo@umdnj edu Daniel A. Gundersen, MA gunderda@umdnj.edu Randal ZuWallack Randal.Zuwallack@macrointernational.com Riki Conrey Frederica.Conrey@macrointernational.com