The “Good Enough” Respondent: Using Sufficiently

The “Good Enough” Respondent:
Using Sufficiently Knowledgeable Adults as
Respondents in a Survey about the Health of
Young Children
American Association for Public Opinion Researchers (AAPOR) 2015
Hollywood, FL
May 15, 2015
Alicia Frasier, NORC at the University of Chicago
Jennifer Vanicek, NORC at the University of Chicago
Andrea Mayfield, NORC at the University of Chicago
Ying Li, NORC at the University of Chicago
Ned English, NORC at the University of Chicago
Joelle Greene, Harder+Company Community Research
Melinda Leidy, First 5 LA
Introduction
• Challenges of screening for surveys regarding young
children’s health and wellness
• Often ideal to have the most knowledgeable respondent (MKR)
• Identifying and contacting the MKR can add screening costs
• Replicated work done by Eisenhower, Immerwahr, Merry,
and Weiss (2012)
"Using an Alternative to “Most Knowledgeable Adult” as a Selection
Rule for Proxy Reporters in a Child Health Survey"
Survey Practice, Vol 5, No 4 (2012) Eisenhower, et al.
http://surveypractice.org/index.php/SurveyPractice/article/view/33/h
tml
• Designed questions to identify a Sufficiently Knowledgeable
Respondent (SKR)
The “Good Enough” Respondent
2
Research Question
• Are Sufficiently Knowledgeable Respondents acceptable
replacements for Most Knowledgeable Respondents in
surveys about young children?
The “Good Enough” Respondent
3
First 5 LA Family Survey
• Best Start Communities
• Place-based initiative with the goal of improving outcomes for
children aged five years old and younger, ensuring that they:
•
•
•
•
Are born healthy
Maintain a healthy weight
Are free from abuse and neglect
Enter school ready to learn
• 14 communities in L.A. County
• Family Survey:
• Parents/caregivers of children ages 0 through 5
• Topics include children’s health, well-being, childcare, child and
adult demographics, and household characteristics
The “Good Enough” Respondent
4
First 5 LA Family Survey
Challenges and Innovations in Case Flow and Case Management
5
First 5 LA Family Survey (cont.)
• Address-Based Sample (ABS), multi-mode design
• Mailed SAQ screener
– Asked for household information
– Attempted to obtain phone number
• Phone follow-up for those with phone numbers
• Field follow-up for portion of those without phone numbers
The “Good Enough” Respondent
6
Screening
• Rostered up to six children in the HH
• Randomly selected up to four children for the interview
• Asked the initial respondent to identify capability to
answer about the selected child
• “Do you know [CHILD] well enough to answer questions about
[his/her] health, what kinds of food [he/she] eats, and [his/her]
general activities?”
• Of those HHs fully rostered only 1.5 percent indicated
they could not consider themselves at least an SKR for all
of the children in their HH
The “Good Enough” Respondent
7
Basics
• 4,232 completed interviews = Child-level
• Multiple child interviews per HH
• Approximately 1.42 children per HH
Number
Number of
of
Frequency
Frequency
children
children
interviews
interviews in
in HH
HH
Percent
Percent
Total interviews
11
1,979
1,979
66.5
66.5
1,979
22
790
790
26.5
26.5
1,580
33
159
159
5.3
5.3
44
49
49
1.7
1.7
2,977
2,977
The “Good Enough” Respondent
33.5%
477
196
4,232
8
SKR vs. MKR
Sufficiently Knowledgeable vs. Most Knowledgeable
Distinguishing SKR vs. MKR status
• All respondents start as SKR status, must further identify
to become classified as MKR
• Assumed to be MKR if only one adult in the household
• “Compared to you, when answering questions about [CHILD]'s
health and daily routines, would you say that the other adults in
the household are equally or more knowledgeable than you?”
The “Good Enough” Respondent
10
Results
• 4,232 completed interviews
• 78 percent identified themselves as an SKR
• 22 percent identified themselves as the MKR
• Biological Mothers (n=2,835)
• 72 percent identified themselves as SKR
• 28 percent identified themselves as MKR
The “Good Enough” Respondent
11
Comparing SKR vs. MKR on Household-Level Indicators
Household-Level Indicator
SKR
(%)
MKR
(%)
Difference
(percentage point)
Food Stamps
38.9
55.0
16.1*
WIC
67.0
72.5
5.5*
CalWorks
16.1
29.6
13.5*
Child Care Assistance
5.8
14.3
8.5*
Child Support
4.9
11.1
6.2*
Social Security
13.5
15.3
1.8
Section 8
4.5
16.1
11.6*
* p < 0.05
The “Good Enough” Respondent
12
Comparing SKR vs. MKR on Child-Level Indicators
Child-Level Indicator
SKR
(%)
MKR
(%)
Difference
(percentage point)
Health Insurance Coverage
97.0
95.8
-1.2
Place to go for Healthcare
98.0
97.8
-0.2
Dentist visit
63.4
65.4
2.0
TV use
54.0
54.9
0.9
Computer use
90.2
90.6
0.4
Fast Food consumption
16.2
15.0
1.2
Sugary Drinks consumption
87.6
85.0
2.6*
* p < 0.05
The “Good Enough” Respondent
13
Item Non-Response – Household-Level Indicators
Household-Level Indicator
SKR
DK response
(%)
MKR
DK response
(%)
Difference
(percentage
point)
Food Stamps
1.78
0.44
1.34*
WIC
1.22
0.29
0.93*
CalWorks
1.48
1.18
0.30
Child Care Assistance
0.74
0.88
-0.14
Child Support
0.61
0.15
0.46
Social Security
1.22
0.15
1.07*
Section 8
0.26
0.00
0.26
* p < 0.05
The “Good Enough” Respondent
14
Item Non-Response – Child-Level Indicators
Child-Level Indicator
SKR
DK response
(%)
MKR
DK response
(%)
Difference
(percentage
point)
Health Insurance Coverage
0.70
0.00
0.70*
Place to go for Healthcare
0.64
0.11
0.53*
Dentist visit
0.94
0.32
0.62
TV use
0.27
0.11
0.16
Computer use
0.15
0.11
0.04
Fast Food consumption
0.15
0.21
-0.06
Sugary Drinks consumption
0.15
0.00
0.15
* p < 0.05
The “Good Enough” Respondent
15
Discussion
• What do we lose or risk if we depend on the SKR
method?
• Hard to tease out
• What do we gain?
• Efficiency
• Dependable data
• What sort of bias might we incur?
• More item non-response
The “Good Enough” Respondent
16
Limitations
• Adults with young children in the HH
• L.A. County respondents
The “Good Enough” Respondent
17
Conclusions
• Most respondents identify as SKR for children in HH
• Only about 25 percent identify as MKR
• Is MKR even a concept that works?
• Small increase in item non-response
• SKR is a legitimate and useful mechanism to gather data
on children under the age of six
• Creates efficiency
• Cuts costs and time for screening
The “Good Enough” Respondent
18
Contact: [email protected]
Thank You!