roundtable on research design, data ...as an industry we are keen to better understand our...
TRANSCRIPT
![Page 1: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/1.jpg)
© 2016 Ipsos. All rights reserved. Contains Ipsos' Confidential and Proprietary information and may
not be disclosed or reproduced without the prior written consent of Ipsos.
1
18th May, 2017
RESEARCH DESIGN,
ROUNDTABLE ON
DATA COLLECTION
& INNOVATION
![Page 2: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/2.jpg)
© 2016 Ipsos. All rights reserved. Contains Ipsos' Confidential and Proprietary information and may
not be disclosed or reproduced without the prior written consent of Ipsos.
2
18th May, 2017
UK RANDOM PROBABILITY
RESPONSE RATES ON
FACE TO FACE SOCIAL SURVEYS
Keith Bolling (Kantar Public) & Kirby Swales (NatCen)
![Page 3: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/3.jpg)
3
EVIDENCE AND CONTRIBUTIONS SUPPLIED BY…
![Page 4: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/4.jpg)
4
CONTENTS
1
2
3
Background: what has been happening to response rates?
Looking beneath the bonnet – levels of effort
Implications and conclusions
![Page 5: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/5.jpg)
5
BACKGROUND : WHAT HAS BEEN HAPPENING TO RESPONSE RATES
01
![Page 6: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/6.jpg)
6
SURVEY RESPONSE RATES BY YEAR: AVERAGE
British Social Attitudes Average
30
35
40
45
50
55
60
65
70
75
80
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Resp
on
se R
ate
ch
an
ge (
%)
![Page 7: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/7.jpg)
7
SURVEY RESPONSE RATES BY YEAR: DETAIL
BARB
British Social Attitudes
Crime Survey for England and Wales
Family Resources Survey
Health Survey for England
Labour Force Survey
Living Costs and Food Survey
National Travel Survey
Average
30
35
40
45
50
55
60
65
70
75
80
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Resp
on
se R
ate
ch
an
ge (
%)
![Page 8: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/8.jpg)
8
SURVEY RESPONSE RATES: CHANGE OVER TIME
-30
-25
-20
-15
-10
-5
0
5
Resp
on
se R
ate
ch
an
ge (
%)
1998/2001* - 2015 2003 - 2015
![Page 9: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/9.jpg)
9
RESPONSE RATES FOR THE HOUSEHOLD BUDGET SURVEY
Finland
France
Lithuania
Poland
Slovakia
Slovenia
Sweden
20
30
40
50
60
70
80
90
100
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Resp
on
se R
ate
ch
an
ge (
%)
![Page 10: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/10.jpg)
10
LONGITUDINAL STUDIES ATTRITION RATES
0
10
20
30
40
50
60
70
80
90
100
1 2 3 4 5 6 7 8 9 10 11 12 13
Resp
on
se R
ate
(%
)
Number of sweeps or waves
Growing Up in Scotland BHPS English Longitudinal Study of Ageing
Millenium Cohort Study National Child Development Study 1970 British Cohort Study
![Page 11: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/11.jpg)
11
POSSIBLE EXPLANATION
Less trust in government, brands and professions
Survey fatigue
Availability
Accessibility
![Page 12: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/12.jpg)
12
LESS TRUST IN GOVERNMENT AND POLITICIANS
0
5
10
15
20
25
30
35
40
45
1986
198…
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
%
Almost never
![Page 13: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/13.jpg)
13
MAJOR FACTORS AFFECTING RESPONSE
Response
rate
Messaging
Channels
Volume
Timing
PARTICIPANT
COMMUNICATIONS
Volumes &
allocation
Management
Reissues
FIELDWORK
SURVEY DESIGN
Sample type
Geography
Clustering
Sample data
Selection method
Interview length
Topic
Mode Training
Motivation
Management
INTERVIEWER/NURSE
Type
Value
INCENTIVESocietal
change
![Page 14: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/14.jpg)
14
LOOKING BENEATH THE BONNET : LEVELS OF EFFORT
02
![Page 15: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/15.jpg)
15
DETAILS OF THREE CASE STUDIES
BARB BSA CSEW
First conducted 1991 1983 1981
Sample design Tightly clustered Clustered Unclustered
PSUs Based on OAs Based on postcode
sectors
Bespoke but based
on LSOAs
Selection Household level
interview
One adult (18+)
randomly selected
One adult (16+)
randomly selected
Assignment size
(issued sample)
c. 17-18 addresses c. 26 addresses c. 32 addresses
Incentives None Conditional voucher
(£10 from 2014)
Stamps sent with
advance letter
Interview length c. 15 mins c. 65 mins c. 50 mins
Fieldwork length Up to 6 weeks Around 4 months Max. 6 months
![Page 16: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/16.jpg)
16
CASE STUDY RESPONSE RATES
51%
46%
75%
72%
69%
63%
30%
35%
40%
45%
50%
55%
60%
65%
70%
75%
80%
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Resp
on
se r
ate
Survey year
BSA CSEW BARB
![Page 17: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/17.jpg)
17
CUMULATIVE RESPONSE RATE AFTER X VISITS NON-CONTACT RATES HAVE BEEN KEPT AT LOW LEVELS
7% 6% 6% 6% 6% 6% 5% 7% 6% 8%
0%
20%
40%
60%
80%
100%
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
4% 4% 4% 3% 3%4% 3% 3% 4% 4%
0%
20%
40%
60%
80%
100%
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
16% 16%13% 10% 10% 9% 9% 9% 9% 6%
0%
20%
40%
60%
80%
100%
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
BSA
BARB
CSEW
![Page 18: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/18.jpg)
18
CUMULATIVE RESPONSE RATEREFUSAL RATES HAVE INCREASED IN THE LAST FEW YEARS
42%39% 39% 40% 40% 41% 41%
46%43%
46%
0%
20%
40%
60%
80%
100%
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
21% 19% 21% 21% 22% 20% 24% 22%26% 25%
0%
20%
40%
60%
80%
100%
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
15% 13% 17% 19% 20% 21% 22% 22% 24% 31%
0%
20%
40%
60%
80%
100%
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
BSA
BARB
CSEW
![Page 19: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/19.jpg)
20
MORE EFFORT BEING PUT IN TO KEEP RESPONSE RATES UP
4.7
3.9
4.1
5.2
4.8
5.2
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
BSA BARB CSEW
2008-9
2015-16
![Page 20: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/20.jpg)
21
MORE EFFORT BEING PUT IN TO KEEP RESPONSE RATES UP
2008/9 2015/16 % increaseExtra calls per
year
BSA (2008-2016) 4.7 5.2 11% c. 4,000
BARB (2009-2015) 3.9 4.8 23% c. 72,000
CSEW (2008-2016) 4.1 5.2 27% c. 55,000
![Page 21: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/21.jpg)
23
INCREASE IN THE PROPORTION OF ADDRESSES WHICH HAVE MORE
VISITS THAN MINIMUM CALL REQUIREMENTS
24%
18%
26%
29% 29%
35%
0%
5%
10%
15%
20%
25%
30%
35%
40%
BSA BARB CSEW
2008-9
2015-16
![Page 22: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/22.jpg)
24
UNPRODUCTIVE ADDRESSES ARE REISSUED TO BOOST RESPONSE
0%
5%
10%
15%
20%
25%
30%
35%
40%
2008 2009 2010 2011 2012 2013 2014 2015 2016
% o
f ad
dre
sses r
eis
su
ed
Survey year
BSA BARB CSEW
![Page 23: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/23.jpg)
25
RESPONSE ON REISSUES IS LOWER ALTHOUGH HAS REMAINED
FAIRLY STEADY
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
2008 2009 2010 2011 2012 2013 2014 2015 2016
Resp
on
se r
ate
Survey year
BSA BARB CSEW
![Page 24: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/24.jpg)
26
CUMULATIVE RESPONSE RATE AFTER X VISITS, 2015-16
28%
42%
46%
57%
67%
67%
46%
63%
72%
0%
10%
20%
30%
40%
50%
60%
70%
80%
0 1 2 3 4 5 6 7 8 9 10 11 12+
Resp
on
se r
ate
After x visits
BSA BARB CSEW
![Page 25: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/25.jpg)
27
RELATIONSHIP BETWEEN RESPONSE RATE AND CALLS
BSA
BARB
CSEW
0%
20%
40%
60%
80%
100%
0%
20%
40%
60%
80%
100%
0 1 2 3 4 5 6 7 8 9
10
11
12
13
14
15
16+
Re
sp
on
se ra
te
% o
f to
tal c
alls
After x visits
Calls Response
0%
20%
40%
60%
80%
100%
0%
20%
40%
60%
80%
100%
0 1 2 3 4 5 6 7 8 9
10
11
12
13
14
15
16
17
18
19
20+
Re
sp
on
se ra
te
% o
f to
tal c
alls
After x visits
Calls Response
0%
20%
40%
60%
80%
100%
0%
20%
40%
60%
80%
100%
0 1 2 3 4 5 6 7 8 9 10
Re
sp
on
se ra
te
% o
f to
tal c
alls
After x visits
Calls Response
![Page 26: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/26.jpg)
30
IMPLICATIONS AND CONCLUSIONS
03
![Page 27: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/27.jpg)
31
CONCLUSIONS
Response rates are falling but the level of effort is increasing. Poses a number
of questions:
What can we realistically control in influencing response?
What is the cost-benefit analysis?
What are the best measures of performance?
![Page 28: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/28.jpg)
32
.aT Y
![Page 29: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/29.jpg)
© 2016 Ipsos. All rights reserved. Contains Ipsos' Confidential and Proprietary information and may
not be disclosed or reproduced without the prior written consent of Ipsos.
33
18th May, 2017
AN OPERATIONS’ PERSPECTIVE
F2F INTERVIEWING IN 2017
Sophie Ainsby (NatCen) & Mark McLaughlin (GfK)
With Caroline Baxter (Kantar Public) & Jeremy May (Ipsos MORI)
![Page 30: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/30.jpg)
34
EVIDENCE AND CONTRIBUTIONS SUPPLIED BY…
![Page 31: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/31.jpg)
35
INTERVIEWING DAYS AND INTERVIEWER PANEL YEAR ON YEAR 2010-2016
170,629
128,861
157,522
125,213
133,607 134,085
118,797
154,633
121,121
97,493
80,000
90,000
100,000
110,000
120,000
130,000
140,000
150,000
160,000
170,000
180,000
2010 2011 2012 2013 2014 2015 2016
Inte
rvie
win
g D
ays
Year
RP Days
RL Days
3,967
3,780
3,698
3,367
3,271
3,002
2,797
2,000
2,500
3,000
3,500
4,000
4,500
2010 2011 2012 2013 2014 2015 2016
Inte
rvie
wers
Year
Interviewers
![Page 32: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/32.jpg)
36
What are our operational challenges today?
F2F 2017
RecruitmentResponse to recruitment
advertising fluctuates with few
responses translating to new
panel members. We face
competition for candidates
from a wide range of
alternative employers offering
less challenging work with
guaranteed income. We have
tried alternative payment
methods but this has largely
been unsuccessful. We can still
build panels but it’s
increasingly challenging to do
so quickly.
RetentionNew people need an
introduction to F2F to help
maintain motivation and build
doorstep skills and experience
and this can be challenging
due to work types available.
Payment by results also has an
impact- on first projects
trainees may see their earning
potential limited.
Response Rate TargetsOn paper response rates remain
as they were 10 years ago.
However, saturation of customer
satisfaction and cold calling
makes target achievement more
challenging than it was. Research
frequently makes headlines, but
there isn’t a great deal of public
awareness regarding the vital role
they play.
![Page 33: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/33.jpg)
37
2016 Recruitment: from applicant to interviewer
Operations/Field Challenges Today
c17K
Respond to
adverts
C1.7K
Invited to Train
1180
Join
interviewer
panel
An average of 7% of
applicants actually became
interviewers in 2016
![Page 34: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/34.jpg)
38
JOINERS & LEAVERS 2012-2016
1444
10211117
1322
11801186
1593
1161
13711495
3698
33673271
3002
2797
0
500
1000
1500
2000
2500
3000
3500
4000
2012 2013 2014 2015 2016
Joiners
Leavers
Panel Size
32% 47% 35% 46% 53%
Leavers as % Total Panel
![Page 35: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/35.jpg)
39
INTERVIEWER LENGTH OF SERVICE
2006 vs 2016
0%
5%
10%
15%
20%
25%
30%
Under 1 1 2 3 4 5 to 9 10+
Years
2006
2016
![Page 36: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/36.jpg)
40
Over the last three years
c£3M spent on
mitigating recruitment
& retention challenges.
This cost is fully
absorbed and not
passed on.
MITIGATING THE ISSUES
Increased support for
new people; mentoring
programmes and
engagement measures
Trying to understand
and address our
interviewers’ challenges
via a cross company
interviewer survey
Cost Support Challenges
Operations Actions
![Page 37: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/37.jpg)
41
As an industry we are keen to better understand our interviewers’ challenges
Produced in conjunction with the MRS, the survey began in 2016. 2017
wave has just been completed
93% of interviewers feel pay levels influence why people choose to stop
interviewing
86% feel the challenges of getting participants to agree to be interviewed is
also a significant factor
Experienced interviewers really enjoy random probability work- 81% rate
interest of subject matter highly
INTERVIEWER SURVEY
![Page 38: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/38.jpg)
42
INTERVIEWER COMMENTS ON WHAT THEY FIND CHALLENGING
“It’s now very
difficult to get
people to take part
in a survey, much
more so than in the
past. Consequently if
being paid per
interview, the rate
per hour is less.”
“People are
becoming
untrusting- newer
interviewers get
disheartened.“
“Working unsocial
hours- especially in
winter.”
“Having to
continually work
evenings and
weekends”
![Page 39: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/39.jpg)
43
OVERCOMING OUR CHALLENGES
Response
Rate Targets
Achievable targets impact
on interviewer earnings
and on retention
With the correct resource
levels we can deliver
Raise awareness regarding
the part the public plays in
collection of this valuable
data
Focus on continued efforts,
make more of the value of
research
Encourage commissioners to
be champions of research
Resourcing and maintaining
a field force grows
increasingly expensive
Supporting this infrastructure
requires investment
Awareness &
Engagement
Costs
![Page 40: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/40.jpg)
44
As an industry we face challenges in attracting people to become interviewers
Interviewer retention also requires continued focus, with increasing
polarisation in experience levels making succession planning a challenge
We can still build panels but it has become more difficult and costly to do so
Raising awareness and encouraging promotion of research can help
Continuing to invest in incentive strategies for target samples
Response rate targets need to reflect underlying changes in society
We plan to continue as individual companies and collectively to work to
address this and to engage with interviewers
IN CONCLUSION
![Page 41: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/41.jpg)
45
.aT Y
![Page 42: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/42.jpg)
© 2016 Ipsos. All rights reserved. Contains Ipsos' Confidential and Proprietary information and may
not be disclosed or reproduced without the prior written consent of Ipsos.
46
18th May, 2017
AND POPULATION INFERENCE
RESPONSE RATES
Patten Smith (Ipsos MORI) & Joel Williams (Kantar Public)
![Page 43: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/43.jpg)
47
Aims
1. Examine relationship between survey response rates and quality of
population inferences
2. Describe relevant empirical work
3. Draw conclusions for future survey practice
Outline
Conceptual framework
Empirical findings
Practical implications for survey commissioners / practitioners
THIS PRESENTATION
![Page 44: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/44.jpg)
48
CONCEPTUAL
FRAMEWORK1
![Page 45: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/45.jpg)
49
Response rate
AAPOR standard definitions: survey response rate is proportion of eligible
population taking part
Quality of population inference
Use Total Survey Error (TSE) framework - generally accepted framework for
assessment of random probability survey quality
Inventory of survey processes and errors identified with each
Places errors on common metric: mean square error (MSE - error variance
from all sources + square of bias (from all sources))
Hard to measure TSE, but framework clarifies quality issues
DEFINING TERMS
![Page 46: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/46.jpg)
50
SURVEY PROCESSES AND TSE
1) Construct (the
information that
you seek)
2) Measurement
(ways to gather the
information) e.g
questions
3) Response
4) Edited response
Survey data
Validity
Measurement
error
Processing error
Sampling error
Coverage error
a) Defining target
population
b) Finding
Sampling frame
c) Drawing sample
d) Collecting data
from respondents
e) Making post-
survey
adjustments
Non-response error
Adjustment error
![Page 47: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/47.jpg)
51
Non-response bias: difference between estimates for survey respondents and
overall population
Non-response bias (NR bias) for variable Y increases with:
1. correlation between Y and likelihood of responding
2. non-response rate
NR bias:
• is variable-specific; wide range of NR bias values in a single survey
• increases as response rate decreases only if non-zero correlation between
response rates and variables
• but not by much unless correlation is substantial
RESPONSE RATE AND NON-RESPONSE BIAS
![Page 48: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/48.jpg)
52
Expressed algebraically, (P=sample member’s propensity to respond (0-1)):
NR Bias( 𝑌𝑅) ≈𝑐𝑜𝑣(𝑃𝑌)
𝑃=
𝐶𝑜𝑟𝑟(𝑃𝑌)×𝑆𝐷(𝑌)×𝑆𝐷(𝑃)
𝑃
RESPONSE RATE AND NON-RESPONSE BIAS
![Page 49: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/49.jpg)
53
EMPIRICAL EXAMINATIONS
OF RESPONSE RATE-TO-2BIAS RELATIONSHIP
![Page 50: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/50.jpg)
54
Relationship between RR and NR bias is open; requires empirical assessment
Two types of empirical study depending on availability of validation data
1. Where good estimate of the “true” value of a variable, compare this with
estimate for survey respondents; difference is estimate of total NR bias
– Aggregate population estimates
– Sample based estimates from “gold-standard” surveys
– Sample frame information about issued cases (direct or linked)
2. Otherwise, assess how survey estimates change with increasing fieldwork
effort (e.g. number of contact attempts, extent of reissuing); assumption - not
a logical necessity! - that this will measure relative NR bias
RESPONSE RATE AND NON-RESPONSE BIAS
![Page 51: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/51.jpg)
55
Ideally, measure total NR bias using benchmark validation data
Often measure relative NR bias because:
– Surveys usually commissioned to study the un-benchmarked!
– Benchmarks usually demographic / structural – nothing on opinions &
very little on behaviour (voting excepted)
– Estimating relative NR bias vs that obtained with a ‘gold standard’ RR
(not 100%!) is a more useful way of judging VFM of each component of
data collection effort
TOTAL AND RELATIVE NON-RESPONSE BIAS
![Page 52: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/52.jpg)
56
US: Groves et al (2006/08, US): meta-analysis of studies of total NR bias
UK:
– Census link study: total bias estimates (demographic only) for six surveys
– Hall et al (2013, UK) - relative bias and FW effort in three surveys
– Sturgis, Williams et al (2016, UK) - relative bias and FW effort in 541 non-
demographic variables in six surveys
– Williams (2016) - relative bias and FW effort in CSEW subpopulations
– Unpublished Ipsos analysis of relative bias in 2009-10 Citizenship Survey
EMPIRICAL EVIDENCE
![Page 53: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/53.jpg)
57
Meta-analyses (30 studies / 319
estimates; 59 studies / 959
estimates)
Measured absolute NR bias
Very low correlation between RR
and NR bias
Greater variation within studies
than between them
Preliminary efforts to identify
survey design features that target
NR bias
GROVES (2006 & 2008 W/PEYTCHEVA, US)
![Page 54: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/54.jpg)
58
Three NatCen Surveys: BSA, FRS and HSE (final RRs 51% - 61%)
Compared demographic and survey variables before and after extended FW
efforts (6+ calls to get contact or reissued after initial refusal)
Almost no significant differences between calibration-weighted estimates before
and after these efforts
(As noted earlier, substantial share of total effort allocated to this reissue phase)
HALL, BROWN, NICOLAAS AND LYNN (2013)
![Page 55: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/55.jpg)
59
Six surveys from 2011-14 (all Kantar Public); all (541) non-demographic items
Relative bias (distance from final estimate) measured after each FW effort level:
• 1 call (RRs=7-22%), 2 calls, 3 calls, 5 calls (cf. final RRs = 55-76%)
• Average error after call 1 = 1.6%pts; 1.1%pts after calibration
• Average error after call 2 = 1.0% pts; calibration barely improves on this
Study estimated that a 4-call rule leads to lowest # of total calls (=lowest cost)
On average, questions about beliefs (and attitudes) tended to respond to FW
effort slightly more than did questions about behaviour
Small number of variables with higher relative bias (internet use, freq. of being in
during day, freq. of going to pub) – but convergence after 3 calls
STURGIS, WILLIAMS & BRUNTON-SMITH (2016, UK)
![Page 56: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/56.jpg)
60
IMPACT OF 1ST, 2ND, 3RD, 4TH/5TH CALLS
![Page 57: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/57.jpg)
61
Subpopulation analysis in
CSEW (e.g. age group) – pre-
reissue v final (both calibrated)
Compares error distribution v
null hypothesis error
distribution
Subpopulation error levels
slightly smaller than total
population error levels (greater
within-group homogeneity?)
WILLIAMS (2016, UK)
![Page 58: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/58.jpg)
62
Occasionally extended FW
efforts can reduce NR bias
more substantially
2009-10 Citizenship survey
reissues (50>>>56% RR)
Formal and informal
volunteering prevalence
decreased by 1% point after
reissuing (CI: 0.5-1.5%)
Fits with US experience also
(e.g. see Pew, 2012)
THE POTENTIAL FOR EXCEPTIONS
![Page 59: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/59.jpg)
63
Absolute bias studies: relationship between RR and NR bias generally very weak
Relative bias studies: extended FW efforts increase RRs, but small impact on
survey estimates
Lack of relative bias should not lead to assumption of no bias but should lead to
questions over VFM of additional DC efforts
For some variables extended FW effort makes a difference – but hard to predict
in advance
Most UK work on relative bias based on f2f interview data – unclear whether
findings generalisable to data collection modes with lower ‘maximum’ RRs
BROAD CONCLUSIONS FROM EMPIRICAL EVIDENCE
![Page 60: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/60.jpg)
64
IMPLICATIONS FOR
SURVEY PRACTICE3
![Page 61: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/61.jpg)
65
Question 1: is this a repeat wave of longitudinal survey?
Response rate maximisation often more important because:
–Cumulative attrition produces shrinkage in sample numbers in key
groups
–Can lead to large losses in precision
QUESTIONS WE SHOULD ASK OURSELVES BEFORE SETTING
RESPONSE RATE TARGETS
![Page 62: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/62.jpg)
66
Question 2: how vulnerable to NR bias are your key variables?
Weighted survey estimates for most variables largely unaffected by
response rate variations within observed ranges
Rational starting assumption that response rate levels will make little
difference
For some variables declining RR has greater impact on NR bias: e.g.
volunteering and web usage
Do you have any such variables? Check literature / earlier survey data
QUESTIONS WE SHOULD ASK OURSELVES BEFORE SETTING
RESPONSE RATE TARGETS
![Page 63: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/63.jpg)
67
Question 3: will marginal reductions in NR bias compromise your
conclusions?
Even with highest achievable response rates, levels of absolute NR bias may
still be substantial
Often trends of greater interest than point estimates. NR bias probably
relatively constant over time - little impact on trend lines
QUESTIONS WE SHOULD ASK OURSELVES BEFORE SETTING
RESPONSE RATE TARGETS
![Page 64: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/64.jpg)
68
Question 4: are there better ways of reducing NR bias?
Blanket increase in RR beyond those obtained by reasonable good practice
poor method for reducing NR bias
Address NR bias in other ways? –reduce correlation between key variables
and response propensity?
Increase FW effort for poor responders differing on key variables using
supplementary variables / paradata to identify them
QUESTIONS WE SHOULD ASK OURSELVES BEFORE SETTING
RESPONSE RATE TARGETS
![Page 65: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/65.jpg)
69
Observed impact of response rates on survey estimates usually modest
Pursuit of highest possible response rates may not be best use of limited survey
resources – may be better ways of addressing survey errors
At times making considerable efforts to maximise response rates can be justified:
Waves 2 – n in longitudinal surveys;
where minimising NR bias critical to conclusions
OVERALL CONCLUSION
![Page 66: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/66.jpg)
70
.aT Y
![Page 67: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/67.jpg)
© 2016 Ipsos. All rights reserved. Contains Ipsos' Confidential and Proprietary information and may
not be disclosed or reproduced without the prior written consent of Ipsos.
71
18th May, 2017
FOR BETTER DATA
STRATEGIC ISSUES TO CONSIDER
Sharon Witherspoon (Campaign for Social Science)
with Debrah Harding (MRS) and Patten Smith (SRA)
![Page 68: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/68.jpg)
72
IMPLICATIONS AND RECOMMENDATIONS
1
2
3
Issues from the previous sessions
Current trends and observations
Initial thoughts for discussion
![Page 69: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/69.jpg)
73
ISSUES FROM THE
PREVIOUS SESSIONS1
![Page 70: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/70.jpg)
74
ISSUES FROM THE PREVIOUS SESSIONS
Face to face survey responses are declining
– Secular trend arising from social changes e.g. declining privacy, trust,
social capital, participant time
– Strains in the labour market e.g. interviewers harder to recruit and retain
– Importance of understanding methodology -- response rates vs. bias in
affecting descriptive estimates and generalising inference
– Requires understanding and EMPIRICAL data
– Choices in use of resources
– Better informed design
![Page 71: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/71.jpg)
75
CURRENT TRENDS AND
OBSERVATIONS2
![Page 72: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/72.jpg)
76
CURRENT TRENDS AND OBSERVATIONS: I
1. Increasing use of other data sources e.g. admin data
2. Primary data collection still needed for DEPTH of information
3. Different methodologies face different challenges for descriptive
accuracy and generalisability: focus here is on FACE TO FACE
4. Accept competition in the research supplier market
5. Social research projects do not yield large profits
6. SHARED interests in robust quality, design, and maintenance of
capacity
![Page 73: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/73.jpg)
77
CURRENT TRENDS AND OBSERVATIONS: II
7. Current tendering often results in specific research design BEFORE data
collection expertise/ experience brought in
Leads to less than optimal designs, lack of empirical pilots, dilemma of
non-compliant bids, lack of specification of key AIMS of research
8. Importance of linking AIMS with value for money in minimising total
survey error
9. Requires move away from viewing response rates as sole (contractual)
measure of data quality
Alternative approaches in tenders, procurement
Alternative approaches to research design
![Page 74: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/74.jpg)
78
SOME INITIAL THOUGHTS
FOR DISCUSSION3
![Page 75: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/75.jpg)
79
DISCUSSION POINTS: SPECIFIC STUDIES
1. Promote earlier expert engagement with research design
Identify key variables and aims
Commission / tender for work on research design – BEFORE main stage
design fixed
Use existing empirical data to consider total survey error
– What additional data – admin. data, paradata, etc. – available to test
impact of non-response?
Pilot different designs if helpful
Will affect timetable and budget allocation, possibly budget size, but
with better outcome
![Page 76: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/76.jpg)
80
DISCUSSION POINTS: GENERAL & STRATEGIC
2. Strategic engagement with data regulators (ICO) about data linkage, use of
department data sets, consents, opt-ins/outs (stressing data protection but
also consent protections)
3. Strategic engagement of commissioners, data collection experts and
procurement to move away from specification of response rates and focus
on agreed PROCESSES to address non-response bias (and other sources of
error)
Better methodologically
Better allocation of resources
Better ethically / ‘politically’
![Page 77: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/77.jpg)
81
OVER TO WIDER
DISCUSSION4
![Page 78: ROUNDTABLE ON RESEARCH DESIGN, DATA ...As an industry we are keen to better understand our interviewers’ challenges Produced in conjunction with the MRS, the survey began in 2016](https://reader033.vdocument.in/reader033/viewer/2022052012/6027c7d9be371935887f176b/html5/thumbnails/78.jpg)
82
.aT Y