HILDA Top-up Sample in Wave 11 Nicole Watson

HILDA Top-up Sample in Wave 11
Nicole Watson
HILDA Deputy Director, Survey Methodology
Why have a top-up
top up sample?
 Current HILDA sample selected in 2001 (n=7800 households)
– Initial sample followed (Permanent Sample Members)
– ‘Following rules’ add others to the sample
• Births to PSMs
• Other parent of births to PSMs
• Immigrants arriving after 2001 (from wave 9)
 First top-up in 2011. Why?
– Undercoverage of population due to immigration
– Various options canvassed in 2006 and decided on general
top-up
• ‘Refresh’
Refresh the ongoing sample
sample. Allows us to compare characteristics of
samples and study cumulative effect of attrition
• Larger combined sample size (n=7300+2000 households)
www.melbourneinstitute.com
Undercoverage over time (% Australian population)
www.melbourneinstitute.com
Top-up
Top
up sample design
 Select 3250 d
dwellings,
ellings 8% out
o t of scope,
scope
66% response = 2000 responding households
 Similar sample design as wave 1
 3-stage clustered design
– Stage 1: 125 Census Collection Districts with
probability
b bilit proportional
ti
l tto size
i ((number
b occupied
i d
dwellings from 2006 Census)
– Stage
g 2: Approx.
pp
24 in-scope
p dwellings
g selected in
each CD
– Stage 3: Up to 3 households selected in each
dwelling
www.melbourneinstitute.com
Overlap of the underlying populations
www.melbourneinstitute.com
Face-to-face
Face
to face interviewers
Experienced
E
i
d
HILDA
interviewers
New-to-HILDA
interviewers
Total
interviewers
Both ongoing and
top-up
p p
70
5
75
Ongoing only
50
15
65
p p only
y
Top-up
5
0
5
Wave 11 total
125
20
145
Wave 10 total
117
15
132
Sample
www.melbourneinstitute.com
Training schedule
July
Aug
Sept
Period 1
Oct
Nov
Dec
Period 2
Jan Feb
Period 3
Top-up
Ongoing
W10
Training
Fieldwork
www.melbourneinstitute.com
Questionnaire design
 Household Form similar to
wave 1
–
–
–
–
–
Name, sex, date of birth
E li h llanguage ability
English
bilit
Employment status
Long term health condition
Relationships
 Dwelling
e g obse
observations
a o so
on
first approach
– type of dwelling
– external condition of the
dwelling
– likelihood that dwelling
contains children under 15
– presence of security
features
•
•
•
•
•
locked gate
security guard / doorman
security door
dangerous dog
no junk mail / hawkers
sign
• grab rails / ramp access
• bars on windows
• roller shutters
– condition of the garden
– type
yp of road the dwelling
g
is on
www.melbourneinstitute.com
Questionnaire design, cont’d
cont d
 Household
H
h ld Q
Questionnaire
ti
i – same as ongoing
i W11
 Person Questionnaire – excludes
– Section L on retirement
– Section M on intentions and plans
– Parts of Section G on family formation
• G27 to G67 on caring for children, grandchildren, and desires and
preferences for more children
– Section E on other labour market activity
 Self-Completion Questionnaire – same as ongoing W11
melbourneinstitute.com/hilda/questionnaires/q11.html
www.melbourneinstitute.com
Interview length
60
50
Miinutes
40
30
Age <=19
Age 20+
20
10
0
1
2
3
4
5
6
7
8
9
10
Wave
www.melbourneinstitute.com
Respondent material
www.melbourneinstitute.com
Weighting methodology
 Two approaches
– Combine estimates – identify  which minimises
variance for combined estimate ( yorig + (1- ) ytopup)
of key variable
• Choice of key variable or average across several
• Weights more variable
– Pool samples – probability of selection
• Need to know probability of individual being selected in
sample they were not selected in
• Weights are less variable
www.melbourneinstitute.com
Weights to be provided
Release 1
2
3
…
10
11
12
13
14
Crosssection
2
3
…
10
11
12
13
14
1-2
1-3
2-3
…
…
…
1-10
2-10
1-11
2-11
10-11
10
11
1-12
2-12
10-12
10
12
11-12
1-13
2-13
10-13
10
13
11-13
12-13
1-14
2-14
10-14
10
14
11-14
12-14
13-14
1&2
1&3
2&3
…
…
…
1&10
2&10
1&11
2&11
10&11
1&12
2&12
10&12
11&12
1&13
2&13
10&13
11&13
12&13
1&14
2&14
10&14
11&14
12&14
13&14
1
Longitudinal
Continuous
Pair
Red=Includes top-up
www.melbourneinstitute.com
Timeframe for top-up
top up
Activity
y
Time
Sample selection and preparation
Oct 2010 – Jun 2011
Fieldwork
July 2011 – Mar 2012
Technical paper on survey design
Sept 2011
Data preparation
Apr - Nov 2012
Data release
Nov 2012
Technical paper on weighting
methodology
gy
Nov 2012
Discussion paper on comparison of
ongoing and top-up samples
Dec 2012
www.melbourneinstitute.com