Survey Data Analysis in Stata: Jeff Pitblado

Survey Data Analysis in Stata
Jeff Pitblado
Associate Director, Statistical Software
StataCorp LP
Stata Conference DC 2009
J. Pitblado (StataCorp) Survey Data Analysis DC 2009 1 / 44
Outline
1
Types of data
2
Survey data characteristics
3
Variance estimation
4
Estimation for subpopulations
5
Summary
Why survey data?
Collecting data can be expensive and time consuming.
Consider how you would collect the following data:
Smoking habits of teenagers
Birth weights for expectant mothers with high blood pressure
Using stages of clustered sampling can help cut down on the
expense and time.
Types of data
Simple random sample (SRS) data
Observations are "independently" sampled from a data generating
process.
Typical assumption: independent and identically distributed (iid)
Make inferences about the data generating process
Sample variability is explained by the statistical model attributed to
the data generating process
Standard data
Well use this term to distinguish this data from survey data.
Types of data
Correlated data
Individuals are assumed not independent.
Cause:
Observations are taken over time
Random effects assumptions
Cluster sampling
Treatment:
Time-series models
Longitudinal/panel data models
cluster() option
Types of data
Survey data
Individuals are sampled from a xed population according to a survey
design.
Distinguishing characteristics:
Complex nature under which individuals are sampled
Make inferences about the xed population
Sample variability is attributed to the survey design
Types of data
Standard data
Estimation commands for standard data:
proportion
regress
Well refer to these as standard estimation commands.
Survey data
Survey estimation commands are governed by the svy prex.
svy: proportion
svy: regress
svy requires that the data is svyset.
Single-stage syntax
svyset
_
psu
_
weight
_
,strata(varname)fpc(varname)
Primary sampling units (PSU)

Sampling weights pweight
Strata
Finite population correction (FPC)
Sampling unit
An individual or collection of individuals from the population that can
be selected for observation.
Sampling groups of individuals is synonymous with cluster
sampling.
Cluster sampling usually results in inated variance estimates
compared to SRS.
Sampling weight
The reciprocal of the probability for an individual to be sampled.
Probabilities are derived from the survey design.
Sampling units
Strata
Typically considered to be the number of individuals in the
population that a sampled individual represents.
Reduces bias induced by the sampling design.
Strata
In stratied designs, the population is partitioned into well-dened
groups, called strata.
Sampling units are independently sampled from within each
stratum.
Stratication usually results in smaller variance estimates
compared to SRS.
Finite population correction (FPC)
An adjustment applied to the variance due to sampling without
replacement.
Sampling without replacement from a nite population reduces
sampling variability.
Example: svyset for single-stage designs
Population 1000
SRS sample 200
Cluster sample 20 (208 obs)
Stratied sample 198
Stratied-cluster sample 20 (215 obs)
Multistage syntax
svyset psu
_
weight
_
, strata(varname) fpc(varname)
_
|| ssu
_

_
|| ssu
_

...
Stages are delimited by ||
SSU secondary/subsequent sampling units
FPC is required at stage s for stage s +1 to play a role in the
linearized variance estimator
Poststratication
A method for adjusting sampling weights, usually to account for
underrepresented groups in the population.
Adjusts weights to sum to the poststratum sizes in the population
Reduces bias due to nonresponse and underrepresented groups
Can result in smaller variance estimates
Syntax
svyset ... poststrata(varname) postweight(varname)
Example: svyset for poststratication
Strata with a single sampling unit
Big problem for variance estimation
Consider a sample with only 1 observation
svy reports missing standard error estimates by default
Finding these lonely sampling units
Use svydes:
Describes the strata and sampling units
Helps nd strata with a single sampling unit
Example: svydes
Handling lonely sampling units
1
Drop them from the estimation sample.
2
svyset one of the ad-hoc adjustments in the singleunit()
option.
3
Somehow combine them with other strata.
Certainty units
Sampling units that are guaranteed to be chosen by the design.
Certainty units are handled by treating each one as its own
stratum with an FPC of 1.
Variance estimation
Stata has three variance estimation methods for survey data:
Linearization
Balanced repeated replication
The jackknife
Variance estimation
Linearization
A method for deriving a variance estimator using a rst order Taylor
approximation of the point estimator of interest.
Foundation: Variance of the total estimator
Syntax
svyset ...
_
vce(linearized)
Delta method
Huber/White/robust/sandwich estimator
Variance estimation
Total estimator Stratied two-stage design
y
hijk
observed value from a sampled individual
Strata: h = 1, . . . , L
PSU: i = 1, . . . , n
h
SSU: j = 1, . . . , m
hi
Individual: k = 1, . . . , m
hij
Y =
w
hijk
y
hijk
V(
Y) =
h
(1 f
h
)
n
h
n
h
1
i
(y
hi
y
h
)
2
+
h
f
h
i
(1 f
hi
)
m
hi
m
hi
1
j
(y
hij
y
hi
)
2
Variance estimation
y
hijk
Strata: h = 1, . . . , L
PSU: i = 1, . . . , n
h
SSU: j = 1, . . . , m
hi
hij
Y =
w
hijk
y
hijk
V(
Y) =
h
(1 f
h
)
n
h
n
h
1
i
(y
hi
y
h
)
2
+
h
f
h
i
(1 f
hi
)
m
hi
m
hi
1
j
(y
hij
y
hi
)
2
Variance estimation
y
hijk
Strata: h = 1, . . . , L
PSU: i = 1, . . . , n
h
SSU: j = 1, . . . , m
hi
hij
Y =
w
hijk
y
hijk
V(
Y) =
h
(1 f
h
)
n
h
n
h
1
i
(y
hi
y
h
)
2
+
h
f
h
i
(1 f
hi
)
m
hi
m
hi
1
j
(y
hij
y
hi
)
2
Variance estimation
Example: svy: total
Variance estimation
Linearized variance for regression models
Model is t using estimating equations.
G() is a total estimator, use Taylor expansion to get

V(
).
G() =
j
w
j
s
j
x
j
= 0
V(
) = D
V{
G()}|
=
b

Variance estimation
Linearized variance for regression models
Model is t using estimating equations.
G() is a total estimator, use Taylor expansion to get

V(
).
G() =
j
w
j
s
j
x
j
= 0
V(
) = D
V{
G()}|
=
b

Variance estimation
Example: svy: logit
Variance estimation
Balanced repeated replication
For designs with two PSUs in each of L strata.
Compute replicates by dropping a PSU from each stratum.
Find a balanced subset of the 2
L
replicates. L r < L +4
The replicates are used to estimate the variance.
Syntax
svyset ... vce(brr)
_
mse

Variance estimation
BRR variance formulas
point estimates
(i )
i th replicate of the point estimates
(.)
average of the replicates
Default variance formula:
V(
) =
1
r
r
i =1
{
(i )

(.)
}{
(i )

(.)
}
Mean squared error (MSE) formula:
V(
) =
1
r
r
i =1
{
(i )

}{
(i )

}

Variance estimation
Example: svy brr: logit
Variance estimation
The jackknife
A replication method for variance estimation. Not restricted to a
specic survey design.
Delete-1 jackknife: drop 1 PSU
Delete-k jackknife: drop k PSUs within a stratum
Syntax
svyset ... vce(jackknife)
_
mse

Variance estimation
Jackknife variance formulas
(h,i )
replicate of the point estimates from stratum h, PSU i
h
average of the replicates from stratum h
m
h
= (n
h
1)/n
h
delete-1 multiplier for stratum h
Default variance formula:
V(
) =
L
h=1
(1 f
h
) m
h
n
h
i =1
{
(h,i )

h
}{
(h,i )

h
}
Mean squared error (MSE) formula:
V(
) =
L
h=1
(1 f
h
) m
h
n
h
i =1
{
(h,i )

}{
(h,i )

}

Variance estimation
Example: svy jackknife: logit
Variance estimation
Replicate weight variable
A variable in the dataset that contains sampling weight values that
were adjusted for resampling the data using BRR or the jackknife.
Typically used to protect the privacy of the survey participants.
Eliminate the need to svyset the strata and PSU variables.
Syntax
svyset ... brrweight(varlist)
svyset ... jkrweight(varlist
_
, ... multiplier(#)
)
Focus on a subset of the population
Subpopulation variance estimation:
Assumes the same survey design for subsequent data collection.
The subpop() option.
Restricted-sample variance estimation:
Assumes the identied subset for subsequent data collection.
Ignores the fact that the sample size is a random quantity.
The if and in restrictions.
Total from SRS data
Data is y
1
, . . . , y
n
and S is the subset of observations.
j
(S) =
_
1, if j S
0, otherwise
Subpopulation (or restricted-sample) total:
Y
S
=
n
j =1
j
(S)w
j
y
j
Sampling weight and subpopulation size:
w
j
=
N
n
, N
S
=
n
j =1
j
(S)w
j
=
N
n
n
S
Variance of a subpopulation total
Sample n without replacement from a population comprised of the N
S
subpopulation values with N N
S
additional zeroes.
V(
Y
S
) =
_
1
n
N
_
n
n 1
n
j =1
_
j
(S)y
j

1
n
Y
S
_
2
Variance of a restricted-sample total
Sample n
S
without replacement from the subpopulation of N
S
values.
V(
Y
S
) =
_
1
n
S
N
S
_
n
S
n
S
1
n
j =1
j
(S)
_
y
j

1
n
S
Y
S
_
2
Example: svy, subpop()
Summary
1
Use svyset to specify the survey design for your data.
2
Use svydes to nd strata with a single PSU.
3
Choose your variance estimation method; you can svyset it.
4
Use the svy prex with estimation commands.
5
Use subpop() instead of if and in.
References
Levy, P. and S. Lemeshow. 1999.
Sampling of Populations. 3rd ed.
New York: Wiley.
StataCorp. 2009.
Survey Data Reference Manual: Release 11.
College Station, TX: StataCorp LP.

Survey Data Analysis in Stata: Jeff Pitblado

Uploaded by

Copyright:

Available Formats

Survey Data Analysis in Stata: Jeff Pitblado

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Survey Data Analysis in Stata: Jeff Pitblado

Uploaded by

Copyright:

Available Formats

Survey Data Analysis in Stata

Primary sampling units (PSU)

G() is a total estimator, use Taylor expansion to get

J. Pitblado (StataCorp) Survey Data Analysis DC 2009 30 / 44

G() is a total estimator, use Taylor expansion to get

J. Pitblado (StataCorp) Survey Data Analysis DC 2009 30 / 44

J. Pitblado (StataCorp) Survey Data Analysis DC 2009 32 / 44

Mean squared error (MSE) formula:

J. Pitblado (StataCorp) Survey Data Analysis DC 2009 33 / 44

J. Pitblado (StataCorp) Survey Data Analysis DC 2009 35 / 44

Mean squared error (MSE) formula:

J. Pitblado (StataCorp) Survey Data Analysis DC 2009 36 / 44

You might also like