multistage sampling

29
1 Multistage Sampling

Upload: garron

Post on 12-Jan-2016

64 views

Category:

Documents


0 download

DESCRIPTION

Multistage Sampling. Outline. Features of Multi-stage Sample Designs Selection probabilities in multi-stage sampling Estimation of parameters Calculation of standard errors Efficiency of multi-stage samples. Introduction. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Multistage Sampling

1

Multistage Sampling

Page 2: Multistage Sampling

2

Outline

Features of Multi-stage Sample DesignsSelection probabilities in multi-stage samplingEstimation of parametersCalculation of standard errorsEfficiency of multi-stage samples

Page 3: Multistage Sampling

3

Introduction

Multi-stage sampling means what its name suggests -> there are multiple stages in the sampling processThe number of stages can be numerous, although it is rare to have more than 3For this topic we will concentrate on two-stage sampling– Also known as subsampling

Page 4: Multistage Sampling

4

Sampling Units in Multi-stage Sampling

First-stage sampling units are called primary sampling units or PSUs.Second-stage sampling units are called secondary sampling units or SSUs.Last-stage sampling units are called ultimate sampling units or USUs.

Page 5: Multistage Sampling

5

4-stage Sampling (example)

Villages EAs Dwelling Persons

B

C

AA

Page 6: Multistage Sampling

6

Your Examples

Estimation DomainsStratificationNumber of stagesSampling units for each stageSample selection scheme in each stageSampling frames used in each stage

Page 7: Multistage Sampling

7

Example: Maldives HIES 2002

Page 8: Multistage Sampling

8

Two-Stage Sampling

Stage One. Select sample of clusters from population of clusters.– Using any sampling scheme, usually: SRSWOR, PPSWR,

LSSStage Two. Select sample of elements within each of the sample clusters.– Language: also referred to as ‘subsample’ of elements within

a cluster– Subsampling can be done also using any sampling scheme

Page 9: Multistage Sampling

9

Most Large-Scale Surveys UseMulti-stage Sampling Because …

Sampling frames are available at higher stages but not for the ultmate sampling units. Construction of sampling frames at each lower stage becomes less costly.Cost efficiency with use of clusters at higher stages of selectionFlexibility in choice of sampling units and methods of selection at different stagesContributions of different stages towards sampling variance may be estimated separately

Page 10: Multistage Sampling

10

Probabilities of Selection

Probability that an element in the population is selected in a 2-stage sample is the product of– Probability that the cluster to which it belongs is

selected at the first stage– Probability that the element is selected at the second

stage given that the cluster to which it belongs is selected at the first stage

Page 11: Multistage Sampling

11

Example: Two-Stage Samples

Page 12: Multistage Sampling

12

Estimation Procedures: Illustrations

• SRS at stage 1 and SRS at stage 2• SRS at stage 1 and LSS at stage 2 (b from B)• PPSWR at stage 1 and SRS at stage 2 (b from B)

Page 13: Multistage Sampling

13

SRS – SRS: Estimation of Total

a

1

a

1s yB

a

AY

a

AY

2

Variance of Estimator

2

1

222 1111

S

BbB

a

AS

AaAYV

A

b

2

1

2

1

1

A

b YYA

S

2

1

2

1

1

B

YYB

S

Estimator of Total

Page 14: Multistage Sampling

14

SRS – SRS: Variance of Estimator

2

1

222 1111

S

BbB

a

AS

AaAYV

A

b

Total variability = Variability among PSUs + Variability of SSUs

Sources of Variation = {PSUs} + {SSUs}

Page 15: Multistage Sampling

15

SRS-SRS: Estimating Variance

2 2 2 2

1

1 1 1 1ˆˆa

b

Av Y A s B s

a A a b B

2

1 1

2 1

1

1

a a

b Ya

Ya

s

Estimator of Variance of Estimator for Total

2

2

1 1

1 1

1

b b

s y yb b

Page 16: Multistage Sampling

16

Each PSU has same number of elements, BSubsample of b elements is selected

a

ya

Y1

1

2 2

ˆ 1 1b wS Sa bV Y

A a B ab

where 2

2

1

1

1

A

bS Y YA

SRS-SRS: Estimating a Mean

22 2 2

1 1

1 1;

1

A B

wS S S YA B

Y

Page 17: Multistage Sampling

17

2

2

1

1 ˆ1

a

bs y Ya

… with variance estimate

2 2ˆˆ 1 1b ws sa b

v YA a B ab

22 2 2

1 1

1 1;

1

a b

ws s s y ya b

Page 18: Multistage Sampling

18

SRS-SRS: Population Mean (1)PSU’s have unequal sizes

11

1ˆa

Y B yaB

2

2 21 1 2

1

1 1 1 1 1ˆ( ) ( ) ( )A

b

BV Y S S

a A aA B b B

2

2 21 1 2

1

1 1 1 1 1ˆˆ( ) ( ) ( )a

b

Bv Y s s

a A aA B b B

Page 19: Multistage Sampling

19

SRS-SRS: Population Mean (2)PSU’s have unequal sizes

21

1ˆa

Y ya

21

1ˆ( ) ( )A

bias Y B B YAB

2 22 2

1

1 1 1 1 1ˆ( ) ( ) ( )A

bV Y S Sa A aA b B

Page 20: Multistage Sampling

20

SRS-SRS: Population Mean (3)PSU’s have unequal sizes

13

1

ˆ

a

a

B yY

B

2 2 23 3 2

1

1 1 1 1 1ˆ( ) ( ) ( )A

bV Y S B Sa A B aA b B

Page 21: Multistage Sampling

21

1

1ˆa

Y ya

2 2

1

1 1 1 1 1ˆ 1 1 1A

bV Y s S ba A ab B A

2ˆˆ bsv Y

a

SRS-LSS: Estimation of Mean

Page 22: Multistage Sampling

22

1

ˆ1ˆ ˆ;a Y

Y Y B ya p

2

1

ˆ1ˆ ˆˆ1

a Yv Y Y

a a p

PPSWR-SRS: Estimation of Total

Page 23: Multistage Sampling

23

Design Effect for 2-stage Sample

If is positive, the design effect decreases as the subsample size b decreases. For fixed n=ab, the smaller the sub-sample size and, hence, the larger the number of clusters included in the sample, the more precise is the sample mean.

ˆ( )ˆ( ) 1 ( 1)ˆ( )

cluclu

srs

V YDEFF Y B

V Y

Page 24: Multistage Sampling

24

Designing a Cluster Sample

What overall precision is needed?What size should the psus be?How many ssus should be sampled in each psu selected for the sample?How many psus should be sampled?

Page 25: Multistage Sampling

25

Choosing psu Size

Often a natural unit– not much choiceLarger the psu size, more variability within a psu– ICC is smaller for large psu compared to small psu– but, if psu size is too large, less cost efficient

Need to study relationship between psu sizes and ICC and costs

Page 26: Multistage Sampling

26

Optimum Sample Sizes (1)

Goal: get the most information (and hence, more statistically efficient) for the least costIllustrative example: PSUs with equal sizes, SRSWOR at both stages

Page 27: Multistage Sampling

27

Optimum Sample Sizes (2)

Variance function2 2b wS S

Va ab

0 1 2C C ac abc Cost function

Minimize V subject to given cost C*

Page 28: Multistage Sampling

28

Optimum Sample Sizes (3)

Minimize V subject to given cost C*

Optimum a=a* and b=b* 2 21 2* /w bb c S c S

2*

01

2 21 2

( )

*

b

b w

SC C

ca

c S c S

Page 29: Multistage Sampling

29

Optimum Sample Sizes (4)Optimum b=b*

21 1

22 2

1* w

b

Sc c rohb

c S c roh