Computing optimal sampling designs for two-stage studies



40


Stata Technical Bulletin


STB-58


in the resulting 12 (2x2x3) different strata are given in the following table.

sex

weight

alive

deceased

male

<60

160^^

8-

60-70

1,083

33

>70

5,418

103

female

<60

440

18-

60-70

407

26

>70     378       14

We decided to sample 10 pilot observations from each stratum. One stratum had only 8 observations available, so all of
these were included in the pilot sample. The resulting pilot data are available in the dataset wtpilot.dta, which can be loaded
as follows and the coding command run to see in which order we should enter the vector of prevalences for the strata.

. use wtpilot

. coding mort sex wtcat

grp.yz

mort

sex

Wtcat

grp_z

nobs

1

0

0

1

1

10

2

0

0

2

2

10

3

0

0

3

3

10

4

0

1

1

4

10

5

0

1

2

5

10

6

0

1

3

6

10

7

1

0

1

1

8

8

1

0

2

2

10

9

1

0

3

3

10

10

1

1

1

4

10

11

1

1

2

5

10

12

1

1

3

6

10

for functions requiring first stage sample sizes∕prevalences

enter these in the order of grp_yz

This tells us that our prevalence vector should be

. matrix prev=(0.02,.134,.670,.054,.05,.047,.001,.004,.013,.002,.003,.002)'

and we can find the design which will optimize (i.e., minimize) the variance of Ivedb subject to a total budget of £10,000.

. optbud mort sex-surg,first(sex wtcat)


prev(prev) var(7) b(10000) cl(2) c2(15)


the second stage sample sizes

----------+.

group(mor I
t sex      I

wtcat)     I

----------+.

1

2

3

4

5

6

7

8

9

10

11

12
----+.

Freq.


10

10

10

10

10

10

8

10

10

10

10

10


please check the sample sizes !

grp.yz

mort

sex

Wtcat

grp_z

prev

n2-pilot

1

0

0

1

1

.02

10

2

0

0

2

2

.134

10

3

0

0

3

3

.67

10

4

0

1

1

4

.054

10

5

0

1

2

5

.05

10

6

0

1

3

6

.047

10



More intriguing information

1. Manufacturing Earnings and Cycles: New Evidence
2. The name is absent
3. Commitment devices, opportunity windows, and institution building in Central Asia
4. SOCIOECONOMIC TRENDS CHANGING RURAL AMERICA
5. Giant intra-abdominal hydatid cysts with multivisceral locations
6. A Hybrid Neural Network and Virtual Reality System for Spatial Language Processing
7. The name is absent
8. Research Design, as Independent of Methods
9. AN ECONOMIC EVALUATION OF THE COLORADO RIVER BASIN SALINITY CONTROL PROGRAM
10. THE RISE OF RURAL-TO-RURAL LABOR MARKETS IN CHINA