Module 8 Survey sample builder

Report
Global Programme to Eliminate Lymphatic Filariasis (GPELF)
Training in monitoring and epidemiological assessment of mass
drug administration for eliminating lymphatic filariasis
TAS
Module 8 Survey sample builder
Module 8 Survey sample builder (SSB)
Learning objectives
By the end of this module, you should understand:
1. how to use the survey sample builder to:
 determine the survey design
 select random clusters and children or households
2. the protocol for TAS
Slide 2
Module 8 Survey sample builder (SSB)
Overview
 The survey sample builder
 Determining the survey design
 Selecting randomized clusters
 Selecting randomized children or households
 Protocol for TAS
Slide 3
Module 8 Survey sample builder (SSB)
Survey sample builder
The survey sample builder is a Microsoft Excel tool that can be used to (i)
automate calculations for determining the appropriate survey design and (ii)
facilitate random selection of clusters and children or households from a list
of randomized numbers.
Evaluation unit
School
Survey
design
Cluster-based
Systematic
Community
Census
Cluster-based
Systematic
Sample size and critical cut-off
Random
selection
Slide 4
Enumeration area
School
Child
Child
Module 8 Survey sample builder (SSB)
Household
Household
Census
Preparation before sample selection
School-based survey
 Communicate with the ministry of education to obtain a
comprehensive list of all primary schools in the defined
EU.

Ideally, the list of schools will be ordered according to
geographical proximity rather than alphabetically. This will allow
better geographical representation of the selected schools in
the EU.
 Obtain an average non-response rate for the schools in
the EU, if available.

Slide 5
The non-response rate gives an estimate of the non-participant
rate and should include school absenteeism, refusals and
inability to collect sufficient blood for diagnostic testing.
Module 8 Survey sample builder (SSB)
Preparation before sample selection
Community-based household survey
 Census enumeration areas are recommended as clusters
if cluster sampling is used; enumeration areas are usually
the smallest areas for which census data are available.
 Obtain a list of all enumeration areas in the EU.

Ideally, the enumeration areas will be listed according to
geographical proximity rather than alphabetically.
 Obtain enumeration area census maps.

Slide 6
These maps can often be obtained from the census department
or bureau of statistics. A fee may be charged to obtain these
maps.
Module 8 Survey sample builder (SSB)
Determining the survey design
Available for download at http://www.filariasis.us./resources.html
Slide 7
Module 8 Survey sample builder (SSB)
Determining the survey design
Slide 8
Module 8 Survey sample builder (SSB)
Determining the survey design
Slide 9
Module 8 Survey sample builder (SSB)
Determining the survey design
Survey sampling
method determined
by data entered into
programme
Slide 10
Module 8 Survey sample builder (SSB)
Selecting randomized clusters and
children or households
Once the appropriate survey design has been determined, the survey
sample builder facilitates selection of clusters and children or households
by generating a list of randomized numbers.
Evaluation unit
School
Survey
design
Cluster-based
Systematic
Community
Census
Cluster-based
Systematic
Sample size and critical cut-off
Random
selection
Slide 11
School
Child
Module 8Child
Survey sample builder
(SSB)
Enumeration area
Household
Household
Census
Selecting randomized clusters
Cluster sampling requires a minimum of 30 schools or
enumeration areas in which children are selected.
1. Obtain a comprehensive list of all primary sampling units (i.e.
schools, enumeration areas) in the evaluation unit.
2. Number them, preferably in order of geographical proximity as
opposed to alphabetical order to achieve better geographical
distribution of the evaluation unit.
3. Once all sampling units have been assigned a number, click the
“Randomize” button.
4. Once the survey design has been determined, a list of random
numbers (e.g. 10, 23, 36, 49 …) is generated. Select the schools or
enumeration areas in the list according to the random numbers (i.e.
10th, 23rd, 36th, 49th …).
Slide 12
Module 8 Survey sample builder (SSB)
Selecting randomized clusters
Directions: Obtain a comprehensive list of all the primary sampling units (i.e. Schools or EAs) in the EU
and number them, preferably in order of geographic proximity. Once all sampling units have been
assigned a number, double check that the values listed below are correct and click the "Randomize"
button. When you are finished you can click "Return to Previous Page" to go back.
Total number of EAs:
No. of EAs (clusters) to be sampled:
487
37
Randomize
Return to Previous Page
The following schools or enumeration areas have been
selected for your study:
10
23
36
49
62
76
89
102
….
Slide 13
Module 8 Survey sample builder (SSB)
A list of random numbers
can be generated to
facilitate the selection of
clusters to be included.
Selecting randomized clusters
 Selection of 5–10 additional clusters is recommended as a
contingency plan if the sample size falls considerably
short of the target after all the clusters have been
surveyed.
 Additional clusters should be selected from a list of all remaining
clusters, excluding those that have already been chosen. The
same random selection process used for the original clusters
should be used.
 The survey sample builder allows selection of additional clusters.
Slide 14
Module 8 Survey sample builder (SSB)
Selecting randomized children or households
After selection of randomized clusters in cluster sampling or after choosing
systematic sampling, children (in school-based surveys) or households (in
community-based surveys) are selected randomly.
 The survey sample builder calculates the appropriate sampling interval.
 The survey sample builder also calculates a random starting number
and the sampling interval (inverse of the sampling fraction) to generate
two numbered lists (A and B) to guide the actual selection of
schoolchildren or households
 For school-based surveys, lists A and B are used to select the children to be
tested.
 For community-based surveys, lists A and B are used to select the houses in
the enumeration areas to be sampled.
Slide 15
Module 8 Survey sample builder (SSB)
Selecting randomized children or households
In this example, the team chose list A; therefore, the first, third, fourth, fifth …
house on the list or map will be selected.
Every 6- and/or 7-year-old child in the house should be tested. The same list
should be used throughout the survey.
Slide 16
Module 8 Survey sample builder (SSB)
Example 1
 Net school enrolment ratio: 78%
 Primary vector: Culex
 Population of students in target grades: 18 945
 Total number of primary schools: 386
 Estimated non-response rate: 15%
Slide 17
Module 8 Survey sample builder (SSB)
Example 1. Survey design
Country:
Name of EU:
Example 1
Example 1
Primary Vector:
Anopheles, Culex, or Mansonia
Demographic data on EU (provided by user)
Population of students in target grade(s):
Total number of schools:
18,945
386
Avg. number of students in target grade(s) per
school:
49
*Note, if any of the above information is incorrect, click on the 'Back to "Start Here"' button and then click
'edit EU information'
Cluster Survey of School Entrants
Survey Sampling Methodology (calculated by program)
Sample Size for Cluster Design
Number of Clusters
1
1,552
38
Sampling Fraction (of children within the schools)1 Test all children in the target grade(s)
Slide 18
Sampling Interval (of children within the schools)1 N/A
Critical Cut-off (maximum # positive ICT/Brugia
Rapid results allowable for country to "Pass")
18
Module 81 Survey sample builder (SSB)
Based on a 15% absentee rate
Example 1. Randomized cluster selection
Directions: Obtain a comprehensive list of all the primary sampling units (i.e. schools or enumeration areas) in the
evaluation unit, and number them, preferably in order of geographical proximity. Once all sampling units have
been assigned a number, double-check that the values listed below are correct, and click the "Randomize" button.
When you have finished, you can click "Return to previous page”.
Total number of schools:
No. of schools (clusters) to be sampled:
386
38
The following schools have been selected for your study:
8
18
29
39
49
59
69
79
90
100
110
…
Slide 19
Module 8 Survey sample builder (SSB)
Example 1. Randomized cluster selection
The following schools
have been selected for
your study:
8
18
29
39
49
59
69
79
90
100
110
…
Slide 20
Module 8 Survey sample builder (SSB)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
…
School name
Woodridge Elementary
Lakeside Elementary
Shadow Rock Elementary
Austin Elementary
Idlewood Elementary
Henderson Mill Elementary
Stone Mill Elementary
Rockland Elementary
Sage Elementary
Oak Grove Elementary
Brockett Elementary
Princeton Elementary
Chestnut Elementary
Rockbridge Elementary
Dresden Elementary
Midvale Elementary
Columbia Elementary
Pine Ridge Elementary
Flatrock Elementary
Snapfinger Elementary
Example 2
 Net school enrolment ratio: 68%
 Primary vector: Anopheles
 Population of students in target grades: 23 128
 Total number of enumeration areas: 284
 Estimated non-response rate: 15%
Slide 21
Module 8 Survey sample builder (SSB)
Example 2. Survey design
Country:
Name of EU:
Primary Vector:
Example 2
Example 2
Anopheles, Culex or Mansonia
Demographic data on EU (provided by user)
Population of 6-7 yrs:
Total number of EAs:
23,128
284
Avg. population 6-7yrs per EA:
81
*Note, if any of the above information is incorrect, click on the 'Back to "Start Here"' button
and then click 'edit EU information'
Cluster Survey of Households
Survey Sampling Methodology (calculated by program)
Sample Size for Cluster Design
Number of Clusters
Slide 22
1
1,552
30
Sampling Fraction (of children within the EAs)1
0.75
Sampling Interval (of children within the EAs)1
1.34
Critical Cut-off (maximum # positive ICT/Brugia
Rapid results allowable for country to "Pass")
1Based on a 15% absentee rate
18
Module 8 Survey sample builder (SSB)
Example 2. Randomized selection of
households
Lists corresponding to the houses selected for sampling out
of all houses in the selected enumeration area
The random number used
to generate these lists is:
0.775367533
List A
List B
1
2
3
4
5
6
12
11
10
9
8
7
Module 8 Survey sample builder (SSB)
1
3
4
5
7
8
9
11
12
13
15
16
…
1
2
4
5
6
8
9
10
12
13
14
16
…
Example 3
 Net school enrolment ratio: 95%
 Primary vector: Aedes
 Population of students in target grades: 2 814
 Total number of primary schools: 24
 Estimated non-response rate: 15%
Slide 24
Module 8 Survey sample builder (SSB)
Example 3. Survey design
Country:
Name of EU:
Primary Vector:
Example 3
Example 3
Aedes
Demographic data on EU (provided by user)
Population of students in target grade(s):
Total number of Schools:
2,814
24
Avg. number of students in target grade(s)
per school:
117
*Note,
if
any
of
the
above
information
is
incorrect,
click on the 'Back to "Start Here"'
every
button and then click 'edit EU information'
Will visit
school in the
evaluation unit
Systematic Sample of School Entrants
Survey Sampling Methodology (calculated by program)
Sample Size for Systematic Sample
Slide 25
1172
Sampling fraction (children within each
school)1
0.49
Sampling interval (children within each
school)1
2.04
Critical Cut-off (maximum # positive
ICT/Brugia Rapid results allowable for
country
to "Pass")
Module 8 Survey sample
builder
(SSB)
Based
on
a
15%
absentee rate
1
7
Protocol for TAS




Define EU
Determine survey site and sampling strategy
Calculate sample size
Prepare lists


Schools (school-based survey)
Enumeration areas (community-based survey)
 Test selected children
 ICT (areas endemic for W. bancrofti)
 Brugia RapidTM (areas endemic for Brugia spp.)
 Interpret results on basis of critical cut-off
Slide 26
Module 8 Survey sample builder (SSB)
Exercise
Using population data from the evaluation unit(s) defined in
the ‘Evaluation unit’ module:

Use the survey sample builder to define:
 the appropriate survey design





the sample size needed
the number of clusters needed for the survey
the sampling fraction
the sampling interval
Use the survey sample builder to:
 select clusters to be included (if necessary)
 generate lists A and B (if necessary)
Slide 27
Module 8 Survey sample builder (SSB)

similar documents