Multiple Indicator Cluster Surveys
Survey Design Workshop
Sampling Overview
MICS4 Survey Design Workshop
Introduction
• MICS is multi-national survey program
– Consistency and comparability between surveys is
important
• MICS based on national stratified multi-stage sample
• Workshop session covers highlights of MICS
sampling, including:
– Identifying effective sampling frame and sample design
– Determinants and calculation of sample size
MICS4 Survey Design Workshop
Importance of correct sampling
procedures
• Sample design will affect inferences of survey
results
• Coverage of frame and appropriate
probability sampling are critical
• Accuracy of survey results depends on both
sampling and non-sampling errors
MICS4 Survey Design Workshop
Key concepts in proper sampling
• Use accepted probability techniques at every
stage of selection
• Use of simple sampling methodology
• Use the latest census as the sampling frame
when available
• Sampling in two stages
• Use of clusters of moderate size
• Use of an adequate sample size
MICS4 Survey Design Workshop
MICS4 prototype sample design
• Manual presents different sampling options
• All options involve two or three-stage cluster
sampling
– Primary sampling units (PSUs) – census enumeration areas
– Listing in sample clusters or use of compact clusters
– Sample households selected from listing
• Option for each country depends on type of frame
and availability of existing samples
MICS4 Survey Design Workshop
Sampling option 1 – use existing
sample
• Most cost-effective approach
• Determine whether existing sample is
appropriate
– Probability sample
• Known probabilities of selection at each stage
– Nationally representative
– Sufficient number of sample PSUs for MICS
– Example – 2008 Mozambique MICS3
MICS4 Survey Design Workshop
Sampling option 1 – continued
• Determine when listing of households in sample EAs
was conducted
– Listing can be used if less than 1 year old
– Generally select new sample of households from listing
– Possible to use selected households from other survey
• Consider response burden, analytical needs for both surveys
• Conduct new listing in existing sample or subsample
of EAs
• Availability of master sampling frame in some
countries
MICS4 Survey Design Workshop
Sampling option 2 –
standard segment design
• Based on standard segments of about 500 persons or 100
households
• Effective procedure when some EAs are too large for listing
• Three-stage sample design
• Measure of size for each EA – calculated number of segments
• EAs selected systematically with PPS
• Larger sample EAs with 2 or more segments subdivided
• One segment selected in each EA for listing, enumeration
MICS4 Survey Design Workshop
Sampling option 3 –
modified segment design
• Three-stage sample design
• Small compact clusters with about 20-30 households
• Measure of size of EAs – calculated number of
compact clusters
• EAs selected systematically with PPS
• Mapping and segmentation in all sample EAs
• One compact cluster selected in each sample EA for
listing, enumeration
MICS4 Survey Design Workshop
Relationship of sampling with other
aspects of survey planning
• Survey budget – may determine maximum
sample size
• Sampling plan should also take into account
required timing of MICS results
• Estimate costs and time related to PSUs –
transportation, listing
• Costs related to sample households – cost of
interviewing (days of enumeration)
MICS4 Survey Design Workshop
Sampling and fieldwork organization
• Effective balance of number of sample PSUs
and cluster size
• Dispersion of sample affects transportation
costs, travel time
• Size of team of enumerators and supervisor
• Interviewer workload
MICS4 Survey Design Workshop
Factors that affect sample size
•
•
•
•
•
Budget and resource constraints
Timing and logistical considerations
Select key indicators
Required level of precision
Need for sub-national estimates
– Regions, provinces, urban/rural domains
MICS4 Survey Design Workshop
Selecting key indicators
• Person variables (rates)
• Important indicator that will yield the largest
sample size
• Estimated rates between 15 and 20%
• Target population that comprises a small
proportion of the total population
– Children aged 12 to 23 months
MICS4 Survey Design Workshop
Sample size determinants –
sampling parameters
• Level of precision – standard error, coefficient of
variation (CV)
• Acceptable level of error - margin of error, relative
margin of error
• Level of confidence – 95%
• Design effect (DEFF) – depends mostly on level of
clustering in the sample
• Average household size
• Proportion of base population in total population
• Adjustment for non-response
MICS4 Survey Design Workshop
Use of previous MICS or DHS results
• Review sample design of previous survey
• Examine sample size and allocation to
geographic domains
• Study level of precision for key indicators
– Design effects
– 95% confidence intervals
• Results can be used to determine sample size
and allocation for new MICS
MICS4 Survey Design Workshop
Formula for sample size calculation
• Two similar spreadsheet templates for MICS4
– Entry cell for each sampling parameter
– Formula for calculating required number of
sample households
• Presentation and discussion of spreadsheets
MICS4 Survey Design Workshop
Sample Size Formula
n
4r (1  r )( deff )(1.1)
_
2
(. 12 r ) ( p)( n)
where
– n is the required sample size, expressed as number of households, for the KEY
indicator
– 4 is factor to achieve 95 percent level of confidence,
– r is anticipated prevalence rate for key indicator,
– 1.1 is factor to raise sample size by 10 percent for potential nonresponse,
– deff is shortened symbol for design effect,
– 0.12r is margin of error to be tolerated, defined as 12 percent of r (12 percent
thus represents the relative margin of error),
– p_ is proportion of total population that smallest group comprises, and
– n is average household size.
MICS4 Survey Design Workshop
Sample size reference table
• Table 4.3 in Chapter 4 of MICS manual
• Can be used for specific set of parameters:
–
–
–
–
–
Target population is 2.5% of total population
DEFF = 1.5
Nonresponse rate is 10%
Relative margin of error is 12%
95% level of confidence
MICS4 Survey Design Workshop
Sample Size (Households) to Estimate Coverage
Rates for Smallest Target Population
Average
Household Size
(number of
persons)
4.0
4.5
5.0
5.5
6.0
estimated rate,
r = 0.25
estimated rate,
r = 0.30
estimated rate,
r = 0.35
estimated rate,
r = 0.40
13,750
12,222
11,000
10,000
9,167
10,694
9,506
8,556
7,778
7,130
8,512
7,566
6,810
6,191
5,675
6,875
6,111
5,500
5,000
4,583
Use this table when your
1. Target population is 2.5% of total population; this is generally children 12-23 months
old
2. Sample design effect, deff, is assumed to be 1.5 and nonresponse is expected to
be 10 percent
3. Relative margin of error is set at 12 percent of estimate of coverage rate, r
MICS4 Survey Design Workshop
Number of PSUs and cluster size
• More PSUs and smaller clusters
– Statistically more efficient – greater dispersion of
sample
– More costly – transportation, listing
• Minimum number of PSUs - 300 to 400 at
national level
– Minimum of 30 PSUs for subnational domains
• Number of sample households per cluster
generally in range of 15 to 30 households
MICS4 Survey Design Workshop