Copyright 2010, The World Bank Group. All Rights Reserved. Part 2 Survey Design Produced in Collaboration between World Bank Institute and the Development.

Copyright 2010, The World Bank Group. All Rights Reserved.

Part 2

Survey Design

Produced in Collaboration between World Bank Institute and the Development Data Group (DECDG)


C. Who should be measured (units)?

What unit of measurement should be used?

• Businesses are inappropriate for labor force surveys.• Not able to provide demographic detail needed• Not able to report on other necessary details needed

for measurement• Enrollment of children• Earnings and occupations of other household

members

www.gwu.edu



What unit of measurement should be used? (continued)

– Households, rather than individuals or families, are generally the best unit of measurement.

• If you survey individuals, you lose information on the characteristics of other people who live in the household.

• Presence and school enrollment of children would be missed

• Impact of a spouse being absent or other family members being present could not be measured.



What unit of measurement should be used? (continued)

• Families are sometimes bigger than households. • Multiple families live together in one household. • People who do not

live together have less

of an impact on labor

force decisions than

those who do• Unrelated people who

live together are also a

household.



Who should be included in the household?

– Anyone who usually lives there and does not have a usual residence elsewhere.

• Include all members of household on roster• Collect all demographic details for each member• Some questions only make sense for people over a

certain age. • Marital status • Employment



Who should be included in the household? (continued)

• It is okay to ask questions for a broader age range than ILO standards because once you have age, you can make tabulations of different age groups.

• Generally, only the civilian non-institutional population is surveyed.

• People living in hospitals and prisons are also usually inaccessible and may not be capable of answering the survey.


C. Who should be measured (units)?Who is the head of household?

• In most cases, the head of the household is the person who manages the income earned and expenses incurred by the household, and who is the most knowledgeable about other members of the household.

• The head of household does not have to be a man or the oldest person.

• The head of household is usually an adult but it may be a child


D. When? Frequency and Timing

How often should the survey be conducted?• Once

• Does the season matter?

Rain, migration, ability to travel,

and availability of respondents

might vary.• More than once

• Annually• Quarterly• Monthly• Seasonally

www.cheetah.org


D. When? Frequency and Timing

How often should the survey be conducted? (continued)

• Annual surveys should be conducted at the same time each year

• Eliminate impacts of seasonal patterns• Quarterly and monthly surveys may allow researchers to

see the impacts of these seasonal shifts


E. Where? Areas

Is the survey intended to be national or subnational?• If subnational:

• What area will it cover?• Why only that area?• Might there be a desire to expand it later?

• If national:• Are there weather, seasonal, or geographic differences

that need to be considered?• Are there tribal or ethnic differences? • Are there different household structures, such as

polygamy?• Are there urban and rural areas?


E. Where? Areas

Whenever possible, a survey should be designed to be used anywhere in the country and by all groups.

This allows for maximum data comparability


F. How will we measure these concepts? (survey type)

Will the survey be self administered?• Self administered surveys require:

• literacy• internet access if they are to be conducted online• timely postal service if they are to be conducted by mail

• These requirements must span all segments of society to avoid bias.

• Efficient means of follow up on non-responses are also required.



Will the survey be interviewer administered?– Interviewer administered surveys:

• require interviewers to be able to reach respondents• do not require respondent literacy

– Interviewer administered surveys may be: • Computer assisted (on a laptop or handheld)• Paper based



Considerations when determining survey method include:– Literacy of respondents and interviewers– Ability to follow skip patterns

• A self administered survey requires the most of respondents

• A paper based, interviewer administered survey requires the most of interviewers

• A computer assisted, interviewer administered survey requires the least of both



Considerations when determining survey method include:– Data quality

• Self administered surveys have the highest risk of errors • Computer assisted surveys eliminate improper skips,

problems with reading handwriting, and errors in data entry from paper questionnaires.

– Expense• Enumerators, travel, printing, data entry staff, postage,

training, etc

Copyright 2010, The World Bank Group. All Rights Reserved. Part 2 Survey Design Produced in Collaboration between World Bank Institute and the Development.

Documents

world bank group

household members

members of household

head of household

world bank institute

child slide

best unit of measurement

unrelated people