Click here for search results

Sampling for Household and Enterprise Survey

January 31, 2003
9:00 AM - 5:00 PM
Presenter: Juan Munoz

The objective of this one-day course is to present the fundamentals of sampling theory, as applied to household and enterprise surveys, using computer simulations rather than mathematical formulae to convey the basic facts intuitively.

The course should be of interest not just to those working on or planning a survey, but to anyone relying on survey data for analysis and policy making.  Understanding the basics of sampling is crucial to judge how reliable and representative are survey results.

The course was organized into five sessions:

  1. Basic concepts of sampling presented as a simulation of simple random sampling for proportions (Example: an electoral poll in a small island). 
    • Population and sample.  What is random sampling? 
    • Truth and estimation: the meaning of sampling error and the distribution of sampling error.
      • How is sampling error related to sample size? 
      • How is sampling error related to the size of the population? 
      • How is sampling error related to the proportion that is being estimated?
  2. Participants simulate various alternative situations in their own computers.
  3. Estimating averages (Example: an income survey in a small island). 
    • How is sampling error related to the distribution of incomes in the population?
  4. Stratification (Example: simulation an electoral poll in an archipelago). 
    • The two main reasons for stratification are:
      • improving the precision of the overall estimate
      • obtaining separate estimates for different groups of the population.
    • These objectives are generally contradictory in practice and arbitration is needed.
    • Case study: The Nepal NLSS-II sample
  5. Two-stage sampling: rationale and motivation (Example: simulation of an electoral poll with respondents clustered into city bocks). 
    • Comparison with a simple random sample: the price to pay is generally a larger error. 
    • The concept of design effect. 
    • Case study: Household Budget and Labor Force surveys in Moldova.

Back to Training Events and Materials

Permanent URL for this page: