Sampling Design and Size Determination - Strive for reliability and validity. Check each decision to determine if it will increase
sampling error or introduce a systematic bias into the data.
- Identify components quite precisely. The population, sample units, and sample
frame must be described with clarity and precision.
- Determine sample size carefully. Consider confidence required, population
variance, analysis techniques, and the resources available for the project.
- Depend upon random sampling. The greater the deviation from random selection,
the less legitimate and accurate the statistical analysis and reports will be.
- Evaluate design refinements. Use stratification, clustering, quota sampling,
and special designs to increase reliability and decrease costs.
- Use trial when necessary. A pilot survey or sequential sampling will indicate
the appropriate sample size if it can’t be determined in advance.
- Identify sources of bias. Be aware of the major sources of bias, but constantly
inspect the design to be sure additional sources of bias aren’t introduced.
- Be creative and confident. Sampling design and size determination is more of
an art than a science!
|
