Choosing Representative Samples

1
Choosing Representative Samples

• Sample Size?
• Calculator
• Sampling Methods

2
Factors indicating Large Sample

• Decision based on survey is serious or costly
• Sponsors demand high level of confidence
• Likely to have high variance
• Many sub-samples needed
• Costs or timing vary little with larger sample
• Time and money is available for a larger sample

3
Factors indicating Small Sample

• Few, if any, major serious consequences
• Sponsors require only rough estimates
• Population is fairly homogeneous
• Analysis to be based on the whole sample
• Costs vary dramatically with sample size
• Budget constraints and/or time constraints

4
Probabilistic Sampling Methods

• Simple Random Sample (Probability sample)
• every member of population has an equal chance
• chance of being selected can be calculated, n/N
• permits the calculation of confidence intervals
• Stratified Random
• when population strata are known (proportions)
• when membership in strata can be clearly
  determined
• randomly sample from each strata to maintain
  known proportions

5
More Sampling Methods

• Non-random (Non-probabilistic)
• Purposive (a Qualitative approach)
• Convenience Sample - a group that already exists
• Snowball Sample - each member chooses another