RESMET

• Multistage random sampling


  • Combines sophistication with efficiency
  • Can be complex and difficult to explain to nontechnical audiences
• Simple random sampling


  • Simple to implement; easy to explain to nontechnical audiences
  • Requires a sample list (sampling frame) to select from
• Expert nonprobability sampling (purposive)


  • As an adjunct to other sampling strategies
  • Experts can provide opinions to support research conclusions
  • Likely to be biased; limited external validity
• Accidental, haphazard, or convenience nonprobability sampling


  • Very easy to do; almost like not sampling at all
  • Very weak external validity; likely to be biased
• Heterogeneity nonprobability sampling (purposive)


  • When you want to sample for diversity or variety
  • Easy to implement and explain; useful when you’re interested in sampling for variety rather than representativeness
  • Won’t represent population views proportionately
• Snowball nonprobability sampling (purposive)


  • With hard-to-reach populations
  • Can be used when there is no sampling frame
  • Low external validity
• Systematic random sampling


  • When you want to sample every kth element in an ordered set
  • You don’t have to count through all of the elements in the list to find the ones randomly selected
  • If the order of elements is nonrandom, there could be systematic bias
• Quota nonprobability sampling (purposive)


  • When you want to represent subgroups
  • Allows for oversampling smaller subgroups
  • Likely to be more biased than stratified random sampling; often depends on who comes along when
• Modal instance nonprobability sampling (purposive)


  • When you only want to measure a typical respondent
  • Easily understood by nontechnical audiences
  • Results only limited to the modal case; little external validity
• Cluster (area) random sampling


  • When organizing geographically makes sense
  • More efficient than other methods when sampling across a geographically dispersed area
  • Usually not used alone; coupled with other methods in a multistage approach
• Stratified random sampling


  • When concerned about underrepresenting smaller subgroups
  • Allows you to oversample minority groups to assure enough for subgroup analyses
  • Requires a sample list (sampling frame) from which to select, and variable(s) to stratify by