Sampling Design and Procedures

Define Target Population - who are you trying to find out about? Age? Geography? Occupation? Marital Status? Children?

Define Sample Frame - Is there a list of potential samples? Is it up to date? Does it have all the information?

Non-Probability Sampling Probability Sampling Sampling Methods
 * Convenience - take what is convenient
 * good - less expensive, quick
 * bad - non-representative, selection bias
 * Judgement - researcher samples what they think should be sampled
 * good - ideal for exploratory research
 * bad - subjective, no generalisation
 * Quota - put a quota on what to sample and then use convenience or judgement sampling
 * good - can control for certain characteristics
 * bad - no assurance of representativeness
 * Snowball - sample one person, ask person to recommend other people like themselves
 * good - can sample rare characteristics
 * bad - time consuming
 * SRS - Simple Random Sampling. See ST201.
 * good - easily understood, projectable
 * bad - low precision and representativeness
 * Systematic - See ST201.
 * good - better representativeness
 * bad - relies on 'order' of data
 * Stratified - See ST201
 * good - increases precision
 * bad - choosing strata, expensive
 * Cluster - See ST201
 * good - easy to implement, cheap
 * bad - imprecise, hard to interpret
 * Multistage - See ST201
 * good - ensures representativeness
 * bad - long process
 * Online
 * Fast, Cheap, Convenient, More honest(percieved respondent anonymity is high), More young males.
 * Internet penetration may not be as high in some areas
 * Lack of sampling frame
 * There may be professional panelists, people who sign up to multiple panels
 * Three methods
 * Access Panel
 * Have a reliable source of people
 * Cheap, quick and have demographic info
 * But not all panelists may be accessible
 * Inaccurate info
 * Deduplicating possible duplicate panelists
 * Different panels give
 * Pop-Up
 * Easy to access many people
 * Cheap and Fast
 * But easy to dismiss
 * Sample is self-selecting
 * Pop-up block software
 * Client Email List
 * Meaningful Sample Frame
 * Client list can be badly maintained
 * Hard to get response
 * Other issues:
 * Weighting data on online surveys are difficult
 * Data quality is a problem
 * Postal
 * Telephone
 * In person