Research Sampling Procedures, Methods, & Issues Rick Carlton Fall 1999
Nonprobability Sampling Provides a statistical basis for verifying that sample is representative of population Every member of the population has a known, nonzero probability of being included in the sample Based on random selection techniques Nonprobability Sampling Sample chosen based on judgement regarding population characteristics and survey needs Limitations on generalizing findings
Probability Sampling Simple Random Sampling Stratified Random Sampling Systematic Sampling Cluster Sampling
Probability Sampling Simple Random Sampling Sampling frame must include all or nearly all members of the population All members have an equal chance of selection No sensitivity to subgroup differences
Probability Sampling Stratified Random Sampling Population is divided into subgroups to measure selected factors Subgroup identification can be through literature or expert opinion
Probability Sampling Systemic Sampling For populations without naturally occuring repetition, choosing systematically (I.e., every other name) can assure randomness in selection
Probability Sampling Cluster Sampling Uses naturally occuring units (i.e., one entire school) Multistage cluster sampling uses random sampling from naturally occuring clusters Increasing number of clusters allows decreasing sample size within each cluster Useful in permitting use of more clusters or more samples, depending on which is easier to access
Nonprobability Sampling Rationale . Helps with hard to identify groups where cooperation and completion of interview or survey may be difficult Assists when sampling groups where some members may be incapable of cooperating or where ethical considerations may limit some data collection techniques Provides pilot study data for making decisions about feasibility of a study
Nonprobability Sampling Convenience Sampling Snowball Sampling Quota Sampling Focus Groups
Nonprobability Sampling Convenience Sampling Ready and available sample Very low (or nonexistant) generalizability
Nonprobability Sampling Snowball Sampling Relies on previously identified members of a group to identify other members of the population Useful when population listings are not available
Nonprobability Sampling Quota Sampling Use subgroups and sample proportionately Requires current records to obtain accurate sample
Nonprobability Sampling Focus Groups Provides in-depth portraits Unperceived uniqueness can skew results dramatically
Purposeful Sampling In Qualitative Research Deviant Case Intensity Typical Case Maximum Variation Stratified Purposeful Homogeneous Critical Case Snowball or Chain Criterion Theory-Based Confirming and Disconfirming Purposeful Random Politically Important Cases Convenience