CRIM 2653 Lecture Notes - Lecture 15: Nonprobability Sampling, Snowball Sampling, Probability Theory
Document Summary
Each case has same chances of selection. Probability of that samples will be normally disturbed. Starts a random point, then choose every kth case. Based on target population and projected sample size. Divide target population into subgroups then randomly choose cases. Cluster sampling selects clusters from target population and then randomly select cases within each cluster for sample good for large, dispersed populations complex and may not be as representative. Repetitiveness distribution of characteristics in sample same as target population if not, subject to bias not for all characteristics relevant characteristics. Much match target population need sampling frame unknown population strengths ability to generalize. Generalizability extending findings to the population broad conclusions from particular applying to larger population. Minimize by increasing sample size sampling bias impedes representativeness under/over-estimates. Sampling bias error in sample creative or analysis problems of representativeness under- and over-estimates researcher bias. Confidence level limit generalizability heterogeneity confidence level expressed as percentage.