SOCI 2P11 Lecture Notes - Lecture 4: Audit Trail, Stratified Sampling, Advantageous
Document Summary
When studying a sample rather than the entire population, a given sample statistic is more likely to differ from the corresponding pop parametre than be equal to it. Difference b/w sample statistic and pop parametre. B/c sampling error can be computed, probability sampling enables use of. Inverse relationship b/w sample size and sampling error inferential statistical tests. Incomplete sampling frame: the people who do respond tend to be demographically different from those who don"t respond i. e. marginalized groups (indigenous communities, migrant workers) If a large non-response rate is suspected, sample size will be larger. Diversity of parent population i. e. large diversity of social class, ethno racial composition etc, we would want a large population size. Probability of each case in the pop being selected for the sample is unknown. Cannot use inferential statistics: sample statistics that we want to use to generalize/infer from.
