PSYC 2001 Lecture Notes - Lecture 14: Oversampling, Internal Validity, Nonprobability Sampling

• School: Carleton University
• Department: Psychology
• Course: PSYC 2001
• Professor: Deanna Whelan

Document Summary

Population: entire set of people/products/animals/ organizations that you are interested in. Sample: smaller set taken from the population. Diffe(cid:396)e(cid:374)t (cid:373)ethods fo(cid:396) (cid:862)how(cid:863) the set is sele(cid:272)ted. Census: sampling every member of the population. Possible bias, may not generalize, threat to external validity. Simple random sampling = equal for everyone. Cluster sampling = everyone in a cluster that is chosen at random. Multistage sampling = two (cid:396)a(cid:374)do(cid:373) sa(cid:373)pli(cid:374)g stages . Random sampling of clusters, random sampling of people within the cluster. Stratified random sampling = strata then random from each strata to represent population. Oversampling: purposely over sampling a group/ subset. Small numbers are rarely generalizable (minimum number needed) Weighting of data to be representative in analyses. Similar to stratified but non-random sampling methods to achieve desired percentages. (cid:862)who(cid:863) is pa(cid:396)t of the study sele(cid:272)ted at (cid:396)a(cid:374)do(cid:373) (cid:894)e(cid:395)ual (cid:272)ha(cid:374)(cid:272)e of pa(cid:396)ti(cid:272)ipati(cid:374)g(cid:895) About 30 (per group) and you can assume normality* If sample is biased, more participants does not fix the problem!