PS295 Lecture Notes - Lecture 7: Cluster Sampling, Simple Random Sample, Sampling Frame
1) Probability sampling:
-a researcher knows the probability that a particular individual will be chosen from the
population
-entire sampling frame is known and accessible
-in an epsem design, all cases in the population have an equal probability of being chosen
-error of estimation is relevant only if the sample is a probability sample
Three probability sampling procedures:
1) Simple random sampling
-each member has equal chance of being selected
-selection of 1 person does not influence selection of another (independence)
-step 1: clearly define population of interest, 2: list all members (sampling frame), 3. select
members from list using random process
-removes biases and generally (but no guarantee, esp with small samples) generates
representative sample and could still result in bias by chance
2) Stratified random sampling
-population first divided into strata (subgroup that shares characteristic)
-useful when population is known to contain distinct subgroups
-step 1: divide into strata, 2. randomy select from each strata
-when randomly selecting, can use equal sample size from each subgroup or proportionate
Document Summary
A researcher knows the probability that a particular individual will be chosen from the population. In an epsem design, all cases in the population have an equal probability of being chosen. Error of estimation is relevant only if the sample is a probability sample. Three probability sampling procedures: simple random sampling. Each member has equal chance of being selected. Selection of 1 person does not influence selection of another (independence) Step 1: clearly define population of interest, 2: list all members (sampling frame), 3. select members from list using random process. Removes biases and generally (but no guarantee, esp with small samples) generates representative sample and could still result in bias by chance: stratified random sampling. Population first divided into strata (subgroup that shares characteristic) Useful when population is known to contain distinct subgroups. Step 1: divide into strata, 2. randomy select from each strata. When randomly selecting, can use equal sample size from each subgroup or proportionate sampling.