ENV330H5 Lecture Notes - Lecture 2: Multistage Sampling, Simple Random Sample, Stratified Sampling
Document Summary
Population of 400 experimental units (i. e. , elements in a 20 20 grid). A)simple random sample of 20 from the population. B)stratified random sample; four strata are established and five samples are randomly taken from within each strata. C)cluster sampling; four clusters are established, two of these are randomly selected and a census is taken from each selected cluster. Randomly sample a few clusters --and then randomly sample observations from within these clusters (rather than all of the cluster). Usually, we use multistage sampling for economic reasons. Stratified sampling = separating the layers then taking nibbles from within each. Cluster sampling = eating a few small, but complete, slices. Benefits: maximum spatial coverage of an area for a given number of samples need no prior information about a site. Sites are selected based on some known characteristic. Selection of sample locations based on professional judgment alone, without any type of randomization. Very site-specific (cannot extrapolate to other sites)