The effect of nonrandom selection of clusters in a two stage cluster design

THE EFFECT OF NONRANDOM SELECTION OF CLUSTERS IN A TWO STAGECLUSTER DESIGN byJASON PARROTTDecember 2012Advisor: Dr. Shlomo SawilowskyMajor: Evaluation and ResearchDegree: Doctor of Philosophy Although not as efficient as simple random sampling, cluster sampling has beenregarded as a valid sampling technique when the researcher is attempting to save cost. However, in order to do so, it is necessary that random selection occurs in all stages ofsampling. This simulation study examines purposeful selection of cluster sampling in thesecond stage of a two stage cluster design. Using Monte Carlo methods, a simulation wasconducted comparing the random selection of both stages of a two stage cluster sample to purposeful selection of the first stage of a two stage cluster sample. The study comparespurposeful selection to random selection by examining the width of the confidence intervalsthat are returned for each simulation. After conducting the study, it was evident that usingpurposeful selection can yield a confidence interval up to nine times greater than that of its