120 likes | 217 Views
SAMPLING. Why sample? Practical consideration – limited budget, convenience, simplicity. Generalizability –representativeness, desire to establish the broadest possible generalizations. What happens when you don’t sample?
E N D
SAMPLING Why sample? • Practical consideration – limited budget, convenience, simplicity. • Generalizability –representativeness, desire to establish the broadest possible generalizations.
What happens when you don’t sample? • Probability sample – everyone has a chance of being included in the sample.
Populations • Defining population of interest • Target population – the population to which the researcher would like to generalize their results. • Sampling frame—operational definition of the population. Is the list of units of analysis from which you take your sample and to which you generalize.
2) Population parameter vs. Sample estimates: • Population parameter – pertains to the population. E.g. the average age of all people is a parameter. • Sample statistic – pertains to the sample – sample characteristics.
Probability Sampling What is the difference between probability and non-probability sampling? • Probability sampling --- every member has an equal chance of being selected. Major advantage: reduces possibility of bias. • Non-probability – we don’t know the probability of selecting a unit into a particular sample.
Types of Sampling Designs Simple Random Sample • Most basic sampling method. Systematic Random Sample • List everyone in population, then start with a randomly selected person, and take every Kth person. • Problem – affected by periodicity.
Stratified Sampling • Ensures that key sub-populations are included in your sample. Rules on stratification • If differences on a dependent variable are large across age, sex, race, etc, then stratify • If differences are small, do not stratify
Disproportionate Sampling • Appropriate whenever an important subpopulation is likely to be underrepresented in a simple random sample or stratified sample. • Suitable as long as the two samples are analyzed separately. • Have to weight if you combine the two samples. Weighting compensates for disproportionate sampling.
Multistage Cluster Sampling • Used where there are no convenient list or sampling frames. • Minimizes travel time to scattered units of data collection
Nonprobability Sampling Quota sampling • Decide on subpopulation of interest and on the proportions of those subpopulations in the final sample. • Quotas ensure that sample is representative of certain characteristics in proportion to their prevalence in population.
Purpose Sampling • You decide the purpose you want informants to serve, and you go out to find some. Good reasons for use of purposive sampling • used in pilot studies • selection of few cases for intensive study • Studying critical cases-- key informants.
Snowball Sampling • Locate one or more key individuals and ask them to name others who would be likely candidates for your research.