Lecture 6: Samples

do most studies use a representative sample?

no, they are often limited/focused to a population

why do researchers want a large sample?

for homogeneity

pros of large samples

increase representativeness and statistical significance accuracy

cons of large samples

cost more (more incentives), harder to get, more time consuming

pros to small samples

easier to obtain, practical to use with homogenous groups

cons to small samples

less generalizable results, harder to obtain statistical accuracy, increased error (not so representative)

sampling

selection of participants

simple random sampling

the ideal method (everyone has a fair chance); ensures representativeness

systematic sampling

systematically obtaining a representative SRS; feasible, not truly random

stratified random sampling

selecting proportionately equal groups based on desired characteristics (matched samples)

cluster sampling

randomly selecting groups within a set of available groups/clusters; practical, but susceptible to geographic bias

convenience sampling

selecting participants who are easy to access; nonrandom (selection bias) and commonly used in research

judgment sampling

selecting participants of interest; nonrandom, but targeted (some bias) and commonly used in clinical research

chain referral sampling

selecting hard to access participants; nonrandom, but targeted (some bias); participants help recruit other referred participatns