Experiment
RANDOM + TREATMENT + MEASURE → two groups
Survey
ASK + SAME QUESTIONS + RECORD answers
Observational
NO TREATMENT → just WATCH & record
Bias
MISREPRESENT → one group MORE LIKELY
Simple Random
EVERYONE = EQUAL chance; random tool
Stratified
DIVIDE by TRAIT → sample from EACH group
Cluster
Pick WHOLE GROUPS; all inside surveyed
Systematic
RANDOM START → every Nth person
Convenience
EASY ACCESS / close by → BIAS risk
Self-Selected
VOLUNTEERS choose themselves → BIAS risk