#power-analysis

Design robust RCT with appropriate statistical power. Study design: 1. Define primary outcome clearly (e.g., change in depression score). 2. Choose randomization method (simple, block, stratified). 3. Blinding strategy (single, double, triple-blind where possible). 4. Control group selection (placebo, wait-list, treatment-as-usual). Power analysis using G*Power: 1. Set α = 0.05, power = 0.80. 2. Estimate effect size from pilot data or literature (Cohen's d). 3. Calculate minimum sample size, add 20% for dropouts. Example: t-test, medium effect (d=0.5), requires n=64 per group, with dropout n=80 per group. Randomization tools: Research Randomizer, RedCap. Registration: ClinicalTrials.gov before recruitment. Monitor for interim analyses and stopping rules.

Design appropriate sampling strategy and calculate required sample size. Sampling methods: 1. Probability sampling: simple random, systematic, stratified, cluster sampling. 2. Non-probability sampling: convenience, purposive, snowball, quota sampling. 3. Mixed methods: sequential explanatory requires smaller qualitative sample after quantitative phase. Sample size calculation: 1. Continuous outcomes: use power analysis with effect size, alpha=0.05, power=0.80. 2. Categorical outcomes: use proportion formulas with expected proportions and margin of error. 3. Longitudinal studies: account for dropouts, multiply by 1/(1-dropout rate). 4. Cluster sampling: design effect multiplier for correlated observations within clusters. Tools: G*Power, R pwr package, online calculators. Survey research: response rates typically 20-30% for online, 40-60% for phone. Adjust target sample accordingly. Report response rates and compare respondents to non-respondents on available characteristics.