Independent groups
Define the difference in original units, outcome variability, allocation ratio, and whether baseline adjustment is planned.
FREE DISSERTATION PLANNING TOOLS
Plan the study.
Show your assumptions.
A defensible sample-size plan connects the research question, focal effect, analysis model, dependence structure, and expected recruitment losses. Use these tools to make those assumptions visible before settling on one number.
TOOL 01
Attrition and recruitment target calculator
Start with the analyzable sample required by your design. Then model each stage separately instead of adding one undocumented attrition percentage.
This tool plans recruitment. It does not calculate the analyzable sample required for your statistical model.
TOOL 02
Cluster design-effect explorer
See how similarity within schools, classrooms, sites, or practices reduces effective sample size. Explore the ICC, average cluster size, and cluster-size imbalance together.
Planning approximation: a design effect is not a complete multilevel or cluster-randomized power analysis. The number of clusters, treatment assignment level, small-sample correction, outcome type, and planned model still matter.
TOOL 03
Power-analysis assumptions checklist
Use this before opening any sample-size software. A calculation is difficult to defend when the analysis, focal parameter, or evidence for the effect is still unclear.
0 of 10 documented
Complete the checklist to identify what should be documented before calculating sample size.
DESIGN-SPECIFIC GUIDANCE
The model changes
the planning problem.
Use the calculation that represents the claim you plan to make.
Regression
Plan the focal coefficient, block, interaction, or change in explained variance—not merely the overall model.
Clustered studies
Vary the number of clusters and people per cluster separately. More people in the same few clusters may add limited information.
Longitudinal studies
Include waves, timing, within-person correlation, random slopes, dropout, intermittent missingness, and the planned time contrast.
Mediation and SEM
Simulate the indirect effect or focal parameter. Global model-fit power does not establish adequate power for every path.
After data collection
Report estimates and confidence intervals. Do not use observed power calculated from the final effect estimate.
REPORTING EXAMPLE
Show how the target was obtained.
“The planned analysis tests [focal parameter] using [model]. Under a planning effect of [value], [nuisance assumptions], a two-sided α of [value], and [allocation or dependence structure], [analyzable sample] provides approximately [power] power. Accounting for eligibility, consent, retention, and usable data, the recruitment target is [target]. Sensitivity analyses produced requirements from [low] to [high].”
NEED THE COMPLETE PLANNING WORKFLOW?
Move beyond
one calculator result.
The premium workflow includes editable R Markdown, power and precision scenarios, clustered and longitudinal planning, factorial and mediation simulations, recruitment outputs, and dissertation-ready text.
Explore the $59 workflow