Calculating the correct sample size before data collection is not optional — it is a methodological requirement that most ethics committees, thesis supervisors, and journal reviewers will ask for. A sample size justification demonstrates that the study was planned with adequate statistical power to detect the expected effect.
Statistical power is the probability that a study will detect a true effect when one exists. Conventionally set at 80%, though 90% is increasingly expected in clinical research. Lower power means a higher chance of a false negative — concluding there is no effect when there actually is one. Underpowered studies waste resources and contribute to a literature full of inconclusive findings.
Effect size is the most important — and most uncertain — input to a sample size calculation. If you are building on existing literature, use the effect sizes reported in previous similar studies. If your study is the first of its kind, use a conservative (small) effect size assumption. An overestimated effect size leads to an underpowered study.
For structural equation models with multiple constructs and paths, rules of thumb are commonly used in the absence of an exact formula. The most cited rule is 10 observations per free parameter estimated. For a model with 20 free parameters, this implies a minimum sample of 200. SmartPLS users can use the PLSpredict function to assess predictive relevance.
We provide sample size calculation as a standalone service or as part of our full thesis support package. Our calculations include the formula, software output (G*Power or R), and a ready-to-use methodology paragraph that justifies your sample size for your research design.
