Definition · Plain-language
Power analysis
Power analysis is a statistical method used to determine the minimum sample size required to detect an effect of a given magnitude with a specified level of confidence.
The step most authors miss
Doing CRediT right? Don’t stop at the statement.
A CRediT statement credits you inside one paper. The recognition CRediT was built for happens when those roles are tied to you, persistently. Sign in with your ORCID — free — and claim your CRediT contributions on casrai.org, the home of the standard. They become a verified, portable part of your identity, not a line that disappears into one PDF.
Free: claim your contributions, then export a journal-ready CRediT statement, schema.org structured data, JATS XML, CSV or BibTeX — and preview your public profile. A membership publishes that profile publicly and verifies the journals you serve.
Understanding statistical power
Statistical power is the probability that a study will successfully detect an effect or relationship if one actually exists in the population. Mathematically, it is defined as 1 − beta, where beta represents the probability of committing a Type II error (a false negative, or failing to reject a false null hypothesis). Traditionally, researchers aim for a power of 0.80 or 80%, meaning the study has an 80% chance of detecting a true effect. Underpowered studies often miss real findings, wasting funding and resources, while overpowered studies may recruit too many participants, raising ethical concerns in clinical research by exposing subjects to unnecessary interventions.
The four interconnected parameters
A power analysis relies on the mathematical relationship between four key parameters: sample size, significance level (alpha), expected effect size, and statistical power. If three of these parameters are fixed, the fourth can be calculated. Alpha is typically set at 0.05, and power is targeted at 0.80. The expected effect size (measured by metrics like Cohen's d or Pearson's r) represents the strength of the relationship in the population, estimated from prior literature, pilot studies, or clinical significance. The primary purpose of power analysis is to calculate the minimum sample size required to detect this expected effect. This process prevents studies from being either too small to find real effects, or unnecessarily large, which optimises research budgets and ensures that experimental designs are statistically valid.
Prospective versus retrospective power analysis
Prospective (or a priori) power analysis is conducted during the planning phase of research, before data collection begins. It is a critical requirement for grant applications and ethical approvals, ensuring that the study is designed with an adequate sample size to yield statistically valid results. In contrast, retrospective (or post-hoc) power analysis is performed after data collection and analysis, often to explain non-significant results. However, methodologists discourage post-hoc calculations because they are directly determined by the observed p-value. Researchers should instead use confidence intervals to show the precision of the estimated effect size. By focusing on effect size estimation and interval precision rather than post-hoc power, researchers can provide more meaningful interpretations of their experimental findings for academic publication.
Key facts
At a glance
- Definition: a method for determining required sample size based on statistical parameters
- Mathematical formula: statistical power is 1 − β, where β is the Type II error rate
- Standard target: researchers commonly aim for a power of 0.80 (80% chance of detecting a true effect)
- Core inputs: requires specifying the significance level (alpha), effect size, and desired power
- Timing: best performed prospectively before collecting data to avoid underpowered studies
- Ethical role: prevents wasting resources on studies that are too small to yield definitive results
Common misconceptions
What people often get wrong
Often heard: A post-hoc power analysis explains why a non-significant result occurred.
Actually: Post-hoc power analysis is mathematically redundant, as it is directly determined by the observed p-value. Researchers should instead report confidence intervals to show the precision of the estimated effect.
Often heard: Statistical power only depends on the number of participants in a study.
Actually: Power depends on the interaction of sample size, alpha, and the magnitude of the effect. A very large effect can be detected with high power using a small sample, whilst a tiny effect requires a massive sample.
Common questions
FAQ
Why is 80% power the standard in scientific research?+
Setting power at 0.80 is a convention established by statistician Jacob Cohen. It represents a reasonable balance between the risk of a Type I error (often set at 5% or 0.05) and the risk of a Type II error (set at 20% or 0.20), reflecting the view that false positives are generally more damaging than false negatives.
How does effect size influence the required sample size?+
Effect size and sample size have an inverse relationship in power analysis. A large effect is easy to distinguish from random noise, requiring a smaller sample. Conversely, a small, subtle effect requires a much larger sample to achieve the same level of statistical power.
Going deeper
