Effect Size and Statistical Power

College Depth 80 in the knowledge graph I know this Set as goal
Unlocks 77 downstream topics
effect-size Cohen-d power sample-size type-II-error

Core Idea

Effect size quantifies the practical magnitude of a relationship or difference, independent of sample size. Cohen's d measures standardized mean differences; r measures correlation magnitude. Statistical power is the probability that a study will detect a true effect when one exists — it increases with larger samples, larger effect sizes, and higher significance thresholds. Low-powered studies produce many false negatives and, paradoxically, inflate effect sizes when they do detect effects. Power analysis before data collection is standard practice for well-designed research.

How It's Best Learned

Given a Cohen's d of .50 and α = .05, use a power table or calculator to find the sample size needed for 80% power. Discuss what happens to false-positive rates when many underpowered studies are conducted.

Common Misconceptions

Explainer

Suppose a study finds that a new therapy reduces anxiety scores by 2 points on a 100-point scale, and this result is statistically significant (p = .01). Should you prescribe the therapy? The p-value tells you the result is unlikely to be a fluke, but it says nothing about whether a 2-point reduction is clinically meaningful. This is the gap that effect size fills.

Effect size measures the *magnitude* of a difference or relationship, independent of how many participants were tested. Cohen's d is the most common measure for comparing two means: it expresses the difference in units of the pooled standard deviation. A d of 0.2 is conventionally small, 0.5 medium, and 0.8 large — though these benchmarks are rough guides, not universal standards. What matters is whether the effect is large enough to be practically important in context. Effect sizes, unlike p-values, are comparable across studies even when sample sizes differ, which makes them the currency of meta-analysis.

Statistical power is the flip side of the same coin. Power is the probability that a study will detect a real effect — that it will return p < .05 when a true effect of a given size exists. Power increases with three things: larger sample sizes (more data = more sensitivity), larger true effect sizes (bigger signals are easier to detect), and more lenient significance thresholds (though raising alpha risks more false positives). The conventional target is 80% power, meaning a 1-in-5 chance of missing a real effect. Most psychological studies prior to the 2010s were severely underpowered, with power estimates often below 50%.

Low power has a pernicious side effect beyond just producing false negatives. When an underpowered study does reach significance, it is often because that particular sample happened to overestimate the true effect. This is the "winner's curse": the effect sizes that survive the significance filter are the lucky, inflated ones. This is why large effects from small studies shrink in replication — the original estimate was atypically large, not representative. Pre-registering a power analysis before data collection constrains researcher degrees of freedom and ensures the study is genuinely equipped to answer its research question.

The practical upshot: always report effect sizes alongside p-values. When reading research, ask whether the study was adequately powered, and treat a non-significant result from a small study as uninformative rather than as evidence of no effect. The replication crisis in psychology was driven in large part by the combination of underpowered designs, flexible analysis practices, and publication bias toward significant results — understanding power and effect size is the foundation for reading that literature critically.

Practice Questions 3 questions

Prerequisite Chain

Counting to 10Counting to 20Understanding ZeroThe Number ZeroCounting to FiveOne-to-One CorrespondenceCombining Small Groups Within 5Addition Within 10Addition Within 20Two-Digit Addition Without RegroupingTwo-Digit Addition with RegroupingAddition Within 100Repeated Addition as MultiplicationMultiplication Facts Within 100Division as Equal SharingDivision as Grouping (Measurement Division)Division: Grouping (Repeated Subtraction) ModelDivision: Fair Sharing ModelDivision as Equal SharingDivision as GroupingBasic Division FactsDivision Facts Within 100Two-Digit by One-Digit DivisionDivision with RemaindersRemainders and Quotients in DivisionDivision Word ProblemsIntroduction to Long DivisionFactors and MultiplesPrime and Composite NumbersEquivalent FractionsRelating Fractions and DecimalsDecimal Place ValueReading and Writing DecimalsComparing and Ordering DecimalsAdding and Subtracting DecimalsMultiplying DecimalsDividing DecimalsDividing FractionsMixed Number ArithmeticOrder of OperationsInteger Order of OperationsVariable ExpressionsCombining Like TermsOne-Step EquationsTwo-Step EquationsSolving Multi-Step EquationsEquations with Variables on Both SidesAngle Pairs: Complementary, Supplementary, and VerticalParallel Lines and TransversalsCorresponding AnglesAlternate Interior AnglesTriangle Angle Sum TheoremExterior Angle TheoremTriangle Inequality TheoremSimilar Triangles: AA SimilaritySimilar Triangles: SSS and SAS SimilarityProportions in Similar TrianglesRight Triangle Trigonometry IntroductionTrigonometric Ratios ReviewRadian MeasureConverting Between Degrees and RadiansThe Unit CircleGraphing Sine and CosineGraphing Tangent and Reciprocal Trigonometric FunctionsDerivatives of Trigonometric FunctionsAntiderivativesIndefinite IntegralsBasic Integration RulesRiemann SumsDefinite Integral DefinitionProbability Density Functions and Continuous DistributionsCumulative Distribution FunctionsContinuous Random VariablesNormal DistributionCentral Limit TheoremConfidence Intervals for MeansZ-Tests and T-Tests for MeansOne-Sample Z-Test for MeansOne-Sample and Two-Sample T-TestsInferential Statistics in PsychologyEffect Size and Statistical Power

Longest path: 81 steps · 408 total prerequisite topics

Prerequisites (5)

Leads To (8)