Table 1.

Hypothetical Examples of Measures of Effect Size

Effect Size	Description	Example
Mean, M	Original units	Mean length of stay in hospital, M = 10 days.
Difference between two means	Original units	The average patient satisfaction increased last year by 0.5, from 3.1 to 3.6 in a 1 to 5 scale.
Median, Mdn	Original units	Median length of stay in hospital, Mdn = 5 days.
Percentage	Units-free	35.5% of patients were African American.
Frequency	Units-free	39 practices implemented a palliative care intervention.
Correlation, r	Units-free	Anxiety scores correlated with depression scores among patients (r = 0.6).
Standardized difference between two means, Cohen's d	Standardized	The average effect of psychotherapy was d = 0.68.
Regression weight, b	Original units	The slope of the regression line for income against age was b = $1,350/year.
Standardized regression weight, β	Standardized	The standardized β-weight for age in the regression was 0.23.
Proportion of outcome variance explained by a model, R²	Units-free	Three variables of age, education, and family status in the multiple regression together explained 48% of the variance of the outcome variable (R² = .48).
Proportion of outcome variance explained by components of a model, partial ω² (Greek omega-squared)	Units-free	The independent variable age accounted for 21% (partial ω² = 0.21) of variance of the outcome variable after controlling for the other predictors.
Risk	Units-free	The risk that a child has a bicycle accident in the next year is 1/45 or 2.2%.
Relative risk	Units-free	A boy is 1.4 times as likely as a girl to have a bicycle accident in the next year.
Odds	Units-free	The odds of infarction among smokers are 2.02; the odds of infarction among non-smokers are 1.25
Odds ratio	Units-free	The odds of infarction among smokers are 1.6 times the odds among non-smokers
Number needed to treat, NNT	Patients	NNT = 14 indicated that 14 people need to be treated with Apixaban to prevent one case of recurrent thromboembolism