TimelyScholarFree Tools › Bayes Factor Calculator
Calculate default Bayes factors from summary statistics to quantify how strongly the data support the alternative hypothesis over the null, or the reverse.
FreeNo account neededYour data stays in your browser
| BF₁₀ | 1.715Evidence for H₁ relative to H₀ |
|---|---|
| BF₀₁ | 0.583Evidence for H₀ relative to H₁ |
| Interpretation | not worth more than a bare mentionKass & Raftery (1995) categories |
| Prior | Cauchy prior on effect size δ, scale r = 0.707 |
Results match the BayesFactor R package (ttest.tstat, correlationBF, proportionBF). Bayes factors depend on the prior, so report its width and consider a robustness check with wider priors. Category labels are rules of thumb, not thresholds.
One-sample or paired t-test, independent t-test, correlation or proportion.
For example t and group sizes, or r and n.
Medium, wide or ultrawide, as in the BayesFactor R package.
A Bayes factor compares how well two hypotheses predict the observed data. BF10 = 6 means the data are six times more likely under the alternative than under the null; BF01 = 1/BF10 expresses support for the null.
Unlike a p-value, a Bayes factor can show evidence for the absence of an effect. Its value depends on the prior placed on the effect size, so report the prior and consider checking how the result changes with wider priors.
Kass and Raftery (1995) describe a Bayes factor of 3 to 20 as positive, 20 to 150 as strong and above 150 as very strong evidence. These are guides, not thresholds.
Yes. The alternative allows effects in either direction.