TimelyScholarFree Tools › Confidence Interval from P-Value
Recover a 95% confidence interval from an estimate and its exact p-value, or work out the p-value from a reported confidence interval. Useful for meta-analysis when studies report only one of the two.
FreeNo account neededYour data stays in your browser
| 95% confidence interval | 0.66 to 0.99 |
|---|---|
| Standard error | 0.1026Of the log estimate |
| z | 2.054From the p-value |
This assumes the result came from a test based on the normal distribution (as for most ratios and large-sample differences). It uses the exact normal quantile rather than the approximation in Altman & Bland; results agree closely. Rounded p-values (such as p = 0.04) give approximate intervals.
CI from p, or p from CI.
Ratios are handled on the log scale.
The estimate with its p-value or 95% CI.
For estimates whose test is based on the normal distribution, the p-value, estimate and standard error are linked: z = estimate/SE. Knowing two of them gives the third, and from the SE the confidence interval.
Ratios such as odds, risk and hazard ratios must be log-transformed first. The method needs an exact p-value: 'p < 0.05' cannot be converted.
Rounded p-values and estimates, or a test based on the t distribution, lead to small differences.