The confidence interval for a difference between 2 proportions or 2 means can be estimated if the P value is reported.
Parameters:
(1) value 1
(2) value 2
(3) reported P value
difference between the 2 values =
= ABS((value 2) – (value 1))
normal distribution test z =
= (-0.862) + SQRT(0.743 – (2.404 * LN(P value)))
standard error =
= ABS((difference between the 2 values) / (z value))
low end 95% confidence interval for difference = (difference) – (1.96 * (standard error))
high end 95% confidence interval for difference = (difference) + (1.96 * (standard error))
