TOSTER
TOSTER copied to clipboard
Equivalence tests for one proportion
Adding a function for equivalence tests for one proportion would be a nice idea. While the Wald method based on the normal approximation is very common, it might be inappropriate. Among other:
- it produces limits out of the possible range of values (below 0 or above 1).
- normally distributed only with large n
For a reference comparing CI calculations by 7 methods, see Newcombe (1998). Among other, Newcombe mention:
- direct calculation of the binominal distribution (very conservative)
- the Wilson Score --> more accurate than the normal approximation, not as conservative as the direct calculation of the binominal distribution AND easy to compute with softwares.
Best,
Marie
Newcombe, R. 1998. Two-sided confidence intervals for the single proportion: Comparison of seven methods. Statist. Medic. 17: 857-72.
This is a nice addition - I will try to work on it when I find time!