[Feature Request]: more methods for CIs for proportions

[DrLynTaylor]

Feature description

I'm developing a comparison page for CAMIS looking at CIs for proportions in R vs SAS: https://psiaims.github.io/CAMIS/Comp/r-sas_ci_for_prop.html. For a CI for a single proportion, I found 2 methods (midp and blaker) that were in SAS but not in cardrx (they are available in package PropCIs). Could they be added into cardrx?

Code of Conduct

• I agree to follow this project's Code of Conduct.

Contribution Guidelines

• I agree to follow this project's Contribution Guidelines.

Security Policy

• I agree to follow this project's Security Policy.

[ddsjoberg]

Thanks @DrLynTaylor ! We could depend on the PropCIs pkg for the calculations, but the package seems to have relatively few downloads from CRAN. Perhaps it would be best for us to implement them ourselves? Can you link to documentation for the methods' calculations, and we can assess if we can have someone write it themselves?

[DrLynTaylor]

Thanks....I don't know why sas implemented these 2 ones and not the Newcombe altman method. I've not seen them used so i think my preference might be to remove the issue for midp & blaker (remove this from Camis too & propCis package if it's not used much), then just add Newcombe altman into cardrx? What do you think?

[DrLynTaylor]

Wonder if we could ask some other stats people for their opinion too? The topic seems quite complex, I'd love a list of just those commonly used in pharma and the reason for each one! That's what I wanted to put into Camis but I got a bit lost in the weeds & put too much in!

[ddsjoberg]

I think if you use it often, that probably means others do as well and we can add it. If we just found it as an option for calculating CIs, and we don't know anyone using it....we can leave it out until someone requests it?

BUT I do love the larger goal of surveying methods used in the field!

We also have an open PR for bootstrapped CIs. I haven't had the bandwidth to review yet. But I think that would be a good addition to your list since it's so commonly used. And it would be even more amazing to include descriptions (as you mentioned) of when each of the bootstrap methods are appropriate.

[DrLynTaylor]

Let me do more research on midp and blaker methods and get back to you. I'll do a short survey as well and see if we can get any consensus on what actually used. I'll also look into the most common CIs for differences as that list may be different too!

[DrLynTaylor]

Done more research into this.... will close this issue as I dont think these two methods are needed.