Consistent API

[ddsjoberg]

let's think up a consistent API for these functions and stick to it. For example, there will be none of the nonesense like prop.test() where you need to sum the counts before passing it to the function.

That said, let's talk about APIs that are pipe-able from a data frame vs a function that accepts vectors. Should we decide on one, or have methods for both?

[statasaurus]

I was having a play with this. I like the vector method you used in the cardx functions. But when you have have two sample and stratification it breaks down. Do you think it will be easier to just always have

ci_function(.data, x, y? strat?, conf.int = 0.95, ect.) 

I like the vectors, but it just won't work for all cases and I can't decide what is more important consistency across the package or ease within a given function