How to deal with p-values?

[ddsjoberg]

OPTIONS

• [ ] don't deal with it at all and leave it to the users.
• [ ] write a function that accepts 'survfit2' objects and can return the results from survdiff(). this would focus on reporting the p-value in a format that makes it easy to report in a figure caption.
  • maybe the function is just called survdiff2() with the only difference is that instead of formula and data arguments, it accepts a survfit2 object.
  • we could export sister functions that report the p-values directly, e.g. logrank() would return a formatted p-value, p_logrank() would return the formatted p-value with the "p" preprended, e.g. p = 0.12.

The first version of the pkg we release could just ignore p-values altogether while we decide on an API?

[ddsjoberg]

I added an outline of what I think could work in PR #29.

Need to think on it more however. If we stick with it, then need to update the documentaton.

[ddsjoberg]

• delete the logrank functions and keep survfti2_p()