Ternary Comparison Operator in Stata?

In my Stata do scripts, I often have to compare dates which may be missing. Unfortunately, 开发者_开发技巧the internal representation of . is the largest possible number of the given range, so the following holds:

5 < .

This can become quite annoying e.g. when checking whether a date is within a certain range:

gen between_start_stop     = . if d == .
replace between_start_stop = 1 if ///
    !missing(d) & !missing(start) & !missing(stop) & ///
    start < d & d < stop
replace between_start_stop = 0 if ///
    ((!missing(d) & !missing(start) & !(start < d)) | ///
     (!missing(d) & !missing(stop) & !(d < stop))

instead of the following:

gen between_start_stop = (start < d) & (d < stop)

Is there a way to use comparison operators that work with ternary logic?

I.e., I would like the following statements to be true:

(5 < .) == .
(. < .) == .
(. < 5) == .
(. & 1) == .
(. & 0) == 0
etc...

---

A couple of suggestions:

1. use inrange() (also look at inlist) to specify ranges instead of a series of < and > statements;
2. you can specify multiple items in missing() or !missing() statements like !missing(start, stop, d) and

3. it really sounds like you want to use cond(), which (using an ex from the help file) can be used to specify multiple conditions in one function:

   g var = 1 if cond(missing(x), ., cond(x>2,50,70))

returns . if x is missing, returns 50 if x > 2, and returns 70 if x < 2

---

The analogy does not get you what you want -- This formulation returns ‘missing’ when a known d is below a known start (even if stop is, here irrelevantly, missing) or a known d is above a known stop (even if start is, here irrelevantly, missing). The correct value in both cases is ‘false’. I have a utility ('validly') which allows 'generate' to access three-valued logic and does what you want -- see discussion on my webpage http://www.nuffield.ox.ac.uk/People/sites/KIM/SitePages/Biography.aspx which has a link to a paper expanding further (but be warned -- that has just been rejected by the Stata Journal as being "far too difficult to understand"