Prepare a numeric vector for Stata-style comparison against a constant

In Stata, numeric missing (.) is internally a very large finite value, so every comparison operator (==, !=, >, <, >=, <=) against a missing operand resolves to a definite TRUE/FALSE — never a further "missing" result. E.g. age_cat == 0 is FALSE (not missing) when age_cat is missing; age < 2 is FALSE; age >= 2 is TRUE. R's ==/ </etc. instead propagate NA when either side is NA.

Usage

stata_num(x)

Arguments

x

A numeric vector.

Value

A numeric vector with every NA replaced by Inf.

Details

stata_num(x) OP k (for a literal constant k) reproduces Stata's behaviour by substituting Inf for NA before comparing — since Inf is larger than any finite k, every comparison operator then resolves the same way Stata's missing-as-largest-value convention would. Only use this at the exact point of comparison (it must not replace the vector's stored value elsewhere); only valid when comparing against a literal, itself never-missing, constant — for comparisons between two possibly-missing variables, resolve missingness explicitly instead (see Pseudocode/00_conventions.md's MISSING note).