Stata-style conditional replace, preserving the existing value on NA

Reproduces Stata's replace x = value if cond: rows where cond is FALSE or NA keep their current value of x unchanged; only rows where cond is definitely TRUE get value. This differs from dplyr::if_else(), which returns NA when cond is NA — correct for a fresh gen-style computation (there is no "existing value" to fall back to), but wrong for a replace, where an unresolved condition must leave the row exactly as it was. Used throughout the sequential-overwrite classification rules (tech_grad, ind_a38, type, FOLLOW/LEAD, ...) documented in Pseudocode/*.md as "later rule overwrites earlier" chains.

Usage

stata_replace(cur, cond, value)

Arguments

cur

The current value of the variable being updated.

cond

A logical vector (may contain NA).

value

The replacement value (recycled as needed) where cond is TRUE.

Value

A vector the same length/type as cur.