Testing for omitted variables should play an important part in specification analyses of statistical "linear form" models. Such omissions may comprise terms in variables that were included themselves (e.g., a quadratic term, or a categorical specification instead of a metric one), interactions...