Is it possible that a highly significant interaction term, does not have significant marginal effects at various levels of the primary explanatory variable? Regards David
