In a cohort study, a questionnaire measures exposure with similar error in participants who later do and do not develop disease. What is a likely effect of this non-differential exposure misclassification?