Text this: Inequalities in the health survey using validation question to filter insufficient effort responding: reducing overestimated effects or creating selection bias?