Quality-adjusted life year weights and treatment bias: Theory and evidence from cognitive interviews