Non-randomized experiments can be biased Measured demand curve P DM is biased compared to PH true demand D People generally choose the amount of insurance they receive Sicker people will choose more insurance because they know they will need PL DM Q more care Bhattacharya,Hyde and Tu-HealthEconomics Bhattacharya, Hyde and Tu – Health Economics Non-randomized experiments can be biased  Measured demand curve DM is biased compared to true demand DT  People generally choose the amount of insurance they receive  Sicker people will choose more insurance because they know they will need more care