First, the results themselves may be underestimates of the trueMedicaid take-up rate as researchers may have only noisy proxiesof Medicaid eligibility. As Remler and Glied (2003) note, programswith asset tests often appear to have low take-up rates in partbecause individuals who appear eligible to the researcher—whooften only has data on individuals’ pre-tax income levels—are notin fact eligible.7Moreover, state-level rules on Medicaid eligibilityvary tremendously—including the deductions allowed in calcu-lating assets and income and how long one must be uninsuredbefore gaining eligibility—further complicating researchers’ effortsto identify who is eligible for the program. In a similar setting,Daponte et al. (1999) find that proxying eligibility with whetheran individual meets the income test creates substantial downwardbias in estimated take-up rates in the food stamp program: onlyabout half of a sample of households that met the income test metthe other requirements in the state of Pennsylvania.