The National Center for Health Statistics (NCHS) has been evaluating health outcome estimates obtained from a recruited probability-sampled web panel. Historically, traditional survey methods including in-person interviews, telephone interviews or mail surveys have been used to collect information about national health outcomes. However, increasing household Internet access in the US has led to new opportunities to conduct online health surveys. Through the Research and Development Survey (RANDS), a web-based panel survey, NCHS has been investigating the use of methods to combine web survey data with high quality reference data, such as the National Health Interview Survey (NHIS), to improve web estimates. Propensity score calibration has been shown to improve the estimates of health outcomes evaluated in RANDS, although this adjustment technique involves many considerations to set up the propensity score model. In this talk, we introduce research surrounding these factors and assess RANDS estimates in comparison to the NHIS.