Title: |
|
Some Relationships and Possible Models for State Estimates from the Medical Expenditure Panel Survey, Household Component |
Description: |
|
In recent work at the Agency for Healthcare Research and Quality, efforts have focused on production of direct estimates of expenditure variables with data from the Medical Expenditure Panel Survey, Household Component (MEPS - HC). For some more common medical expenditures for larger states this work has been very successful. However, quality direct estimates cannot be made for many important variables for many
states. In order to produce better state estimates for many variables, model based small area estimates will be necessary. To produce such estimates, models need to be developed to serve as the basis of these types of estimates.
In this paper relationships among estimates for individual states are examined over time as well as the relationships between estimates for individual states and national estimates over time. We show there are high rank correlations among state estimates over time. Further we show, that most estimates for individual states correlate well with national
estimates over time. The former correlations indicate that values for individual states are somewhat fixed in relation to those of other states over time. The latter correlation indicates that change for individual states closely follows the national change over time.
Finally, we briefly outline how this knowledge might be useful in the development of a Bayesian small area estimation process and propose work proceed in this direction in the efforts to develop better state level estimates with the MEPS - HC data. |
Agency: |
|
Rockville (MD): Agency for Healthcare Research and Quality |
Back
Email Updates
