Methodology and Validation
The method of generating small area estimation (SAE) of the measures is multilevel statistical modeling and post-stratification (MRP) framework. The methods and validations are listed on this website.
Before the 2023 release, the parameters of the multilevel regression model for each measure were applied to the Census population data categorized by age, sex, and race/ethnicity. A Monte Carlo simulation was then used to draw 1,000 random samples to generate the distribution of the estimates and construct 95% CIs. The simulation assumed that the random error for the random effects varied within each of the population categories. Beginning with the 2023 release, the assumption in the Monte Carlo simulation approach was changed so that the random error for the random effects varied only within counties. As a result, the estimated CIs may be wider than previous releases. This updated approach was adopted because it produces CIs similar to the 95% credible intervals generated using hierarchical Bayesian estimation via Markov Chain Monte Carlo, and the approach is computationally efficient. More information is available here: Constructing Statistical Intervals for Small Area Estimates Based on Generalized Linear Mixed Model in Health Surveys.