For example, black drivers were more than twice as likely to be searched during vehicle stops than white drivers, even after investigators considered non-race variables.
local ARIC and SAHS models with one combined race variable for Malays and Asian Indians had better model fit than their counterparts with two separate race variables.
Other information on English proficiency, ethnicity, and race variables were lacking.
We explored interactions between the level 1 race variables and socioeconomic (Medicaid and uninsured) and geographic (metropolitan and state) variables.
The R gradually improved after adding age or race variables, or both, although the degree of improvement was not large.
For local SAHS models, the Akaike Information Criterion AICC) for models with one and two race variables were 588.6 and 590.6, respectively.
For the local version of the ARIC and SAHS models, we initially had two separate dummy variables for Malays and Asian Indians, but these two dummy variables were combined when the goodness of fit for model with one combined race variable is better than using two separate race variables.
Prior to the modeling process, we recoded the Race variable into Black (20.3%), Hispanic (47.8%), White (27.9%) and Other or Unknown (3.7%).
Given that the sample contains no Hispanics and whites are the omitted category, the share other race variable refers to the share that is non-Hispanic, non-black, and non-white.
If one or two covariates have an absolute standardized difference above 10% after matching (such as the Hispanic and White race variable at 14.94%), the two groups are still considered to be sufficiently balanced overall, and analysis can proceed so long as the average standardized difference is below 10% (Rosenbaum and Rubin 1983).
