|
Abstract:
|
The Asthma Call-back Survey (ACBS) is conducted with BRFSS respondents who reported a child asthma diagnosis through the Random Child Selection (RCS) and Asthma Prevalence Modules. Since the ACBS sample size is limited by the RCS module sample size, most states did not have sufficient sample in a single year to produce reliable weights and required combining and reweighting multiple years of ACBS data. In 2011, BRFSS started using the "raking" methodology to weight the data. However, the ACBS continued to use post-stratification because the year that states started collecting cell phone data varied from 2011 to 2015. To incorporate yearly data variation and be consistent with BRFSS, we used raking to reweight the combined child ACBS data. We tested and defined a set of variables (year of survey, cell phone participation, sex, age, and race) as raking margins. The collapsed criterion for margins was a minimum 2.5% of sample. We used a tolerance of 0.025% points and 75 as the maximum number of iterations. Finally, we compared the estimates of key measure variables and the mean square errors from post-stratification and raking to evaluate the reliability of the raking methodology.
|
