Hi,
We are trying to reconcile the percentage of persons with disabilities in the 5-year 2015-2019 ACS for Colorado with the percentage of persons with disabilities in the 2019 BFRSS.
The percentage from ACS table B26108 shows that approximately 11% of all persons and 13% of persons age 18+ have disabilities. In contrast, the BFRSS percentage for "Calculated Disability" is 20%.
The BRFSS is a survey of persons age 18+
I constructed a confidence interval around the ACS estimates, but the BRFSS value is outside these bounds.
Does anyone have ideas about why these two estimates differ?
I'd appreciate any thoughts about how to reconcile these differences.
TIA
AB
This could be due to a difference in the overall definition of disability between the two surveys or in the way the questions are asked then counted (eg, "are you disabled" versus "do you have any the…
Following up on Cliff Cook's response, can Adam Bickford research and post how the disability is defined in both surveys? That would be beneficial for the record for the rest of us. (Mind you, there…
This could be due to a difference in the overall definition of disability between the two surveys or in the way the questions are asked then counted (eg, "are you disabled" versus "do you have any the following issues...", which is then treated as the basis for disability).
This is not the only discrepancy of this type. There is often a wide gap between the number of children who receive special education services through an IEP versus those counted as disabled by the ACS.
Following up on Cliff Cook's response, can Adam Bickford research and post how the disability is defined in both surveys? That would be beneficial for the record for the rest of us. (Mind you, there's also NHIS, the National Health Interview Survey, a nationwide in-person survey that is generally considered to be higher quality than BRFSS but of course smaller sample size; see https://nhis.ipums.org/nhis/. It will probably have yet another different definition of disability and correspondingly a different estimate).
Adding to the discussion, there are also fielding, sampling, non-response bias, and weighting differences between the two surveys. Here's a brief summary of non-response bias and weighting differences identified in 2013: www.mathematica.org/.../using-american-community-survey-disability-data-to-improve-the-behavioral-risk-factor-surveillance