Home / Topics / Research Design / Statistical Analysis / Selecting Statistical Analyses / Traditional Measurement of Racial Disparities

Traditional Measurement of Racial Disparities

Posted on December 5, 2011

Benjamin Cook discusses the traditional way of measuring racial disparities.

 

So I think the bulk of research in racial ethnic disparities in health care have either looked at unadjusted differences or race coefficients. So I'll explain some more about that. A simple comparison -- and you'll see these in the agency for Health Care, Research, and Quality national disparities report. They put out one every year. And these are simple means. They use nationally representative data sets.

They just get percentages of, for example, mental health care for those with depression, let's say, or who are likely to have depression. And there's a percentage for African Americans, and there's a percentage for whites, and there's a percentage for Latinos, and if they're sampled, there's a percentage for Asian Americans and a sample percentage for Native Americans, and those are compared. And so that's a very simple unadjusted comparison. So that's one very common way that disparities are measured.

The other very common way that disparities are measured is that a lot of variables are put into a regression model. So now on the left-hand side as a dependent variable, the variable that you're sort of interested in modeling, you put any mental health care, for example. And on the right-hand side, you put indicators of an individual's race or ethnicity and then a number of other variables: their health status, their mental health state, their income, their education, where they live, their employment, their marital status, et cetera, age, sex.
And then what that regression model will do is give you sort of independent contributions to any mental health care. It'll model any mental health care and give you all of those other variables, independent contributions.

So, typically, what is often done for racial disparities research is let's look at the individual, independent contribution of race/ethnicity and see whether that's a significant coefficient. Is that coefficient different from zero? And if it is... if the African American coefficient is greater than zero or less than zero and the comparison group is whites, then you'll say African Americans have greater or less mental health care than whites. Adjusting for all of those other things.

So the inherent definition of that very common statistical technique is that a disparity is something that you have a difference between whites and African Americans, for example, after you've adjusted for everything else that you have in your data set. So you equalize whites and African Americans on income. You equalize them on their geography and their income and marital status, and then what's left, that is a disparity.

Viewing Preferences

 

Downloads

 


Excerpted from an interview with the researcher conducted at the 2011 NHSN Conference held in Miami, FL.

Feedback

Please note that the feedback is viewed only by 4researchers staff and is not intended for communication with individual contributors.

 

Use the form below to submit feedback about this article. If you would like a response, please be sure to include your e-mail address.


More About "Selecting Statistical Analyses"

2:45

Analysis in Randomized Trials

Analyzing Count Data

Show All...

 

More From Benjamin LĂȘ Cook (bio)

3:44

Combining Cross-Sectional and Longitudinal Data

1:29

Equalizing Variables Across Groups

Show All...