Racial Disparities in Mental Health: A Data-Driven Learning Guide

Application

This exercise explores the relationship between race and reported levels of depression using crosstabulation and linear regression.

Crosstabulation

Depression was measured with the question, "how often do you feel so sad that nothing could cheer you up?" (V226) The response options were "very often," "fairly often," "not too often," "hardly ever," and "never." Fewer than 10 percent of respondents said "very often" or "fairly often," so we created a new variable (called V226NEW) that combined these two categories and reversed the coding so that higher numbers meant higher frequency of depression.

Begin by examining the distribution of responses to V226NEW. This will show you how many people chose each answer category. What percent of people said that they hardly ever felt sad and how many people was that?

In this survey, respondents were asked to identify the race that best describes them. Look at the distribution of responses to this variable (V118). As you can see, very few selected Asian, American Indian, Hispanic, or other. For this exercise we dropped these respondents in order to compare only Black and White respondents. V118NEW is the recoded where Black is coded as "1" and White is coded as "0".

To see whether Blacks and Whites report feeling sad with similar frequencies, look at a crosstab of V226new with V118new. What do you find? What percent of Whites report feeling sad very or fairly often? What percent of Blacks do? What about never feeling sad?

Multiple Regression

One concern when studying race and mental health is that the relationship between race and depression may be caused by a third factor. Next we will use multiple regression to control for potential confounding factors and isolate the relationship between race and reports of depression. In addition to race and depression, we will include the following measures (the links take you to frequency distributions of each):

• Education level (V1001new) will be included in a measure of socioeconomic status. Previous studies have shown that depression rates may be higher among people with lower socioeconomic standing.
• Marital status (V107new) will be included because research has shown that rates of depression tend to be higher among single people than married people.
• Discrimination (V125new)

Note, we recoded education (V1001) into V1001new which collapses the answers from specific years of schooling completed to categories that represent less than high school, high school degree, and so on. We recoded V107 so that respondents were considered either married or not married rather than retaining all categories of "not married" (divorced, single, etc.). Lastly, we removed the two people who answered "Don't know" to the question about discrimination, creating V125new.

Can you think of other factors that might confound the relationship between race and feelings of depression?

First, we start by including only race and frequency of feeling sad in the regression. How would you interpret the outcome? How would you compare the regression results to the results of the crosstabulation you conducted? (Hint: to understand regression results, look at the B value and the Probability. The B value tells the strength and direction of the relationship and the probability tells you if the relationship is significant. If you need more help with the interpretation, see the Interpretation Guide on the next tab.

Next the control variables are added to the regression.

What do you find? How would you interpret the coefficient on race? How has the coefficient changed compared to the previous model without control variables? How would you interpret the coefficients on the other variables in the model? What can you say about the relationship between race and depression overall based on these models?

Note: The online data analysis system (DAS) used on this site uses a system called Survey Documentation and Analysis (SDA), developed and maintained by the Computer-assisted Survey Methods Program (CSM) at the University of California, Berkeley. Documentation for DAS/SDA can be found on their Web site.

Goal & Concept - Dataset - Application - Interpretation & Summary - Bibliography