Here’s how we did this query. It is similar to most of the queries we used in our analysis.
For the opening example in our story, we combined all of the Home Mortgage Disclosure Act data for Virginia for 2012 and 2011 into a single table in Microsoft Access:
That table had records on 184,463 home loan applications. One of the fields showed whether the application was approved or denied; another field showed the race or ethnicity of the loan applicants:
Other columns also showed the applicants’ income and the size of the loan they were seeking (in thousands of dollars):
In Access, we designed a cross-tabulation query. We told Access to summarize the data by crossing the “race-ethnicity” column with the “approved-denied” column. To make sure we were “comparing apples to apples,” we told Access to do this only for records in which the applicants’ income was between $85,000 and $90,000 and in which the loan amount requested was between $200,000 and $225,000:
When we ran the query, Access showed how many loan applications were approved and denied for each racial/ethnic group:
Lastly, we copied that data into Excel to compute the denial rates — the percentage of the loans that were denied for each racial-ethnic group:
That showed that the denial rates for African American and Latino applicants were about three times the denial rate for non-Hispanic white applicants.
Almost all of the queries we ran showed that loan applications from black and Hispanics were more likely to be rejected than applications from whites. In many cases, the denial rates for African Americans and Latinos were approximately double the denial rates for Caucasians. This was true in Maryland as well as in Virginia.