March 25: This post has been corrected.

On Tuesday, we launched an interactive database containing the addresses of more than 6,900 homes whose owners say they were built with tainted drywall.

Information about the addresses in our database came from two sources: a lawsuit filed in New Orleans federal court, and county property tax data. The court records include 4,166 plaintiff complaints representing at least 3,700 homes. The tax data came from 44 counties, representing 4,649 addresses in Florida, Louisiana and Virginia. Special programs have been established in those states to provide tax relief for drywall victims.

We found 1,232 duplicate addresses in the two data sets, representing 2,111 homes. After we accounted for multiple-family dwellings and condominium complexes, the documents revealed that 6,944 individual homes are affected by tainted drywall.

In order to get an accurate count of homes, we first needed to make sure that homes that appear in both lists were counted only once. The two lists used different typographic rules when listing addresses -- for instance, abbreviations weren't consistent between lists -- and some homes in unincorporated areas were listed inconsistently.

We passed each address through the Google Maps Geocoding API to standardize them. We stored the standardized addresses in our database only if Google was able to confirm each address at its highest degree of confidence ("rooftop").

Having standardized the addresses, we were able to find duplicate addresses across the two lists. In cases where we couldn't standardize addresses with the highest degree of confidence, we did a plain text match between the lists.

When an address represented multiple homes -- an apartment building, for example -- we used the number of units reported in the county appraiser data rather than the lawsuit data, because we believe the county appraiser data to be more accurate in this regard.

Federal Data

For some aggregate statistics, we used the database of homes compiled by the Consumer Products Safety Commission, which is based on consumer complaints the agency received. However, because the CPSC data contains city and state and not individual addresses, we were only able to include them as part of state and county aggregates.

Because the CPSC refused to provide specific addresses for the homes in its database, we placed the location of each tainted home in the geographic center of the city that the CPSC listed. This may have caused some inaccuracy in county aggregates in cases where a city spans more than one county.