Activity Number:
|
491
- Methodology and Utilization of Administrative Data
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 6, 2020 : 10:00 AM to 2:00 PM
|
Sponsor:
|
Government Statistics Section
|
Abstract #311006
|
|
Title:
|
Using Bayesian Improved Surname Geocoding (BISG) to Classify Race and Ethnicity in Administrative Employment Data by Industry: A Validation Study
|
Author(s):
|
Ada Harris*
|
Companies:
|
US EEOC
|
Keywords:
|
Race estimation;
Bayesian Improved Surname Geocoding (BISG);
missing data
|
Abstract:
|
The ability to accurately classify an individual’s race and ethnic group is critical to analyzing racial and ethnic disparities in employment. The Investigative Analytics Team (IAT) within the Equal Employment Opportunity Commission (EEOC) currently uses BISG race estimation techniques when race/ethnicity is missing from administrative employment data provided by employers. This study validates the use of the Bayesian Improved Surname Geocoding (BISG) estimation method to produce probabilistic estimates of race/ethnicity to examine racial disparities and assess variations by industry. The BISG uses a person’s Census surname and geography to produce a set of probabilities that a given person belongs to a set of six mutually exclusive racial/ethnic groups. The BISG method is validated using a large sample of administrative employment data classified by industry of applicants or employees who self-report their race/ethnicity. This study also explores using a first name list from internal EEOC data to improve the classifier.
|
Authors who are presenting talks have a * after their name.