Comparing Manual and Automated Industry and Occupation Coding: Accuracy and Cost from the Perspective of the California Health Interview Survey
Royce Park
UCLA Center for Health Policy Research
David Grant
UCLA Center for Health Policy Research
Matt Jans
UCLA
Marisol Frausto
UCLA Center for Health Policy Research
John Rauch
Westat
This study compares manual coding with an automated, computer-assisted coding system developed by the National Institute for Occupational Safety and Health (NIOSH). California Health Interview Survey (CHIS) I&O coding traditionally involves human coders that review and categorize respondent job titles based on verbatim text entries by CATI interviewers. The manual coding scheme uses 2010 Census occupation codes and the 2012 North American Industry Classification System (NAICS). The NIOSH Industry and Occupation Computerized Coding System (NIOCCS) uses an automated coding algorithm to assign I&O codes to text entries. A user can submit multiple records (batch-mode) via a Web interface (http://wwwn.cdc.gov/niosh-nioccs/). We randomly selected 1,000 manually-coded cases from 2013-2014 CHIS and processed them with the online NIOCCS system. Preliminary results suggest a clear benefit from using the NIOCCS as it substantially reduces the time and resources necessary to complete the coding, both in person-hours and project duration. Our final analysis compares reliability of each coding system, and assesses their success for industry and occupation codes separately.