Nonresponse Bias in the Survey of Occupational Injuries and Illnesses

• Activity Number: 542
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Government Statistics Section
• Author(s): Erin Huband*+ and Patrick Bobbitt
• Companies: and Bureau of Labor Statistics
• Keywords: Non-response bias ; logistic regression

Abstract

The Bureau of Labor Statistics Survey of Occupational Injuries and Illnesses collects data to estimate the number and rates of work-related injuries and illnesses. Participation by private sector employers is mandated by the Occupational Safety and Health Act of 1970. For state and local government establishments, however, state laws determine whether the survey is mandatory. While private sector response rates reflect their mandatory nature, response rates for states in which response is voluntary are low. To determine whether the survey suffers from bias attributable to non-response, government units were classified as either "likely" or "unlikely" respondents using a logistic regression model. Rates of injuries and illnesses for these groups were then compared to determine whether non-response bias exists, and if so, could be measured and corrected. This paper describes the methodology used for this analysis and presents some preliminary results from the analysis.