Department

University of Tennessee at Chattanooga. Dept. of Psychology

Publisher

University of Tennessee at Chattanooga

Place of Publication

Chattanooga (Tenn.)

Abstract

Occupational injuries extract a large social and economic toll on workers and companies alike. Behavior-based safety (BBS) is a method for reducing workplace accidents by identifying at-risk behaviors and reinforcing safe behaviors. BBS requires the direct observation and collection of specific behaviors (pinpoints) by managers and coworkers to be effective. Observers record their findings on behavioral checklists that contain free- and fixed-response items. Literature shows that behavioral checklists are more effective when done at a measured pace, when responses to the checklists are varied, when greater context is given by free response, and when checklist items are more specific. While there has been previous research demonstrating reductions in injuries associated with the quantity of behavioral observations, there has not been much focus on quantitatively evaluating the quality of observation reporting and the impact on injuries. This study will investigate checklist quality through natural language processing, a technique which uses machine learning to accomplish human-like language processing, and assess how that quality moderates the relationship between number of observations and incident prevention. Text analytics will assess presence or absence of free-text, length of free-text, and quality of checklist prompts and free text (use of action words and temporal words).

Date

10-16-2021

Subject

Industrial and organizational psychology

Document Type

posters

Language

English

Rights

http://rightsstatements.org/vocab/InC/1.0/

License

http://creativecommons.org/licenses/by/4.0/

Share

COinS
 

Quality or quantity: using natural language processing as to assess safety checklist quality

Occupational injuries extract a large social and economic toll on workers and companies alike. Behavior-based safety (BBS) is a method for reducing workplace accidents by identifying at-risk behaviors and reinforcing safe behaviors. BBS requires the direct observation and collection of specific behaviors (pinpoints) by managers and coworkers to be effective. Observers record their findings on behavioral checklists that contain free- and fixed-response items. Literature shows that behavioral checklists are more effective when done at a measured pace, when responses to the checklists are varied, when greater context is given by free response, and when checklist items are more specific. While there has been previous research demonstrating reductions in injuries associated with the quantity of behavioral observations, there has not been much focus on quantitatively evaluating the quality of observation reporting and the impact on injuries. This study will investigate checklist quality through natural language processing, a technique which uses machine learning to accomplish human-like language processing, and assess how that quality moderates the relationship between number of observations and incident prevention. Text analytics will assess presence or absence of free-text, length of free-text, and quality of checklist prompts and free text (use of action words and temporal words).