1. Investigation of discrepancies in South Carolina traffic collision forms
- Author
-
Jackson Wegmet, Nathan Huynh, Luu Van Le, Hai Ngoc Duong, Minh Cong Chu, Mahyar Madarshahian, and Chowdhury Siddiqui
- Subjects
Police crash report form ,Traffic collisions ,Structural equation modeling ,Multiple linear regression ,Misclassification ,Discrepancies ,Transportation and communications ,HE1-9990 - Abstract
The aim of this study is to improve the accuracy of information recorded in the South Carolina traffic collision forms. To accomplish this, it examines 200 forms containing information about fatal crashes in work zones between 2014 and 2020 to determine how many discrepancies exist between the written narrative and other fields. In addition to obtaining these statistics, this study seeks to identify factors that influence discrepancies. To test the hypothesis that crash complexity and weather influence the investigating officer’s level of processing (a theory developed by Craik and Lockhart in 1972), and consequentially his/her ability to complete the traffic collision form accurately, a structural equation model (SEM) is developed. The SEM is used to explain the relationships between measured variables and latent variables and the relationships between latent variables (crash characteristics, weather conditions, and level of processing). SEM results show that increases in collision speed, number of units, number of events, and temperature resulted in an increase in the number of words and characters written in the narrative, whereas increases in precipitation and humidity resulted in a decrease in the number of words and characters written in the narrative. Notably, the number of discrepancies was not statistically significant, suggesting crash and weather-related factors do not affect an officer’s reporting accuracy. A multiple linear regression model is also developed to identify factors that influence a form field’s frequency of discrepancies. The form field’s level of difficulty and its number of inputs are found to be statistically significant.
- Published
- 2024
- Full Text
- View/download PDF