Predictive Modeling of Errors in Child Restraint System Use

• Activity Number: 476 - Distracted Driving and Other Transportation Considerations
• Type: Contributed
• Date/Time: Wednesday, July 31, 2019 : 8:30 AM to 10:20 AM
• Sponsor: Transportation Statistics Interest Group
• Author(s): Elizabeth Petraglia* and Doreen De Leonardis and Amy Benedick
• Companies: Westat and Westat and Westat
• Keywords: CRS; child passenger safety; car seats; predictive modeling; clustered data

Abstract

Properly used and installed child restraint systems (CRSs) save children’s lives in motor vehicle crashes. However, research shows that most parents and caregivers make one or more errors when installing or using a CRS. The current study asked 150 participants to select the appropriate CRS for a doll representing a child of a given age, height, and weight, properly install the CRS in a vehicle, and secure the doll in the CRS. The installations were then checked by a certified technician to record any errors. Each participant was asked to repeat this process with four different dolls, in a randomly assigned order, vehicles and seating positions. For each doll, the participant was provided with a variety of CRS types from which to choose.

Previous analyses have shown differences in error rates by trial-level characteristics such as doll, CRS type, or vehicle. This paper uses predictive modeling for clustered data to determine which, if any, participant-level characteristics (e.g., age, sex, race, number of children, experience) are predictive of more errors or of certain types of errors. The findings may be used in practice to target outreach or education.