Improving the repeatability and reproducibility of belt fit measurement with 6YO and 10YO ATDs

Abstract

In previous work, researchers at the University of Michigan Transportation Research Institute (UMTRI) developed a method for quantifying the belt fit provided by belt-positioning boosters by measuring the belt location relative to the six- and ten-year-old Hybrid-III dummies. In another study, the torso and lap belt scores obtained by this method were found to be closely related to the belt fit obtained by similar-size children across a wide range of booster and belt conditions. The Insurance Institute for Highway Safety (IIHS) adapted the UMTRI procedure to develop a rating system for booster belt fit, but adoption by other labs has been slowed by difficulty in obtaining repeatable results. The current study was undertaken to improve the repeatability and reproducibility across operators of the procedure. The modifications made by IIHS were studied and most incorporated, and a number of other issues were examined through pilot testing. A revised version of the procedure was tested with repeated measurements by three operators in six boosters. The results were analyzed to quantify the variance associated with the operators, the installation of the booster and dummy, and the routing of the belt. The results show that trained operators can perform the procedure with minimal systematic bias across boosters. The variability within booster varies considerably, with some boosters producing higher precision measurements due to the design of their belt-routing features. For any particular booster, the booster and dummy installation process accounts for about half the variability in the belt fit scores with the remainder due to variability in the belt routing and other measurement variability. Based on these findings, multiple measurements of belt fit in each booster are recommended to establish the desired level of confidence in the true belt fit. Straightforward statistical methods involving confidence intervals are recommended for establishing objective test methods. More testing will be needed to determine the reproducibility of the method across laboratories.