The purpose of the current study is to investigate whether there was a statistically significant relationship between vertical roof intrusion and the probability of occupant ejection in rollovers that are likely to be covered by Federal Motor Vehicle Safety Standard No. 216 (FMVSS No. 216). If such a relationship did exist, FMVSS No. 216 might affect the number of occupant ejections in rollovers.

The study applies thirty six different statistical models to crash data to model the probability of occupant ejection using a number of explanatory variables, including the amount of vertical roof intrusion. The data is on vehicle occupants who were involved in relevant rollover crashes, and is taken from NASS CDS for years 1997 to 2006 (n = 5,562). Though the study considers a number of different models, it does not find a statistically significant relationship between vertical roof intrusion in relevant rollovers and the probability of complete occupant ejection. When ejections of any degree are considered (whether complete, partial, or of unknown degree), there was a statistically significant relationship in some subpopulations.

Given that no relationship has been found between the amount of vertical roof intrusion and the probability of complete occupant ejection, increasing roof strength is unlikely to impact the number of complete occupant ejections. The study is limited to occupants in rollovers that are likely to be covered by FMVSS No. 216, and to occupants for whom key data, such as the amount of vertical roof intrusion, are available.