The defendant expert’s reliance on the theoretical injury risk derived from experimental cadaver testing data and two dummy crash tests lacks generalizability to real world, but there were other technical problems with the use of the experimental data, as well.

First, the expert failed to note that the relative risk between what he estimated to be “a little less than 50%” and “around 80%” from the graph reproduced previously in this chapter is 1.6 (80% is 1.6 times 50%), and this is insufficient to allow for the conclusion that the use of a belt would have resulted in a lesser injury to the occupant, on a more probable than not basis. Using these values, only 37% of the total injury risk, or 0.6 of the 1.6 relative risk, would be attributable to the failure to use the seat belt, and thus the legal standard for what is more probable than not would not be met by this evidence.

Further, the expert failed to note the error rate within the data that he relied upon. An examination of the data that the risk curve was based on revealed a total of 54 cadaver skull impact experiments. A reanalysis of the same data, including a 95% confidence interval, was performed and is depicted in Fig. 11.10.