Using US fatality data from 1975 to 2003, Levitt (2008) shows that child safety seats do not significantly reduce fatalities for children aged two to six as compared to standard seat belts. Although we were unable to gain access to the original programs and dataset used, we were able to replicate Levitt's (2008) findings almost exactly. We extend Levitt (2008) by showing that the findings also hold for the years 2004 to 2011 despite changing driver characteristics and restraint use patterns. We fail to find evidence that SUVs provide additional safety for children.
scientific replication Steven Levitt child safety seats improper use SUVs fatalities FARS