The safety of children in motor vehicle crashes is a major concern. Although Child Restraint Systems (CRS) are required by law for their protection, children are still exposed to the risk of injuries ranging from minor to fatal. The effect of restraint use is studied under different risk scenarios consisting of some possible contributors to injury risk:​ the restraint use, impact type, injury severity, and age of crash involved children. The data are analyzed at rather a micro level to estimate the relative risks associated with risk scenarios and test for possible risk factor interactions. Specifically, children of age groups:​ infants, 1 to 3, 4 to 8, and 9 to 12 year olds, who were either uninjured, or sustained minor to fatal injuries in frontal, side, rear-end, or rollover crashes, formed the study population. Some data concerns are also raised in course of the study.

The analysis dataset is extracted from the National Automotive Sampling System– Crashworthiness Data System (NASS-CDS). The study population is segmented, based on three injury risk factors:​ age group, restraint use, and impact crash mode. Clusters of data are identified in which the quantity of data are limited or contains insufficient ‘information’, thereby suggesting the importance of collecting more data in certain segments of the population. Injury risk factors may have an individual as well as joint influence on the outcome (injury severity) of a crash. The significance of the overall association between these factors is tested by the contingency analysis. This, however, provides only a broad picture of the phenomenon. Configural frequency analysis is used to identify the factor-based clusters of the children population that show strong to complete absence of factor association. The estimates of the relative risks associated with different clusters are obtained to compare the two groups of children:​ restrained and unrestrained. In general, the restrained children were found much safer against injuries.