Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Page 99 Appendix D State-level MNCS model prediction vs. performance The panels in this Appendix compare the predicted number of fatalities for each state with the number actually experienced, computed as the ratio between observed and predicted. The MNCS (model not considering state) model was used for the purpose, because the goal was to compare how states differed from the prediction, in order to identify states that had more or fewer traffic fatalities than predicted by the model. In the panels, the horizontal line represents no difference from predicted. The dashed line represents the value obtained by dividing the observed number of traffic fatalities by the number of fatalities predicted by the MNCS model. Lines above the horizontal line represent a greater number than predicted, and those below represent fewer traffic than predicted. The greyed band represents the 90% confidence interval of the ratio. Small states with relatively few traffic fatalities tended to have more volatility in their number of traffic fatalities, so the ratios were less stable. Predictions for larger states tended to be more stable and showed less fluctuation. For several states, the ratio was close to the horizontal line (of no difference), showing that the MCS model prediction was close to the outcome. In these states, the model explained most of the variation. These states included Alabama, Colorado, Florida, Mississippi, North Carolina, and Pennsylvania. In other states, the ratios were persistently less than one. These states may have some features or programs that consistently resulted in a safer traffic environment. These states include Indiana, Massachusetts, Michigan, Minnesota, New York, Ohio, and Vermont. In other states, the dash line was consistently above the horizontal line, indicating that the number of fatalities was persistently greater than predicted by the MNCS model. A comparison of states that had fewer traffic fatalities than predicted by the model with the states that had more would be useful to identify characteristics, programs, and interventions that contributed to the difference in safety environment.