TY  - JOUR
PY  - 2023//
TI  - Crash factors analysis resulted child fatalities using logistic regression model
JO  - AIP conference proceedings
A1  - Mahmud, S. M. Sohel
A1  - Islam, Monzurul
SP  - e040004
EP  - e040004
VL  - 2713
IS  - 1
N2  - Children fatalities due to road traffic crashes are a concerning issue for a long time in Bangladesh, particularly in an urban area like the megacity Dhaka. Several studies have been found in the literature on this problem but almost all of them are mainly focused on the magnitude, trends, general characteristics etc. using police-reported crash data. This study attempts to identify the distinct characteristics of crashes that resulted from at least a child fatality than the other crashes where no child fatalities occurred through a systematic approach. To do that logistic regression modelling techniques are employed using press media reported data from 2018 to 2020. The analysis has focused on the propensity of different factors involved in child fatalities and identified the sticking factors mainly related to characteristics of crashes including time, crash type, vehicular involvement, the significance of the geographical area, child as a pedestrian hit by different vehicles etc. Altogether 82 relevant factors are tested, and 24 variables are found significant. The factors that come upon positively correlated are daytime, autumn season, April month, Major fatal crash, hit pedestrian, hit an object, sideswipe, pedestrian hit by a car, autorickshaw and motorized three-wheelers. On the other hand, Dhaka city, bus, truck, car with light 4 wheelers, pedestrian hit by motorcycle are negatively correlated i.e., the variable decreases the likelihood of child fatalities in a crash. Hit pedestrian by car is the most dominant and strongest child fatality risk factor with odds of 16.7. <br><br>RESULTS in this comparative analysis and modelling would shed light on child traffic safety improvement strategies and policy interventions. Finally, the paper has ended up with some safety improvement policy recommendations and the way forward for future research directions.<p />  <p>Language: en</p>
LA  - en
SN  - 0094-243X
UR  - http://dx.doi.org/10.1063/5.0129760
ID  - ref1
ER  -