Page 102 - tmp_Neat

The study has identified the socioeconomic characteristics of neighbourhoods where road

users live and where crashes occur on road-user injury severity involving heavy vehicle

crashes. To enhance heavy-vehicle safety, it is important to emphasise that these
neighbourhood socio-demographic characteristics should be used in addition to the

information provided by the standard collision hotspot analysis.

6.1.3 Heavy-vehicle angle collisions

On the other hand, study three (Chapter 5) examined factors contributing to occupant injury
severity experienced in angle collisions involving heavy vehicles using binary, scobit

(skewed logistic) and random parameter logit models . This study provides an understanding

of factors contributing to heavy-vehicle angle crashes. In addition, in this study, some
advanced crash severity models such as binary logit, skewed logistic and random parameter

logit models were compared to determine the advantages and weaknesses of these models.
The key findings of factors contributing to heavy-vehicle angle crashes and the statistical

model comparisons are summarised below:

 Severe injury in occupants is more likely to relate to females, younger people

and older adults, occupants not wearing safety restraints and being ejected from
the vehicle, lone drivers, occupants in vehicles which experienced major and

extensive damage, vehicles that catch fire, motorcycles and vehicles impacted on

the right front door area, and occupants in crashes at night, crashes on high-
speed roads and crashes that attended by police.

 In terms of the models' goodness-of-fit, the Scobit model fits the data better than

the standard binary logit and random parameter logit models. This finding was
partially to be expected, due to the possible violation of the symmetry

assumption due to the imbalance in the dependent variable. In addition, uniform

distribution was slightly better than a normal distribution in the random
parameter logit model.

82