Page 17 - tmp_Neat

Chapter 2 Literature review

2.1 Introduction

Road collisions can be analysed based on crash severity and frequency of collisions (Chiou &
Fu, 2013; Jung et al., 2014; Mooren et al., 2014). Crash frequency models predict collisions

which occur at a particular site, while crash severity models predict drivers’ and passengers’

injury severity when involved in collisions. Crash injury severity has become important
among road safety researchers, since it can examine the direct factors influencing the injury

of occupants involved in crashes (Jung et al., 2010). Mathematical models are used in road
safety studies owing to their capability in producing a solid statement for each parameter

(Hughes et al., 2014).

This chapter is divided into seven sections. In the next section, the existing crash severity
models are summarised, while Section 2.3 summarises road collisions involving heavy

vehicles. Next, Section 2.4 provides a literature review of the existing studies focusing on
heavy vehicle crashes at intersections and mid-blocks, while Section 2.5 summarises studies

on the impact of neighbourhood socioeconomic characteristics on crashes. Finally, Section

2.6 provides the limitations of the existing studies and Section 2.7 summarises this chapter.

2.2 Crash Severity Models

Crash severity data have been used in various methods to analyse the injury severity level of

each person (e.g. driver and passengers) or the most seriously injured person involved in
crashes. Crash severity models are mainly categorized as foolows: binary outcome models,

ordered discrete outcome models, and unordered multinomial discrete outcome models
(Savolainen et al., 2011).

The two major types of crash severity models normally used for modelling provide either an

ordered discrete outcome or an unordered multinomial discrete outcome. The ordered
discrete outcome of the dependent variable is categorized as ordinal from low injury to higher

6