of left-turn crashes at intersections
may obscure the real relationship between the crash causes
(i.e., intersection characteristics, etc.) and their effects (i.e., leftturn
crashes, etc.). However, at signalized intersections, traffic
flows and signal operations of different approaches are interactive
at intersections; therefore, disaggregating intersections
into approaches will introduce correlation among observations
from the same intersection. In this study, left-turn crashes were
investigated at the approach level by conflicting patterns based
on geometry and traffic-related explanatory variables using the
appropriate statistical models, which are able to analyze correlated
crash frequencies.
1.1. Factors affecting left-turn crash occurrence
Several studies have attempted to quantify the effects of
traffic flow, intersection geometric design features, and traffic
control and operational features on left-turn crash occurrence.
0001-4575/$ – see front matter © 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aap.2007.04.006
X. Wang, M. Abdel-Aty / Accident Analysis and Prevention 40 (2008) 76–88 77
Poch and Mannering (1996) fitted an approach level left-turn
crash frequency model and identified that left-turning movement,
opposing approach volume, type of traffic control, type of
left-turn signal, speed limit, and sight distance all have an effect
on left-turn crashes. Pernia et al. (2002) identified that overall
traffic, number of lanes on a major road, and the presence of
a median are significant. Hauer et al. (1988) assumed that the
frequency of collisions is related to the traffic flows to which
the colliding vehicles belong and not to the sum of the entering
flows.
On signalized intersection approaches, left-turning traffic
can be treated in one of three ways: “permissive”, “compound”
(“protected/permissive” or “permissive/protected”), and
“protected”,1 and numerous studies have been conducted for
evaluating their safety effect. Agent (1987) found where “permissive”
phasing was replaced by “protected/permissive”, the
number of left-turn crashes usually decreased, except on one
approach, where the speed limit was larger than 45 mph.
Upchurch (1991) compared crash rates (left-turn crashes/million
left-turning vehicles) for different left-turn phasing and found
that “compound” signal has a higher crash rate than “permissive”
phasing. Many researchers have reached the conclusion
that “protected/permissive” phasing has more left-turn crashes
than “protected” phasing in before-after studies (Benioff and
Rorabaugh, 1980; Warren, 1985). Lee et al. (1991) found that
there was no significant difference in crash experience between
leading and lagging operations.
It has been found that as the width of a median increases, the
sight distance for left-turning vehicles decreases significantly
(Harwood et al., 1996; Yan and Radwan, 2004). For unprotected
left-turn traffic at signalized intersections, vehicles turning left
from opposing left-turn lanes often restrict each other’s sight
distance (Joshua and Saka, 1992; McCoy et al., 1992). Joshua
and Saka (1992) pointed out that the minimum value of the offset
can be zero, but cannot be a negative value in practical design,
which would results in unsafe conditions.
1.2. Crash modeling
strategyPernia et al. (2002) fitted an
本论文由英语论文网提供整理,提供论文代写,英语论文代写,代写论文,代写英语论文,代写留学生论文,代写英文论文,留学生论文代写相关核心关键词搜索。