TTE 4274:

Transportation Engineering Systems

Instructor: Sabreena Anowar, PhD

Department of Civil, Environmental and Construction Engineering
University of Central Florida

1
Today’s Outline
• Road safety

2
Crash Data
• Police reported crash data is the most reliable and
primary source of crash data
• The national crash databases are usually compiled from
the police reported crash records
• Crash report form: A pre-printed standardized form on which
information can be recorded either in narrative form or on a
checklist
• Other documents:
• Photographs and sketch diagrams of the scene (these are usually
attached to the report),
• Witness information (statements)

• Secondary source: Hospital data, insurance data

Crash Report Form
Florida Crash Report Form
• Crash report forms:
• Florida Department of Transportation (FDOT) Crash
Analysis Reporting System (CARS)
• Signal Four Analytics (S4A) databases
• FDOT CARS and S4A are long and short forms of
crash reports in the State of Florida, respectively
• The long form crash report includes higher injury severity
level or crash related to criminal activities (such as hit-
and-run or Driving Under Influence)
• The Short Form Report is used to report all other types of
traffic crashes
Analysis of Crash Data
• Accident statistics are measures (or estimates) of the
number and severity of accidents
• The reasons for analyzing crash data are to:
• Identify patterns that may exist
• Determine probable causes/factors
• Compare safety conditions among different locations
• The insight gained help develop countermeasures
that will reduce the rate (occurrence) and severity of
future crashes
Analysis of Crash Data
• To facilitate the comparison of safety conditions
among different locations, one or more of the
following procedures have been used:
• Direct comparison of number of crashes
• Direct comparison of crash rates
• Crash patterns
• Statistical comparison
Analysis of Crash Data
• Direct comparison of number of crashes
• Involves computing the number of crashes for the same
duration before and after the implementation of the safety
countermeasure or computing the number of crashes for
the same time period at different locations.
• The comparison may be done by computing the
percentage change in the number of crashes from
which some inferences can be made.
• Disadvantage:
• This procedure does not consider several factors (exposure), such
as the volume at the location or locations
Analysis of Crash Data
• Direct comparison of crash rates
• These rates are determined on the basis of exposure data
• Accident rate generally fall into two broad categories
• Population-based rates
• Area population
• Number of registered vehicles
• Number of licensed driver
• Highway mileage
• Exposure-based rates
• Vehicle-miles traveled
• Vehicle-hours traveled
Analysis of Crash Data
• Direct comparison of crash rates
• Commonly used rates are
• rate per million of entering vehicles (RMEVs) and
• rate per million vehicle-miles (RMVM).

𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟
𝑅𝑀𝐸𝑉 = ∗ 106 [for intersection]
𝐴𝑝𝑝𝑟𝑜𝑎𝑐ℎ 𝐴𝐷𝑇∗𝑑𝑎𝑦𝑠/𝑦𝑒𝑎𝑟

𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟
𝑅𝑀𝑉𝑀 = 𝑑𝑎𝑦𝑠 ∗ 106 [for intersection]
𝐴𝑝𝑝𝑟𝑜𝑎𝑐ℎ 𝐴𝐷𝑇∗ ∗𝑚𝑖𝑙𝑒𝑠 𝑖𝑛 𝑠𝑒𝑐𝑡𝑖𝑜𝑛
𝑦𝑒𝑎𝑟

Where, ADT = Average Daily Traffic Volume

Analysis of Crash Data
• Direct comparison of crash rates
• Commonly used rates are
• rate per million of entering vehicles (RMEVs) and
• rate per million vehicle-miles (RMVM)

𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟
𝑅𝑀𝐸𝑉 = ∗ 106 [for roadway section]
𝐴𝐷𝑇∗𝑑𝑎𝑦𝑠/𝑦𝑒𝑎𝑟

𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟
𝑅𝑀𝑉𝑀 = ∗ 106
𝑑𝑎𝑦𝑠
𝐴𝐷𝑇 ∗ ∗ 𝑚𝑖𝑙𝑒𝑠 𝑖𝑛 𝑠𝑒𝑐𝑡𝑖𝑜𝑛
𝑦𝑒𝑎𝑟
[𝑓𝑜𝑟 𝑟𝑜𝑎𝑑𝑤𝑎𝑦 𝑠𝑒𝑐𝑡𝑖𝑜𝑛]

Where, ADT = Average Daily Traffic Volume

Analysis of Crash Data
Example 1: Several intersections in Orlando had an apparent
increase in collisions last year. One such intersection – Research
park and Alafaya – may need some special attention. It had 13
crashes. The four leg of the intersection had two-way ADT values
of 9671, 2893, 9506 and 2611 vehicles per day last year.
Calculate the intersection’s crash rate, so that it may be compared
with the rate for other similar intersections.
Analysis of Crash Data
• Solution
Approach ADT = (9671+2893+9506+2611)*0.5 = 12340.5

𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟
𝑅𝑀𝐸𝑉 = ∗ 106
𝐴𝑝𝑝𝑟𝑜𝑎𝑐ℎ 𝐴𝐷𝑇 ∗ 𝑑𝑎𝑦𝑠/𝑦𝑒𝑎𝑟

13
= ∗ 106
12340.5 ∗ 365

= 2.886
Analysis of Crash Data
Example 2: A new state highway safety programs offers funds to
improve local roads with RMVM values greater than 1.00. A 6.1
mile section of University Blvd. in Orange county had six crashes
last year. The two way ADT was 755 vehicles per day. Does the
road qualify for the state Program?
Analysis of Crash Data
• Solution
𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟
𝑅𝑀𝑉𝑀 = ∗ 106
𝑑𝑎𝑦𝑠
𝐴𝐷𝑇 ∗ ∗ 𝑚𝑖𝑙𝑒𝑠 𝑖𝑛 𝑠𝑒𝑐𝑡𝑖𝑜𝑛
𝑦𝑒𝑎𝑟

6
= ∗ 106
755 ∗ 365 ∗ 6.1

= 3.57

RMVM = 3.57 > 1. So, the road qualify for the sate program
Analysis of Crash Data
• Direct comparison of crash rates
• Disadvantages of crash rate approach
• Although the use of crash rates considers the effect of exposure,
it does not take into consideration other factors, usually referred
to as confounding factors, that may affect the occurrence of
crashes.
• Research has also shown that it tends to be biased toward low
volume sites.
• Care should therefore be taken in making conclusions by simply
comparing crash rates.
Analysis of Crash Data
• Crash Patterns
• Two commonly used techniques to determine crash
patterns are
• Expected value analysis
• The analysis is carried out by determining the average number of a
specific type of crash occurring at several locations with similar
geometric and traffic characteristics
• Cluster analysis
• Identification of a particular characteristic from the crash data
obtained at a site
Analysis of Crash Data
Example 3 : Data collected for three consecutive years at an
intersection study site show that 14 rear-end collisions and 10
left-turn collisions occurred during a 3-year period. Data
collected at 10 other intersections with similar geometric and
traffic characteristics give the information shown in the Table.
Determine whether any type of crash is overrepresented at the
study site for a 95% confidence level (Z = 1.96).
Analysis of Crash Data
Analysis of Crash Data
Solution :
Average number of rear-end collisions at 10 control sites = 7.40

Standard deviation of rear-end collisions at 10 control sites = 1.50

Expected range (95% confidence level) = 7.40±1.5×1.96 = 10.34

Number of rear-end collisions at study site = 14

Rear-end collisions are therefore overrepresented at the study

site at 95% confidence level, since 14 > 10.34.
Analysis of Crash Data
Solution :
Average number of left-turn collisions at 10 control sites = 6.90

Standard deviation of left-turn collisions at 10 control sites = 3.07

Expected range (95% confidence level) = 6.90±3.07×1.96= 12.92

Number of left-turn collisions at study site = 10

Left-turn collisions are not overrepresented at the study site

since 10 < 12.92.
Analysis of Crash Data
• Statistical Comparison
• There are several statistical tests that can be used to
compare the level of safety between two or more sites
• The tests are
• the t-test for comparison of two means
• the Proportionality test for the comparison of two proportions
• the KruskalWallis H test
• the Empirical Bayes Method
• The first three of these four tests are used for Hypothesis
Testing
Analysis of Crash Data
• Statistical Comparison – Hypothesis Testing
• Hypothesis Testing is a procedure used to make inferences
on populations based on samples from these populations
• In hypothesis testing, the analyst first assumes that the
means of two or more independent samples are equal.
This assumption is referred to as the Null Hypothesis -
This is a claim that is initially assumed to be true
• The Alternative Hypothesis nullifies this assumption - it is
a statement that contradicts the null hypothesis

A STATISTICAL HYPOTHESIS is a statement about the parameters of one or

more populations.
Analysis of Crash Data
• Statistical Comparison – Hypothesis Testing
• Let us assume
Null hypothesis 𝑯𝟎 : 𝝁𝟏 = 𝝁𝟐
Alternate hypothesis 𝑯𝟏 : 𝝁𝟏 ≠ 𝝁𝟐

Because the alternative hypothesis specifies values of 𝜇1 that

could be either greater or less than 𝜇2
Crashes per year,
it is called a two-sided alternative hypothesis
Analysis of Crash Data
• Statistical Comparison – Hypothesis Testing
• Let us assume

Null hypothesis 𝑯𝟎 : 𝝁𝟏 = 𝝁𝟐 Or Null hypothesis 𝑯𝟎 : 𝛍𝟏 = 𝝁𝟐

Alternate hypothesis 𝑯𝟏 : 𝝁𝟏 > 𝝁𝟐 Alternate hypothesis 𝑯𝟏 : 𝝁𝟏 < 𝝁𝟐

Because the alternative hypothesis specifies values of 𝜇1 that is

greater/less than 𝜇2
Crashes per year,
it is called a one-sided alternative hypothesis
Analysis of Crash Data
• Statistical Comparison – Hypothesis Testing
• We will always state the null hypothesis as an equality
claim
• When the alternative hypothesis is stated with the < sign
• the implicit claim in the null hypothesis can be taken as ≥
• When the alternative hypothesis is stated with the > sign
• the implicit claim in the null hypothesis can be taken as ≤
Analysis of Crash Data
• T-test
• The estimated means and/or variances of the populations
obtained from the data sets are used to test the hypothesis
by computing the test statistic T which is then compared
with a similar value t obtained from the theoretical
distribution
Analysis of Crash Data
• T-test
• The theoretical value of t depends on the degrees of
freedom and the level of significance used for the test, and
whether the test is for a one- or two-tail test
• Degree of freedom for the t-distribution is (n1+ n2-2)
• The level of significance is the probability of rejecting the
null hypothesis when it is true
• The most commonly used value in traffic safety studies is
5 percent although 10 percent is sometimes used
• When the test is one-tail, the t value selected is for 𝛼
and when it is two-tail, the t value selected is 𝛼/2
Analysis of Crash Data
• T-test
• The region of rejection of the null hypothesis

Critical Region Acceptance Region Critical Region

t-table
Analysis of Crash Data
Example 4 : An engineer wishing to test whether large trucks are
significantly involved in crashes on rural two-lane highways than
on rural multilane highways, provided data for a period of 5 years
on randomly-selected rural two-lane and multilane highways in
her district, as given in Table below. These sections are each of
the same length, with similar large truck percentages, AADT, and
posted speed limits. Using the t-test, determine whether it can be
concluded that large-truck-involved crashes on rural two-lane
highways are significantly higher than those on rural multilane
highways at a significance level of 5%.
Analysis of Crash Data
t-table
Analysis of Crash Data
• Proportionality Test
• This is used to compare two independent proportions.
• For example, it can be used to compare the proportion of
fatal and injury crashes at an unsignalized intersection
with that at signalized intersections.
• In this case, the null hypothesis is

• The alternative hypothesis is

Analysis of Crash Data
• Proportionality Test
• The test statistic is calculated from equation below and
the value obtained compared with 𝑍𝛼, the standard normal
variant corresponding to a significance level of 𝛼.
Z-table
Analysis of Crash Data
Example 5: Table below gives the number of fatal and injury
(F& I) crashes and property damage only (PDO) crashes that
occurred over the same period at randomly-selected unsignalized
and signalized intersections with similar approach volumes and
geometric characteristics. Using the proportionality test,
determine whether, based on this data set, it can be concluded that
the proportion of fatal and injury crashes is significantly higher at
unsignalized intersections than at signalized intersections at a 5%
significance level.
Analysis of Crash Data
Example
Z-table
Hazardous Locations
• A primary function of an accident record system is to
regularly identify locations with an unusually high
rate of accidents and/or injuries and/of fatalities.
• Hazardous locations are sites where crash
frequencies, calculated on the basis of the same
exposure data, are higher than the expected value for
other similar locations or conditions.
Hazardous Locations
• A technique that is used to identify possible
hazardous locations is known as
• the critical rate method (CR) method
• CR method incorporates the traffic volume to
determine if the crash rate at a particular location is
significantly higher than the average for the type of
facility.
Hazardous Locations
• The critical crash rate method involves the following
expression.
Hazardous Locations
• The crash ratio of actual crash occurrence for the
segment being studied with respect to the critical rate
is determined.
• Locations with critical crash ratio greater than 1 are
considered hazardous.
Hazardous Locations
Example 5 : An urban arterial street segment 0.2 mile long has
an average annual daily traffic (AADT) of 15,400 veh/day. In a
three-year period, there have been eight crashes resulting in death
and/or injuries and 15 involving property damage only. The
statewide average crash experience for similar types of roadway
is 375 per 100 million vehicle miles (MVM) for a three-year
period of which 120 involved death and/or injury and 255 caused
property damage only. Is the 0.2 mile long street segment
hazardous? In identifying hazardous locations, consider that a
single death/injury crash is equivalent to three property
damage crashes. Use a 95% confidence level.
Hazardous Locations
Hazardous Locations
 Transportation Safety Analysis

Transportation Safety Analysis

Statistical analysis by Conflict Naturalistic Driving

using police reported technique driving study Simulation
crash database study

Crash Crash
frequency severity
analysis analysis
Transportation Safety Analysis

Total Crashes

Fatal Injury Crashes

Major Injury Crashes

Minor Injury Crashes

No Injury Crashes
Typical Crash Types
• Based on severity
• Fatal
• Personal injury
• Incapacitating/major
• Non-incapacitating/minor
• Possible injury/minimal
• Property damage only (PDO)
Typical Crash Types
• Based on the road users involved
• Motor vehicle crashes
• Non-motorized crashes
• Pedestrian
• Bicyclist
• Heavy vehicle crashes
Typical Crash Types
• Based on the number of vehicles involved
• Single vehicle
• Ran-off-road collisions
• Collisions with fallen rocks or debris in the road
• Rollover crashes – tripped or untripped
• Collisions with animals
Typical Crash Types
• Based on the number of vehicles involved
• Multivehicle (two or more)
• Rear end
• Sideswipe
• Angular
• Head on
Typical Crash Types
1 Rear-ender
Rear-end
2 Rear-ended

3 Near-sideswipe
Sideswipe
4 Far-sideswipe

5 Near-angular

Angular 6 Far-angular

7 Short-side angular

Head-on 8 Head-on
References
 https://safety.fhwa.dot.gov/hsip/
 https://www.massdot.state.ma.us/Portals/8/docs/designGuide/CH_6_a.pdf
 Fundamentals of Transportation Engineering: A Multimodal Systems
Approach. By Jon D. Fricker and Robert K. Whitford. Pearson
International Edition.
 Traffic Engineering. By Roger P. Roess, Elena S. Prassas and William R.
McShane. Fourth Edition.
 Traffic & Highway Engineering. By Nicholas J. Garber and Lester A. Hoel.
Fourth Edition.