FDOT State Safety Office

Crash Reduction Analysis System Hub (CRASH)

Information Guide

June 4, 2012 RL Fitzgerald

Contents
Crash Reduction Factors Spreadsheet ..................................................................................................................................................................................................... 3
Project Evaluation and Selection Method in CRASH ................................................................................................................................................................................ 4
CRF Estimation ..................................................................................................................................................................................................................................... 4
CRF Estimation Example....................................................................................................................................................................................................................... 5
Test of Significance .............................................................................................................................................................................................................................. 6
Multiple Improvements at Single Project Locations ............................................................................................................................................................................ 6
Crash Reduction Factors Spreadsheet
FDOT Safety Office Crash Reduction Analysis Safety Hub (CRASH) program is a safety improvement tools used by FDOT to conduct benefit-cost analyses and
calculate crash reduction factors (CRF). Users are able to lists and download the CRF onto an Excel spreadsheet with the following:

1.
1 As of Date: Date of spreadsheet creation and extraction from the Crash Reduction Analysis Safety Hub (CRASH) program.

2.
2 ID: Improvement Type (countermeasure) identification reference number.

3.
3 Improvement: Improvement Type (countermeasure) description.

4.
4 Number of Projects: Number of projects existing in the database for the corresponding improvement type.

5.
5 Total, Fatal, Injury… (Heading): Crash occurrence type that are selected for evaluation and used for CRF calculation.

6.
6 Crash Reduction Factor: Factors generated by the CRASH program that are used to estimate the effects the corresponding countermeasure has on the
6 crash occurrence type.

7.
7 Statistical Significance (“does is meet the minimum reduction threshold ‘yes’ or ‘no’?”): “Yes” represents that the CRF percentage (absolute value) is
= equal or higher than the minimum significant percent reduction; and “No” represents the CRF percentage (absolute value) that is lower than the
7 minimum significant percent reduction.
6
1

3 4
2

7 6
Project Evaluation and
Selection Method in CRASH
The selection of project locations for safety improvements in CRASH is based on the benefit-cost analysis. In
general, the selected project locations should have the highest benefit-cost ratios and these ratios should be at least
greater than one. The estimated benefit for a specific project is calculated as the estimated number of crashes that
can be prevented as a result of the project implementation, multiplied by a cost-per-crash value. The estimated crash
reduction is in turn estimated as the number of crashes before project improvement multiplied by a crash reduction
factor (CRF) for the specific type of project being evaluated. The estimated cost, on the other hand, includes such
costs as right-of-way (ROW), structure, signal/signing, etc., with consideration for lifecycle and interest rate. Both
the estimated benefit and estimated cost are annualized before they are used to calculate the final benefit-cost ratio.

CRF Estimation
CRFs in CRASH are estimated based on the so-called before-and-after method. This method simply estimates a
CRF as follows:
Crash Rate Before - Crash Rate After
CRF   100
Crash Rate Before

The crash rates for both before and after the implementation of a project are calculated as:
Total Number of Crashes
Crash Rate 
Exposure

where the “Exposure” is usually calculated in million vehicle miles (MVM) of travel, as follows:

Project Section Length in Miles  Mean ADT  Years  365 Days

Exposure 
1,000,000

Thus, the complete formula for calculating a crash rate becomes:

Total Number of Crashes  1,000,000

Crash Rate 
Number of Days  ADT  Project Section Length in Miles

For an intersection or spot improvement, the intersection influence area, typically taken as the ±0.05 miles from the
center of an intersection (i.e., 0.1 mile), is treated as the project section length. However, this influence area may
take on other numbers depending on the type of safety improvement project.

Each crash record would typically include the corresponding average daily traffic (ADT). Therefore, an
approximation of the mean ADT can be calculated as:

Summation of Individual ADTs Associated with each Crash

Mean ADT 
Total Number of Crashes

1
CRF Estimation Example
CRFs are generally calculated based on multiple projects in which the same types of project improvements were
applied. CRASH recommends the use of a minimum of five historical projects for CRF estimation. A warning is
provided when the number of historical projects falls below five.

To illustrate how a CRF is calculated based on multiple projects, statistics for a two-project example are
summarized in the before-and-after tables below:

Before Statistics:
Project 1 2 Total
Total Crashes 332 160 492
Project Section Length 2.3 1.9
Mean ADT 15,836 13,523
Study Period 3 3
Exposure 39.822 28.135 67.957

After Statistics:
Project 1 2 Total
Total Crashes 174 113 287
Project Section Length 2.3 1.9
Mean ADT 15,638 15,630
Study Period 3 3
Exposure 39.384 32.518 71.902

For the 332 accidents associated with Project #1 before project implementation, the mean ADT can be calculated as
follows:
14,935  14,935 ...  16,040  17,794
Mean ADT   15,836
332
The corresponding exposure can then be calculated as follows:

Project Length  Mean ADT  Years  365 Days 2.3  15,836  3  365
Exposure    39.822
1,000,000 1,000,000

The exposure for Project #2 is calculated in the same fashion. Total exposure is then calculated by summing the
exposures from both projects. The “before” crash rate is calculated by dividing the total number of crashes before
project implementation by the total exposure from both projects, as follows:
492
Crash Rate Before   7.240
67.957

The “after” crash rate is similarly calculated:

287
Crash Rate After   3.992
71.902

The CRF for the specific type of improvement project for total crashes can then be calculated as:

Crash Rate Before - Crash Rate After 7.240  3.992

CRF   100%   100%  45%
Crash Rate Before 7.240
CRFs for specific crash type or crash severity can be similarly calculated by including only the specific types of
crashes in place of the total number of crashes above.

2
Test of Significance
The Poisson Comparison of Mean Test is used to determine if the crash reduction is statistically significant (i.e.,
significantly better, significantly worse, or no significant change). The formula for the Poisson Test based on a 95%
confidence level is applied as follows1:

 2.326 b  0.16  0.35 

R     100

 b 

where
R = Minimum significant percent reduction, and
b = Total number of crashes before project implementation.

A CRF is said to be significant when its percentage is equal or higher than R. When a positive CRF (i.e., a project is
expected to reduce crashes) is greater than R, it is said to be “significantly better”. On the other hand, when the
absolute magnitude of a negative CRF (i.e., a project is expected to increase crashes) is greater than R, it is said to
be “significantly worse”. A CRF is said to result in “no significant change” if its absolute magnitude is less than R.

A project type would typically be considered for potential safety improvement at a location when its overall CRF is
expected to cause significant crash reduction. However, there are cases when it is desirable to trade a reduction in
more severe types of crashes as a result of a project improvement type for an increase in the total number of crashes.

Multiple Improvements at Single Project Locations

Multiple types of project improvement are often applied to a project location. In this case, a composite CRF that
reflects the total crash reduction is calculated. CRASH uses the following commonly used formula to calculate the
composite CRF:

CRF  CRF 1  1  CRF 1CRF 2  1  CRF 11  CRF 2 CRF 3  ...

where
CRF = Overall composite CRF,
CRF1 = CRF for the first project improvement,
CRF2 = CRF for the second project improvement, and
CRF3 = CRF for the third project improvement.

1
Methods for Evaluating Highway Safety Improvements, National Cooperative Highway Research Program
(NCHRP) Report 162, Transportation Research Board, National Research Council, Washington, D.C., 1975.