FHWA's Interactive Highway Safety Design Model (IHSDM) can readily measure the nominal safety of a road by comparing its design features to prevailing design criteria.

Nominal and Substantive Safety Thresholds

One can readily measure the nominal safety of a road by comparing its design features (lane width, shoulder width, sight distance, curvature, grades, roadside features) to prevailing design criteria. One module of FHWA's Interactive Highway Safety Design Model (IHSDM), a Policy Review Module, is intended to do just that. The vision of FHWA is that the IHSDM will become a standard best practice as a design diagnostic and decision making tool. Appendix F provides an overview of FHWA's IHSDM.

Similarly, one can measure or characterize an existing highway's substantive safety (i.e., define the nature and extent of the safety problem) by determining the frequency, type, severity, and other characteristics of crashes, as well as other information (most importantly, its traffic volume). Here, best practices call for comparing the actual performance of a road with some established benchmark or comparison figure.

The expected safety performance of any road is strongly related to its context, defined by the following:
* Traffic volume
* Location (rural, urban, suburban)
* Functional classification (controlled access, arterial, collector, local)
* Facility type (two-lane, multi-lane undivided, multi-lane divided)
* Terrain (mountainous, rolling, level)
* Roadway segment (mid-block or typical section, intersection, including type of intersection traffic control)
* Surrounding land use (number of driveways, commercial versus residential; associated pedestrian activity)

Typical best practices are to compare the safety performance of a particular highway with a relevant statewide average or expected value for that facility type. Thus, a meaningful review of a two-lane rural highway would involve comparing it to other similar two-lane rural highways (not to all highways or other highway types). Most states compile statistics that describe the mean crash rate, characteristics of crashes (multi-vehicle, single vehicle) and their severity (percent resulting in an injury or fatality) to enable judgments about substantive safety.

Exhibit F-4 (CSD_153)
Representative Accident Rates by Highway Types
The values in Exhibit F-4 (CSD_153) are representative and for general reference only. National statistics for crash rates by different highway types are not available. Care should be taken in comparing statistics from different states, as there may be many differences between states in matters such as reporting levels (minimum crash severity requiring a police report to be filed with the state), data quality, and even definitions of severity, type, or other features. Also, geographic and climate differences can produce differences in overall crash rates between states. Best practices generally call for using a state-specific database or table. Appendix F contains a description of Iowa DOT's "best practice" safety data analysis tool - SAVER: E5, and a discussion of fundamental architecture of the Design Decision Support System from NCHRP Report 430. See also NCHRP Report 430 for more information on crash data quality issues.

Another method for determining the substantive safety of a highway is to compare its performance with accepted crash prediction models. FHWA's Interactive Highway Safety Design Model established models for predicting crashes for two-lane rural highway segments and intersections. Other models published in the technical literature provide insights as to expected performance. They also provide means of testing or describing the expected effects of a different design alternative, and of quantifying safety impacts of a design decision.

Exhibit F-5 (CSD_163) illustrates how knowledge of nominal and substantive safety can influence the overall approach to problem definition and solution. Every highway segment or project can be categorized as being nominally safe or unsafe; and as substantively safe or unsafe. A two-by-two framework thus captures all possibilities. Highway or road projects that may be nominally unsafe but substantively safe may be candidates as 3R projects (assuming a significant mobility issue is not present), which implies less stringent design criteria. Or, for such projects the designer may be more willing to accept a design exception if the context warrants this. Projects that involve a road that is known to be substantively unsafe but nominally safe require special targeted effort to deal with the safety problem. For highways or roads that are both nominally and substantively unsafe, reconstruction to full standards and a reluctance to accept a design exception may be appropriate.

Finally, a project involving a proposed new road has by definition no existing substantive safety performance. For such projects a focus on nominal safety - adherence to design criteria is the best approach. Designers should be reluctant to plan and design a road on newly acquired right-of-way with significant geometric design exceptions.