Crash Frequency, Exposure, and Rates

Crash Frequency, Exposure, and Rates

Define the numerator and period

Crash frequency counts events in a stated place and period. An annual average divides a multi-year total by the number of years, provided the periods and definitions are comparable. Count people separately from crashes. A single fatal crash can have several fatalities, and an injury crash can contain occupants with different injury statuses.

Frequency describes the amount of harm or events an agency must address. It does not normalize travel exposure. Rates divide a count by a defined exposure measure, helping answer a different question. Neither measure automatically establishes a treatable site condition, and a high rate at a low-volume site can be unstable.

State what is counted: all reported crashes, fatal and injury crashes, fatal persons, or a particular collision type. State whether the period includes construction, a changed reporting threshold, or a different road configuration. Precision in definitions is more useful than a formula applied to mismatched inputs.

Calculate segment exposure with consistent units

For an illustrative segment with stable AADT, length, and a 365-day annual approximation:

$$\text{VMT}=\text{AADT}\times365\times L\times Y$$

$L$ is segment length in miles and $Y$ is years. AADT is vehicles per day. Thus the product is vehicle-miles, not vehicle counts alone. Use an appropriate time-varying estimate if volume or configuration changed. Document whether the AADT represents both directions so the same traffic is not counted twice.

For $C$ crashes, the rate in crashes per million vehicle-miles is:

$$R=\frac{C\times10^6}{\text{VMT}}$$

If exposure is already expressed as million vehicle-miles, the formula is simply crashes divided by that number. Do not multiply by a million twice. The same distinction applies to intersection exposure: actual entering vehicles and millions of entering vehicles are different denominator units.

Work a segment and an intersection example

A 4-mile segment with AADT 5,000 has 12 crashes over three years. Approximate exposure is $5,000\times365\times4\times3=21,900,000$ vehicle-miles, or 21.9 million. The rate is 12/21.9 = 0.548, about 0.55 crashes per MVMT. This is not 0.55 per 100 million VMT; that unit would give about 54.8.

At an intersection, use total entering volume from the relevant approaches, avoiding double counting departures as new entries. Suppose 20,000 vehicles enter daily and 30 crashes occur over three years. Exposure is 21.9 million entering vehicles. The rate is 30/21.9 = 1.37 crashes per million entering vehicles. A segment rate and intersection rate have different exposure units and should not be ranked together without a justified method.

These examples illustrate unit handling. The official RSP1 exam is qualitative, so the key skill is interpreting what the metric measures and whether its inputs are suitable.

Interpret rates without assuming linearity

Dividing frequency by exposure is useful descriptively. Comparing rates across very different volumes, however, implicitly treats proportional exposure as an adequate adjustment. Crash frequency may vary nonlinearly with volume, facility, geometry, and crash type. Some models have a linear exposure component; others do not. There is no universal rule that every SPF volume exponent is below one.

Low exposure can magnify the rate produced by one event. For example, one crash over 73,000 vehicle-miles gives about 13.70 crashes per million VMT. Forty crashes over 16,425,000 vehicle-miles gives about 2.44. The first site has a higher rate, but the second has many more events. Neither calculation identifies the feasible treatment benefit by itself.

Use comparable peer groups, multiple years, severity, uncertainty, and appropriate predictive methods. Do not dismiss a low-volume site solely because its events are few, or rank it first solely because the denominator is small. Systemic analysis can identify meaningful risk even without a crash cluster.

Choose exposure for the user and question

Vehicle-miles describe motor-vehicle travel. Pedestrian or bicycle counts, trips, crossing opportunities, or distance may be more relevant for specific user-risk questions. Population rates describe public health burden but are not a direct measure of risk per trip. Registered-vehicle rates answer another question. Select and label the denominator explicitly.

A low pedestrian crash total may occur where few people walk because the crossing is intimidating or inaccessible. Observed exposure can reflect suppressed demand. Combine current counts with route needs, land use, and qualitative information rather than concluding that low counts prove safe design.

Use proportions and overrepresentation

A crash-type proportion is the count of that type divided by a stated total. Comparing it with a suitable peer proportion can identify an investigative priority. For example, 26 wet-weather crashes among 38 roadway-departure crashes is about 68.4%. Compared with an illustrative peer share of 18%, that is about 3.8 times the peer proportion.

This is not automatically a statistically significant result or proof that low friction caused the crashes. Check sample size, weather exposure, definitions, facility comparability, and confounding. The denominator must match: wet-weather departures among departures cannot be directly compared with wet-weather crashes among all events without explanation.

Keep the denominator visible

• Segments: vehicle-miles include distance and time.
• Intersections: entering vehicles exclude double-counted departures.
• User-specific questions: select relevant trips, crossings, or distance.
• Population measures: describe burden, with different interpretation from trip risk.

Report the result in decision terms

Present frequency, rate, severity, and uncertainty together where useful. Explain the measure's purpose and limitation. A defensible screening statement identifies sites for further review; it does not announce that the highest rate is necessarily the greatest investment opportunity. FHWA Road Safety Fundamentals on measuring safety supports selecting data and measures according to the decision.