USA Banner

Official US Government Icon

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure Site Icon

Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

U.S. Department of Transportation U.S. Department of Transportation Icon United States Department of Transportation United States Department of Transportation
FHWA Highway Safety Programs

Speed Study Data Collection

Data collection is an integral part of setting speed limits. Adequate planning and coordination must occur to ensure the data collection process is as complete, efficient, and effective as possible. This section describes typical activities that highway agencies will undergo to plan and implement a data collection effort. Several types of data, including speed, crash, and roadway environment information, are vital to this process. The ITE Manual of Transportation Engineering Studies44 provides guidance in this regard.

The data collection requirements depend on the methodology selected by a jurisdiction in setting posted speed limits. The Safe Systems approach, for instance, requires very little data collection since it is based on very basic road design parameters (e.g., number and frequency of accesses, presence of a raised median, etc.) and general traffic characteristics (e.g., type and frequency of road users). The data collection effort is relatively minor.

The Optimal Speed Limit methodology has a more intensive data collection effort. While the data required for the particular roadway under study is generally manageable, there is a large volume of local data that is required to calibrate the prediction equations and models that are used assessing the societal impacts of the different speed limit alternatives. A discussion concerning the models and their calibration is beyond the scope of this document. Project-specific data that is required as input to these models is detailed in the following subsections of this chapter.

The remainder of this chapter describes the collection process for data that is most often used in the engineering and expert systems methodologies. The exact data needs are determined by the method employed by the road authority—more or less data than described herein may be required.

Data Collection Planning and Coordination

Speed zoning studies are conducted to evaluate safety issues and identify appropriate speed limits for specific roadway segments. In addition to actual travel speeds, there are several other types of information that may be appropriate input to the process of setting speed limits. Therefore, coordination within an agency performing the study and with other agencies that may have additional information may be needed to ensure all the appropriate inputs are considered. Crash data, recent and planned roadway or adjacent land use changes, and even anecdotal information can be obtained from safety, planning, enforcement, and other stakeholders. The data collected will be used to examine the speeds of free-flowing traffic, as well as information on roadway geometry, crash characteristics, land use, and access. The studies provide details regarding some or all of the following:

  • Average annual and hourly vehicular, bicycle, and pedestrian traffic volume.
  • Traffic speeds for each flow direction by hour of day.
  • Road design elements that may be crash factors, such as horizontal and vertical road curvature, access points, drainage, pavement condition, sight distance restrictions, roadside objects, signage, markings and delineation, etc.
  • Road lighting and traffic control devices, including signals, signal timing, and STOP signs.
  • Summary of crashes and crash causes over a multiyear period.
  • Plans for expected new development, changes in the type of development, or major closing of existing development that may change the traffic flow characteristics in the future.
  • Recommendations for the speed limit.19

When planning the data collection activity, it is important to document and control any aspect of the collection that might have an impact on the measured speed. Measurable physical features, roadway surface characteristics and conditions, and traffic characteristics and control are items to be inventoried. If conditions are not relatively consistent throughout the zone under study, consideration can be given to splitting the study area into shorter sections. For example, if the road transitions from a 2-lane to a 4-lane divided facility, or from on-street parking to no parking, or from rural agricultural land use to a commercial or residential land use, then speed samples are typically taken in each section. Factors such as roadway lighting and delineation are reflective of road geometry and land use, but are not necessarily factors that warrant splitting a study area into shorter sections.

Variables considered for documentation of the site include (but may not be limited to):

  • Location and roadway configuration.
  • Lanes, delineation, shoulders, medians, grade separation, roadside objects, driveways or entrances, curvature or grades, lighting, etc.
  • Posted speed limit.
  • Weather (limiting measurements to fair weather is preferable).
  • Direction.
  • Restricted sight distance.
  • Pedestrian activity.
  • Cyclist activity.
  • Date and time (in a common format among collections).
  • Traffic control devices (regulatory and warning).
  • Type and condition of pavement surfaces.
  • Businesses, advertising, or residential developments.
  • Proximate schools and school routes.
  • Surrounding area changes in travel habits or influences.
  • Vegetation changes.

In most instances, the variables collected by a particular road authority are dependent on the methodology used in that jurisdiction to set speed limits, the expected effect of the variable on operating speeds, and available resources.

When undertaking data collection efforts, it is important to understand if there are activities or conditions outside of the study area that may affect measurements, including construction or maintenance activities in the area, road closures, detours, the presence of enforcement, and whether proximate schools are in session.

Study Area

A speed zone study can be initiated in response to a public request for a speed limit review, as a result of network screening (for crash prone locations), or for any other reason. In all situations, a general study area is identified through the initial request or data analysis. The study area can then be divided into homogeneous sections for analysis. A homogeneous section is one where:

  • The roadside development is consistent (residential vs. commercial; type and frequency of businesses and driveways, etc.); and
  • The roadway features are consistent (lane widths, medians, shoulders, surface roughness, curvature, intersection spacing, etc.).

The data collection area will typically extend 500 feet beyond each end of the proposed speed zone in order to include nearby features. These features will help to determine the homogeneity of the proposed speed zone, and whether the study area limits should be extended. It may be helpful to take photographs of features in the intended speed zone and the extended study area, as they may be helpful in describing any concerns within the study area.

A scaled area map, sketch, or aerial view is usually developed to show the study area and the field conditions. Generally, a speed zone study used to support a request for alteration of a speed limit would include this exhibit to identify the location of the proposed zone and any features of interest. A strip map, or line diagram, is an example of an appropriate format for the exhibit, and details the information that can be shown on the map. The data points can be collected using a Geographic Information Systems (GIS) unit, which helps improve the accuracy of the strip map. Figure 6 shows an example of a strip map that is appropriate for a speed zone study. Table 9 shows the information that should be shown on a strip map.

(Extended Text Description: Figure 6. An Example of a Strip Map of a Study Area Showing Existing Conditions. The example map and diagram are for example purposes only, with the following basic features to illustrate the concepts of a study area. At the top of the study area is a map of a road showing open, scattered business and scattered residential areas around an intersection. The road shows a few area details like a motel and office building, with various indicators showing placement of speed limit signs and remarks to remove some speed limit signs. Under the overhead map is a diagram with details about the roadway width, striped lanes, traffic signal data, average daily traffic, observed speed-critical, observed speed-pace and existing signed zone, showing arrows and ranges of speed data. Underneath the diagram is a proposed speed limit graph, accident rate and distance in meters. A second map example is included at the bottom of this figure, which appears as a continuation of the upper map and diagram, from left to right. The second example is labeled State of California Department of Transportation - Traffic Operations Program and Speed Zone Survey. The map in the diagram shows several intersections running through solid business, solid residential and scattered residential areas. Various indicators show placement and removal of speed limit signs along the route. At the bottom of the map is a diagram of various distances and speeds indicating the speed limit sign placement, as well as a graph of proposed speed limits. This example map and diagram has a note at the bottom that reads, "Note: this scaled figure represents a 11 in X 36 in size sheet." Again, this figure serves as an example to illustrate a study area showing existing conditions.)

Figure 6. An Example of a Strip Map of a Study Area Showing Existing Conditions. (Source: CalTrans, 2009)


Table 9. Information to Show on Strip Map

Information Item


Name and highway number of the route to be zoned

Show all names and/or highway numbers.

Indicate sections to be zoned with a wide center line on the strip map.

Cross section

Width of the roadway/lanes.

Pavement markings.

Number of lanes.

Parking restrictions.

Crossroads, cross streets, and driveway access points

Show all names and highway numbers.

Limits of the speed zone

Indicate reference marker, milepoint, control, and/or section numbers.

Adjoining speed zone(s) of connecting map(s)

Note speed limit information for adjoining roadway sections.

Limits of any incorporated city or town

Show reference marker, milepoint, control, and section numbers.

Names and approximate limits of the developed area of unincorporated towns

Indicate by "Beginning of Developed Area" and "End of Developed Area" under the heading, "Development"— not as "City Limits."

Urban districts

Indicate any urban district clearly under the heading "Development."

The territory contiguous to and including any highway or street which is built up with structures devoted to business, industry or dwelling houses, situated at intervals of less than 100 feet for a distance of 0.25 mile or more on either side.

Schools and school crossings

Show schools abutting the highway and those in the vicinity of the highway.

Show location of schools.

Show all school crosswalks.

Traffic signals

Show location of existing devices to aid in proper spacing and placement of speed zone signs.

Important traffic generators

Show all factories, shopping centers/malls, and any other establishments that attract large volumes of traffic.

Ball bank readings

Show readings for each direction of travel for all curves.

Railroad crossings

Indicate the number of tracks and type of grade crossing protection (crossbucks, cantilevers, crossbucks with signals, gates).

Show the name of the railroad at each crossing.


Indicate if the roadway on the bridge is narrower than the roadway on either side of it.

Source: Adapted from the Texas Department of Transportation.19


Speed Data Collection

The result of the speed data collection effort is an accurate picture of the range of vehicles and driver behavior in the study area. In addition to collecting spot speeds of vehicles traveling through the study area, test runs can be used to confirm free-flow speeds and compare study area speeds to adjacent areas outside the speed zone. These data, combined with other crash, roadway environment, and enforcement information, feed into the data analysis and the determination of the speed limit as discussed in the next section.

Vehicle Speeds

A variety of methods are available to measure speeds. These methods can generally be grouped into three categories based on the installation location of the collection equipment:

  • Manually-operated, handheld devices that are portable and can be used in most places (e.g., stopwatch, radar gun, and lidar gun).
  • In-road devices that are installed into or on top of the roadway surface (e.g., pneumatic road tube).
  • Out-of-road devices that are installed overhead or to the side of the roadway surface (e.g. radar recorders).

The advantages and disadvantages for several common speed collection devices are shown in Table 10, and should be considered when selecting a device for use at a particular location.

Ideally, data collection:

  • Uses techniques that capture typical traffic behavior without affecting it.
  • Collects free-flow vehicles and ignores platoons (less than 5 seconds separation from the vehicle ahead*).
  • Collects vehicle type along with the speed so that speed profiles for different vehicle types can be identified, if desired.

The vehicles checked should be only those in which drivers are choosing their own speed or are free-flowing. When a line of vehicles moving closely behind each other passes the speed check station, only the speed of the first vehicle is checked, since the other drivers may not be choosing their own speed. Vehicles involved in passing or turning maneuvers are not to be checked, because they are probably driving at an abnormal rate of speed. Turning lanes, or other special lanes, are not normally used to collect speed data.

* Some analysts prefer to discard a speed measurement if a vehicle is following another vehicle within five seconds, as the lead driver may be slower than they would ordinarily be traveling in an open road situation.


Table 10. Advantages and Disadvantages of Speed Collection Devices (Adapted from Reference 43)


Data Collected


Equipment Cost*



Radar Recorders

Instantaneous speed, traffic volumes, vehicle class, traffic flow gaps**



Little labor required to collect and tabulate data: can collect data for long periods of time; other traffic-related data may be collected at the same time; can be used when snowplows may be present without risk of damage; less visible to traveling public than road tubes

User cannot randomly select vehicles for data set; some devices may not accurately collect data for multi-lane roadways and/or determine directionality of observed vehicles; equipment-intensive method; maintenance/ calibration required

Pneumatic Road Tube

Instantaneous speed, traffic volumes, vehicle class, traffic flow gaps**



Little labor required to collect and tabulate data; can collect data for long periods of time; other traffic-related data may be collected at the same time

Visible to traveling public which may change driver behavior; user cannot randomly select vehicles for data set; use discouraged when snowplows may be present; most equipment-intensive method; maintenance/calibration required

Laser Gun

Instantaneous speed



Equipment is easily portable; user controls vehicles sampled as a more focused laser beam limits the number of readings for non-target vehicles as compared to radar

Cosine error limits horizontal/vertical deployment; scopes and sights may not be user-friendly; laser beams more sensitive to environmental variances than radar; maintenance/calibration required

Radar Gun

Instantaneous speed



Equipment is easily portable; user controls vehicles sampled; accurate data collection method; widespread equipment availability has lowered its cost

Cosine error limits horizontal/vertical deployment; closely-spaced and larger vehicles may create readings for non-targeted vehicles; maintenance/ calibration required


Travel time over a distance



Little equipment to purchase and maintain; easy to perform data collection process

Labor-intensive; collects time data that needs to be converted to speed data; typically low accuracy

*Equipment costs reflect the initial purchasing costs of the equipment and not future maintenance and calibration costs.

**The amount of additional data collected varies for each device. Consult the device's user manual for a better understanding of the capabilities.


Due to different physical and operational characteristics of trucks and buses, data for these vehicles are usually recorded separately. If separate speed limits are believed warranted for large trucks or other vehicle classifications, a separate count and analysis of these vehicles may be needed.

Data collection forms help organize data for the spot speed study. Appendix G (Example Speed Study Forms) contains a sample "Vehicle Spot Speed Study" data collection sheet from the Florida DOT.30

The speed profile for a particular road section can only be estimated by measuring individual speeds through a spot speed study. Prior to conducting these studies, the minimum number of vehicles for which speed data are needed to sufficiently estimate speed parameters should be estimated. The minimum number of vehicles required to accurately estimate the speed profile is dependent on the level of confidence required for the statistical analysis of the data. The ITE Manual of Transportation Engineering Studies44 presents the following equation to calculate the minimum sample size for estimating the 85th percentile speed:

Please see Extended Text Description below.

(Extended Text Description: Equation: N=(S2K2 (2 + U2)) / 2E2)


= minimum number of measured speeds

= estimated sample standard deviation, mph

= constant corresponding to desired confidence level

= constant corresponding to the desired percentile speed

= permitted error in the speed estimate, mph

The Manual of Transportation Engineering Studies provides tables for determining the values for S, K, and U. Table 11 shows an example sample size calculation for estimating the 85th percentile speed, using this formula.44


Table 11. Example Calculation of Sample Size for Study to Determine 85th Percentile Speed


The average standard deviation is rounded to 5.0 mph

S = 5.0 mph

The desired confidence level is 95 percent

K = 1.96

The study will determine 85th percentile speed

U = 1.04

The permitted error is 2 mph

E = 2 mph


Sample Size

N = 37


Using a sample size of 37, the 85th percentile speed can be determined within 2 mph at a 95 percent confidence level.


Performing the same calculation with a permitted error of 1 mph (E = 1 mph) results in a sample size of 148. A common sample size for many jurisdictions is 100 vehicles. Assuming a standard deviation of 5 mph, and using a 95 percent level of confidence, the 100 vehicle sample size will yield between a 1 and 2 mph error in the 85th percentile speed, and it makes calculation of the 85th percentile fairly simple (refer to Appendix H).

Table 12 lists the sample sizes and sample periods used by three States. Most States use 100 or more vehicles in each direction for each station. Since meeting the minimum collection data on low-volume roads can be difficult, adjustments on the sample size can be made based on the duration of the collection period. On highways carrying low traffic volumes, the speed checks at any one station are usually discontinued after two hours, even if a minimum of 100 vehicles have not been recorded.


Table 12. Sample Sizes and Data Collection Periods Used by Three States


Sample Size



100 or more vehicles in each direction should be checked at each station.

On highways carrying low traffic volumes, the checks at any one station may be discontinued after two hours although a minimum of 100 vehicles have not been timed.


Record speeds of 100 vehicles for each direction of travel.

Observation need not exceed one hour even if less than 100 vehicles are recorded traveling in each direction.


A minimum of 125 cars in each direction, at each station.

Discontinue after two hours if radar is used, or after four hours if a traffic counter that classifies vehicles by type is used—even if 125 cars have not been timed.


The Manual of Transportation Engineering Studies formula for determining a spot speed sample size is premised on a random sample of vehicles over the course of the time. Since the analyst is usually stationed at the study site for a limited time, the speed data is actually assembled from a cluster sample. Cluster sampling generally increases the variability of sample estimates above that of simple random sampling, and for this reason cluster sampling usually requires a larger sample than simple random sampling to achieve the same level of accuracy. Therefore, sample sizes that are slightly larger than those predicted by the Manual of Transportation Engineering Studies formula would increase the accuracy of the 85th percentile speed estimate. Furthermore, the times at which the spot speed sample is conducted should include observable speeds that are representative of the operating speeds for all times of the day.

If automated collection of speed data is employed, then it is possible to collect data for extended periods. Collecting data for a 24-hour period will account for variation in traffic patterns and will allow for determination of different speed limits for different times of the day, if needed. For example, a time-limited school zone speed limit or a nighttime speed limit.

Care must be exercised when using automated data collection to ensure that only free-flow speeds are collected, and that data collection units are placed sufficiently far from intersections and other points of access where vehicles that are accelerating/decelerating may influence the speed profile.

Speed check stations need to be located to show all the important changes in prevailing speeds. The data collector should pick a location that will not influence the behavior of the drivers. Table 13 shows recommendations for speed check stations for three States for both urban and rural areas. While these States provide some guidance in the form of set distances between speed check stations, it is important to remember that it is not the distance between stations that is critical— rather it is the changes in the road, traffic, and environment that may lead to different speed profiles and operating speeds. Distances between speed check stations may be increased or decreased from those provided accordingly.

Figure 7. Radar Setup. Please see Extended Text Description below.

(Extended Text Description: Figure 7. Radar Setup. Photo depicting speed measurement from an unobtrusive, undetectable position with the person taking the measurement obscured at the side of an overpass as a car passes underneath.)

Figure 7. Radar Setup.


Radar speed meters, which operate on the Doppler principle, or lidar, which operates on a laser principle, are normally used for making manual speed checks. These devices typically operate from the power of an automobile battery and give direct readings of vehicle speeds which are accurate to within 2 mph (3 km/h).19 The operating instructions for the radar unit will provide factors for calibration, optimum distance of survey, and optimum angle of survey. Speed measurement should be done in an unobtrusive, undetectable manner so as to obtain a sample of normal traffic speeds. If the radar operation is detected by drivers, there is the potential for the data to be biased as drivers change their speeds.45 Figure 7 shows an example of a radar operation setup.

Automatic speed classification equipment technology may be used in determining vehicular speeds for use in calculating 85th percentile speed. Examples of technologies are counter-classifiers with the capability of classifying vehicles, determining vehicular speeds, and differentiating the gap between vehicles. These devices may include video imaging, tube counters, magnetic counters, inductive counters, etc.19 Figure 8 shows an example of a portable traffic analyzer. With automatic data collection equipment, speed data is normally collected at sites for at least a 24-hour period.

Figure 8. Portable Traffic Analyzer. Please see Extended Text Description below.

(Extended Text Description: Figure 8. Portable Traffic Analyzer. Photo depicting a vehicle approaching a portable traffic analyzer which is placed on the ground to collect spot speed data.)

Figure 8. Portable Traffic Analyzer.


The reason for collecting spot speed data is to estimate the free-flow speed of a facility for use in setting speed limits. Ordinarily, it is only necessary to collect spot speed data once during the time-of-day, and the day-of-week that will yield the best estimate of free-flowing speeds. Collecting speed data at more than one time-of-day or day-of-week is dependent on the analyst's confidence in the single measurement representing the true free-flow speed for the facility. Additional spot speed studies at a single location may also be conducted if the analyst is considering a variable speed limit, or a speed limit that is time-limited (i.e., a school zone speed limit).

Table 13. Speed Check Stations for Three States


Speed Check Station Layout Information



  • Speed check stations should be strategically located to show all the important changes to municipalities; speed check stations should generally be located at intervals not to exceed 0.25 miles, depending upon the locality and the uniformity of physical and traffic conditions. Much closer spacing than this may be necessary to obtain an accurate picture of the speed pattern.


  • In rural areas, the spacing of speed check stations may be at much greater intervals provided they properly reflect the general speed pattern. There should be at least one observation for each direction of travel in each zone of a different numerical limit.



  • Speed checks may be taken with any device that will indicate vehicle speed with an accuracy of +/-10 percent.
  • Speed checks should be taken at the 1/3 point (total of four checks) for zones 0.25-1.00 mile in length, and at 0.5-0.75 mile intervals for zones over 1 mile in length.



  • Should generally be located at intervals of 0.25 mile or less if necessary to ensure an accurate picture of the speed patterns.
  • Should be located midway between signals or 0.2 miles from any signal, whichever is less, to ensure an accurate representation of speed patterns.
  • Should take into account locality, and the uniformity of physical and traffic conditions may be determined by trial runs through the area if volumes are too low or if a recheck of speeds is all that is needed.
  • Should be checked midway between interchanges on the main lanes of expressways and freeways.


  • May be at intervals greater than 0.25 mile, as long as the general speed pattern is followed and may only be necessary at each end and the middle point if the characteristics of the roadway are consistent throughout the entire section.
  • May be determined by trial runs through the area if the characteristics of the roadway are consistent throughout the entire section and a speed check in that section indicates that 125 vehicles cannot be checked within the two hours if radar is used, or after four hours if a traffic counter that classifies vehicles by type is used.

Speed Test Runs

The purpose of the test runs is to generate an operating speed profile and ensure that measured spot speeds are representative of speeds throughout the section.

The general idea is to perform several runs at free-flow speeds as a way to confirm the speed data collected for use in determining 85th percentile speed and compare spot speeds to the test run speeds for the full study section. When planning test runs, in general:

  • Test runs should be made by driving as fast as it is comfortably safe.
  • Test runs should be made so that other traffic will not delay the test car.
  • The speed should be recorded at a range of 0.10 to 0.25 mile (.15 km to .45 km) interval or more.
  • The average speed of three test runs should be determined in each direction.29

An alternative methodology for conducting a speed test run is the floating car method, i.e., following cars and recording their speeds or journey times through the study area. This method allows an assessment of a driver's free-flow speed, and not the desired speed of the person conducting the survey (as this might differ from the general population).

To counter arguments that the 85th percentile spot speed studies are not representative of the operating speeds along the entire street, a test run speed profile can be combined with the spot 85th percentile speeds to obtain an 85th percentile speed profile.51 The speed profile is established by an individual driving the road in his or her usual manner, while an observer records the time for the vehicle to travel a set interval (e.g., 100 m). Then the following procedure can be used to develop an 85th percentile speed profile (the data in Table 14 is referenced to demonstrate method):

  • For each location where a spot speed is measured, comparison factors are calculated by dividing the 85th percentile speed by the test run speeds and the average test run speeds for the same location. In the example, there are two locations where spot speeds were measured— Station 0+600 and Station 1+600.
  • The variation of comparison factors for each test run is determined separately. In the example, there are only two comparison factors for each test run, so the difference between the two factors is used as the measure of variance.
  • The comparison factors for the test run with the lowest variation are then averaged and this average factor becomes the correction factor. In the example, Test Run 1 has the lowest variation.
  • The average test run speed for each location is multiplied by the correction factor to yield an estimated 85th percentile speed for each location.51

Table 14. Example of Using Speed Test Runs to Confirm 85th Percentile Speeds (km/h)



Test Run

Comparison Factor



Spot Speed 85th Percentile







Estimated 85th Percentile Speed










































































































Correction Factor






Data Analysis

The ITE Traffic Engineering Council Technical Committee surveyed the speed zoning practices used by agencies across the United States. The committee collected speed zoning guidelines from 40 States and conducted 124 surveys with ITE members serving as traffic engineers in State and local agencies. Based on the results of the survey, the most important factors considered for recommending a speed limit are: 85th percentile speed; followed by roadway geometry, crash exposure, and roadside development.17 This section discusses the compilation of speed and crash data used to develop the inputs to the speed limit setting process.

85th Percentile Speed

The Manual on Uniform Traffic Control Devices (MUTCD) lists the current speed distribution of free-flowing vehicles as a primary factor to consider when establishing speed limits. The MUTCD also states that the speed limit should be within 5 mph (8 km/h) of the 85th percentile speed.15

The 85th percentile speed is the speed at or below which 85 percent of the free-flowing vehicles travel, and has traditionally been considered in an engineering study to establish a speed limit. Traffic engineers have assumed that this high percentage of drivers will select a safe speed on the basis of the conditions at the site. The 85th percentile speed is considered the first approximation for the speed limit.

The Ohio Department of Transportation collects vehicle speeds even if it is not possible to observe freeflow conditions. Then the 85th percentile speed of all vehicles is increased 5 to 10 mph (8 to 16 km/h) to approximate the free-flow 85th percentile speed. If the 85th percentile speed of several speed checks varies considerably, the 85th percentile speeds are averaged, or the most representative speed is selected.19

A convenient way to determine speed percentiles is a frequency distribution table. An example, with an explanation of how to use it, is provided in Appendix H.

10 mph (16 km/h) Pace

The speeds of individual vehicles on a highway vary. Speed dispersion refers to this spread in vehicle speeds. The 10 mph (16 km/h) pace is the ten mile-per-hour range of speeds containing the greatest number of observed speeds and is a measure of speed dispersion. It is described by both the speed value at the lower end of the range and the percentage of all vehicles that are within the range; and, thus, is an indicator of speed dispersion.

A normal speed distribution contains approximately 70 percent of the vehicles within the pace, with approximately 15 percent of the vehicles below and 15 percent above the limits of the pace speed. The upper limit of the 10 mph (16 km/h) pace speed is therefore approximately the 85th percentile speed in most cases. However, the upper limit of the pace speed may vary from the 85th percentile speed, depending on the distribution curve of the vehicle speeds.

There is general agreement that the safest conditions occur when all vehicles at a site are traveling at about the same speed.11

Crash Data

Crash data are typically considered in establishing speed limits. The factors potentially contributing to the crashes should be examined to determine whether speeding was involved, or whether the speeds were too high for a specific condition or feature. Speed contributes to the severity of a crash, and sites with a history of severe injury or fatal crashes may be locations with high levels of speeding. The National Highway Traffic Safety Administration (NHTSA) considers a crash to be speeding-related if the driver was charged with a speeding-related offense or if an officer indicated that racing, driving too fast for conditions, or exceeding the posted speed limit was a contributing factor in the crash.46

Crash data may be accessed via State, local, county or community-wide databases. These databases may be housed within State agencies including the Department of Safety, Department of Transportation, or State Highway Patrol, within local law enforcement agencies, or within the court system. In some cases there may be some information missing from the database. For example, if a database is not GPS-based, or does not include detailed information regarding the location of a crash, it may be difficult to locate the site. In some communities the data is still maintained in paper format, and manually reviewing paper records is time consuming and costly. Coordinating field data collection and crash analysis prior to the start of the program is one way to minimize costs.

A review of crash data can show whether the study area has a higher than average crash experience, and whether the portion of crashes that appear to be related to speeding is higher than average. The MUTCD recommends reviewing reported crash experience for at least a 12-month period.15 Twelve months is considered a short-term crash count and is insufficient as a basis for making sound safety decisions. The implications of crash frequency fluctuation and variation of site conditions are often in conflict. On one hand, the year-to-year fluctuation in crash frequencies tends toward acquiring more years of data to determine the expected average crash frequency. On the other hand, changes in site conditions can shorten the length of time for which crash frequencies are valid for considering averages. This conflict between crash data variations and changing site conditions requires considerable judgment in selecting an analysis period.

Typically, road authorities review crash data for a three- to five-year period.

When crash data are collected, it is important to consider the following factors before interpreting existing data or gathering new data:

  • Only consider data collected on road segments that are within the study area.
  • Gather categorized details about the site geometry, traffic control devices, signage, weather, lighting conditions, time of day, day of week, interaction with other vehicles, etc.
  • Minimize the amount of emphasis placed on individual (severe) crashes rather than trends or clusters of crashes.
  • Differentiate between mid-block and intersection crashes.
  • When comparing crash datasets, ensure that the same filtering criteria are used to develop the datasets.
  • Watch for data format issues between data sources and collection periods to avoid difficulties in coding and analysis.
  • Although data should be sanitized for driver identity before reporting, maintaining a link to the raw data source until the data is ready to be published or used for the last time will make it possible to go back to the source easily to supplement and verify details for analysis and report generation.
  • Note extenuating circumstances that may preclude or overshadow typical trends and analyses and be prepared to filter or caveat them (e.g., long-term construction zones, new developments, major changes or reconstruction, etc.).

Some measure of the crash experience of the study site can be developed and compared to the average of that measurement for similar sites in a jurisdiction. Examples are crash frequency, crash density, and crash rate. Some agencies factor crash severity into safety analyses by giving a higher weight to injury and fatal crashes. The average of this frequency, rate, or other measure, for all roadways of a similar type (such as urban 4-lane undivided arterials) in a State would be a good value for comparison. A discussion of how the analyzed data are used in determining speed limits is presented in the next section.

Isolating the effect of one crash factor, such as speed, can be a challenge. Often it is difficult to identify the role of speed in crashes, and for this reason it is thought that speed-related crashes are often under-reported.47 For this reason, all crashes may be considered in setting speed limits.

A crash diagram is often prepared as a part of the safety analysis to help identify patterns and trends in the crash data.

The state-of-the-art in crash data analysis and determining the safety performance of a facility is contained in the Highway Safety Manual. This document provides analytical tools and techniques for estimating the expected crash risk of different facilities, and it can also be used in assessing the safety effects of a change in the posted speed limit.