Road Diet Informational Guide
5. Determining if the Road Diet is Effective
Post-implementation evaluation of the Road Diet will determine safety, operational, and livability impacts. Impacts associated with roadway conversions include the following:
- Safety (e.g., crash frequency/type/severity, pedestrian-vehicle conflicts)
- Travel speeds (e.g., average travel time, mean/85th percentile speeds, percent of vehicles traveling at high speeds)
- Arterial level of service, delay, queuing
- Intersection operations (e.g., turn delays; v/c ratios; signal operations)
- Traffic volume, including diversion to parallel routes
- Corridor operations including transit operations and similar, the two-way left-turn lane operations, and the ability to evaluate “stopped traffic” in one through lane
- Pedestrian and bicycle safety and operations
- Economic impact / livability.
For example, Seattle DOT conducts follow-up studies after implementation to determine the effects on each treated corridor. Specifically, the department compares the before-and-after conditions for the following:81
- Volume of the principal street’s peak hour capacity
- Speed and collisions
- Traffic signal level of service
- Volume of traffic on parallel arterials
- Travel times
- Bicycle volumes.
5.1 Safety Analysis of a Road Diet
The process of implementing significant (and often controversial) changes in roadway geometry such as Road Diets often incorporates a formal safety evaluation plan to assess crash effects and other safety impacts.
5.1.1 Data Needs
Practitioners typically use police-reported crashes for periods before and after changes have been implemented to conduct observational before-and-after studies. Typically a minimum of 3 years of crash data before and after treatment is preferred, although shorter time periods may be used to assess initial crash outcomes. Crash data can either come from State or local police agencies, State or local DOTs, or State DMV offices. In addition to crash data, traffic volume data is desirable to account for vehicle exposure, thus allowing the safety analysis to compute crash rates before and after treatment. Beyond crash studies, safety analysis can include field evaluations of pedestrian-vehicle conflicts and bicycle-vehicle conflicts, in which case the data needs include well-defined and reliably collected observational measures of road user behavior.
Two basic types of observational evaluations are used to estimate associated safety impacts:82
Before-and-After Studies. Observational before-and-after studies are the most common approach used in safety effectiveness evaluation. An observational before-and-after study requires crash data and volume data from both before and after implementation. These studies can be conducted for any site where changes have been made; however, if a site was selected for an improvement because of an unusually high short-term crash frequency, evaluating this site may introduce the regression- to-the mean (RTM) bias. It is likely that even if no improvement was made, the crash experience would decrease (regress to the mean). Thus, RTM effects can be mistaken for the effects of crash countermeasures. Empirical Bayes techniques account for the effect of regression-to-the mean, but require appropriate statistical knowledge to apply.83 The Highway Safety Manual has been developed to assist practitioners and researchers to conduct robust observational before-after studies that provide results to support decision-making.84
Cross-Sectional Studies. Cross-sectional studies involve studying a treatment where there are few sites where a treatment was implemented, but there are many sites that are similar except they do not have the identified treatment. In some cases, evaluations have been performed only after the fact, and all data were not available for the performance measure during the before period. In such cases, cross-sectional studies may be necessary. These studies might also be necessary when the evaluation needs to account explicitly for effects of roadway geometrics or other related features by creating a CMF function rather than a single value for a CMF. Limitations exist when using a cross-sectional study; for example, confidence in the results may not be high since trends over time are not taken into account, and the inability to account for RTM, which threatens the validity of the results, especially if treated sites were selected because they were identified as high-crash locations. The Highway Safety Manual has been developed to assist practitioners and researchers to conduct robust cross-sectional studies.
5.1.2 Observational Before-and-After Studies of Road Diets
This section focuses on observational before-and-after studies, which are most applicable to State and local evaluations of Road Diet implementations.
A before-and-after study is used to estimate the crash effects associated with implementation of a traffic safety measure such as a Road Diet. The change in crash occurrence is estimated from the change in crash frequency between the periods before and after the implementation of the Road Diet. Before-and-after safety analyses can also consider changes in crash rates, which account for estimated traffic volumes during the before and after periods. Crash outcomes associated with Road Diet implementation can include the following:
- Change in the annual number of crashes on the corridor
- Change in the crash rate per million vehicle miles traveled
- Change in the severity of crashes that occur (e.g., percent of crashes that involve either any type of injury, or serious injuries)
- Change in certain targeted crash type(s) associated with Road Diet implementation
- Sideswipe
- Left-turn related
- Pedestrian-related or bicycle-related
- Right angle
- Changes in the number of crashes occurring during the peak-hours.
To account for changes in crashes unrelated to the safety treatment (e.g., overall traffic volume trends, changes in traffic laws, weather, economic conditions), a proper before-and-after study should incorporate an untreated comparison group that is similar in nature to the treatment group. For a before-and-after evaluation of a Road Diet, the comparison group might be comprised of one or more similar, untreated (four-lane, undivided) roads located in the same geographic region.
When planning a comparison group before-and-after safety evaluation, it is important to include a sufficient
number of crashes to enable the expected change in safety to be statistically detectable. Four variables impact the sample size requirements:
- The size of the treatment group, in terms of the number of crashes in the before period
- The relative duration of the before and after periods
- The likely crash reduction (CR) value (expected crash reduction or desirable reduction)
- The size of the comparison group in terms of the number of crashes in the before and after periods.
After the treatment and comparison sites have been identified and the before-and-after crash data assembled, the next step is to conduct the crash analysis. A number of methodologies and statistical procedures are available to analyze before-and-after crash data. These range in complexity and ease of use. Note that some basic forms of before-and-after studies (e.g., naïve before/ after, before/after with yoked pairs) are not recommended due to issues with the statistical soundness of results.
Observational Before-and-After Evaluation Using a Comparison Group. Observational before-and-after studies can incorporate non-treatment sites into the evaluation by using a comparison group (or control sites). A comparison group typically consists of non-treated sites that are comparable in traffic volume, geometrics, and other site characteristics to the treated sites but which do not have the improvement being evaluated. Crash and traffic volume data should be collected for the same time period for both the treated sites and the comparison group.85
Safety data analysis statistical techniques are available to address regression-to-the-mean and other limitations of before-and- after evaluations. Regression-to-the-mean is the natural variation in crash data. If regression-to-the-mean is not accounted for, the conclusions of a before-and-after study could be erroneous. Many of the methods in the Highway Safety Manual account for regression-to-the-mean and can result in more effectively identifying the safety effect of installing a Road Diet on a particular corridor.86
Empirical Bayes (EB) Before-and-After Safety Evaluation Method. From the Highway Safety Manual, “[This] method can be used to compare crash frequencies at a group of sites before and after a treatment is implemented. The EB method explicitly addresses the regression-to-the-mean issue by incorporating crash information from other but similar sites into the evaluation. This is done by using a Safety Performance Function (SPF) and weighting the observed crash frequency with the SPF-predicted average crash frequency to obtain an expected average crash frequency.”87 Recommended data include 10-20 sites at which the treatment has been implemented, 3-5 years of before-installation crash and traffic volume data, 3-5 years of after-installation crash and traffic volume data, and Safety Performance Functions for the treatment site types.
5.1.3 Surrogate Measures of Safety for Road Diets
In addition to conducting formal safety assessments of Road Diets using data-driven analysis techniques based on pre- and post- installation crash data, surrogate measures of safety can provide valuable feedback to State and local agencies regarding both actual and perceived safety outcomes. A surrogate measure of safety can provide information on the level of safety of a location or system using information other than crash data.
Traffic Conflicts. One such surrogate measure involves the analysis of traffic conflicts before and after Road Diets are implemented. A traffic conflict is defined as a traffic event involving the interaction of two or more road users, at least one of whom takes evasive action such as braking or swerving to avoid a collision.88 Examples of pedestrians taking evasive action to avoid crashes include pedestrians jumping back or running out of the way of an approaching vehicle. A traffic conflict survey is a systematic method of observing and recording traffic conflicts and other events associated with safety and operations. With regard to conducting conflict analyses for Road Diets, agencies might focus on before-after changes in the numbers/rates of rear-end conflicts, sideswipe conflicts, and motor vehicle conflicts involving pedestrians and bicyclists.
Speed. Both speed magnitude and speed variability can have an effect on safety and, in the absence of observational crash data, provide information to determine relative safety of the corridor. Because high travel speeds increase the risk of crashes as well as crash severity, it is important to determine whether Road Diets help to reduce speeding. Likewise, because inconsistent travel speeds between vehicles can increase the risk of rear-end and sideswipe crashes, it is important to determine whether Road Diets help to reduce speed variation.
Level of Comfort. Another surrogate measure of safety involves “level of comfort,” a subjective measure which is especially applicable for bicyclists and pedestrians for Road Diet projects. The concept of road user comfort in transportation engineering is not new. For example, the parameters used to establish the minimum horizontal curve radius are the maximum side friction factor and maximum rate of superelevation. Values for the maximum side friction factor are based on driver comfort, not on physical side friction supply and demand relationships. The result is a significant “margin of safety.”89 With regard to assessing the level of comfort for Road Diets, options include conducting systematic visual assessments of pedestrian and bicyclist interaction with motor vehicles and conducting interviews with sufficient samples of non-motorized road users.
5.2 Operational Analysis
The operational effects of Road Diets have been summarized to some degree, but the research is limited to a relatively small number of publications. The literature shows that a properly located and designed Road Diet can result in maintained traffic operations. The general objective of this section will be to discuss ways in which Road Diet operation can be measured.
5.2.1 Analyzing Vehicle Operations
Traffic Volumes. Before-and-after studies should examine if changes occur in daily traffic and peak hour traffic. Evaluate potential changes to determine if there was diversion as a result of a Road Diet installation or if variations from year to year may be the result of background traffic changes. A broader downturn in the economy may result in lower traffic volumes, but patterns going back several years should also be examined for longer-term trends.
Level of Service. Evaluate the level of service of arterial segments and intersections. The facility type that carries the most leverage is based on factors such as signal spacing and segment length. For intersections, the overall LOS should be considered, but the analysis should also drill down to determine how LOS changes for individual movements at an intersection approach. Consider the LOS guidelines for each jurisdiction when determining whether a certain level of vehicular LOS degradation is acceptable. This requires weighing safety benefits as well as improved LOS or QOS for pedestrians and bicyclists. Corridor LOS is generally determined by traffic flow. Intersection LOS is measured by average vehicle delay.
Speed. Practitioners should evaluate the actual speed change (if any) as a result of the Road Diet. Data are collected through the use of before-and-after speed studies using radar, tubes or a
pace car. It is important to collect and compare average speed, 85th percentile speed and speed paces in 10 mph increments. This last group is important to determine if the number of high-end speeders has been reduced.
Two-Way Left-Turn Lane Operation. The addition of a TWLTL will improve operations for through vehicles by removing turning vehicle from the through lane and reducing the uncertainty it causes. Left turning traffic may have additional delay since all through vehicles are in one lane, which could result in fewer gaps. This depends on gaps created by traffic signal timing, on- street parking maneuvers, and vehicles stopping for pedestrians crossing the street.
Queue Lengths. This measure is closely related to signalized intersection LOS described above. It may increase due to only one through lane, but this could be offset due to left turning vehicles no longer queuing in a through lane. Signal spacing needs to be considered so that queues do not extend to the upstream intersection. This may only be a concern for higher volume corridors with closely spaced signalized intersections. Modeling the before and after conditions can provide guidance as to expectations relating to vehicle queue lengths. Signalized intersections in the corridor may need to be re-timed to provide optimal progression.
Trucks, Slow-Moving Vehicles, and Buses. Reducing the number of through lanes from two to one in each direction may create an impact if there are grade changes or if heavy vehicles such as buses, semi-trucks or farm equipment are present. Bus stop placement and the transit policy for whether or not to stop in-lane is also a consideration for Road Diet operation. Give special consideration to these heavy vehicles driving through a corridor and also using the Road Diet corridor circulation to side streets. This is described further in the section below.
Turning Traffic. The Road Diet may make it easier for larger vehicles to make right turns with small curb radii by increasing the effective radius due to the addition of a bike lane. The vehicle mix needs to be considered for each location. Some intersections may not need to accommodate larger semi-truck traffic as they may only be present at such an infrequent interval that it is not an issue. The land use type and demand for smaller single unit type vehicles should also be considered.
5.2.2 Non-Motorized Operations
Non-motorized operations can be measured with respect to pedestrian accessibility and bicyclist use along the corridor. Two studies reported increased bicycle and pedestrian usage along the corridor after a Road Diet conversion.90, 91, 92
Pedestrian Wait Time. Study the wait time for pedestrians crossing at unsignalized intersections and pedestrian “comfort” with crossing the corridor. A before-and-after study of pedestrian crossing behavior can be challenging because many pedestrians may avoid crossing a four-lane undivided arterial due to the level of discomfort or perceived safety issues. Pedestrians may choose to cross exclusively at signalized intersections if there are few gaps in traffic.
Vehicle Yield/Stop Compliance Rate for Pedestrians Crossing the Street. The Road Diet eliminates the risk of the “multiple vehicle threat” pedestrians can face when crossing two lanes of traffic traveling in the same direction. The term describes a scenario in which the first vehicle stops for the pedestrian but a vehicle in the second adjacent lane does not or fails to see the pedestrian in enough time to stop. The prevalence of this problem can be measured in the before and after conditions.
Increased Bicyclist and Pedestrian Volumes. Pedestrians and bicyclists may avoid traveling on a four-lane undivided arterial due to discomfort or perceived safety concerns with no dedicated bicycle lanes or pedestrian facilities. They may switch to a street that has been reconfigured due to increased comfort or perception of improved safety that clearly delineated bicycle lanes and pedestrian facilities (e.g., sidewalks, fewer lanes to cross, or pedestrian refuge islands) can provide.
Some bicyclists may not find a bike lane adjacent to a vehicle lane comfortable enough, which is why the use of a buffered bicycle lane or protected lane is advisable when the street cross section provides enough room. The buffering can come in the form of either a painted barrier between the bike lane and the vehicle lane, a raised barrier, or, in some cases, by placing the bike lane against the curb and placing the parking lane between the bike lane and the vehicle through lane.
5.2.3 Tools and Methods to Evaluate Impacts
Input Requirements. The data needed for this analysis consists of intersection turning movement counts, daily traffic volumes by direction, and operating speed information. If these volumes have been observed to create delay in the before condition, visually observe delays caused by mid-block, left-turning traffic at driveways. The physical characteristics and complexity of corridor determine how detailed the analysis should be; some corridors may only require corridor analysis while others will need analysis of signalized intersection operations. The traffic volume along the corridor, transit operations, and the number of access points will all help determine whether the analysis procedures presented in the 2010 Highway Capacity Manual are sufficient or whether a macro- (such as Synchro) or micro-level computer simulation (such as VISSIM) is needed to determine the projected outcome of a Road Diet.
Output Provided. The output provided will depend on the tool used for analysis. The factors to consider depend on the type of analysis and the questions posed.
Complexities with Analyzing Three-lane Sections. The intersection analysis should be straightforward, but practitioners must ensure field conditions are accurately analyzed between signalized intersections, too. Some of the factors to consider are parallel parking maneuvers using a through lane, buses maneuvering into and out of a bus stop (whether it is along the curb or in the lane), left-turning vehicles (from stopping in the through lane to slowing to enter the two-way, left-turn lane), cross-street traffic looking for a gap to turn or cross the arterial, and pedestrians crossing the street at unsignalized intersections. It is helpful to observe the corridor operating conditions in the four-lane, undivided configuration to determine a “baseline” condition and see where existing conflict points are and what causes them prior to evaluating the corridor in the “after” condition to determine how overall conditions have changed.