82 ■ Transportation Research Record 1773

Paper No. 01- 0511

Modeling the Roadside

Walking Environment
Pedestrian Level of Service

Bruce W. Landis, Venkat R. Vattikuti, Russell M. Ottenberg,

Douglas S. McLeod, and Martin Guttenplan

A method is needed to objectively quantify pedestrians’ perception of mode is far more complex in comparison with that of the motor vehi-
safety and comfort in the roadside environment. This quantification, cle mode. Whereas operators of motor vehicles are largely insulated
or mathematical relationship, would provide a measure of how well road- in their travel environment and hence are influenced by relatively few
ways accommodate pedestrian travel. Essentially, it would provide a mea- factors, the pedestrian is relatively unprotected and is subject to a
sure of pedestrian level of service (LOS) within a roadway environment. host of environmental conditions.
Such a measure of walking conditions would greatly aid in roadway In general, planners and engineers have not yet come to consensus
cross-sectional design and would help evaluate and prioritize the needs on which roadway environment features have statistically reliable sig-
of existing roadways for sidewalk retrofit construction. Furthermore, nificance to pedestrians. There have been several recent initiatives by
the measure can be used to evaluate traffic-calming strategies and planners to develop “walkability audits”; however, these measures
streetscape designs for their effectiveness in improving the pedestrian generally include the myriad features of the entire roadway corridor
environment. Such a measure would make it possible to merge pedestrian environment (including conditions at intersections) and they have not
facility programming into the mainstream of transportation planning, yet been statistically tested or widely applied. There is consensus that
design, and construction. To meet the need for such a method, as well as pedestrians’ sense of safety and comfort within a roadway corridor is
to fulfill a state mandate to establish levels of service standards for all based on a complex assortment of factors including the following:
transportation modes, the Florida Department of Transportation
sponsored the development of the Pedestrian LOS Model. The model • Personal safety (i.e., the threat of crashes),
was developed through a stepwise multivariable regression analysis of • Personal security (i.e., the threat of assault),
1,250 observations from an event that placed 75 people on a roadway • Architectural interest,
walking course in the Pensacola, Florida, metropolitan area. The Pedes- • Pathway or sidewalk shade,
trian LOS Model incorporates the statistically significant roadway and • Pedestrian-scale lighting and amenities,
traffic variables that describe pedestrians’ perception of safety or com- • Presence of other pedestrians, and
fort in the roadway environment between intersections. It is similar in • Conditions at intersections.
approach to methods used to assess automobile operators’ level of service
established in the Highway Capacity Manual. The complexity of the issue, however, should not deter attempts to
model pedestrians’ response to the roadway environment, even if it is
for one aspect or component of a roadway corridor. Elected repre-
In recent years there have been initiatives in metropolitan areas
sentatives, public officials, and transportation planners and engineers
throughout the United States to create more livable communities in
need to be able to determine a roadway’s performance with regard to
which walking and bicycling are encouraged and accepted as legiti-
mate forms of transportation. Characteristic of these efforts is the accommodating pedestrian travel. Roadway designers need solid guid-
reintroduction of bicycle lanes and sidewalks to the streetscapes, ance on how to better design pedestrian environments: how far side-
complete with street furniture, landscaping, pedestrian-scaled light- walks should be placed from moving traffic, what types of buffering
ing, and other features making the public right-of-way more inviting or protective barriers are needed and when they should be used, and
for people to travel by bicycle or on foot. The transportation planning how wide the sidewalk should be.
and engineering community has recently been attempting to provide The purpose of this study, therefore, is to focus on, and identify
analysis and design methods to help create more “livable” streets and those factors in the right-of-way that significantly influence the pedes-
roadway environments. trian’s feeling of safety and comfort. The collection of these factors
Historically, compared with the level of research done for motor- into a mathematical expression, tested for statistical reliability, pro-
ized transportation, there has been relatively little study and analysis vides a measure of the roadway segment’s level of service (LOS) to
of the factors that affect the quality of the walking environment. pedestrians. This measure evaluates the conditions along roadway
Evaluating the performance of a roadway section for the walking segments between intersections. A key application of this measure
is to help planners and roadway engineers make informed deci-
sions when designing or choosing the appropriate cross section for
B. W. Landis, V. R. Vattikuti, and R. M. Ottenberg, SCI, 18115 US Highway 41 North,
Suite 600, Lutz, FL 33549. D. S. McLeod and M. Guttenplan, Florida Depart-
any given roadway—a cross section that meets pedestrians’ basic
ment of Transportation, Systems Planning Office, 605 Suwannee Street, MS 19, need to feel safe and comfortable while walking. As such, the mea-
Tallahassee, FL 32399-0450. sure presented in this paper is one piece of the puzzle, albeit an impor-
Landis et al. Paper No. 01-0511 83

tant one—many other factors influence a pedestrian’s (enjoyment of modates their travel is from a perspective of safety or comfort. “It’s
the) walking experience. These factors should be studied further to a dangerous place to walk” or “it’s fairly safe and comfortable” is the
improve the body of knowledge on this subject. way they express their views of the roadway. This measure is the sub-
The researchers of this study acknowledge that intersection con- ject of our research, hence this paper. Considering only the roadway
ditions have a significant bearing on the pedestrians’ total roadway environment (i.e., excluding intersection conditions), the factors
corridor experience, and must also be studied. Further, they believe thought to significantly affect pedestrians’ sense of safety or comfort
that a measure(s) must be developed to be combined with this road- include the following:
way segment performance measure. In fact the research sponsor, the
Florida Department of Transportation (DOT), is using this research • Presence of a sidewalk,
team to develop intersection performance measure(s) as Phase II of • Lateral separation from motor vehicle traffic,
this study. FHWA is beginning a similar study initiative. • Barriers and buffers between pedestrians and motor vehicle
traffic,
• Motor vehicle volume and composition,
MEASURES OF THE PEDESTRIAN • Effects of motor vehicle traffic speed, and
ENVIRONMENT • Driveway frequency and access volume.

Dan Burden, a leading national advocate for more walkable com- The perception of safety or comfort is a qualitative measure of
munities and transportation systems, spoke for many when he said effectiveness recognized by the 1994 Highway Capacity Manual.
pedestrians in the roadside environment are subjected to a multi- The manual states, “The concept of level-of-service uses qualitative
tude of factors significantly affecting their feeling of safety, com- measures that characterize operational conditions within traffic the
fort, and convenience. These factors may be classified under three stream and their perception by (the facility users) . . . descriptions of
general performance measures describing the roadside pedestrian individual levels of service characterize these conditions in terms of
environment: (a) sidewalk capacity, (b) quality of the walking envi- such factors as speed and travel time, freedom to maneuver, traffic
ronment, and (c) pedestrian’s perception of safety (or comfort) with interruptions, and comfort and convenience for the facility type.”
respect to motor vehicle traffic. These three measures are briefly With respect to measures of effectiveness, the manual states, “For
outlined below. each type of facility, levels of service are defined on the basis of one
The first performance measure, sidewalk capacity, was developed or more operational parameters that best describe operating quality
in the early 1970s by Fruin (1). His method, as formalized in the High- for the facility type” (2, p. 1–5). This is the direction of our (measure
way Capacity Manual (2), is the only established method of quan- of effectiveness) effort to model the roadway walking environment.
tifying sidewalk capacity. However, this performance measure is Therefore, a calibrated, transferable model is needed to objectively
limited in its applicability. It evaluates only conditions for an existing reflect the perceived safety or comfort of pedestrians along a roadway
(or a planned) sidewalk and then only from the perspective of “walk- segment using measurable traffic and roadway variables. In response
ing space” or effective sidewalk width available to the pedestrian. to this need, the Pedestrian LOS Model outlined herein has been
Additionally, it cannot be used to evaluate and prioritize roadways for developed. The model is objective, transferable, and applicable at
sidewalk retrofit construction, a widespread need in the United States the roadway segment and, ultimately, when combined with an inter-
today. This is an important limitation. It is estimated that typically less section LOS measure, it is applicable at the facility corridor and net-
than 20 percent of the collector and arterial networks of U.S. metro- work levels. It evaluates roadside walking conditions whether there is
politan areas have sidewalks. Furthermore, it is estimated that less a sidewalk or not. It can also demonstrate the impact of adding or
than approximately 3 percent of roadways have pedestrian activity improving sidewalks. It uses common, measurable traffic and roadway
levels that can be effectively measured by Fruin’s capacity method. variables for economy of data collection, accuracy, and reliable and
Currently, there is no established approach for the second measure, repetitive application. The model is designed to evaluate a roadway
that of the quality, or enjoyment aspect, of the walking environment. segment; it does not include intersections and their complex condi-
Several researchers and a number of planners have proposed qualita- tions, which are the subject of separate research initiatives.
tive measures of the total quality of the walking experience. Their
approaches include numerous qualitative assessments relating to the
pedestrian’s enjoyment of the walking experience (e.g., convenience DESIGN OF RESEARCH
of the walking experience and the perception of personal security).
Works by Sarkar (3, 4), Khisty (5), Dixon (6), Crider (7), and others This research initiative by Florida DOT placed people in actual traf-
are examples of methods that include a mixed combination of some fic and roadway conditions to obtain real-time feedback. Although a
factors of all three performance measures. However, most of these virtual reality, or simulation approach, was briefly considered by
methods require the presence of a sidewalk to be applicable. And researchers because of its advantage of safety to the participants, it
although the qualitative measure of a pedestrian’s enjoyment of the was not pursued because it was not possible to include or replicate all
walking experience is important to provide a complete picture of the response stimuli of the roadway environment. Accordingly, a special
walking environment and to design an “inviting” sidewalk, it is a sep- event was created to place a significant number of people on a walk-
arate measure of effectiveness and must be developed and calibrated, ing course consisting of typical roadways in a typical U.S. metropol-
if possible, separately from the sidewalk capacity or safety perception itan area. The purpose was to obtain their real-time response to the
measures. roadway environment stimuli and to create and test a mathematical
The third measure, the perceived safety or comfort (with respect relationship of measurable factors to reflect the study participants’
to the presence of motor vehicle traffic) has not, until now, been quan- reactions. It should be noted that the research was designed to elicit
tified as a stand-alone performance measure. The common expression responses from participants walking individually, not in pairs or
of pedestrians concerning how well a particular street or road accom- groups. The following sections outline this approach.
84 Paper No. 01-0511 Transportation Research Record 1773

Participants condominium units and other forms of attached dwelling units. Some
course segments had single-family homes directly fronting them. Por-
Nearly 75 people participated in the first (i.e., the course-walking) tions of the course passed through traditional grid street patterns; other
portion of the study. The participants represented a broad cross sec- parts ran through curvilinear street forms. Neighborhoods represented
tion of age, gender, experience level, and geographic origin. Partic- a mix of income levels.
ipants’ ages ranged from 13 to 69. Because of the potential hazards
of walking in urban-area motor vehicle traffic, children younger than
age 13 were not permitted to participate. The gender split of the study Participant Response
group was 47 percent female and 53 percent male. The researchers
and sponsor sought participant diversity in both geographic origin The real-time data collection activity of the study was promoted as an
and walking experience. Accordingly, the study test course was event titled the FunWalk for Science, with prize drawings and gifts
located in Pensacola, Florida—a metropolitan area with significant as incentives for participation. Volunteer participants were recruited
in-migration. The average participant had lived in areas other than using a broad-based, areawide multimedia approach that included
the Pensacola Bay region for most [approximately 73 percent] of newspaper notices and articles, radio announcements, and direct mail-
their lives. ings by and to numerous organizations and businesses. Displays with
There was a considerable range of walking experience among the brochure-registration forms were deployed at area retail sports outlets,
participants. A significant number made relatively few walking trips health clubs, colleges, government office lobbies, major employers,
(hence, mileage), and some reported that they walked extensively and bicycle shops.
virtually every day of the week. Average distances walked per week The need for a large number of volunteer walkers mandated a
ranged from a low of 1.6 km (1 mi) to a high of 79 km (49 mi). weekend testing period. Accordingly, the FunWalk for Science was
scheduled for the morning of one of the busier (from a traffic-volume
standpoint) Saturdays of the year in Pensacola, March 18. To ensure
Walking Course that all participants experienced uniform motor vehicle traffic vol-
umes, the event was run during a single time block in the midmorn-
A walking course was designed to subject participants to a variety of ing. Participants first updated or completed registration forms that
traffic and roadway conditions. It included road segments with traffic included a variety of demographic questions. They were then briefed
and roadway conditions typical of U.S. metropolitan areas. Approxi- in groups as to the purpose and rules of walking the course. Follow-
mately 8 km (5 mi) in length, the looped course consisted of 24 road ing the briefings, walkers were sent to two starters who released them
segments (48 directional segments) with near equal lengths, but with onto the course individually at 1-min intervals, in opposite directions.
varying traffic and roadway conditions. Although most of the seg- Although the participants were briefed on the course configuration
ments were collector and arterial roads, some were local streets. Dur- and had instructions for completing the response cards, course proc-
ing the walking event stage of the study, traffic volumes ranged from tors were deployed at strategic points throughout the course. The
a low average daily traffic (ADT) of 200 to a high ADT of 18,500. proctors consisted of staff from the West Florida Regional Planning
The percentage of heavy vehicles [as defined in the Highway Capac- Council, Florida DOT, the University of Florida, SCI, Inc., and a
ity Manual (2)] ranged from 0 to 3 percent. Traffic running speeds number of regional bicycle and pedestrian coordinators from
ranged from 25 to 125 km/hr (15 to 75 mph). The roadway cross sec- throughout Florida. The proctors ensured that temporal spacing
tions included two to four lanes in forms of one-way, undivided, between walkers was maintained and that participants were inde-
divided, and continuous left-turn median lane configurations. The pendently completing the response cards as they walked each seg-
walking course included both curb and guttered as well as open shoul- ment. Participants were encouraged to reflect on their accumulating
der cross-sectioned roadbeds. Some segments had striped shoulders, experience and regrade any previously walked segments as they pro-
and some included designated bicycle lanes. ceeded through the course.
There were a variety of typical metropolitan area roadside con- The study’s purpose was to evaluate the quality, or LOS, of the
ditions in the course. For example, some segments were urban in roadway segments, not the intersections. Accordingly, participants
character with mixed combinations of on-street parking, landscaped were instructed to disregard the conditions at intersections and their
buffers, street trees, and buildings adjoining the sidewalks, with immediate approaches. They were also encouraged to exclude from
structures and awnings covering the sidewalks. Some segments their consideration the surrounding aesthetics. They were to include
were more suburban or rural in nature with roadside characteristics only conditions in, or directly adjoining, the right-of-way. The par-
ranging from no sidewalks to sidewalks directly adjoining the travel ticipants evaluated on a 6-point (A to F) scale how safe and comfort-
lanes, to sidewalks with intervening buffers of widths ranging from able they felt as they traveled each segment. Level A was considered
0 to 7.6 m (25 ft). the most safe and comfortable (or least hazardous). Level F was con-
The walking course passed through a spectrum of land develop- sidered the least safe and comfortable (or most hazardous).
ment forms and street network patterns found in U.S. metropolitan
areas. Retail commercial development forms ranged from large retail
shopping centers to small convenience strip centers. Some segments REDUCTION AND ANALYSIS OF DATA
had office buildings or other professional service establishments
fronting them. Other land uses included churches, auto dealerships, The study design yielded approximately 1,700 initial observations
banks, sit-down and fast-food restaurants with drive-throughs, pro- coincident with a myriad of traffic and roadway conditions through-
fessional and personal care businesses, car repair shops, and light out the walking course. The resulting data were compiled into both
industrial areas. spreadsheet and Statistical Analysis Software (SAS) program data-
In the residential portions an array of development forms directly bases for extensive analyses. Response outliers and trends were
adjoined the course. Residential dwellings included apartment and identified resulting in 1,250 observations and 21 roadway sections
Landis et al. Paper No. 01-0511 85

(42 directional segments) available for further analysis of the specific The factors listed above were considered the most probable pri-
effect of traffic and roadway variables. mary factors affecting pedestrians’ sense of safety. As such, they are
An interesting response trend was identified, ultimately determined the basis for the preliminary structure and testing of the Pedestrian
to be that of response (or scoring) fatigue. A slight diminishing scor- LOS Model represented in the following mathematical expression:
ing trend was evident. Course length was not a factor (the average
total duration of the participant’s course experience was approxi- Pedestrian LOS = a1 f (lateral separation factors)
mately 2 h) due to the clearly constant slope of the response trend. Pre-
sentation order of the segments was not a source of the trend either, + a2 f ( traffic volume)
because the course presented a variety of traffic, roadway, and urban + a3 f (speed, vehicle type)
forms in a random distribution. Because the participants walked the
course in two direction groups, averaging the responses allowed for + a4 f (driveway access frequency
removal of the fatigue trend, thus Pearson Correlations among the and volume) + an f ( xn ) + L + C (1)
traffic and roadway variables and stepwise regression of the depen-
dent variable were possible using the nonbiased (averaged) responses Researchers conducted stepwise regression analyses using the
for correlation. 1,250 real-time observations. Numerous variable transformations
and combinations of the factors were tested. Table 1 shows the best
model form and its terms, coefficients, and T-statistics. The correla-
MODEL DEVELOPMENT tion coefficient (R 2 ) of the best-fit model is 0.85 based on the aver-
aged observations from the 42 directional segments (see Figure 1 for
Several Pearson Correlation analyses were run using the SAS pro- a plot of predicted pedestrian LOS versus mean observed values).
gram on a variety of traffic and roadway variables. Not surpris- The coefficients are statistically significant at the 95 percent level.
ingly, several variables exhibited some colinearity. However, the Thus, the following model was developed:
colinearity was not enough to preclude the inclusion of some col-
Pedestrian LOS = −1.2021 1n (Wol + Wl + f p × %OSP + fb
inear variables into the model because of notable exceptions. For
example, although in some cases the presence and width of side-
walks and buffers cor-related with increasing speed, in many cases × Wb + fsw × Ws ) + 0.253 1n(Vol15 L )
they did not, reflecting that the current practice of roadside design
(or provision of sidewalks and buffers) is not consistent with pro- + 0.0005 SPD2 + 5.3876 (2)
viding a uniform level of pedestrian safety and comfort through-
out transportation systems. where
A “long list” of potential primary independent variables influenc- Wol = width of outside lane (feet),
ing pedestrians’ sense of safety or comfort within the roadway was Wl = width of shoulder or bike lane (feet),
generated and then tested (along with numerous other potential fac- fp = on-street parking effect coefficient (= 0.20),
tors) in the stepwise regression portion of the model’s development. %OSP = percent of segment with on-street parking,
The long list was generated based on the following: (a) results of the fb = buffer area barrier coefficient (= 5.37 for trees spaced
Pearson Correlation analyses; (b) variables (and model terms) identi- 20 feet on center),
fied by group consensus and confirmed during the development of the Wb = buffer width (distance between edge of pavement and
earlier Roadside Pedestrian Conditions Model [developed for the sidewalk, feet),
Tampa metro area’s Hillsborough County Metropolitan Planning Ws = width of sidewalk (feet),
Organization Pedestrian Plan (8)], which is currently the basis for Vol15 = average traffic during a 15-min period,
several major metropolitan area pedestrian plans; and (c) extensive L = total number of (through) lanes (for road or street),
iterative testing of segment groupings with common levels of inde- SPD = average running speed of motor vehicle traffic (mph),
pendent variables (wherein additional variables were identified and
that potentially could further explain the variation of the dependent fsw = sidewalk presence coefficient
variable—the pedestrians’ ratings of safety and comfort). The = 6 – 0.3Ws. (3)
resulting long list of primary factors included, but was not limited to
the following:
TABLE 1 Model Coefficients and Statistics
1. Lateral separation elements between pedestrians and motor
vehicle traffic, including
• Presence of sidewalk,
• Width of sidewalk,
• Buffers between sidewalk and motor vehicle travel lanes,
• Presence of barriers within the buffer area,
• Presence of on-street parking,
• Width of outside travel lane, and
• Presence and width of shoulder or bike lane;
2. Motor vehicle traffic volume;
3. Effect of (motor vehicle) speed;
4. Motor vehicle mix (i.e., percentage of trucks); and
5. Driveway access frequency and volume.
86 Paper No. 01-0511 Transportation Research Record 1773

FIGURE 1 Residual plot of predicted and standardized residuals.

The Pedestrian LOS Model equation was created with a statistical confirms, the value of a sidewalk varies according to its location and
significance at the 95 percent level. The factor “driveway access fre- buffering (i.e., the lateral separation) relative to the motor vehicle
quency and volume,” although included in the stepwise regression traffic. In general, as the lateral separation increases, the pedestrian’s
analyses, was not found to be statistically significant at that level. comfort or sense of safety also increases (see Figure 2). Additionally,
Table 2 may be used as a basis for stratifying the model’s numer- when a barrier such as on-street parking, line of trees, or roadside
ical result into a pedestrian LOS class when it is applied to a partic- swale is present in the buffer area between motor vehicle traffic and
ular roadway segment. It should be noted that this stratification was the pedestrian, the pedestrians’ sense of protection, hence safety, is
predetermined because the responses gained in the study were based improved (see Figure 3). Finally, the frequency of parked cars, trees,
on the standard U.S. educational system’s letter grade structure (with or an increase in the depth of the intervening roadside swale would
the exception of Grade “E”). further improve the sense of safety.
The mathematical expression that reflects these elements of lateral
separation, barriers, buffers, and presence of a sidewalk follows:
DISCUSSION OF MODEL TERMS
LS = Wol + Wl + f p × %OSP + fb × Wb + fsw × Ws ( 4)
Terms of the calibrated model were developed and refined through
extensive variables transformation testing and regression. The fol- Examples of how the lateral separation elements are used to quan-
lowing briefly outlines some of the aspects of the terms and how the tify some typical roadway cross sections follow.
dependent variable responds to them. Figure 4 shows a curbed cross section with no vertical barriers in
the horizontal buffer area between the travel lane and sidewalk. Note
that there is no on-street parking, therefore the %OSP term equals
Presence of a Sidewalk and Lateral Separation
0. Thus for this scenario, the lateral separation term is given by the
Having a safe, separate place to walk alongside the roadway is fun- following:
damental to pedestrians’ sense of safety and comfort in the roadway
environment. This sense of safety or comfort is strongly influenced LS = Wol + Wl + fb × Wb + fsw × Ws (5)
by the presence of a sidewalk. Furthermore, as the calibrated model
In the case in which there is on-street parking, as illustrated in Fig-
ure 5, its effect as a barrier is quantified as in Equation 6. Note that
there is no striped shoulder or landscape buffer, therefore the Wl and
TABLE 2 Level of Service Categories

FIGURE 2 Effect of lateral separation.

Landis et al. Paper No. 01-0511 87

FIGURE 5 Lateral separation with on-street parking.

ber of through lanes), it uses Ld (total number of directional through

lanes).

Vol15
Traffic volume = ×D (8)
Ld

where
Ld = total number of directional (through) lanes (for road or
street), and
D = directional factor
This effect on the walkers in the study was found to be statistically
significant. Transformations of this variable and subsequent stepwise
FIGURE 3 Typical barriers within the roadside buffer. regressions revealed that at lower traffic volumes, changes in the inde-
pendent variable produced significant changes in the dependent vari-
able. At higher volumes, however, there was less sensitivity; hence,
Wb terms equal 0. Thus, the lateral separation term is simplified to the the natural log mathematical form of this term.
following:

LS = Wol + f p × %OSP + fsw × Ws (6) Effect of Speed

This section introduced the elements of lateral separation and Similarly, the speed of motor vehicle traffic was confirmed as sig-
their mathematical expression. The next sections describe the other nificantly affecting pedestrians’ sense of safety. As speed increases,
two statistically significant terms of the Pedestrian LOS Model. pedestrian discomfort increases. It was determined that the depen-
dent variable had an exponential relationship with the average run-
ning speed of the motor vehicle traffic, somewhat similar to that
Motor Vehicle Volume relationship discovered during the development of the Bicycle Level
of Service Model (9), which has been incorporated into Florida’s
The frequency of motor vehicles passing pedestrians, represented by multimodal level of service analysis guidelines (10).
the outside lane volume, was also found to be a significant factor.
As passing frequency increases, the pedestrians’ feeling of safety
decreases. The effect of traffic volume is calculated by the following: Driveway Access Frequency and Volume

Vol15 Along a roadway segment, uncontrolled vehicular access to adjoining

Traffic volume = ( 7) properties (i.e., driveway cuts) was thought to reduce pedestrian sense
L
of safety. This transverse feature represents a similar “turbulence” or
The equation above assumes a 50/50 directional distribution. In hazard to the pedestrian as to motor vehicle operators. Accordingly,
cases in which the directional distribution is other than 50/50, Equa- as the number of driveways increases, a corresponding decrease in the
perceived safety to the pedestrian was expected. Affecting this per-
tion 8 (below) should be used. The difference between the two is that
ception of safety is the volume of vehicles accessing the driveways.
Equation 8 uses a directional factor and instead of using L (total num-
However, stepwise regression analyses revealed that this effect was
not statistically significant at the 95 percent confidence level.

FINDINGS AND APPLICATIONS

The result of this initial research sponsored by Florida DOT is the

development of a reliable, statistically calibrated pedestrian level of
service model suitable for application not only in Florida metropoli-
tan areas, but also throughout North America. The Pedestrian LOS
Model provides a measure of a roadway segment’s performance with
FIGURE 4 Buffers and sidewalk. respect to pedestrians’ primary perception of safety or comfort; as
88 Paper No. 01-0511 Transportation Research Record 1773

such it serves as the basis for Florida DOT’s statewide multimodal REFERENCES
(particularly for the pedestrian mode) LOS evaluation techniques.
However, it can also be used to greatly influence roadway cross- 1. Fruin, J. J. Pedestrian Planning and Design. Metropolitan Association
sectional design and can help in the evaluation and prioritizing of the of Urban Designers and Environmental Planners, Inc., New York,
1971.
needs of existing roadways for sidewalk retrofit construction, appli- 2. Special Report 209: Highway Capacity Manual, 3rd ed. (1994 update).
cations for which the model’s precursor, the Roadside Pedestrian TRB, National Research Council, Washington, D.C., 1985.
Conditions Model, has been successfully used. For example, trans- 3. Sarkar, S. Determination of Service Levels for Pedestrians, With Euro-
portation planners and engineers can now establish a target pedestrian pean Examples. In Transportation Research Record 1405, TRB,
LOS and use the model to test alternative roadway cross-section National Research Council, Washington, D.C., 1993, pp. 35–42.
4. Sarkar, S. Evaluation Method for Comfort Requirement in Outdoor
designs by iteratively changing the independent variables to find the Pedestrian Spaces. Presented at the 75th Annual Meeting of the Trans-
best combination of factors to achieve the desired LOS. The model portation Research Board, Washington, D.C., 1996.
thus provides roadway designers with solid guidance on how to bet- 5. Khisty, J. C. Evaluation of Pedestrian Facilities: Beyond the Level-of-
ter design pedestrian environments: how far sidewalks should be Service Concept. In Transportation Research Record 1438, TRB,
placed from traffic; what types of buffering or protective barriers are National Research Council, Washington, D.C., 1994, pp. 45–50.
6. Dixon, L. B. Bicycle and Pedestrian Level-of-Service Performance
needed and when; how wide the sidewalk should be; and so on.
Measures and Standards for Congestion Management Systems. Pre-
Finally, the Pedestrian LOS Model, when coupled with the capacity sented at the 75th Annual Meeting of the Transportation Research
(Fruin) measure and a quality performance measure (i.e., a walkabil- Board, Washington, D.C., 1996.
ity audit, in the case of an existing sidewalk, to assess the enjoyment 7. Crider, L. Safe Ways to Schools Program. Florida Department of
and convenience of the walking experience) “completes the picture” Transportation, 1998.
of the roadside walking environment. 8. The Hillsborough County MPO Pedestrian Plan. Hillsborough County
Metropolitan Planning Organization, Tampa, Fla., 1998.
9. Landis, B. W., V. R. Vattikuti, and M. T. Brannick. Real-Time Human
Perceptions: Toward a Bicycle Level of Service. In Transportation
Research Record 1578, TRB, National Research Council, Washington,
ACKNOWLEDGMENT D.C., 1997, pp. 119–126.
10. McLeod, D. S. Multimodal Arterial Level of Service. In Transportation
The authors wish to thank Jennifer Toole of SCI, Inc., the West Research E-Circular E-C018: Fourth International Symposium on High-
way Capacity, TRB, National Research Council, Washington, D.C., 2000,
Florida Regional Planning Council, Drs. Linda Crider and Rhonda pp. 221–233.
Phillips of the University of Florida, and the state and regional bicycle
and pedestrian coordinators of Florida who assisted in this study. Publication of this paper sponsored by Committee on Pedestrians.