Office of Planning, Environment, & Realty (HEP)

Data Collection and Modeling Requirements for Assessing Transportation Impacts of Micro-Scale Design: Executive Summ

Chapter 2.0 Micro-scale Design Variables

Although the transportation planning literature identifies more than 300 variables that could describe a certain aspect of micro-scale design (MSD), the variables can be grouped along several dimensions to create a manageable set. These variables can be grouped according to their physical properties, their statistical attributes, and their policy connections. Several of the commonly used classifications include:

• Variables that describe the density, diversity and design characteristics of a particular site or neighborhood (sometimes called the three D's of MSD);
• The mathematical and statistical attributes that include whether or not the variable is linear or non-linear, and whether or not the independent variable is ordinal, cardinal or binary; and
• Description of how we observe the variable, which includes whether our observations suggest that we see the variable subjectively or objectively, and whether we see it as a single element or as a surrogate for some other combination of characteristics, including composite variables.

These distinctions carry some weight in the development of travel behavior models since many modelers shy away from "subjective" variables and the so-called objective variables frequently are surrogates or only a partial measure of the impact of a subjective variable.

We are frequently concerned about the stability of subjective variables, although they can be well behaved if the evaluator is provided with clear direction. Researchers have also suggested that choices are based on personal, subjective responses to all MSD elements, even those that we call "objective." Another category for MSD variables includes composites, such as transit-oriented design (TOD) or neo-traditional neighborhood design (NTND). These terms frequently come into use when we find it difficult to identify just which characteristics of the environment influence our travel choices.

2.1 Micro-Scale Design Elements

Just what are MSD elements, and how do they affect travel behavior? For purposes of travel demand analysis, we are including those site-specific and urban design elements in the man-made environment that appear to affect travel choices, including the choices of when, where, how, and by what route to travel. These elements include sidewalks, pedestrian-oriented street systems with protected intersection crossings, location of structures relatively close to the sidewalks, and parking control and location that foster or support walking and transit use. A common thread running through all these elements is their relatively small, human scale and their association with individual building sites.

Transportation impacts of MSD and mixed-use development often result in conflicts between regional and local government regulations. Regional, state, and federal governments are interested in transportation impacts and the resulting congestion levels, often far from the particular site being developed. Local governments are concerned with the economic vitality of the development.

The desirable attributes of independent variables for travel models include simplicity, stability and their high correlation with dependent variables such as number of trips or choice of mode. Independent variables should also have a recognizable and statistically defensible correlation with the policies they represent. After identifying more than 300 variables it seemed important to classify them according to dimensions that will help us select the most efficient estimators for travel demand models.

The following sections describe the results of recent research on the most frequently analyzed MSD variables and the relationship of the MSD variables to travel behavior. Research often includes analysis of several variables at one time, making it difficult to isolate the impact of each element. There is no universal agreement on the significance of MSD elements. Therefore, the list of those most frequently cited has been arranged in alphabetical order.

2.2 Accessibility and Connectivity

Accessibility and connectivity are widely used in transportation planning to provide a measure of satisfaction and transportation system effectiveness, and as an integral means of establishing winners and losers when evaluating alternative transportation systems. Lacking a well-established lexicon, accessibility and connectivity have been used as the label for a number of different concepts regarding the ability to reach a satisfying number of activities.

2.2.1 Definition

Accessibility usually is defined as the number of destinations or attractions within a defined reach, either distance or travel time. Usually "more" is considered better. Actually "enough" is the appropriate target, although that seems to be somewhat elusive.

Connectivity refers more to the ease with which destinations may be reached because the locations are well connected and, hence, more accessible. High levels of connectivity imply smaller grid pattern networks and facilities that enhance pedestrian travel. In both these ways, accessibility is generally increased.

Most accessibility and connectivity measures use land use information such as households and employment, with zone-to-zone travel times. They differ, however, in their construction and, consequently, in their translation of policy into changes in travel behavior. Two example measures will be discussed here: Accessibility Measures 01 and 02.

2.2.2 Means of Measurement

Accessibility 01, a typical accessibility measure, is computed by estimating the number of jobs, households, or retail employment within a certain distance or travel time from the site in question. This measure (sometimes called the "cutoff accessibility," which sums the number of households or jobs within a certain travel time) has probably been used more in travel demand studies than the other variables with the same name. It is fairly easy to calculate and quite easy to understand. Normally, the measure is stated in words such as, "the number of retail employees within 15 minutes of transit travel time." Unfortunately, the measure is a "cliff" measure and it is possible that adjacent zones can have very different values. Cliff measures tend to be a creation of our techniques used to measure accessibility as a function of time or distance from a given traffic zone to other zones. It can occur when the number of the destination points lie just beyond the cutoff distance or travel time results in adjacent zones having disparate numbers of destination points.

Accessibility 02 is estimated by summing, for all destination zones, the product of the land use information and a function of travel time; in some cases, the function of travel time is time raised to a negative power (such as 2). This second accessibility measure, (sometimes referred to as total accessibility), is not a cliff variable, but it is not intuitively easy to visualize. For example, one study used the following to measure accessibility to employment:

Accessibility (zone I) = Sum for all destination zones (employment * highway time -2.0 )

In this accessibility measure, employment located "far" from the origin zone is given a small weight, while employment close to the origin zone is given a large weight. For example, a zone with 200 employees ten minutes from the origin zone would produce a value of 2.0 while a zone with 200 employees twenty minutes from the origin zone would produce a value of 0.5. When the product of land use and time function are summed, the close zones obviously provide the greatest contribution to the accessibility measure. This is an excellent measure of accessibility; its major deficiency is the lack of a simple description of the measure to the general public. This measure functions as a relative measure (i. e., zone x is twice as good as zone y).

The accessibility measure, of course, does not have to use employment as the size variable but can use any variable associated with the analysis. Transportation planners speak of trips being produced at their normal origin point, such as a home or a household, and attractions as the ends of trips that are attracted to non-home locations, such as retail or office spaces. The travel time for the accessibility measure does not have to be highway travel time, but it also can be transit time, non-motorized time (walk or bicycle time) or any other measure of zone-to-zone impedance. For example, a multi-modal accessibility might use the Log Sum measure from the mode choice logit model. The analyst may also wish to consider using some equivalent time measures. For example, the out-of-vehicle travel time might be weighted more (typically, 2.5 times) than the in-vehicle travel time.

Accessibility and connectivity may provide measures of compactness and decentralization. For example, Miller, E. J. and A. Ibrihim define the "combination of the physical distribution of activities and the activity patterns of people over time and space" as urban form. They propose simple measures of density, decentralization and structure to represent the physical component of urban form. Based on statistics for the greater Toronto area, regression models were developed to include the home-based work vehicle/ kilometers per worker as a function of a set of urban structure variables. These included the distance to the central business district (CBD), the distance to the nearest high-density employment zone outside the central area, accessibility to rail or subway stations, the number of jobs within a 5-kilometer radius of the zone centroid divided by the population within that same 5-kilometer radius (a jobs/ housing ratio surrogate), the number of jobs within 5 kilometers of the job centroid normalized to the largest observed value, and the zone population density in terms of thousands of persons per square kilometer.

Accessibility has sometimes been measured by a function of travel impedance, time, distance, or cost, and the amount of activity (the number of employees or square footage of activity space). (Handy, S.) We can think of the factors that contribute to accessibility in at least two sets. The first relates to the separation between activities, and the second relates to the nature of destinations within the available set and covers a wide range from the amount of activity to the quality of the shopping center design. Urban designers can suggest potentially important qualitative factors in the description of urban form.

2.2.3 Potential for Describing Travel Behavior

The previously discussed study by Miller and Ibrihim concluded that:

• Centralization or compactness matters (that vehicle miles traveled (VMT) per worker increased with distance from the center);
• A system of high-density employment/ activity centers would appear to reduce travel when compared to sprawl (this supports the concept of a multi-nuclear city);
• Other than the impact of high-density employment centers in the suburbs, the job/ housing balance was not found to have a significant impact; and
• "Population density appears to be more of an intermediate variable rather than a strong causal variable in the explanation of variations in VMT per worker across the urban area."

Accessibility as a framework for characterizing urban form leads to the following conclusions:

• High levels of accessibility should be associated with shorter average travel distances.
• More activity should lead to greater variety in the range of options, therefore high accessibility.
• A travel budget is suggested where residents in low accessibility areas compensate for longer trip distances by taking fewer trips.
• Residents in high accessibility areas make more trips because they are easier and have a greater variety of potential destinations.
• In high accessibility areas, residents may have more viable options to walking.
• Walk trips may replace driving for some trips, or walk trips may be in addition to driving trips high accessibility induces travel.

Accessibility has also been defined as the "intensity of the possibility of interaction" (Hansen, 1995, as quoted in Handy, Understanding the Link, etc., as above).

The research supporting the above accessibility concept involved case studies of four neighborhoods in the San Francisco Bay area. Their selection was based on three factors:

• Location within the region and accessibility to regional centers,
• One traditional and one typical neighborhood in each of two areas, and
• Socio-economic characteristics of the residents in the neighborhoods.

The research involved the following neighborhood features:

• Characterization and evaluation of urban form,
• A description of the neighborhoods based on urban design characteristics, including:
• Sidewalk width/ size of streets
• Building setback
• Variations in building materials
• Orientation of buildings to the street
• Building design
• The nature of human activity which, in turn, influences the perception of the pedestrian environment and thus influences the choice to walk.

Grid street patterns and small blocks encourage walking and transit use but only if there are suitable destinations within acceptable maximum walk distance and/ or attractive transit service. In this case, function does not always follow form. Grid-street patterns, as seen in Figure 2.2-1, provide connectivity superior to that of do cul-de-sac plans and reduce VMT (McNally, 1964).

Source: Planning for Transit-Friendly Land Use: A Handbook for New Jersey Communities. Prepared for the Federal Transit Administration. June 1994.

2. 3 Balance

2.3.1 Definition

Within the MSD lexicon, balance suggests an ordered efficiency that minimizes travel as a part of daily activities. Since the reported research has not put forth a precise measure, balance is frequently expressed as the job/ housing ratio.

2.3.2 Means of Measurement

Although the regional level of jobs per household may be 1.5 or slightly higher, the ratio for individual jurisdictions may vary dramatically. The ratio for the jurisdiction with the CBD may be considerably higher than for other counties because of the concentration of jobs in the CBD. On the other hand, counties (farthest from the CBD) may have considerably lower job/ housing ratios, and even below 1.0. Values near the regional average (1. 6 for Washington, DC, for example) should be reflected in the minimum commuting distances, on the assumption that the jobs match the household labor force.

2.3.3 Potential for Describing Travel Behavior

The reported research has been inconclusive, but the few existing studies report that regional travel decisions seem to be relatively insensitive to changes in the job/ housing ratio. It is possible that this is partly due to the labor force skills not matching nearby job requirements.

2.4 Density

2.4.1 Definition

Much of the popularity of using density to explain variation in transit and non-motorized travel is the ease of collecting and computing of data.

Density measures are used in travel studies as they tend to imply measures of urban "congestion." For example, it could be inferred that a high population density would make it more difficult to garage an automobile, which would increase the implicit cost of owning an automobile. It could also be inferred that a high employment density would increase the cost of the land and therefore would increase the price of parking an automobile with a consequent increase in the use of transit.

2.4.2 Means of Measurement

The measure of density of development is widely used in relationships between MSD elements and travel behavior. It takes on a number of different forms:

• Persons per acre or square mile
• Households per acre or square mile
• Employment, employees, or jobs per acre or square mile

The land area used to calculate the density may be:

• Total area of zone or census area
• Office or retail development and/ or zoned area
• CBD or other specified sub-region
• Transit station/ stop area (physical boundary or psychological limit)

In the crudest sense, we may use U. S. Census data (for population and households) and select land areas from TIGER files. Employment data are generally less readily available; but in many models only the CBD density is needed, and an approximation can often be calculated from state or local employment statistics.

2.4.3 Potential for Describing Travel Behavior

The density variable( s) has desirable properties. It is a useful, predictive variable, but not a causal variable: as it increases, so does transit use, walk and bicycle access to transit and increased non-motorized travel. Assertions that increased density increases transit use, walk access to transit trips and non-motorized travel must be accompanied, however, by carefully defined assumptions about the design features of the development and the transportation system performance (see Figure 2.4-1).

Research in Portland, Oregon (Sun, Wilmot & Kasturi, 1998) into the fundamental relationship between land use and transportation asserts that households in high-density locations will make fewer and shorter trips. However, many of the households in high density development areas are relatively small and generally make fewer trips. Another issue is that the surveys may only be counting vehicle trips. The Washington, DC area employment center data collected by Douglas & Evans (1998) found that workers in high density areas actually make more work-related trips than their colleagues in suburban work places, but their short walking trips were not always counted.

Research (Frank & Pivo, 1994) concluded that the relationships at the tract level between density and mode choice are relatively weak. Using data from the Puget Sound area (Seattle, Washington), the authors examined the relation between mode choice (single-occupant vehicle (SOV), transit and walking) for work trips and shopping trips versus urban form variables, which in this case refers to population and employment density and land use mix. The density calculations used the gross area of the census tract. The mix of land uses was calculated using the entropy function described in Section 2.5, using seven land use types, which resulted in a maximum value of 0.794. The maximum population density was 47 residents per acre and the maximum employment density was 401 employees per acre. The study concluded that:

• The relationships between employment density, population density, land use mix, and SOV usage were consistently negative for both work and shopping trips.
• The relationships between employment density, population density, land use mix, and transit use and walking are consistently positive for both work and shopping trips.
• As employment density increases to more than 75 employees per acre, there is a significant shift from SOV use to transit and walking.
• The reduction in SOV travel was less significantly associated with population density than with employment density.

Figure 2.4-1 Retail Density and Imporved Access (Connectivity)

Source: Planning for Transit-Friendly Land Use: A Handbook for New Jersey Communities. Prepared for the Federal Transit Administration. June 1994.

The study concluded that although the relationships can be measured at the census tract level, the relationships are relatively weak (see Table 2.4-1). This suggests that further research on land use mix of smaller geographic units would be more sensitive to the relationships with mode choice.

Table 2.4-1 Correlation Coefficients Between Urban Form and Mode Choice Variables
WORK TRIPS
Travel Behavior Variables Employment Density Population Density Mixing of Uses
% SOV-0.26--0.13
% Transit0.590.190.15
%Walk0.430.340.21
SHOPPING TRIPS
% SOV-0.15--
% Transit0.440.16-
% Walk0.240.31-

Source: L. D. Frank and G. Pivo, Impacts of Mixed Use and Density on Utilization of Three Modes of Travel: Single-Occupant Vehicle, Transit, and Walking. In Transportation Research Record 1466, Washington, D. C., 1994.

Another assessment of density impacts was based on research comparing five San Francisco Bay area communities with household densities ranging from 3.8 to 117 households per residential acre. Population densities range from 2 to 52 persons per total acre. Based on these statistics for 1998 auto use, Holtzclaw concluded that a 100% increase in density is associated with a 30% reduction in vehicle travel, both per capita and per household. While not actually measured, the reduction in VMT was attributed to accessibility to commercial and retail properties and transit service.

By itself, density does not appear to have a major impact on travel decisions. This is particularly true in a mono-cultural development such as a housing tract where the residents have walk access only to other residences. Residential density is frequently highly correlated with the transportation prevailing at the time of development. For example, many of the inner suburbs surrounding major cities were built in the early part of the 20th Century when walking and transit were much more important travel modes for everyday use. In an attempt to provide decent accessibility, lots were small, sidewalks were plentiful, and commercial areas were within tolerable walking distance from most houses. Such development is conducive to non-motorized travel and provides greater travel freedom for children, for teenagers who do not yet drive, and for elderly people who no longer drive.

It is important to recognize that it is not density that is driving these decisions, but the accessibility of destinations within tolerable walking distances and comfortable and attractive urban design for pedestrian and bicycle use. The use of area type and, in some cases, density, may depend on the assumption that increased density brings with it increased diversity and/ or pedestrian and transit-oriented infrastructure. Older developments also carry with them different architecture, different construction and integrity, and a location closer to downtown where people go for dining and entertainment in a cultural and recreational atmosphere. A higher level of transit service is usually found in such areas.

2.5 Diversity or Mix of Land Uses

Diversity of land uses and dense development often lead to reduced vehicle trip making, more walk trips, and attractive urban settings. A mixed-use development may provide the desired diversity and spatial arrangement of activities. Individuals with similar settlement patterns may differ in the activities they choose, the locations they choose for these activities, and the way they choose to travel to desired locations. This difference in behavior is attributed to some extent to differences in urban forms.

2.5.1 Definition

The Urban Land Institute has a structured definition of mixed-use development that describes some of the relevant attributes but does not satisfy all of the requirements for predicting behavior patterns. Some of the mixed-use developments that fit one or more definitions include:

• A mix of commercial and retail spaces which, because of their operating times, can share parking spaces (such as hotels and offices) or share access roads
• A mix of retail and residential spaces
• A mix of residential, retail, commercial, and transportation facilities that produce a sustainable and somewhat enclosed community

Such mixed uses were incorporated in the "constructs" used in the MSM Study (Middlesex, Somerset, Mercer Regional Council, 1992). The purpose of the study was to develop settlement patterns that would combine future development in ways that would minimize vehicular traffic and/ or optimize the satisfaction of activity needs with a minimum of motorized transportation.

2.5.2 Means of Measurement

Measuring the success, vitality or "health" of a mixed-use development is a challenge. On one hand, we are searching for surrogates, or simple indicators, such as temperature and blood pressure used in a medical exam, to indicate the vitality of a settlement. At the same time, we are looking for thresholds to give us some indication as to whether a development is healthy.

One scale for evaluating the mix of uses is whether a certain proportion of residents can fulfill a major proportion of their weekly shopping needs within walking distance of their residence. To determine this, we can measure the proportion of households that have to fulfill their shopping needs, and the proportion of shopping needs that have to be fulfilled in order to satisfy the neighborhood mixed-use concept.

Measurers would need to examine pedestrian friendliness and mixed-use character at each end of a home-to-work trip. In other words, both the home and work ends need to have facilities that result in a mixed-use development.

Two concepts that are used in evaluating mixes of use include assessment of parcel files, as a basis for analysis, and accessibility measures, which calculates the total employment or retail employment within one mile or within a particular time limit from the residence.

2.5.3 Potential for Describing Travel Behavior

The arrangement of diverse land use appears to have a strong influence on activity patterns and, thus, on travel patterns. Cervero, 1991, proposed a diversity function that is dimensionless, but little work has been reported on market response to the mixed-use developments. This early work related land use mix and suburban activity center characteristics relating the percentage of transit or carpool use as a function of parking availability, mixed use (a dummy variable) or tenancy (another binary variable where 1= multitenant, and 0= a single tenant in a building). The most highly correlated variable was building height, the number of stories in office buildings that was highly correlated with the percent of work trips made by mass transit. The parking supply had a weak but noticeable effect on automobile and transit use; as parking availability increased automobile use went up and transit use declined. Data were not available for disaggregate modeling.

There are several methods for considering mixed-use areas. At a fairly macro level, this is the mix of residential units and employment within a short distance, e. g., less than 5 miles. At the more micro level, this could be mixing residential and commercial establishments in the same block. At the macro level, major policy decisions can affect land uses and have a major effect on travel. A macro level mixed-used policy would attempt to "match" workers and jobs in fairly small markets. This would reduce "bedroom" communities and increase areas with a reasonable "match" between labor force and employment. The separation of residential areas and commercial areas can increase trip length, change modal market shares, and substantially increase VMT.

In estimating the distribution of travel, the travel times between areas are considered a major variable in the determination of trip length. But the relative location of the productions (where the workers live) and the attractions (where the workers work) can affect the average trip length to a much greater degree than travel times. At the micro-level, the mixing of residential and commercial establishments can increase the propensity to walk and may also promote shorter non-work vehicle trips. A micro-level mixed-use policy would attempt to match residences with "convenient" type retail areas (including cleaners, grocery stores, bookstores, etc.) within a fairly small area (within walking distance). This type of policy is much more difficult to implement since it runs counter to new super-large and commercially efficient stores.

On the residential side, many people do not want commercial establishments close to their residences. Separating "mixed use" from some density measures and accessibility measures is difficult, since the density and accessibility measures make use of the mixed-use information. Therefore, many effects of mixed-use policies can be captured in the other measures, especially accessibility measures. But policies on mixed use can be an extremely effective method of reducing traffic and should be seriously considered both at the macro-and micro-scale levels.

2.6 Neighborhood and Transit-Oriented Design Factors

2.6.1 Definition

This section reviews the research into the impacts that neighborhood and transportation facility design factors may have on travel, including TOD, NTND, street geometry, and provisions for sidewalks and bikeways. The discussion identified the micro-scale effects of these elements on travel, in addition to the more macro-scale effects that these design factors would have. For example, TODs and NTNDs might increase development densities and increase walk and transit accessibility. Since development patterns, either transit-oriented or neo-traditional neighborhoods, involve a combination of strategies, it is sometimes more effective to use illustrations. Figures 2.6-1 and 2.6-2 illustrate a discouraged and a preferred layout of development surrounding a major or minor intermodal transfer facility. Figure 2.6-3 illustrates the sidewalk orientation and parking locations suggested to provide transit-supportive and pedestrian-supportive development. These attributes of TODs and NTNDs are generally the result of research topics and should be revised as more research takes place.

TOD designs should incorporate land use configurations that minimize the impedance to access transit stops and stations in addition to increasing densities near stations. This might include placing parking lots behind buildings instead of in front of them, having covered and lit walkways at transit stops, and providing clear transit directions in easily identified locations. NTND would include micro-level mixed-use design coupled with good walk and bicycle access systems. In addition, the MSD of NTNDs is intended to promote walk and bicycle trips as well as increase visiting within the neighborhood.

2.6.2 Means of Measurement

Design factors are difficult to measure for the base year and even more difficult to specify for future years. Most current evaluation procedures include weighting and rating schemes that require a fairly detailed on-site evaluation of the area with a great deal of subjective and professional judgment. Hopefully, the increasing use of GIS systems to store urban data will allow these procedures to become more objective and efficient. In both cases, either detailed on-site evaluation or GIS analysis, the most probable overall measure for design factors will be a rating system, ranging from a low number (such as 1) for a design that does not minimize travel or promote transit or walk trips, to a high number for a design that is instrumental in promoting use of transit and walk trips. With GIS systems, it may be possible to estimate the walk time to bus stops more accurately and identify whether these travel times can be made on enclosed walkways or sidewalks.

Figure 2.6-1 Transit-Friendly Station Area Development

Source: Planning for Transit-Friendly Land Use: A Handbook for New Jersey Communities. Prepared for the Federal Transit Administration. June 1994.

FIGURE 2.6-2 TOD Density and Diversity

Source: Planning for Transit-Friendly Land Use: A Handbook for New Jersey Communities. Prepared for the Federal Transit Administration. June 1994.

Figure 2.6-3 Elements of Pedestrian and Transit-Oriented Development

Source: Planning for Transit-Friendly Land Use: A Handbook for New Jersey Communities. Prepared for the Federal Transit Administration. June 1994.

2.6.3 Potential for Describing Travel Behavior

TOD variables tend to reflect those attributes of station area development that research has suggested contribute to transit attractiveness. Many of these measures are subjective in nature and composite in structure that is, they describe an area as being friendly to transit, using judgement and descriptive indices.

In a study of access to rail station access in San Francisco and Chicago (Davis) were interested in the role of the built environment in explaining the mode of access. They summarized the distances to which walking predominates, as shown in Table 2.6-1 below:

TABLE 2.6-1 SUMMARY OF THE INFLUENCE OF DISTANCE ON MODES OF ACCESS AND EGRESS AMONG CLASSES OF BART STATIONS
Distance Up to Which Walking Predominates Mode of Access Beyond Walking Distance Mode of Egress Beyond Walking Distance
Station Class Home-End Access Work-End Egress Dominant Secondary Dominant Secondary
San Francisco Office Center3,000 ft4,000 ftTransit---Transit---
San Francisco Commercial/Civic Center4,000 ft3,300 ftTransitKiss-n-RideTransit---
Downtown Oakland3,800 ft3,600 ftTransitKiss-n-RideTransit---
Urban Districts3,300 ft3,600 ftTransitDrive-alone/Kiss-n-RideTransitBicycle
Suburban Centers2,700 ft3,300 ftPark-n-RideTransit/Kiss-n-RideTransitPassenger Pick-up
Low Density Areas2,900 ft2,900 ftPark-n-RideTransit/Kiss-n-RideTransitPassenger Pick-up

Source: Davis, J., R. Cervero, and S. Seskin. Mode of Access to Rail Transit. Cooperative Research Program, Transportation Research Board, Washington, DC.

In their analysis they define the catchment area as one-half mile from the rail station. The independent variables included the households and employees per acre, the percent of land area in commercial and residential uses and the entropy index of land use mixes. The study calculated the percentage change in probability in using each access mode associated with an increase of either one household or one employee per acre. The higher household densities give a higher proportion of walking trips to rail stations and a lower proportion by car. They found that an increase of one household per acre resulted in about two percentage points increase in the probability of walking, with a similar decrease in the probability of driving.

When looking at transit-oriented development, it is necessary to:

• Check the level of transit accessibility.
• Examine the level of service of transit, such as whether it's local bus or express bus.
• Examine whether or not there is neighborhood transit.

Using data from the BART Passenger Survey in the fall of 1992 Loutzenheizer, 1996, analyzed the reason for the variation in walking access trips among the various stations. The conclusion was that the variation was in great measure due to the design of the station and access facilities. At the time of the survey, walk access at shared individual BART stations varied from 3% to 74%. A series of analyses using logit models resulted in the following significant findings:

• Individual characteristics (income, age, education, job classification) influence walking more than urban design and station area characteristics.
• Walking distance is the most significant factor in the choice to walk.
• Males are more likely to walk than females.
• Population and dwelling density, while appearing significant when analyzing station area characteristics alone, are insignificant in a combined model taking into account individual characteristics.
• When stations are located in downtown corridor areas with office domination, there is a low incidence of walking to rail transit.
• Station areas with a strong retail-oriented environment produced the greatest proportion and incentive for walking.
• High incomes and the availability of a car are the strongest disincentives to walking, after distance and gender.

It will be noticed that the NTND and TOD developments circumvent this problem to some extent by presuming that the urban design guidelines employed during site planning provide a superior level of transit accessibility for the entire development. Today, actual implementation of these ideas has led to some mixed results, in part because the transit accessibility, while necessary, is not sufficient to ensure travel patterns that make use of the transit service and walking patterns provided by the development itself.

2.7 Pedestrian-Oriented and Bicycle-Oriented Development

The infrastructure elements required to support pedestrian and bicycle travel are emerging as we enter the 21st Century. Major elements are pathways, amenities, and safety-related fixtures that provide a secure and attractive series of paths in our urban areas.

2.7.1 Definition

Pedestrian-and bicycle-oriented development are the natural environment for MSD elements. The essential ingredients are a mixture of uses clustered together with acceptable pathways for walking or biking. Many of these attributes are implicitly included in the neo-traditional neighborhood and transit-oriented developments described in Section 2.6. In this section, the emphasis is more on complete trips made by non-motorized travel, principally walking and biking.

The initial infrastructure needed for bicycle commuting is relatively low-cost; provision of a few lockers, a few paths, and maybe a shower or two. Cyclists also need protection and safety, although the cyclists themselves may pick the fastest route, not necessarily the safest and most scenic.

2.7.2 Means of Measurement

Many transit-friendly factors are subjective with survey questions scaling from one to three or one to five. An agreed-upon definition of levels of pedestrian friendliness is crucial, so that ratings could be replicated if performed by different planners using the same definitions of service levels. There is an emerging consensus that a composite measure is probably needed and that research should concentrate on accurate definitions.

There is a wide range in the levels of some indicators. Developing local neighborhood travel would be easier if a parcel-level system were used. In the past, a network for pedestrian travel has been thought to be superfluous because of the short trips and the average size of the transportation analysis zone (TAZ). The need for pedestrian trip data is now more appreciated because of the tendency towards planning more expensive walk projects and congestion mitigation and air quality (CMAQ) programs. A walk origin-destination matrix could essentially be the major travel matrix with more trip ends occurring within cells on the main diagonal plus a few cells on either side of the diagonal.

The bicycle-oriented development variables describe infrastructure attributes that support bike use. They also may act as surrogates for community and employer attitudes toward bicycles and serious use for other than recreation trips. Bicycle planning and travel forecasting presents some unique challenges, partly because there are numerous small cyclist populations with characteristics and preferences that are still not well understood by most urban and transportation planners. There are recreational cyclists, business/ commuting cyclists, and daredevil, high-speed cyclists who take risks and travel along the fastest route. Although cyclists today make up a small proportion of the total regional trips, models of their choices can be calibrated. The EPA believes that cycling will be more important in the future, and bicycle facilities are eligible for CMAQ funds.

2.7.3 Potential for Describing Travel Behavior

Only a few MPOs include walk trips explicitly in their model process. Even those that estimate walk trips often use a model based on a crude relationship between trips and regional location, such as the CBD. In that process, the trip ends are discarded after the trip generation step, and are generally not traded for vehicle trips in response to policy changes.

Historically, home interview surveys either ignored short trips or explicitly directed the respondent to exclude trips of short duration (in minutes or blocks), thus under-counting numerous walk trips. One argument for this practice was the absence of a need for this information when planning roads and transit lines. Walkers were considered to be non-users. Likewise, those under 16 or over 70 years of age without driver's licenses were thought not to have a major impact on the auto driver trip pool. What is lost in this restricted type of data collection is the increased mobility and accessibility provided by urban environments.

This study recommends that current and future home interview surveys should collect as much non-motorized travel data as possible, even though their models might not use the information at the moment. If these surveys are to be the only source of household data for the study area for the next 5 to 10 years, then the details collected should be as complete as possible.

Figure 2.7 Pedestrian/Transit-Friendly Streetscapes

Source: Planning for Transit-Friendly Land Use: A Handbook for New Jersey Communities. Prepared for the Federal Transit Administration. June 1994.

Updated: 3/25/2014
HEP Home Planning Environment Real Estate
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000