The Effects of Landuse

The Effects of Land Use

Preface

This report presents the analytical results of a larger project undertaken for
the Federal Highway Administration by Cambridge Systematics, Inc.
investigating the "Effects of Land Use and Demand Management on Traffic
Congestion and Transportation Efficiency." Prior tasks examined the
literature that was available on this subject and explored the utility of
either using or augmenting existing databases. In support of this previous
work, JHK and Associates, Inc., assessed the potential of using the employment
site dataset developed as part of NCHRP Project 3-38(2), "Travel
Characteristics of Large-Scale Suburban Activity Centers." It was determined,
though, that the site characteristics contained in this NCHRP dataset were not
sufficient to support investigation of the interactive effects of land use and
travel demand management policies on an employee's commuting behavior.

The work reported herein represents an ambitious program of data collection
and analysis with respect to employment sites located in the Los Angeles
metropolitan area. Activities were carefully designed so that the most
interesting land use and urban design variables could be tested to determine
their influence on travel behavior. Data analysis was carried out by
Cambridge Systematics, including preparation of this report. Primary
contributors included Arlee Reno, Susan Moses, John Suhrbier, Eric Paquette,
Anne Martin, Krista Rhoades, and Yoram Shiftan. An initial phase of the
project was performed by Sam Seskin.

The data collection design portion of the work was performed by Elizabeth
Deakin of the University of California at Berkeley and the firm of Deakin,
Harvey, Skabardonis, with advice from other team members and FHWA. The actual
data collection was conducted by graduate students from UCLA, under the
supervision of Ms. Deakin. Site information, a comprehensive set of
Regulation XV data, advice on site selection, and review of the technical
results was provided by Professor Genevieve Giuliano of the University of
Southern California's School of Urban and Regional Planning, serving as a
consultant to Cambridge Systematics, Inc. Data was gathered for the
employment site itself, for its immediate surroundings, and for the general
environs. All project work was carried out under the overall direction of
Fred Ducca of the Federal Highway Administration.

Copies of the compiled "Los Angeles Land Use/TDM" dataset, Accession No. PB95-
500427 can be obtained either from the U.S. Department of Commerce, National
Technical Information Service, Springfield, VA 22161; or by contacting:

Cambridge Systematics, Inc.
150 CambridgePark Drive
Suite 4000
Cambridge, MA 02140
Telephone: (617) 354-0167
Fax: (617) 354-1542
Table of Contents

Preface vii

1.0 Introduction 1-1
1.1 Overview 1-1
1.2 Background 1-2

2.0 Approach 2-1
2.1 Overview 2-1
2.2 Selection of Variables to be Analyzed 2-1
2.3 The Southern California Regulation XV Database 2-4
2.4 Survey Instrument Design 2-6
2.5 Sample of Employment Sites Surveyed 2-9
2.6 On-Site Data Collection 2-11
2.7 Data Quality Assessment 2-11
2.8 Data Analysis 2-12

3.0 Findings 3-1
3.1 Change in Mode Share After TDM Implementation 3-1
3.2 Impacts of Transportation Demand Management Strategies 3-3
3.3 Impacts of Land Use and Urban Design Characteristics
on Work Trip Mode Choice 3-5
3.4 Summary of Impacts of Land Use and Urban Design
Characteristics 3-21
3.5 Land Use and Urban Design Characteristics of Sites with a
High Walk or Bicycle Mode Share 3-23
3.6 Sites with Low Single-Occupant Vehicle Use 3-25

4.0 Conclusions 4-1

Appendix A
Site Survey Data Collection Form

Appendix B
Procedures and Instructions Provided to Data Collection Staff
List of Tables2.1 Data Elements # General Environs of the Site 2-7

2.2 Data Elements # Site Area (One-quarter mile radius) 2-8

2.3 Data Elements # Work Place Characteristics 2-10

2.4 Distribution of AVR, Location, and Industry Characteristics 2-13

2.5 Distribution of Employer Size and Trip Reduction Incentives 2-14

2.6 Distribution of Transportation Mode Shares at Implementation
and One Year Later 2-15

2.7 Composite Land Use/Urban Design Variables 2-22

3.1 Change in Work Trip Mode Share 3-2

3.2 Categories of TDM Strategies 3-2

3.3 Impact of Financial Incentives on Mode Share 3-4

3.4 Impact of TDM Strategies on Drive Alone Mode Share 1989-1991 3-4

3.5 Impact of TDM Strategies on Ridesharing 1989-1991 3-6

3.6 Land Use and Urban Design Feature: Land Use Mix 3-8

3.7 Land Use and Urban Design Feature: Availability of Convenience
Services 3-12

3.8 Land Use and Urban Design Feature: Availability of Convenience
Services 3-14

3.9 Land Use and Urban Design Feature: Accessibility 3-16

3.10 Land Use and Urban Design Feature: Perception of Safety 3-18

3.11 Land Use and Urban Design Feature: Aesthetics of Area 3-20

3.12 Comparison of Drive Alone Shares Among Sites with Financial
Incentives and Alternative Land Use Characteristics 3-22

3.13 Characteristics of Sites with a High Walk or
Bicycle Mode Share 3-24

3.14 Sites Having a Low Percentage of Single-Occupant Vehicle
Commuters 3-25
List of Figures
2.1 Overall Approach to Data Collection and Analysis 2-2

2.2 Factors Influencing Work Trip Mode Choice 2-3

2.3 Analysis of Data 2-17

2.4 Principal Components Analysis 2-20

1.0 Introduction

1.1 Overview
There is considerable current interest in the effects of urban design and land
use characteristics on the transportation choices made by commuters. The
underlying assumption is that these employment site characteristics have an
important influence on a person's willingness to commute by transit,
ridesharing, bicycling, or walking - modes other than driving alone. Further,
the selection of transportation demand management (TDM) strategies that an
employer may choose to implement should be a function of surrounding site
characteristics, and that the combination of site characteristics and TDM
strategies can have a positive interactive effect in influencing an employee's
choice of commute travel mode. While the effectiveness of travel demand
management strategies, implemented both individually and in combination, has
been investigated, relatively little empirical work has been done to evaluate
the interactive effects of land use and TDM strategies on commuting behavior.

For this project, an integrated database of land use characteristics and
travel demand management strategies was developed for a sample of specific
employment locations in the Los Angeles urban area. The integrated database
was constructed by adding land use and site information, developed through
field observation, to the "Regulation XV" dataset of the South Coast Air
Quality Management District (SCAQMD). The SCAQMD dataset includes information
about aggregate employee travel characteristics, and the incentive programs
offered by employers. This integrated database was then analyzed to explore
the interactions that may exist between travel demand management programs,
land use, urban design characteristics, and employee mode of travel. The
primary objective was to develop conclusions about the combined impacts of
land use and travel demand management strategies on employee travel behavior.

Information was collected regarding the land use and urban design
characteristics of a work site, the set of transportation incentives provided
to the employees by the employer at that site, and the mode of travel by
employees both before and after implementation of the transportation
incentives for the trip between home and work. Data were collected and
analyzed for individual employment sites. Data were not available regarding
the characteristics or travel behavior of individual employees at a given work
site, only the aggregate distribution of modal shares. Information was not
available in the dataset about the residential end of the work trip.
Similarly, data on midday travel, trip chaining, or other related topics were
not available.

The second section of this report describes the methodological approach
utilized, including the specific data collection and analysis procedures.
Findings of the statistical analyses are presented in the third section. The
effects of various travel demand management strategies were examined both
individually and in combination with land use characteristics. General
conclusions are presented in the final section. The overall finding is that
an interaction effect does indeed exist. The effectiveness of programs of
travel demand management measures is increased at those locations where
supportive land use and urban design characteristics also exist.

The results presented here represent an initial or preliminary analysis of an
extensive dataset; considerable additional analysis is possible and is
encouraged. The integrated employment site, land use/transportation database
represents a valuable product by itself. Previously existing datasets do not
include descriptions of both land use and travel demand management programs
for individual employment sites.

1.2 Background
An examination of trends in travel behavior shows increases in the number of
workers per household, licensed drivers per household, vehicles per household,
vehicle trips, and average trip length. Both employment and housing are
growing at faster rates in suburban than at central city locations, with the
majority of employment growth now occurring at suburban employment centers.
Auto occupancies are declining and the percentage of single-occupant commuting
is increasing. Overall, vehicle miles of travel, in most areas, are growing
at a higher rate than either employment or population. The result of these
trends is an increasing level of traffic congestion in urban and suburban
areas across the country, particularly during peak commute periods.
Congestion often exists in newly developing suburbs and semi-rural areas, as
well as in central business districts and older residential neighborhoods.

A response to these increasing levels of congestion has been a desire to
increase the effectiveness with which existing transportation resources are
utilized. While this response has included both the expansion of highway
capacity and increasingly sophisticated systems of traffic engineering and
control, a variety of transportation demand management measures also have been
implemented throughout the country. TDM programs typically include a variety
of employer provided incentives aimed at inducing commuters to rideshare, use
public transportation, walk or bicycle to work. TDM incentives include
ridesharing and transit subsidies, preferential parking for rideshare
vehicles, rideshare matching services, facilities for bicyclists (e.g., bike
storage, showers and lockers), award programs, and a variety of other
miscellaneous strategies.

Over the past two years, experience with transportation demand management
(TDM) measures has grown rapidly. For example, it has been found that transit
works best in areas with moderate to high densities at both the work and the
home end of the trip; ridesharing is most effective for long trips destined
for large centers where "matches" can be easily found.

The importance of land use policies and patterns as determinants of travel
conditions and choices has increasingly been hypothesized in the last few
years, and efforts to avoid and manage traffic problems through more forceful
land use planning have received attention. It is believed that the
effectiveness of a particular transportation demand measure depends, in large
part, on its application in a supportive environment. For example, commute
alternative programs are most successful in settings having a mix of uses to
which midday trips easily can be made on foot. Where supportive conditions
are absent, or when conditions tend to cancel out (as when ridesharing is
"encouraged" but convenient free parking is guaranteed), efforts at demand
management generally are far less successful.

Project-level planning and zoning controls are one way of affecting
transportation through the land development process; measures include
exactions and impact fees to help pay for transportation improvements, and
conditional approvals requiring on-site traffic mitigation programs.

With respect to urban design, a better matching of transportation to the
specific land uses it serves often is advocated. Design objectives aim for
environments in which buildings are clustered and uses are mixed; a balance of
housing, jobs, and services is available; and the streets, sidewalks, transit
stops, and bike facilities are designed to be transit-oriented and pedestrian-
friendly. At the regional or large-area level, policies such as urban limit
lines, infill incentives, level of service requirements, and control of
activities at the rural fringe are considered as means of channeling growth to
areas best able to accommodate it, pacing growth to reflect infrastructure
availability, and providing incentives to manage travel demand, simultaneously
protecting valuable farmland and open space.

An important issue that has arisen is how best to design transportation
strategies so that these transportation actions are fully reflective of an
area's land use characteristics, and therefore have maximum impact in
controlling congestion. Analogously, the urban design characteristics of an
employment site can be adjusted so that they are more fully supportive of the
particular transportation demand management strategies to be utilized at that
particular location than otherwise might be the case. It is argued that by
matching carefully conceived packages of transportation demand management
programs with land use plans and actions, greater transportation benefits
should accrue, such as reduced congestion and improved air quality. Since
relatively little work has been done that explicitly sorts out the respective
contributions of transportation measures and land use controls, or that
identifies and quantifies the synergistic effects of joint land use and
transportation actions, this project was undertaken to develop this specific
kind of quantitative information.

The underlying hypothesis of this project, therefore, is that land use and
urban design characteristics of work sites affect employee work trip mode
choices. Furthermore, these land use and urban design characteristics may
interact with various employer-based transportation demand management (TDM)
strategies to alter commuter work trip mode choice. That is, similar TDM's
may cause different changes in mode choice as a result of the mix of land use
and urban design characteristics present at different work sites. In order to
test these hypotheses, it is necessary to define and quantify the specific
aspects of land use and urban design which may actually influence mode choice.

It is further hypothesized that the general quality, ambiance, or environment
of a work site, rather than any individual characteristic, helps determine
mode choice. It is not necessarily an individual land use or urban design
characteristic that influences mode choice, rather it is their combination
which create a work site environment. For instance, the presence of a vacant
lot may not by itself cause a person to choose a different work trip mode of
travel. However, that vacant lot may influence a commuter's perception of the
overall safety of the work site, with mode choice being affected more by
overall safety considerations than by any one particular land use
characteristic. In this example, the commuter's perceived safety is an
influential site characteristic in determining mode choice, while the presence
of a vacant lot (in combination with other individual land use and urban
design characteristics) influences the perception of safety. Acknowledgement
of this relationship is essential to defining the factors which contribute to
mode choice.
2.0 Approach

2.1 Overview
The underlying objective of the analytical portion of this project was to
statistically analyze employee commuter behavior at individual employment
sites to determine any differences that may exist that are a function of both
the particular transportation demand management incentives and the particular
land use and urban design characteristics that exist at that site. As
indicated in Figure 2.1, the first step involved the definition of the
particular variables to be tested.

Assembly of a dataset containing information on the desired variables then
constituted the second, and difficult, step of the analysis process. It was
determined that no existing dataset contained the desired information on both
TDM and land use. Consequently, it eventually was decided to build upon the
results of earlier analyses of the data developed for the Southern California
Regulation XV Employee Trip Reduction Ordinance that had been conducted by
Giuliano, Hwang, and Wachs, (1992), (Begin Footnote: Giuliano G., Hwang K.,
and Wachs M., "Employee Trip Reduction in Southern California: First Year
Results," June 1992 End Footnote) which are primarily oriented to travel
demand management (TDM) variables. For a subset of the sites contained in the
dataset developed by Giuliano et al., detailed land use and urban design data
were collected. The data on urban design and land use were then analyzed in
conjunction with the transportation demand management measures that existed at
that employment site to determine their respective influence on the choice of
travel mode for commuting. The site specific land use/TDM dataset is far more
detailed than any data on employment sites and commuter travel patterns
previously developed or evaluated. The following subsections describe the
analytical methodologies utilized, addressing each of the steps identified in
Figure 2.1.

2.2 Selection of Variables to be Analyzed
Work trip mode choice is affected by many factors. Giuliano, Hwang, and Wachs
(1992), in their analyses of the Southern California Regulation XV data,
classified the factors known to influence commuters' choice of modes into
three broad categories: employee characteristics, intraorganizational
characteristics of the work place, and environmental factors (Figure 2.2).
Employee characteristics were defined as including factors such as workers'
commute distances, incomes, and levels of automobile ownership. Similarly,
organizational characteristics defined include factors such as the need in
certain industries for all workers to be at work simultaneously, and the
degree to which certain jobs require traveling (i.e., sales).

Environmental factors, as defined by Giuliano et al., include transportation
supply characteristics, such as the availability of alternative modes and mode
attributes (e.g., travel time and cost, parking availability, and congestion);
factors that have been shown through prior research to have a strong influence
upon commuters' choice of modes. The approach in this project was to build
upon the Giuliano et al classification by expanding upon the definition of
'environmental factors' to also incorporate a range of land use and urban
design variables.

Land use encompasses factors relating to the spatial pattern of urban
development, as well as the distribution of different activities within an
urban area. Land use factors hypothesized to influence commuters' choice of
modes included accessibility to services, particularly the mix and intensity
of services within walking distance of the work place, and employment density.
It was hypothesized that increasing the mix and intensity of services within a
convenient walking distance encourages the use of alternative modes (e.g.,
transit, ridesharing, biking, walking) by increasing the feasibility and
desirability of making midday trips without using a personal vehicle.
Although employment density was not hypothesized to affect commuters' choice
of modes directly, it was included because of its strong relationship with the
level of transportation services provided, the range and intensity of other
services provided, and urban form characteristics.

Urban design relates to the physical characteristics of specific sites. These
characteristics include architecture, streetscape, and site layout, and can
influence the way people perceive urban environments and, in turn, the way
urban environments make people feel. Urban design features hypothesized to
affect commuters' choice of travel modes included characteristics that would
enhance the aesthetic appearance of the work place, and those features
contributing to feelings of comfort and safety. It was hypothesized that
these factors encourage the use of alternative modes for the work trip in two
ways: 1) by increasing the desirability of using an alternative mode for the
work trip itself; and 2) by increasing the desirability of fulfilling midday
trip needs without the use of a personal vehicle.

2.3 The Southern California Regulation XV Database
Extensive data on travel demand management measures and employee travel to
work sites had been collected in association with the South Coast Air Quality
Management District's (SCAQMD) Regulation XV Trip Reduction program. After
examining the use of other potential datasets and data collection
opportunities, a decision was made to focus data collection efforts upon
augmenting the existing Southern California dataset with the desired
information on land use and urban design characteristics. (Begin Footnote:
The SCAQMD Regulation XV dataset included travel demand management variables
and employee travel behavior, but no land use variables at the employment
site. Thus, additional data collection was necessary to add site-specific
land use variables.

An alternative dataset considered was one that included land use variables and
employee travel behavior information at a national sample of employment sites
having over five million square feet of existing floor space. Developed by
JHK and Associates, Inc. as part of NCHRP Project 3-38(2), "Travel
Characteristics of Large-Scale Suburban Activity Centers," this dataset had
the disadvantage of lacking information on the specific travel demand
management information that existed at each employment site.

An exploration of the feasibility of adding travel demand management variables
to the NCHRP dataset indicated that it was very difficult for people to either
remember or estimate what kind of travel demand management actions or programs
had been in effect at the time the data on travel behavior was collected. For
the Regulation XV database, not only was it judged to be easier to assume that
overall site variables had not changed since the travel data was collected,
but the travel data was more recent and included different years of travel
behavior and travel demand program information for each site. Since the South
Coast data promised more opportunities and fewer difficulties, a decision was
made by FHWA and the project team that available resources would be best
devoted to enhancement of the South Coast dataset by adding the land use
variables to the already existing travel demand management and travel behavior
information. End Footnote)

Under the SCAQMD's Regulation XV program, employers having 100 or more
employees at any work site within the Los Angeles Metropolitan area are
required to develop and implement a trip reduction program aimed at obtaining
a specified Average Vehicle Ridership (AVR) target. AVR is defined as the
ratio of the number of employees arriving between 6:00 a.m. and 10:00 a.m., to
the number of motor vehicles used by the employees. Regulation XV does not
require employers to meet their AVR targets, but employers can be fined for
not implementing the trip reduction programs.

The Regulation XV trip reduction ordinance was enacted by the South Coast Air
Quality Management District in 1988 with implementation phased in over a
period of years. Employers having 500 or more employees were to submit plans
during the first year. Participation by employers of 200 or more was
scheduled for 1989, and employers of 100 or more were phased in during 1990
and later years. In January 1994, the SCAQMD decided in the form of Rule 1501
to undertake a review of implementation experience with Regulation XV. The
intent of the review is to identify options that could both reduce
implementation costs to individual employers and increase effectiveness in
terms of the magnitude of emission reductions that are being achieved.

As a part of the Regulation XV program, information was collected for each
work site. Data included work site characteristics, time series data
describing initial and follow-up trip reduction measures implemented, mode
share, and AVR information. Data pertaining to individual employees, however,
was not included. The resulting information was incorporated into a database
as part of the investigation completed by Giuliano, Hwang, and Wachs [1992] of
the results of the first year of the Regulation XV program. Two basic
criteria were used for inclusion of a work site in the database: 1) the site
had to have submitted and received approval for first and second year trip
reduction plans as of August 1991; and 2) data for the site had to pass
certain tests for logic and consistency. The resulting dataset covered 1,110
work sites, or 27 percent of the 4,032 total sites that had received approval
for their plans as of August 1991.

Many employers and especially larger employers in Los Angeles County, already
had implemented a variety of TDM measures prior to enactment of the
Regulation XV ordinance by the SCAQMD. Since all TDMs are not, therefore, a
direct result of Regulation XV, the estimates of effectiveness derived from
the SCAQMD database will be an underestimate, all else being equal, of actual
impacts if true before and after data were available.

2.4 Survey Instrument Design
The primary aim in designing the survey instrument was to identify urban
design and land use variables that would provide a comprehensive picture of
the work site and its environs. A secondary aim, though, was to supplement
some of the data already contained in the Regulation XV dataset, particularly
information on parking and transit accessibility. Given the desire to test
hypotheses about the relationship between urban design variables and work trip
mode share, a particular emphasis in designing the survey instrument was
placed upon identifying variables that would describe the 'friendliness' of
the area to specific modes, particularly walking, bicycling, transit, and car
and van pools. Another emphasis was upon defining measures of accessibility
to services, including both the mix and intensity of services within walking
distance of the site.

To meet these data collection objectives, a survey instrument was designed to
gather information at three distinct levels. The first level of information
focused upon the general environs of each work site ranging from one-half
square mile to two square miles from the work site area. Data collection at
this level was focused upon understanding the 'friendliness' of the area to
specific commute alternatives. Specific data elements defined for collection
at this level were grouped into two categories: land use and street
characteristics, as shown in Table 2.1.

The second level of data collection involved the area within one-quarter mile
of the work place. Information gathered at this more detailed level was
designed to provide an understanding of the feasibility and desirability of
satisfying midday trip needs by walking. Data items collected described the
accessibility of the site to services and the quality of the travel paths
between the work place and services. Although some information collected at
this level covered topics similar to those addressed within the site environs,
such as land use and street characteristics, the information collected at this
level was much more detailed. Examples of specific data elements collected at
this level are shown in Table 2.2.

The third scale of data collection was the work site itself. Data gathered at
this level focused upon understanding the general ambiance of the site and its
immediate surroundings. The intent of work-site data collection was to
understand the environment that employees would encounter both in arriving at
a particular work site and in choosing to spend their lunch hour within the
immediate vicinity of the site. A summary of the data items collected for the
work site is shown in Table 2.3.

A copy of the final survey instrument used for field data collection is
provided as Appendix A to this report. Given the volume and complexity of the
data to be collected, an extensive instruction sheet was prepared for use by
field staff along with the survey instrument. A briefing also was held to
instruct field staff. Both the instructions and survey form were refined
based upon a pretest that was conducted at several sites. A copy of the the
instructions provided to the field staff is provided as Appendix B.

2.5 Sample of Employment Sites Surveyed
The full Regulation XV database covers the entire South Coast AQMD regulatory
area, a 13,000-square mile region encompassing Los Angeles County, Riverside
County, Orange County and the non-desert portions of San Bernardino County.
Given the extent of urban design data desired as well as the need for on-site
data collection, collecting on-site data from firms across such a large area
would have been prohibitively time consuming. Consequently, it was necessary
to focus the data collection efforts on a sample of the 1,110 sites in the
Regulation XV database. To determine the desired sampling approach, the 1,110
sites were examined to determine the range in those factors thought to most
strongly affect urban form: location, land use, scale, and employment
density.

The 1,110 sites were grouped by zip code and mapped by location to determine
the extent of geographic clustering. After the sites were mapped, the
clusters and isolated sites were analyzed to determine representation along
the other factors, including types of land uses, scale, and employment
density.

The majority of the 1,110 sites, approximately 68 percent, contained in the
Regulation XV database are located within Los Angeles County. Twenty-one
percent of the sites are located within Orange County, and the remaining sites
are fairly evenly divided between San Bernardino and Riverside Counties.

Mapping of sites illustrated that there was substantial geographic clustering.
Major clustering occurred within downtown Los Angeles, Hollywood, and in the
major employment centers in the corridors stretching from the downtown Los
Angeles area to western Los Angeles, to Santa Monica, and to the Los Angeles
International Airport. Other major clusters were found in Orange County,
within the vicinity of Santa Ana and the Orange County Airport.

Given both the distribution of sites geographically and the logistical
implications of on-site data collection, the decision was made to focus the
data collection efforts on the sites within Los Angeles County. Based upon
the mapping of sites and the analysis of their representation of the range in
land uses, scale and employment densities, a sample of 330 work sites was
selected from the 761 work sites within Los Angeles County.

2.6 On-Site Data Collection
For the 330 selected employment sites, data collection was completed from
March through July 1993. Data were collected by graduate students in urban
planning, urban design, and related fields from UCLA. Since some of the data
to be collected, such as pedestrian activity or traffic levels, were items
that varied by time of day and day of the week, surveyors were instructed to
observe this type of information during standard weekday business hours. This
ensured that observations made were consistent with the environment that an
employee would encounter during the standard work day. Data collection
included completion of the survey form, diagrams, maps, and photographs of
each site.

2.7 Data Quality Assessment
Three different means were utilized to evaluate the quality of the collected
data. Coding of the data provided the opportunity to examine maps of the
sites along with the data and surveyor comments provided on the survey form.
Through the coding process, a small number of data elements were identified
where there were clearly differences in interpretation among field staff or
simply a lack of understanding of the information desired. Missing values for
certain data elements also provided clues as to information that was not
understood by the surveyors.

A second opportunity for assessing data quality was provided by independently
sending two different teams of data collectors to collect the same information
at five of the selected sites. With the exception of certain data elements
that could be expected to vary with time and day of observation, such as level
of traffic or pedestrian activity, the data collected by one team of data
collectors under an 'ideal' scenario would have exactly replicated the results
of the other. Although the size of the subsample, five, was too small to
gauge data quality with any statistical significance, it did help identify
particular data elements that could be of potential concern.

The final test of data quality was a series of logic tests undertaken to check
specific variables, and the relationships between variables, for
reasonableness and consistency. For example, tests were completed to
determine the number of sites identified both as having several different land
uses within the vicinity and as 'single use.' These types of tests, checking
for multiple conditions that would not reasonably be expected to occur
simultaneously, also helped to assess the quality of the data that had been
collected.

Based upon these assessments of data quality, a small number of variables were
identified as being of potential concern. Even though some variables were
thought to be reasonably objective to measure, there were some cases where it
was clear data collectors did not understand the particularly information
desired. Examples of this first type of problem included items such as floor-
to-area ratio (FAR), block and parcel size, number of bus lines, street type,
and street set-back. For example, FAR values were missing in over 98 percent
of the site observations.

Some data elements proved to be problematic simply because they proved to be
more subjective to measure than originally expected. Among this group of
variables were items such as land use grain, and presence of trees. Although
the difficulty with some of these data elements, such as the presence of
trees, was unexpected, it is possible that different data collectors set
different thresholds for the frequency with which a characteristic needed to
occur (e.g., number of trees) before they would indicate a characteristic was
present at the site.

Certain data elements also proved to be more difficult to measure than
anticipated, even though they are not inherently subjective or abstract
concepts. Examples include the presence of certain services within walking
distance of a site. Follow-up visits to sites indicated that, in some
situations, landscaping obscured particular land uses. To thoroughly observe
the presence of all services within walking distance in large suburban
locations having extensive landscaping proved to require an extensive amount
of time to completely walk within one-quarter mile of each building and site
and examine the specific services offered. A quick review of signage and
streetwall frontage proved insufficient for such site locations.

2.8 Data Analysis
Distribution of Site Characteristics
The first part of the data analysis involved an examination of the employer
and transportation characteristics of the subsample of 330 sites used in the
urban design survey. As tabulated in Tables 2.4, 2.5, and 2.6, the
characteristics examined include the degree of variation in AVR, site
attributes, trip reduction incentives, mode share, and change in mode share
between implementation and one year following implementation. Table 2.4
identifies the percentage of sites by AVR, location, industry type, and size
classifications. Table 2.5 tabulates the mix of trip reduction incentives
utilized. Table 2.6 examines and compares AVR and mode share.

These tables also present a comparison of the 330 and 1,110 site databases,
noting statistical significance (at the 0.95 level), to provide an indication
of the difference between the 330-site subsample and the larger 1,110-site
sample. However, just as the 1,110 work sites do not necessarily reflect a
random sample of the employment sites subject to the Regulation XV ordinance,
the 330 subsample was not intended to be fully representative of the
characteristics of the full 1,100-site database. While there are differences
between the two datasets, as noted below, the overall conclusion is that the
330 sample is approximately representative of the SCAQMD database as a whole.

As visible from Table 2.4, the 1.3 AVR target was under represented in terms
of statistical significance and the 1.5 AVR target was statistically over-
represented within the 330-site subsample. The AVR targets are based upon
geographic location. The AQMD assigned low-density areas an AVR target of
1.3, while developed urban and suburban areas are targeted for 1.5 AVR and the
Los Angeles CBD has an AVR target of 1.75. Several geographic areas within
the Regulation XV Los Angeles metropolitan area differ in their
representativeness within the two datasets. As discussed in Section 2.5,
Sampling, the data collection effort was limited to locations within Los
Angeles County. Therefore, San Bernardino, Riverside, and Orange Counties are
not represented in the 330-site subsample, and sites including north,
southwest, east, and remote west Los Angeles County and Long Beach are over-
represented compared to the larger 1,110-site sample. As indicated in
Table 2.4, manufacturing was the only type of industry with a significant
variation between the two samples. Both the size of the work site, based upon
number of employees, and trip reduction incentives for the 330-site subsample
were representative of the 1,110-site sample.

The comparison between the 1,110-site and the 330-site subsample presented in
Table 2.6 identifies statistically significant differences for bus share at
implementation of plan, and for carpool share and bus share one year later.
Based upon the calculated means for mode share, the bus share is over-
represented in the 330-site subsample for both implementation years. This
difference presumably is due to the restriction of the data collection to Los
Angeles County. Areas such as San Bernardino and Riverside Counties, that are
contained in the larger 1,110-site sample, do not have comparable bus systems
to those existing in Los Angeles County.

Overview of Analysis Methodology
The land use, urban design, and transportation incentive data collected for
the 330 employment sites were analyzed using a combination of Principal
Components analysis and standard analysis of variance statistical techniques.
As shown in Figure 2.3, the first step in the analysis involved a refinement
of the hypotheses concerning the effects of land use and urban design
characteristics and TDM incentives on the commute trip mode choice. Principal
Components analysis was then used to identify groups of land use variables
having similar impacts. The intent was to derive a small number of composite
variables that could be used to capture the overall characteristics of a site.

Standard analysis of variance techniques were used next to examine the effects
of groups of land use and urban design characteristics and trip reduction
incentives both individually and in combination. The findings presented in
Section 3.0 focus on this phase of the data analysis. Finally, groups of
employment sites having certain transportation characteristics in common were
analyzed to determine if these sites also shared any land use or urban design
characteristics. It is in this context that the subset of sites having a high
percentage of people that commute by either walking or bicycling are examined
as a part of the Section 3.0 findings.

Identification of Composite Land Use and Urban Design Variables
It was hypothesized that the interactions of individual site characteristics
could be very significant. For instance, the presence of a sidewalk and the
level of area traffic each may have some influence in measuring the
accessibility of services. A sidewalk may enhance accessibility slightly,
while increased traffic may inhibit accessibility slightly. However, an area
which combines high traffic and no sidewalk may have much lower accessibility
than would be expected given that each individual influence is slight. Thus,
an analysis methodology is required that allows for the many potential
interactions of a site's individual land use and urban design variables to be
combined into a meaningful estimate of site characteristics.

The method of Principal Components was used to create composite variables.
(Begin Footnote: Morrison, D.F., Multivariate Statistical Methods, 2nd
Edition, New York; McGraw-Hill Book Company, 1976. End Footnote) (Begin
Footnote: An alternative methodology considered was to test the individual
land use and urban design characteristic variables using standard analysis of
variance techniques. One could determine which variables had the greater
influence on mode share and the change in mode share after the implementation
of TDM's. These variables could then be called proxies for the groups of site
qualities. For instance, if sites that had vacant lots present nearby
experienced a lower than average shift from driving alone to transit use, one
might infer that the presence of vacant lots signifies an unsafe area. It
seems reasonable that commuters would be less likely to shift from driving
alone if their safety is in jeopardy. Similarly, variables that would be
proxies of the other site characteristics could be tested.

It was concluded that there were serious difficulties with this technique
because of the large number of variables involved. With this many variables,
it would be extremely difficult to test all potential interactions. If only
five variables are tested as proxies for each of five possible site qualities,
there are more than 30 combinations of the variables and their interactions
which would have to be tested for each quality. There are also problems of
severe multicollinearity (correlation among supposed independent variables)
when testing high level interactions which make selection of the appropriate
variables an unwieldy task. Another problem is the high number of variables
compared to the relatively small number of observations.

Another methodological option considered was to develop an index of each of
the site qualities based on individual characteristics. Creating an index a
priori to measure the presence of groups of these features at a given site is
not feasible for two reasons. First, to do so requires assumptions about the
relationships between the individual characteristics and the composite feature
being examined. Second, there is potentially a high degree of correlation
among the individual characteristics. Each poses serious difficulties to
accurate identification of influential factors. End Footnote) Briefly,
Principal Components analysis can reduce a large number of variables into a
smaller set of uncorrelated composite variables. The Principal Components
method creates composite variables, called principal components, which are
orthogonal linear combinations of the initial variables. For p different
initial variables, a total of p different principal components can be formed.
The linear combinations are formed sequentially to explain as much of the
variability in the initial variables as possible. The first principal
component explains the most variation in the initial variables; the second
principal component explains the second most variation; etc. Because there is
correlation among the initial variables, this allows the bulk of the variation
to be explained in relatively few principal components.

Principal Components analysis is perhaps most easily understood geometrically.
Figure 2.4 shows a cluster of points for n observations of variables x and y.
The first principal component explains the most variation, so it will be the
linear combination that spans the line shown by P1. Each of the n
observations will receive a principal component score, p1i where i is one of
the n observations, that is a linear combination of the values of each of the
n values xi, yi.

The value of the first principal component, P1i, represents a point on the
line P1. Similarly, the second principal component (in this case the only
other possible) explains the second most variation in the cluster, so it will
be the result of the linear combination that spans the line shown by P2. The
value of the second principal component, p2i, will represent a place on the
line P2. Finally, the principal components by definition are orthogonal
(perpendicular), and thus they will be perfectly uncorrelated. This
eliminates any potential problems with multicollinearity.

While this interpretation is applicable for any number of variables, it is
most easily visualized in the two-variable instance. If there are three
initial variables, there is a three-dimensional cluster of points which would
be spanned by three orthogonal principal components. While Principal
Components analysis cannot be visualized beyond the three-variable case, the
same properties hold true.

Because a principal component is a linear combination of the individual
variables, it can easily be seen how each variable influences the principal
component by looking at each individual coefficient. In the individual
example shown in Figure 2.4, P1 = -.707x + .707y. This means that as x
increases, the value of the first principal component decreases. The
principal component increases in value when the value of y increases. In this
case, the coefficients are equal so each variable has the same impact on P1.
In most cases, the coefficients would not be equal. The variable with the
larger coefficient has the greater impact on the value of the principal
component.

The particular site characteristics determined to influence commuter mode
choice are the perception of safety, the accessibility of services for walking
on midday trips, the availability of convenience services, the mix of
surrounding land uses, and the aesthetics of the area surrounding the work
site. Each of these site characteristics results from a mix of the individual
land use and urban design characteristics of the area, contained in the data
base, and also some unmeasured characteristics. It is possible to estimate
the general site characteristics by combining individual land use and urban
design characteristics. For example, the augmented Regulations XV data
includes variables to help capture the "friendliness" of the area to various
commute alternatives (i.e., variables that measure land use and urban design
characteristics within two miles of the work site), the ease of satisfying
midday trip needs by walking (i.e., variables that measure land use and urban
design characteristics within one-quarter mile of each work site), and the
general ambiance of the site and its immediate surroundings (i.e., variables
that measure the land use and urban design characteristics of the work site
itself). By combining land use and urban design variables in meaningful ways,
quantifiable variables of the more general site characteristics, such as
safety, can be established. These composite variables can then be used to
quantify the influence of the various site characteristics on commuter mode
choice. Furthermore, the interaction of land use and urban design variables
with TDM incentives can be tested to determine additional impacts of
implementing TDM's in particular environments.

Table 2.7 lists the individual land use and urban design characteristics from
which the principal components were developed. These reduced variable groups
were compiled based on a combination of statistical analysis and expert
opinion. They were the variables deemed most important to each of the
principal components. Diagnostic measures were performed on the independent
component variables to test for the potential problems of heteroscedasticity
and multicollinearity. These tests indicated that the data were within
acceptable parameters, and that remedial measures were not warranted.

The mix of land uses was represented by a principal component of variables
which indicated the presence of residential, office, retail, and personal
services within one-quarter mile of the work site. As the variety of these
land uses increases, the "mix" of land uses increases.

The presence of convenience services was measured by a principal component of
variables that indicated the presence of restaurants, banks, child care
centers, dry cleaning, drug stores, and post offices within one-quarter mile
of the work site. As the number of these services increase, the value of the
component measure increases.

The accessibility of services was measured by a principal component of
variables indicating the presence of four or more services within walking
distance, sidewalks, transit service, and the level of traffic around the
site. As each of these variables increases, the value of the principal
component increases.

The perception of safety was measured by a principal component of variables
which indicate the presence of street lighting, vacant lots, and sidewalks,
and the level of pedestrian activity in the area. Street lighting, sidewalks
and increased pedestrian activity raise the value of the principal component
and the presence of vacant lots reduces it.

The aesthetic level of each site was measured by a principal component of
variables indicating the presence of trees and shrubs, wide sidewalks and
graffiti around the site. Trees and shrubs and wide sidewalks enhance the
level of aesthetics, while graffiti diminishes aesthetic appeal.

Determining Effectiveness of Commuting Behavior
With principal components developed, empirical testing of their impacts on
mode choice was possible. Standard analysis of variance techniques were used
to determine if the prevalence of the different principal components
corresponded to different mode shares. (Begin Footnote: Neter, John,
Wasserman, William and Kutner, Michael H., Applied Linear Statistical Models:
Regression, Analysis of Variance and Experimental Design, 3rd Edition, Boston:
Richard D. Irwin, Inc., 1990. End Footnote) For example, the drive alone
share was compared in areas with principal components that indicate high
safety with those that were not as safe. If the drive alone shares were found
to be different by an amount that was statistically significant at a .95 level
of confidence, then safety was considered a significant influence on the drive
alone share. (Begin Footnote: The determination of statistical significance
used throughout the analyses is based on a difference of means t-test having a
95 percent level of confidence. The difference in the means of two values is
compared taking into account the standard errors in their respective values,
where the standard error is an indication of the precision of the mean.
Imprecise estimates of mean will have relatively large standard errors,
whereas precise estimates will have relatively small standard errors. By
comparing the standard errors with the difference between the means, one can
determine the likelihood of the difference being the result of random
imprecision of the estimated means. Testing with 95 percent confidence
implies that the likelihood is less than five percent that the difference
between two values is simply due to random error.

The test is applied by comparing the difference in the two means to the sum of
their respective standard errors. If the sum of the standard errors is larger
than the difference between the means, the difference in the means may be due
to random errors in estimation. If the size of the difference is larger, then
it is unlikely (less than five percent likely in this case) that the
difference is due to random error, and is instead an indication that with a 95
percent level of confidence that real differences exist between the two
factors being compared. End Footnote)

Similar analysis of variance techniques were employed to measure the impact of
the TDM incentives. Mode shares were compared in areas with and without
different types of transportation incentives. TDM incentives were categorized
into one of four groups: financial incentives; assistance programs; flexible
work hours; or awards programs. Significant differences in the mode share
were attributed to the impact of a TDM. Measures which did not cause a
statistically-significant shift in mode share were considered unsuccessful.
Trip reduction incentives were analyzed both individually and in combination.

Finally, the interaction affects of land use characteristics and TDM
incentives were evaluated. An interaction effect captures impacts that may
not have been found by combining the land use with the TDM impacts in a simple
additive manner. Mode shares in areas having both particular land use
features and TDM incentives were compared to their expected levels given the
features and incentives taken individually. Statistically significant
differences were attributed to the interaction of the land use of urban design
features with the trip reduction program.
3.0 Findings
This section presents findings related to the impact of land use and urban
design characteristics when combined with transportation demand management
(TDM) strategies on work trip mode share. To understand this combined impact,
it is important first to identify the impact of transportation demand
management measures alone, as well as land use and urban design
characteristics alone. Section 3.1 identifies, for the entire sample, the
change in mode share between Year One of the analysis (i.e., pre-Regulation
XV) and Year Two of the analysis. Section 3.2 describes the impacts of
individual TDMs as well as groups of TDMs on mode share across the 330-site
dataset.

Hypotheses regarding how land use and urban design characteristics interact
with different categories of TDMs to affect work trip mode choice are
evaluated in Section 3.3, examining five different categories of land use and
urban design characteristics. Mode shares in areas that display particular
land use and urban design characteristics are compared to sites that do not
have these characteristics. In addition, changes in mode share between sites
that share the same land use characteristics but have adopted different types
of TDMs are identified. Shifts in mode share when TDMs are introduced in
areas that exhibit similar land use and urban design characteristics are
documented. These shifts are compared to changes in mode share in areas that
do not exhibit these characteristics. A summary of these findings is
presented in Section 3.4.

The chapter concludes with an examination of sites having two special
characteristics. Section 3.5 evaluates conditions at employment sites having
a higher than average walk or bicycle mode share in order to identify the land
use and urban design characteristics of these sites that may be unique.
Section 3.6 then examines sites having a low level of single-occupant vehicle
commuting.

3.1 Change in Mode Share After TDM Implementation
Table 3.1 displays the average change in work trip mode share between the base
year 1988, and the first year after implementation of TDMs for the 330-site
dataset. Over the period, the drive alone share decreased from 76.2 to 71.4
percent, an absolute change of 4.8 percent. This shift was more than
accounted for by an absolute increase of 5.4 percent for ridesharing, which
accounted for 13.4 percent of all work trips in the base year, and 18.8
percent in Year Two. In relative terms, ridesharing increased by 40.3
percent, a notable increase which can be attributed to the introduction of
TDMs.

Both transit and walk/bicycle mode shares decreased (-0.2 percent and -0.4
percent respectively) over the period. (These shifts were not significant at
the 95 percent confidence level.) It appears that commuters using these modes
switched to ridesharing. This shift may be accounted for by the fact that
there were many more categories of incentives offered exclusively for
carpooling (10) versus transit (2) and walk/bicycle (4). The rideshare
incentives also may have provided bigger financial rewards than those offered
for transit and walk/bicycle. (The database does not provide information
regarding the magnitude of individual financial incentives.) (Begin Footnote:
Analysis of variance tests indicate that seasonal changes (i.e., weather) do
not significantly affect the bike/walk or transit mode shares, or AVR. End
Footnote)

The average vehicle ridership (AVR) increased slightly from 1.22 in the base
year to 1.25 after employer implementation of the their TDM programs. This is
a relative increase of 2.5 percent.

3.2 Impacts of Transportation Demand Management Strategies
A total of 31 different transportation demand management strategies were
implemented by employers within the survey sample. As shown in Table 3.2,
these strategies can be grouped into five categories: financial incentives,
flexible work schedules, assistance programs, award programs, and other
strategies. Financial incentives include all TDM strategies that provides an
employee with a financial reward for participation. TDM strategies that
provide employees with opportunities to alter work schedules to avoid a five
day per week rush hour commute are classified under the category of flexible
work schedule. Strategies included in the category of assistance programs are
those that provide information regarding alternative modes, and help with ride
matching. Award programs include prize drawings and recognition in company
publications. The category of "other" includes three strategies that do not
logically fall under any of the first four categories. Financial
disincentives in the form of pricing measures were not implemented at any of
the employment sites, and thus are not listed in Table 3.2.

As a group, financial incentives were the only strategies that showed a
statistically significant impact on mode share. Table 3.3 displays the shift
in drive alone and in ridesharing when financial incentives were absent
compared to when they were present. At sites where financial incentives were
absent, the average drive alone share decreased by 1.7 percent in the period
after implementation of TDM strategies compared to the before period. In
contrast, the drive alone share decreased by 6.4 percent when financial
incentives were offered. The carpool share increased by 3.2 percent in the
absence of financial incentives, compared to 5.7 percent when financial
incentives were present. Financial incentives did not have a statistically
significant impact on the shift in transit or other modes.

Individual financial incentives that resulted in a statistically significant
shift from the drive alone share were bicycle subsidies, vanpool seat
subsidies, vanpool subsidies, transit subsidies and other employee benefits.
The differences in the drive alone share with and without these subsidies are
shown in Table 3.4. Two individual financial incentives, vanpool seat
subsidies and transit subsidies, are statistically significant in influencing
a shift to carpooling over the period (Table 3.5). It is unlikely, though,
that transit subsidies, per se, are actually increasing the carpool share.
Instead, the increase in carpool share is likely being caused by carpool
incentives which are correlated with the offering of a transit subsidy. That
is, firms offering transit subsidies are also providing ridesharing subsidies.
While vanpool subsidies probably are actually increasing the vanpool share,
there are indications in the data that some employers report vanpooling as
part of carpooling.

Individual incentives in the categories of assistance programs, award
programs, flexible work schedules, and other had small impacts on mode share
that were not statistically significant. Assistance programs as a group, when
offered in conjunction with financial incentives, did have a statistically
significant impact on the change in AVR over the study period, leading to the
conclusion that assistance programs help to facilitate the effectiveness of
financial incentives.

3.3 Impacts of Land Use and Urban Design Characteristics on Work Trip Mode
Choice
As described in the previous chapter, the technique of principal components
analysis was used to develop composite variables to describe areas with land
use and urban design characteristics. Composite variables derived using
principal components analysis were then used in subsequent analyses to
identify the impacts of these individual land use characteristics on work trip
mode choice, as well as the impacts of combinations of land use
characteristics and TDM strategies on mode choice. Each of the identified
land use characteristics was matched with each category of TDM (e.g., mix of
land uses and financial incentives; preponderance of convenience services and
assistance programs; etc.) It was hypothesized that the change in mode share
away from drive alone would be significantly greater when TDMs and land
use/urban design characteristics were combined than when TDMs were implemented
at sites that did not exhibit the land use characteristics. Analysis of
variance techniques were used to test the interactive impacts of the composite
variables describing land use characteristics and TDM strategies on mode
share. The results of these analyses are described below.

Areas Characterized by a Mix of Land Uses
It is commonly hypothesized that as the number and mix of land uses in close
proximity to a work site (i.e., within one-quarter mile) increase, work trip
mode shares may shift away from the single-occupant vehicle toward alternative
modes. This shift may occur because workers are able to make midday trips
(both business and personal) by foot, bicycle, or public transit. In
addition, employees who live within walking distance of their work may be able
to walk or bicycle to their jobs. It also is hypothesized that the relative
effectiveness of TDMs is increased at sites with a diverse mix of uses. As
the number and strength of TDMs are increased, the mode share may shift even
further away from the single-occupant vehicle toward alternative modes.

The analysis conducted partially support these hypotheses. Statistical
analysis reveals financial incentives are the only category of TDMs that
significantly affect mode shifts in areas having a mix of land uses. (Begin
Footnote: Reference the discussion of statistical significance on page 2-23.
End Footnote) Table 3.6 shows the difference in share for each mode based on
changes in land use mix, and whether or not financial incentives were offered.
(Begin Footnote: In Tables 3.6 through 3.11 and throughout the analysis,
flexible work schedules are treated as a "mode of travel" consistent with
their treatment in the original SCAQMD Regulation XV dataset. The percentages
contained in the five cells - drive alone; transit; car and van pool; flexible
work schedules; and bicycle, walk, other - sum to 100 percent.

For the TDM incentives identified in a particular table, the results reflect a
comparison of post- and pre-implementation data. For the land use or urban
design characteristic identified, the comparison is between sites with and
without that particular characteristic. End Footnote) For each mode share
and for Average Vehicle Ridership (AVR), the table shows the percentage of
workers using that mode at sites with:

* A limited mix of land uses where no financial incentives were offered;
* Sites with a limited mix of uses where financial incentives were offered;
* Sites with a broad mix of land uses, but no financial incentives; and
* Sites with both a broad mix of land uses and financial incentives
available.

The table also shows the standard error associated with each mode share.

As shown in Table 3.6, when the absence or presence of financial incentives is
held constant, land use mix does not impact drive alone mode share to a degree
that is statistically significant. That is, in areas with no financial
incentives offered, there is no significant difference in the drive alone mode
share between sites with a limited mix of land uses and sites with a broad mix
of land uses. This is also true for sites where financial incentives are
offered.

Conversely, when land use mix is held constant, the introduction of financial
incentives does have a significant impact on drive alone mode share. At sites
with a minimal mix of land uses, the drive alone share decreases by 5.5
percent from 77.2 to 71.7 percent when financial incentives are offered.
Similarly, when financial incentives are offered at sites characterized by a
diverse mix of land uses, the drive alone mode share decreases by 4.4 percent
from 75.2 to 70.8 percent. While in terms of the percentage change in drive
alone mode share, the impact of incentives is greatest in areas without a mix
of land uses, it is important to realize that the drive alone mode share in
mixed uses areas is smaller to begin with. Thus, while the percentage change
is smaller, financial TDMs offered at sites with a mix of land uses will
result in a lower drive alone mode share than when offered at sites without a
mix of uses. (Begin Footnote: In interpreting the results of Table 3.6 (and
also Table 3.7-3.11), it is important to note that the impacts of financial
TDMs and of a particular land use or urban design characteristic are not
cumulative. For example, in Table 3.6, the difference in mode share between
the upper left cell (sites without either financial incentives or a mix of
land uses) and the upper right cell (sites without financial incentives but
with a mix of land uses) cannot be simply added to the values of the lower
left cell (sites with financial incentives but without a mix of land uses) to
determine the impact of the lower right cell (sites having both a mix of land
uses and offering financial incentives). For Tables 3.6-3.11, there are some
instances when the difference in mode share between sites without either TDMs
or a particular land use/urban design characteristic and sites with both TDMs
and that land use/urban design characteristics exceeds the sum of the parts,
while in other cases this difference is less than the sum of the parts.
Factors such as the size of financial incentive offered, as well as unmeasured
urban design and land use variables (e.g., density) may influence these
differences. End Footnote)

When the drive alone mode share at sites with neither a mix of land uses nor
financial incentives is compared to the drive alone mode share at sites with
both a mix of uses and financial incentives, the difference in drive alone
share is 6.4 percent (77.2 percent versus 70.8 percent, respectively). This
difference indicates that there is indeed an interactive effect between land
use mix and financial incentives that result in a smaller work trip drive
alone mode share when both are present.

While land use mix alone does not significantly impact the drive alone share,
it does result in shifts in the transit mode share. Among sites that do not
offer financial incentives, sites with a mix of land uses have a 1.9 percent
higher transit mode share than do sites without a mix of land uses. Among
sites where financial incentives are offered, sites with a mix of land uses
have a 3.5 percent greater transit mode share for work trips than areas
without a mix of land uses.

For sites characterized by both a limited mix of land uses and no financial
incentives, the transit mode share is 2.8 percent lower than at sites with
both a diverse mix of uses and the presence of financial incentives. In areas
characterized by a limited mix of land uses, the introduction of financial and
TDM incentives appears to shift trips from transit to rideshare and flexible
work schedule, resulting in a lower transit mode share than if no incentives
are offered. One reasonable explanation for this shift is that more of the
incentives encourage ridesharing than transit use. In addition, it is likely
that transit service is more limited and less convenient in areas having a
limited land use mix. Thus, when financial incentives are introduced at sites
with a limited mix of land use, transit riders are induced to shift to carpool
options. When both a broad range of land uses and financial incentives are
present, transit regains a portion of the ridership lost to ridesharing in
areas with limited land uses. When land use mix is held constant, financial
incentives do not significantly impact transit mode share.

As with the drive alone mode share, differences in the land use mix alone do
not significantly impact shifts in ridesharing. The addition of financial
incentives does result in significant shifts toward ridesharing both at sites
with a limited land use mix (+5.3 percent) and at sites with a mix of land
uses (+4.7 percent). The rideshare mode share was greatest (18.7 percent) at
sites with limited land uses and financial incentives. The slightly lower
share (17.7 percent) at sites with both a mix of land uses and financial
incentives can be accounted for by the higher share of transit mode share (6.4
percent). Thus, the interactive effect of both land use mix and incentives on
ridesharing results in a significant positive change relative to limited use
sites with no financial incentives, but the impact of the interaction is
tempered by shifts from ridesharing to transit.

The share of workers using flexible work hours is significantly higher at
limited land use sites where financial incentives are offered (4.0 percent)
than at either sites with a similar land use mix but no financial incentives
(2.4 percent) or sites with both a mix of land uses and financial incentives
available (1.9 percent). It is possible that these results occur because in
areas with a limited land use mix, transportation options are also limited so
that flexible hours becomes a more attractive options for both employers and
employees. As land use mix intensifies, other transportation options (such as
transit and rideshare opportunities) increase, and workers opt for these
option at the expense of flexible work hours.

An analysis of the bicycle/pedestrian mode share for different combinations of
land use mix and financial incentives does not reveal any statistically
significant differences in mode share. While the percentage of people walking
and biking to work is higher where land use is mixed, the differences are not
significant with a 95-percent level of confidence. The use of bicycling,
walking and other travel modes actually decreases with the introduction of
financial incentives and other TDMs. This is an indication that the
particular mix of incentives selected by employers had the unintended impact
of increasing AVR by reducing the tendency of people to bike and walk.

In areas without financial incentives and a limited land use mix, the average
vehicle ridership is 1.218. This increases to 1.271 in areas with both a
diverse land use mix and financial incentives. Since this is a statistically
significant increase, the interaction between land use mix and financial TDMs
can be considered positive in reducing the dependence on the single-occupant
vehicle.

Areas Characterized by the Availability of Convenience-Oriented Services
The hypothesis tested was that at sites characterized by the presence of
convenience-oriented services, workers will commute using alternative modes of
transportation more frequently than at sites where convenience services are
absent. It was reasoned that when at least four types of convenience-oriented
services (such as restaurants, banks, child care centers, dry cleaners, drug
stores and post offices) are present, workers will be able to conduct personal
business and run errands during the work day without the use of an automobile.
It was further hypothesized that when TDMs are provided in an area having a
number of convenience-oriented services, the drive alone mode share will
further decrease in favor of transit, ridesharing, bicycling and walking.

Statistical analysis revealed that two categories of TDMs, financial
incentives and assistance programs, each significantly affected mode shifts at
sites characterized by convenience-oriented services. Table 3.7 shows
differences in share for each mode based on the availability of convenience
services and the level of financial incentives offered. For each mode share
and for AVR, the table shows the percentage of workers using that mode at
sites with:

* Limited availability of convenience services and where no financial
incentives are offered;
* Sites with limited convenience services, but where financial incentives are
offered;
* Sites having a mix of convenience services, but no financial incentives;
and
* Sites with both a mix of convenience services and financial incentives
available.

The table also shows the standard error associated with each mode share.

When the absence or presence of financial incentives is held constant, there
is no significant difference in the drive alone mode share between sites
without convenience-oriented services and sites having a mix of convenience-
oriented services. However, the introduction of financial incentives does
significantly shift the work trip mode share away from drive alone both at
sites with few convenience-oriented services (-4.3 percent) and at sites
characterized by the presence of convenience-oriented services (-5.6 percent).
The drive alone mode share is 7.1 percent higher at sites without either
convenience services or financial incentives than at sites that both offer
financial incentives and have a preponderance of convenience services. This
indicates that the interaction effect of financial incentives and the presence
of convenience-oriented services results in the greatest overall shift from
drive alone to other modes.

The shift from drive alone can be accounted for by gains in both transit and
rideshare. Interestingly, that portion of the shift that accrues to transit
results from changes in the land use mix. Across all sites without financial
incentives, the transit mode share increases from 3.7 to 6.1 percent (+2.4
percent) when the land use mix shifts from limited convenience-oriented
services to a mix of convenience-oriented services. Across sites where
financial incentives are available, the difference in transit mode share
between sites without and with services is 3.7 percent. Differences in the
availability of financial incentives do not significantly impact transit mode
share when the availability of convenience-oriented services is held constant
across sites.

Conversely, the absence or presence of convenience services by itself does not
significantly affect ridesharing. Instead, the availability of financial
incentives significantly affects ridesharing both at sites characterized by
limited convenience-oriented services, and at sites characterized by a mix of
convenience-oriented services. At sites without convenience services, the
rideshare mode share increases from 13.4 to 18.6 percent (+5.2 percent) when
financial incentives are present. When incentives are made available at sites
with a selection of convenience-oriented services, the rideshare mode share
increases by 5.0 percent from 12.5 to 17.5 percent. The slightly lower share
for rideshare at sites with convenience-oriented services can be explained by
the higher percentage of transit share at these sites.

The impact of TDMs on alternative mode shares is influenced by the land use
characteristics of the sites. At work sites with limited convenience-oriented
services, TDMs favor ridesharing. At sites with both financial incentives and
convenience-oriented services, both ridesharing and transit benefit from the
combination. While the impact on ridesharing is less than at sites without
convenience services, the combined affect of the shift to transit and
rideshare results in the greatest decrease in the drive alone share.

At sites without financial incentives, the bike/walk mode share does increase
significantly when convenience oriented services are present. However, this
increase is lost when financial incentives are offered, suggesting that
financial incentives encourage ridesharing at the expense of biking and
walking. The presence of convenience-oriented services and financial
incentives has no significant impact on the percent of people working a
flexible hours schedule.

Neither the availability of convenience services nor the availability of
financial incentives alone significantly alter the AVR. However, the
interactive affect of the presence of convenience services and financial
incentives does result in a significant shift from 1.224 at sites without
either convenience services or financial incentives, to 1.286 at sites with
both.

For employment sites that provide a mix of convenience-oriented services, TDMs
that provide assistance to employees in identifying feasible alternatives to
driving alone significantly affect shifts in mode share. Table 3.8 displays
the impacts of assistance programs on mode share when combined with
convenience-oriented services.

The table shows that assistance programs result in a small but significant
shift away from the drive alone mode share (-2.6 percent) when offered at
sites with limited convenience-oriented services. At sites with a mix of
convenience services, the drive alone mode share changes from 75.2 percent
without assistance programs to 69.9 percent with assistance programs, a
decrease of -5.3 percent. A total of 6.8 percent fewer employees drive alone
at sites having both assistance programs and convenience services than in
areas with neither assistance programs nor convenience services. This
indicates an interactive effect between convenience oriented services and
assistance programs causing a shift of trips away from the single-occupant
vehicle. Both transit and rideshare realize statistically significant gains
to account for this shift from drive alone. Changes in bike/walk and flexible
work hours occur, but with one exception are not statistically significant.
The mode shifts that occur when assistance programs are offered parallel those
that occur when financial incentives are offered, with similar changes in AVR.

Sites Characterized by a High Level of Accessibility
It was hypothesized that work sites providing easy access for transit users,
pedestrians and bicyclists, and with easy access to nearby services may
realize a smaller drive alone mode share than sites with lower accessibility.
Moreover, when TDMs are offered at sites having high accessibility, the drive
alone share will decrease further. Statistical analysis revealed that the
only category of TDMs to significantly impact changes in work trip mode share
when combined with site accessibility are financial incentives. The impacts
of financial incentives and accessible sites on work trip mode share are
displayed in Table 3.9.

The table shows that accessibility alone does not statistically affect drive
alone share. That is, when the availability of financial incentives is held
constant, there is no significant difference in the drive alone mode share
between sites characterized as lacking accessibility and those characterized
as providing accessibility. Conversely, at sites where financial incentives
are available, there is a significant reduction in the drive alone mode share
both for sites not characterized as accessible (-4.3 percent), and at sites
characterized as having a high level of accessibility (-5.5 percent). When
sites without either access or financial incentives are compared to sites with
both access and financial incentives, the drive alone share decrease from 76.4
to 70.5 percent, a change of -5.9 percent. This change is greater than that
realized simply by the addition of financial incentives, and indicates that
accessibility and financial incentives interact to produce a greater impact on
the reduction in the drive alone mode share.

Accessibility appears to impact transit share significantly regardless of the
presence of TDMs. When financial incentives are not present, the transit
share is 2.0 percent greater for accessible sites than for inaccessible sites.
When TDMs are available, this difference increases to 3.3 percent. This gain
in transit mode share occurs at the expense of drive alone, rideshare and
flexible work hour shares. When land use characteristics are held constant,
financial incentives alone do not create a significant shift in the transit
mode share. However, sites that combine both financial incentives and a high
level of accessibility show the highest transit mode share (6.3 percent).

As with other land use characteristics analyzed, when the availability of
financial incentives is held constant, a difference in accessibility by itself
is not significant in altering the amount of ridesharing. It is the
introduction of financial incentives, whether at low or high access sites,
that impacts ridesharing. At low access sites, when financial incentives are
present the rideshare mode share increases from 13.8 to 18.8 percent (+5.0
percent). At accessible sites, the addition of financial incentives results
in a change in ridesharing from 13.0 to 17.7 percent (+4.7 percent). At high
access sites, ridesharing is less than at low access areas because transit
captures a greater share. Ridesharing and transit combine to create a
cumulative reduction in driving alone that is greatest for high access sites
with financial incentives.

Financial incentives result in a reduction in the bike/walk mode share in low
access areas. The likely reason for this reduction is that the financial
incentives are geared toward ridesharing and transit. Areas with poor access
are not particularly pedestrian or bicycle friendly, and it appears that the
financial incentives offered in these areas are sufficient to induce cyclists
and pedestrians to switch to alternative modes. In areas that are accessible,
this shift away from the walk/bike mode is not statistically significant when
financial incentives and other TDMs are introduced. The percent of workers on
flexible work hour schedules decreased at sites having a higher level of
accessibility, and was not affected in a statistically significant manner by
financial incentives

There is a significant increase in AVR between areas having low accessibility
and no incentives (1.225 AVR) and sites having both high accessibility and
financial TDMs (1.272). This change appears to reflect an interactive effect
between accessibility and financial TDMs.

Areas Characterized as Safe
It was hypothesized that in areas perceived as safe, the drive alone mode
share would be smaller than in areas where safety concerns may exist. This
would occur because employees would be more willing to walk from transit to
their job sites, and to make midday trips by foot or transit. For the
purposes of this analysis, sites were considered to have a higher level of
safety if they were characterized by sidewalks, street lighting, pedestrian
activity, and by the absence of vacant lots.

As with many of the other land use categories analyzed, financial incentives
are the only category of TDMs that result in significant shifts in work trip
mode shift when combined with the perception of safety. Table 3.10 shows
difference in mode share based on changes in the perception of safety and the
level of financial incentives offered.

When financial incentives are not available, there is a 3.9 percent difference
in the drive alone share between areas with low safety compared to areas that
are perceived as safe. The availability of TDMs helps this shift. Both at
sites that lack the perception of safety, and those that are perceived as
safe, the introduction of financial incentives leads to a decrease in the
drive alone mode share (-5.8 percent and -4.5 percent, respectively). The
interactive affect between the perception of safety and the availability of
financial incentives result in an even larger shift away from the single-
occupant vehicle for commuting to work (-8.4 percent). Sites that are not
perceived safe and that do not offer financial incentives have an above-
average drive alone share of 79.0 percent, while sites perceived as safe at
which financial incentives are available have a 70.6 percent drive alone mode
share.

At sites having a low level of safety, the shift away from drive alone when
financial incentives are offered is accounted for by the 5.6 percent increase
in ridesharing (from 12.8 to 18.4 percent). At sites perceived as safe, there
is a 4.4 percent change in ridesharing when financial incentives are offered.
None of the other modes show any significant shifts when the perception of
safety is held constant and financial incentives and other TDMs are
introduced.

When the level of TDMs is held constant, changes in the perception of safety
result in a significant change for the transit and walk/bike mode shares.
When TDMs are available, the transit mode share increases by 1.8 percent from
3.6 to 5.4 percent. The bike/walk mode share increases by 1.5 percent between
sites not characterized as safe and those that are perceived as safe. This is
a large shift, given that the bike/walk mode share accounts for less than 4.0
percent of all trips even at sites perceived as safe.

There is an 8.4 percent decrease in the drive alone mode share between sites
having a low level of safety and without TDMs, and sites that are perceived as
safe and that offer TDMs. This change is accounted for by significant
increases in mode shares for the transit, rideshare, and walk/bike. The
perception of safety is the only land use characteristics that results in
positive shifts in share for more than two of the model alternatives. The
perception of safety and the presence of TDMs seem to interact to achieve the
larger shift away from the drive alone mode share.

There is also a significant increase in AVR between areas without financial
TDMs and a low level of perceived safety compared to sites that are perceived
as safe and that also have financial TDMs available (AVR of 1.206 versus
1.263).

Areas Characterized by an Aesthetically Pleasing Urban Setting
It was hypothesized that sites located in an aesthetically pleasing
environment would have a lower than average drive alone mode share. Using the
Principal Components analysis technique, a group of variables including street
noise, lots of signs, an aesthetic appearance, and landscaping combined into a
single composite variable that characterized a site as aesthetically pleasing.
Only financial incentives were found to have a statistically significant
impact on mode share when combined with aesthetic characteristics. Table 3.11
shows the difference in share for each mode based on aesthetics and the level
of financial incentives offered.

The drive alone mode share was statistically affected by both financial
incentives and by the presence of an aesthetically pleasing work site. For
those sites not characterized as aesthetically pleasing, the drive alone mode
share is 4.7 percent lower at work sites where financial incentives are
offered than at sites without such incentives (72.3 percent and 77 percent,
respectively). Similarly, for sites that are aesthetically pleasing, the
presence of financial incentives decreases the drive alone mode share from
72.5 percent to 66.6 percent.

When the presence of financial incentives is held constant, site aesthetics
has a statistically significant influence on drive alone mode share. At sites
without incentives, the drive alone mode share is 77 percent for "less
aesthetic" urban sites, and 72.4 percent for "more aesthetic" sites. The
combined impacts of both financial TDMs and aesthetically pleasing urban sites
is of particular note. Sites without either financial incentives or an
aesthetically pleasing quality have an average drive alone mode share of 77.0
percent, while sites with both a high level of aesthetics and financial TDMs
have an average drive alone mode share of only 66.6 percent. This is the
lowest drive alone mode share for any of the land use and urban design
characteristics evaluated in this analysis.

Once again, introduction of TDMs has no significant impact on the transit mode
share when land use characteristics are held constant. The transit mode
share, however, does shift significantly when the presence of TDMs is held
constant and the level of aesthetics changes. When no TDMs are available, the
transit share increases from 3.9 to 7.8 percent. At sites that offer
financial TDMs, the average transit share increases from 4.2 to 8.3 percent
for sites that are aesthetically pleasing. This is the highest transit mode
share identified for any of land use or urban design characteristics.

As was true with the other land use and urban design characteristics
evaluated, ridesharing varies with the level of financial incentives
available. At sites that are not aesthetically pleasing, ridesharing
increases from 13.3 to 17.9 percent when financial incentives are introduced.
In aesthetically pleasing urban areas, ridesharing changes from 13.9 to 18.9
percent when financial incentives are introduced. As with transit,
ridesharing is higher for aesthetically pleasing sites than for any of the
other categories of land use and urban design characteristics evaluated.

Unlike the other land use and urban design characteristics analyzed, there
does not appear to be a tradeoff between transit usage and ridesharing when
financial incentives are present at aesthetically pleasing sites. Instead,
the interactive affect between aesthetics and financial incentives yields
increases in both transit and ridesharing, thus reducing the drive alone share
to the low of 66.6 percent.

Due to the large shift away from the drive alone mode share, sites
characterized as aesthetically pleasing and where financial incentives are
offered achieve an AVR of 1.337. This AVR is higher than that of any other
combination of financial incentives and land use characteristics. It also
appears to be a result of the interaction between the land use characteristics
and the TDMs, as opposed to either one individually. One can conclude,
therefore, that the presence of an aesthetically pleasing setting is important
in improving the effectiveness of TDMs.

3.4 Summary of Impacts of Land Use and Urban Design Characteristics
To further understand the impact of individual land use and urban design
characteristics, the differences in the drive alone share are compared between
sites without each land use characteristic and sites exhibiting that land use
characteristic (Table 3.12). In this analysis, financial incentives are
present at all sites.

The biggest change in the drive alone share (-5.7 percent) occurs between
sites that lack an aesthetic urban quality and sites having an aesthetic urban
quality. Sites with an aesthetic urban quality realize the lowest drive alone
share (66.6 percent) of sites displaying one of the urban design
characteristics under study. The next lowest drive alone share is achieved at
sites with a preponderance of convenience services (69.6 percent). The drive
alone share at sites having a high accessibility to services is 70.5 percent,
while at sites perceived as safe the drive alone share is 70.6 percent. At
sites with a mix of land uses, the drive alone share is 70.8 percent.

Since land use and urban design characteristics cannot always be easily
changed, it may be difficult to change mode share simply by introducing a new
land use or urban design characteristic to an existing employment site.
However, given that this analysis reveals that land use and urban design
characteristics do impact commute mode share, communities may wish to
encourage developers of new employment sites to incorporate land use and urban
design characteristics that support a lower drive alone mode share into their
site designs. These characteristics also can be introduced as part of a major
site rehabilitation or modernization project.

It also is important to note that precise causality cannot be measured due to
the limitations of the database. For example, areas with a mix of land uses,
good accessibility, and lots of pedestrian traffic may also be areas with
higher than average density and transit service. However, density and level
of transit service cannot be adequately measured with the existing data so the
impacts of these factors on mode share cannot be evaluated separately. Other
factors that may influence the results but for which information was not
available include parking costs and availability at work sites, weather, and
the magnitude of employer-provided financial incentives.

3.5 Land Use and Urban Design Characteristics of Sites with a High Walk or
Bicycle Mode Share
The share of work trips made by walking and bicycle as a percentage of the
total work trips in the data set is small. This makes identification of work
site characteristics that encourage utilization of these modes difficult. To
understand the characteristics of sites that have a high walk or bike mode
share, sites with a walk or bike mode share of greater than 10 percent were
identified and evaluated separately.

Table 3.13 displays the land use characteristics and TDM measures available at
sites having a combined bicycle and walking mode share that is greater than 10
percent either before or after implementation of Regulation XV. Twenty-five
sites or 7.6 percent of all observations, spread throughout the study area had
a walk/bike mode share greater than 10 percent prior to implementation of the
Regulation XV trip reduction measures. After implementation, though, the
walk/bike mode share declined to 8.3 percent, 6.2 percent lower than the pre-
Regulation XV figure of 14.5 percent. As a whole, this group displays a
greater percentage of sites having land use and urban design characteristics
that encourage alternative modes of travel for the work trip than is true of
the entire 330-site data set. The level of TDMs offered at theses sites,
however, is below the average for the entire data set. Furthermore, a smaller
percentage of these sites offer financial subsidies for walking and bicycling
than is true for the entire data set. The walk/bike share at these sites may
have declined because incentives promoting ridesharing and transit were
offered to employees, with corresponding fewer incentives supporting walking
and bicycling being available.

The second column of Table 3.13 summarizes characteristics of sites having a
walk/bike mode share greater than 10 percent after implementation of the
Regulation XV measures. In contrast to the column one sites, the percentage
of employees walking or biking increased at these locations from an average of
12.3 to 13.9 percent. The land use and urban design characteristics of these
sites more closely parallel those found to encourage alternative modes than
either the data set as a whole or the sites with a pre-Regulation XV walk/bike
mode share of 10 percent or more. The percentage of sites offering financial
incentives was comparable to that for the entire data set. Furthermore, the
percentage of these sites that offered walk and bike subsidies was well above
the average for the complete data set. Bicycle racks were also more common at
these sites. It appears that both land use and urban design characteristics
that encourage alternative modes and the provision of TDMs that are
specifically designed to be supportive of bicycling and walking can be
effective.

3.6 Sites with Low Single-Occupant Vehicle Use
The analysis identified 21 work sites at which 50 percent or fewer of the
employees commuted by single-occupant vehicle (SOV). Within the context of
the 330 work sites sampled, these sites provide one indicator of a practical
"upper bound" for AVR and utilization of non-drive alone travel modes for the
trip between home and work. As such, they serve as a point of reference
against which the potential of other sites may be judged. Table 3.14 shows
the average of the commute mode shares for these 21 sites. Ridesharing
accounts for the majority of non-SOV commuting, having a 36 percent modal
share. Transit also carries a large proportion (12.8 percent) of non-SOV
commuters, a level that is almost three times the transit mode share for the
sample as a whole. Average vehicle ridership is 1.68, compared to the base
year average of 1.22.

These modal shares, however, do not appear to be the result of a particular
mix of land use design, and TDM strategies. While employees at these 21 sites
are the least reliant on single-occupant automobiles, the TDM incentive levels
and the land use/urban design characteristics at the sites are not
significantly different from the average for all sites in the sample. Factors
other than those analyzed apparently account for the low SOV mode share.

4.0 Conclusions
The findings presented in the previous section lead to the following general
conclusions that can guide implementation of land use, urban design, and
transportation demand management strategies in urban settings.

1. Financial Incentives are Important as Part of a TDM Strategy
A successful travel demand management strategy should be built around a
core of financial incentives, regardless of the land use and urban design
characteristics of a particular site. As a group, financial incentives are
the only TDM strategies that consistently result in a statistically
significant reduction in the drive alone mode share. At sites where
financial incentives are not included among the TDMs offered, the drive
alone mode share decreased by 1.7 percent over the study period compared to
6.4 percent decrease in the drive alone mode share at sites where financial
incentives are included among the TDMs offered. For each land use
category, financial incentives account for the majority of the reduction in
the drive alone share. Individual financial incentives that resulted in a
statistically significant shift from driving alone were transit, vanpool,
and bicycle subsidies and other employee benefits.

2. Specific Land Use and Urban Design Characteristics Influence Mode Choice
Urban design and land use characteristics that can be controlled by public
officials and private business working in a cooperative partnership can
influence a person's choice of commuting mode. The findings demonstrate
that the availability of TDM strategies and transportation alternatives,
combined with opportunities to accomplish midday errands without having to
drive, reduce the use of single-occupant vehicles for commuting.

The data reveal that when financial incentives are present, the greatest
reduction in the drive alone share is realized in areas with an
aesthetically pleasing urban character. The drive alone mode share at
these sites is at least three percent less than at sites exhibiting any of
the other land use characteristics analyzed. This appears to be a result
of the availability of alternative modes (e.g., transit service), and the
quality of the environment. Sites with a preponderance of convenience-
oriented services realize the next greatest reduction in the drive alone
mode share, followed by sites with good access to services, sites with the
perception of safety, and sites with a mix of land uses.

3. A Positive Interactive Effect Exists Between Land Use Characteristics and
Financial Incentives
Travel demand management strategies have a larger influence on reducing the
drive alone mode share than do land use characteristics, when each is
considered individually. However, the findings further reveal that there
is a positive cumulative impact on increasing average vehicle ridership
(AVR) and reducing drive alone mode share when both financial incentives
and one of the five land use characteristics analyzed are present. When
both are present, the increase in AVR is always greater (by at least 2.5
percent) than when TDMs are present in an area without any of the land use
characteristics, or when TDMs are absent from sites where the land use
characteristics are present.

The impacts on mode share, however, are not linearly additive as further
TDMs as well as land use and urban design characteristics are included at a
site. The cumulative effect is less than the sum of the parts.

In implementing a regional TDM strategy, efforts should focus on areas that
exhibit at least one of the land use characteristics studied as there is a
greater potential for increases in the AVR in these areas. Consideration
of this interactive effect when designing a TDM strategy may result in a
more effective and efficient program. Adoption of policies that support
compatible development of work sites with the land use and urban design
characteristics found to encourage alternative modes is warranted.

4. Tradeoffs Exist Between Ridesharing, Transit, and Walk/Bike
Modal decisions are made not only between driving alone and alternative
modes, but also among available alternative modes. The TDM programs
examined are most beneficial in increasing the level of ridesharing. This
increase in ridesharing, however, results not only from a decrease in
driving alone mode, but also comes at the expense of transit, walking, and
bicycling. Transit and walk/bike mode shares are highest at sites with
supportive land use and urban design characteristics. This further
indicates that mode choice is influenced by both land use characteristics
and the availability of TDMs.

From a policy standpoint, it is important to understand these tradeoffs
when designing a transportation management program. For example, a TDM
strategy that increases ridesharing at the expense of transit, walking, or
bicycling may not be supportive of broader regional transportation policies
or goals. By understanding the tradeoffs that may occur given particular
land use characteristics, a TDM program can be designed to strengthen
incentives that will encourage the full range of available non-drive alone
modes. It may be effective to focus TDM strategies on ridesharing in areas
that do not exhibit land uses that are supportive of transit, walking, and
bicycling, TDMs that support transit and walk/bike should be featured in
areas where the land uses are supportive of these modes.

5. Employer-Provided Transportation Assistance Programs are Most Helpful
at Sites Having a Variety of Convenience-Oriented Services
Employer-provided transportation assistance programs have a small but
statistically significant impact on reducing the drive alone modal share
(-5.3 percent) and increasing the AVR (from 1.223 to 1.285) at sites having a
mix of convenience-oriented services. Assistance programs were not found,
by themselves, to have a significant impact on either the drive alone share
or AVR at sites with other land use characteristics. For sites having a
high level of convenience-oriented services, a TDM strategy featuring
assistance programs should be successful in helping to achieve increases in
AVR. A program that includes assistance programs but not financial
incentives, though, will have a smaller positive impact than a program that
includes financial incentives.

6. Selected Individual Sites Attain High Levels of Non-Drive Alone Commuting
While the average level of walking and biking over all the sites surveyed
is 5.4 percent, selected sites have post-implementative mode shares that
are two and one-half times this level. These sites are characterized by
land use and urban design characteristics that encourage alternative modes
of travel for the work trip. Furthermore, these sites offer financial
incentives in the form of walk and bicycle subsidies that are well above
the average for all the sites analyzed.

Twenty-one of the 330 sites examined have less than half of their employees
commuting by driving alone, leading to an average vehicle ridership of 1.68
compared to the overall average of 1.25. Ridesharing accounts for the
majority of alternate mode commuting, achieving a share of 36 percent, with
transit accounting for 12.8 percent of the work trips at these sites.
These figures provide one indication of the practical upper bound that may
be achievable in terms of the distribution of commuting mode shares and the
level of average vehicle ridership.

7. Transferability of Results
The impetus for the implementation of TDM strategies at many employment
sites within the Los Angeles metropolitan area has been the Regulation XV
trip reduction ordinance enacted by the South Coast Air Quality Management
District. Many Los Angeles area employers, though, had implemented a
diverse range of TDMs for a variety of reasons prior to enactment of the
Regulation XV ordinance. An evaluation of TDM strategies at these
particular locations based only on a Regulation XV "before" condition,
therefore, may understate their level of effectiveness since the TDM
measures already would have been in place.

The impetus for the provision of TDM measures or supportive urban design
characteristics is not relevant to an analysis of their effectiveness.
Similar results should be obtained independent of the factors motivating
their implementation.

The data used in this analysis are specific to Los Angeles county, and thus
reflect the particular socioeconomic and geographic characteristics of that
particular portion of the Los Angeles metropolitan area. There are,
however, numerous urban areas in the U.S. that are similar to this portion
of Los Angeles in terms of their land use characteristics, densities,
socioeconomic characteristics, and commute trip travel characteristics.
The results of this study are directly applicable to the development of TDM
programs for these areas.

It is recognized, though, that the drive alone mode share is higher and
that the development density is lower in the Los Angeles metropolitan area
than in many older areas in the United States. For these areas, the
results of this study are considered a conservative estimate of the
interactive effects of land use and transportation demand management
strategies on mode choice. Areas having land use characteristics that are
more supportive of alternative modes of transportation could have higher
levels of effectiveness than reported here. In addition, the results are
transferable to other urban areas in terms of the relative ranking of
importance of the land use and TDM factors analyzed.

Table 2.1 Data Elements # General Environs of the SiteLand Use

7 Land use mix
7 Predominant single land use
7 Special features or notable
sites
7 Building types

Street Characteristics

7 Identification of the main
streets
7 Traffic levels
7 Presence of sidewalks
7 Landscape quality

Table 2.2 Data Elements # Site Area (One-quarter mile radius)Land Use

Land Use Mix
7 Horizontal
7 Vertical

Presence of Specific Land
Use Types
7 Residential
7 Office
7 Retail
7 Heavy industrial
7 Light industrial
7 Auto-related
7 Institutional
7 Open space
7 Parking (off-street)
7 Personal services
7 Business services

Services

Presence, Frequency, and
Distance to Specific
Services
7 Restaurants/coffee shops
7 Groceries
7 Banks/ATM machines
7 Parks/open space
7 Child care
7 Dry cleaning/laundry
7 Drug stores
7 Entertainment:
movies,videos, etc.
7 Haircuts
7 Health
club/exercise/dance
7 Copies
7 Post office
7 Travel agent
7 Parking lot
7 Parking structure

Street Characteristics
7 Street type
7 Median
7 On-street parking
7 Level of traffic
7 Street layout
7 Mix of Traffic
7 Noise level

Streetwall Characteristics
7 Building set-back
7 Quality of streetwall
7 Adjacent uses
7 Signage
- Parcel use
- Unrelated to use (e.g.,
billboards, graffiti)

Sidewalk Characteristics
7 Presence of sidewalk
7 Pavement type
7 Level of maintenance
7 Sidewalk zones
- Tree/shrub planting
strip
- Arcades/awnings
- Street furniture

Pedestrian Characteristics
7 Types of pedestrians
7 Extent of pedestrian
activity

Landscaping Characteristics
7 Presence of trees
7 Tree size and spacing
7 Shade effect

Table 2.3 Data Elements # Work Place CharacteristicsParcel and Block
Characteristics
7 Block form
7 Block density
7 Floor-to-area ratio (FAR)
7 Parcel size
7 Number of parcels in block
7 Block dimensions

Building Characteristics
7 Building size
7 Architectural style
7 Aesthetic appearance
7 Building materials
7 Building set-back
7 Orientation
7 Scale
7 Building maintenance

Transportation Characteristics
7 On/Off-street parking
7 Distance to bus stop
7 Distance to rail transit
7 Width of sidewalk at main
entrance
7 Typical sidewalk distance in
area
7 Number of cyclists

Table 2.4 Distribution of AVR, Location, and Industry Characteristics Percent of Sites
Statistically
330 Subsample 1,110 Sample Significant

AVR Target
1.3 0.3 2.7 Yes
1.5 97.0 93.4 Yes
1.54 0.0 0.1 No
1.63 0.0 0.1 No
1.75 2.7 3.7 No

Sub Areas
(1) Los Angeles County Central 13.2 11.7 No
(2) Los Angeles County West 14.1 6.0 Yes
(3) Los Angeles County South 4.0 4.6 No
(4) Los Angeles County Southwest 16.0 8.1 Yes
(5) Los Angeles County East 12.6 5.4 Yes
(6) Los Angeles County Remote West 7.1 3.4 Yes
(7) Los Angeles County Remote Northeast 7.4 9.3 No
(8) San Fernando Valley 5.8 6.5 No
(9) Burbank,Glendale,Pasadena 6.1 7.9 No
(10) Long Beach 13.8 4.8 Yes
(11) San Bernardino County 0.0 5.9 Yes
(12) Riverside County 0.0 4.7 Yes
(13) Orange County North 0.0 13.0 Yes
(14) Orange County South 0.0 8.1 Yes
(15) Remote Area 0.0 0.5 Yes

Area
L.A. Central (1) 13.2 11.7 No
L.A. County (2,3,4,5,7,9,10) 73.9 46.2 Yes
Suburb (6,8,11,12,13,14,15) 12.9 42.1 Yes

Industry (SIC Code)
Ag/Fo/Fi/Mi 0.6 0.6 No
Construction 0.0 0.2 No
Manufacturing 30.5 36.3 Yes
Tran/Comm/Util 19.7 17.2 No
Whole/Retail 13.8 13.5 No
Fire 7.4 5.4 No
Services 23.7 22.6 No
Public Office 4.3 4.2 No

Business
Manufacturing 30.5 36.3 Yes
Service/Service Related 49.2 45.7 No
Others 20.3 18.0 No
Table 2.5 Distribution of Employer Size and Trip Reduction Incentives
Percent of Sites
Statistically
330 Subsample 1,110 Sample Significant

Size
250 or less employees 27.4 28.6 No
251 or greater employees 72.6 71.4 No

Incentives
Preferential Parking Area 71.6 67.7 No
Transit Subsidy 53.2 49.0 No
Guaranteed Ride Home 52.0 47.8 No
Prize Drawings 47.7 48.2 No
Bike Racks 41.0 43.0 No
Regional Commuter Management
Agency Matching 40.7 36.9 No
Information Booths 36.1 31.9 No
Flexible Work Hours 34.3 31.7 No
Commuter Information Center 32.4 27.1 Yes
Other Marketing Elements 28.7 24.7 No
Carpool Subsidy 27.8 29.0 No
New Hire Orientation 27.8 25.8 No
Employer Based Matching Service 25.7 26.3 No
Compressed Work Week Program 25.1 21.4 No
Other Employee Benefit 24.2 23.7 No
Showers and Lockers 22.9 21.9 No
Other On-Site Services 19.9 16.2 No
Cafeteria/ATM/Postal/Fitness Center 19.6 19.2 No
Company Owned/Leased Vanpool 19.0 15.9 No
Walk Subsidy 18.7 18.6 No
Preferential Parking - Carpool 17.8 16.4 No
Bike Subsidy 17.1 17.7 No
Vanpool Subsidy 14.7 13.9 No
Auto Service 13.8 13.8 No
Special Interest Group 11.3 12.8 No
Recognition in Newsletter 10.4 12.9 No
Commuter Fairs 9.5 11.7 No
Telecommuting Program 8.9 8.8 No
Other Financial Subsidy 8.0 8.0 No
Additional Time Off With Pay 6.1 7.1 No
Rideshare Parking Subsidy 4.3 2.5 No
Introductory Transit Pass Subsidy 4.0 5.5 No
Vanpool Seat Subsidy 3.7 3.6 No
Other Facility Improvement 3.4 3.2 No
Passenger Loading Area 2.4 1.7 No
Preferential Parking - Vanpool 1.2 1.8 No
Other Parking Management 1.2 2.1 No
Transportation Allowance 1.2 0.5 No
Employee Parking Subsidy 0.9 0.7 No
Childcare Services 0.9 1.2 No
Table 2.7 Composite Land Use/Urban Design Variables Independent Principal
Variables Component

Offices within 1/4 mile of site
Residential development within 1/4 mile of site
Retail development within 1/4 mile of site Mixed Use
Personal services within 1/4 mile of site
Open space (parks) within 1/4 mile of site

Restaurant(s) within 1/4 mile of site
Bank(s) within 1/4 mile of site
Child care within 1/4 mile of site
Dry cleaners within 1/4 mile of site Convenience Services
Drug store(s) within 1/4 mile of site
Post office within 1/4 mile of site

Presence of numerous services (4 or more)
Frequency with which certain services are present
Presence of sidewalk Accessibility of Services
Traffic volume
Transit stop

Absence of graffiti
Presence of trees and shrubs in the sidewalk zone Aesthetics
Wide sidewalks
Minimal building setbacks

Absence of vacant lots
Pedestrian activity Safety
Sidewalks
Street lighting

Table 3.1 Change in Work Trip Mode Share Implementation Absolute Relative
Period Percent Percent
Mode Base Year Year Two*** Change Change

Drive Alone 76.2 71.4 -4.8 6.3
Rideshare* 13.4 18.8 5.4 40.3
Transit 4.6 4.4 -0.2 -4.3
Other** 5.8 5.4 -0.4 -6.9
Average Vehicle
Ridership (AVR) 1.22 1.25 0.03 +2.5

* Carpool and vanpool.
** Bicycle, walk, etc.
*** The second year varies for each site within the sample, based on when TDMs were
implemented at the site. All second year data were collected between 1990 and 1992.

Table 3.2 Categories of TDM Strategies
Financial Incentives7 Transportation allowance7 Bike subsidy
7 Carpool subsidy
7 Introductory transit pass
subsidy
7 Other financial subsidy
7 Vanpool seat subsidy
7 Transit subsidy
7 Vanpool subsidy
7 Walk subsidy
7 Rideshare parking subsidy
7 Additional time off with pay7 Other employee benefits

Flexible Work Schedules7 Flexible work hours7 Telecommuting program7 Compressed work week program7 Compressed work week program

Assistance Programs
7 Commuter information center7 Commuter fairs7 New hire orientation
7 Other marketing elements
7 Special interest group
7 Regional commuter management
agency matching
7 Employer-based matching
service
7 Information booths7 Company owned/leased vanpool7 Other parking management

Award Programs
7 Prize drawing - free meal
certificate7 Recognition in news letter

Other7 Child care service7 On-site services (e.g.,
cafeteria, health club, post
office)7 Auto service
Table 3.3 Impact of Financial Incentives on Mode Share
Shift in Commute Mode Shares
1989-1991
Percent Change Percent Change
When Financial When Financial Difference
Incentives are Incentives are (Incentives Present #
Mode Absent Present Incentives Absent)

Drive Alone -1.7 -6.4 -4.7

Carpool/Vanpool 3.2 5.7 2.5

Table 3.4 Impact of TDM Strategies on Drive Alone Mode Share
1989-1991* Percent Change Percent Change Difference
in Drive Alone in Drive Alone (TDMs Percent #
Strategy When Absent When Present TDMs Absent)

Bicycle Subsidy -4.4 -7.1 -2.7

Vanpool Seat
Subsidy -4.7 -10.1 -5.4

Transit Subsidy -3.2 -6.3 -3.1

Vanpool Subsidy -4.4 -7.7 -3.3

Other Employee
Benefits -3.9 -8.0 -4.1

* Includes strategies that, when present, result in a statistically significant shift from the
drive alone mode share at the 95 percent confidence level.
Table 3.5 Impact of TDM Strategies on Ridesharing1989-1991* Percent Change Percent Change Difference
in Carpool/Vanpool in Carpool/Vanpool (TDMs Present #
Strategy When Absent When Present TDMs Absent)

Vanpool Seat
Subsidy 4.7 10.3 5.6

Transit Subsidy 4.0 5.6 1.6

* Includes TDM strategies, that when present, result in a statistically significant shift from the
drive alone mode share at the 95 percent confidence level.
Table 2.6 Distribution of Transportation Mode Shares at Implementation and One
Year Later
At Implementation One Year After Implementation
Mean Mean
Statistically Statistically
330 Subsample 1,110 Sample Significant 330 Subsample 1,110 Sample Significant

Average Vehicle Ridership 1.213 1.208 No 1.245 1.243 No
Drive Alone Share 0.762 0.766 No 0.714 0.714 No
Carpool Share 0.125 0.134 No 0.174 0.188 Yes
Vanpool Share 0.009 0.007 No 0.013 0.011 No
Bus Share 0.046 0.038 Yes 0.043 0.036 Yes
Other Mode Share 0.033 0.031 No 0.028 0.028 No
Telecommuting Share 0.005 0.004 No 0.003 0.003 No
Compressed Work Week Share 0.020 0.018 No 0.024 0.020 No

Table 3.12 Comparison of Drive Alone Shares among Sites with Financial Incentives
and Alternative Land Use Characteristics
Percent Drive Alone
Sites with Land Sites with Land Absolute
Use Characteristics Use Characteristics Percent
Land Use Characteristics Missing Present Change

Mix of Land Uses 71.7 70.8 -0.9
Accessibility to Services 72.1 70.5 -1.6
Preponderance of Convenient Services 72.4 69.6 -2.8
Perception of Safety 73.2 70.6 -2.6
Aesthetic Urban Setting 72.3 66.6 -5.7

Table 3.13 Characteristics of Sites with a High Walk or Bicycle Mode Share
Sites with Pre-Regulation Sites with Post-Regulation
XV Walk/Bike Share XV Walk/Bike Share All
>10% >10% Sites

Numbers of Observations 25.0 14.0 330.0
Pre-Reg XV Walk/Bike Share 14.5% 12.3% 3.3%
Post-Reg XV Walk/Bike Share 8.3% 13.9% 2.8%
Difference # Pre-Reg to Post-Reg XV -6.2% +1.6% -.5%

Site Characteristics
Aesthetic Urban Environment 35.0% 24.0% 16.0%
Perceived Safe 88.0% 92.0% 72.0%
Access to Services 68.0% 79.0% 53.0%
Mix of Land Uses 68.0% 71.0% 52.0%
Numerous Convenience-Oriented Services 56.0% 64.0% 35.0%
Financial Incentives 56.0% 64.3% 66.1%
Assistance Programs 72.0% 71.4% 82.1%
Flexible Work Schedules 44.0% 35.7% 47.6%
Award Programs 36.0% 42.9% 50.3%
Bicycle Subsidies 12.5% 23.1% 17.1%
Walk Subsidies 16.7% 38.5% 18.7%
Bike Racks 45.8% 61.5% 41.0%

(landuse.html)