Adjusting ITEs Trip Generation Handbook For Urban
Adjusting ITEs Trip Generation Handbook For Urban
Adjusting ITEs Trip Generation Handbook For Urban
net/publication/276474455
CITATIONS READS
26 4,209
3 authors, including:
Some of the authors of this publication are also working on these related projects:
Assessing Cool Corridor Heat Resilience Strategies for Human-Scale Transportation View project
All content following this page was uploaded by Kelly Clifton on 19 April 2016.
Christopher D. Muhsc
Portland State University
Abstract: This study examines the ways in which urban context affects vehicle trip generation rates across
three land uses. An intercept travel survey was administered at 78 establishments (high-turnover restaurants,
convenience markets, and drinking places) in the Portland, Oregon, region during 2011. This approach was
developed to adjust the Institute of Transportation Engineers (ITE) Trip Generation Handbook vehicle trip rates
based on built environment characteristics where the establishments were located.
A number of policy-relevant built environment measures were used to estimate a set of nine models
predicting an adjustment to ITE trip rates. Each model was estimated as a single measure: activity density, num-
ber of transit corridors, number of high-frequency bus lines, employment density, lot coverage, length of bicycle
facilities, presence of rail transit, retail and service employment index, and intersection density. All of these
models perform similarly (Adj. R 2 0.76-0.77) in estimating trip rate adjustments. Data from 34 additional sites
were collected to verify the adjustments. For convenience markets and drinking places, the adjustment models
were an improvement to the ITE’s handbook method, while adjustments for restaurants tended to perform
similarly to those from ITE’s estimation.
The approach here is useful in guiding plans and policies for a short-term improvement to the ITE’s
Trip Generation Handbook. The measures are useful for communities seeking to develop local adjustments to
vehicle trip rate estimates, and all could be calculated from spatial data available in most locations. The paper
concludes with a discussion on what long-term improvements to the ITE’s Trip Generation Handbook might
entail, with further implications in planning and practice.
1 Introduction
The Institute of Transportation Engineers (ITE) Trip Generation Handbook (Institute of Transporta-
tion Engineers 2004) and Trip Generation: An ITE Informational Report (Institute of Transportation
Engineers 2008)1 began collecting data on vehicle trip rates in the 1960s and focused on single-use,
vehicle-oriented trip rates in suburban sites in the United States. Despite its widespread use, the ITE
Trip Generation Handbook lacks the supporting information to apply these rates across a range of ur-
ban contexts. Today, there is national interest in building an evidentiary database that supports the
estimation of trip generation for site-level analysis in transportation impact assessments (TIAs) for new
developments located in urbanized areas that support multimodal transportation options such as infill
locations, transit-oriented developments (TODs), or mixed-use developments (Lee et al. 2011, Daisa et
al. 2009). For locations that support greater non-automobile mode shares, ITE recommends that local
1
For the remainder of this paper, ITE’s Trip Generation Handbook will refer to both the handbook and the accompanying
informational report.
a
kclifton@pdx.edu, bkcurrans@pdx.edu, cmuhs@pdx.edu
The Journal of Transport and Land Use is the official journal of the World Society for Transport and Land Use
(WSTLUR) and is published and sponsored by the University of Minnesota Center for Transportation Studies.
6 JOURNAL OF TRANSPORT AND LAND USE 8.1
rates be established via data collection: “If the site is located in a downtown setting, served by significant
public transportation...the site is not consistent with the ITE data” (Institute of Transportation Engi-
neers 2004). ITE acknowledges the limitations of the Trip Generation Handbook dataset in the omission
of transit, non-motorized transportation facilities, mixed-land uses, and density but currently offers little
guidance on how to address these shortcomings.
Planners are challenged to plan for the transportation impacts of infill, mixed use, and TODs with-
out adequate evidenced-based guidance. Many jurisdictions have to balance the warnings on the limited
applicability of the rates given in the ITE Trip Generation Handbook and the expense and effort involved
in collecting local data (Nelson\Nygaard Consulting Associates 2005, E. Lerner-Lam et al. 1992, Badoe
2000, Fleet and Sosslau 1976). Despite evidence that a more compact urban form, access to transit, and
a greater mix of uses generates fewer and shorter vehicle trips, local governments are often compelled
to use current ITE trip generation rates to evaluate transportation impacts and calculate transportation
system development charges. The expense of collecting local data, combined with the lack of alterna-
tive sources of information or empirically tested methods may lead to the application of the ITE’s Trip
Generation Handbook in areas outside the scope of its guidelines for use.
When analysts ignore the impacts of transit, pedestrian infrastructure, bicycle facilities, and urban
settings on vehicle trip generation, vehicle trips may be vastly overestimated. High vehicle trip estimates
increase the amount of vehicle-oriented development, necessitating other automobile priority measures.
More vehicle use, greater capacity, abundant parking supply, faster travel times, and fewer automobile
alternatives are all related to overestimating vehicle trip rates. Further, developers may be hesitant to lo-
cate new developments in infill or TOD areas because of the increased impact from fees that may result
from the overestimation of vehicle trips (Cervero and Arrington 2008, Ewing et al. 2011).
Alternatively, some communities rely on the application of adjustments to the ITE rates for differ-
ent contexts (e.g., a 10 percent reduction in ITE rates in areas with high-frequency transit service) (Vir-
ginia Department of Transportation 2010, City of San Diego 1998, City of Rockville 2011). However,
these reductions are sometimes derived without empirical basis and applied arbitrarily. In these cases,
“rule-of-thumb” adjustments may lead to the under-estimation of vehicle trips, which can lead to the
undersupply of automobile infrastructure, potentially resulting in inadequate parking, congestion and
delay. Thus, accurate estimation of vehicle trips for new development is important for both short-term
accommodation of transportation system users as well as planning for the maturity of urban develop-
ments over the long term.
To remedy this need, this paper presents a methodology to adjust the ITE Trip Generation Hand-
book rates to better reflect the relationship between land use, transportation, and travel demand for
specific land-use types located in various urban settings. The project collected data (using counts and
intercept surveys) on person and vehicle trip rates for a limited number of land-use types (restaurants,
24-hour convenience markets, and drinking places). These new trip rates were compared to the ITE
rates for the same land-use category and establishment size, and from this a set of models for adjusting
the ITE rates as a function of built environment attributes was estimated and verified using additional
data collected from establishments at other locations. Although developed from a limited number of
land-use types using data collected in the Portland, Oregon, metropolitan area, the findings punctuate
the need for adjustment of the ITE trip rates for infill and urban contexts; the analysis points to the key
built environment attributes that can be used to define context; and the models provide a replicable ap-
proach that can be expanded to other locations and other land-use types.
The remainder of this paper is organized as follows. The next section summarizes the state of
the practice in how communities are estimating vehicle trip rates for different urban contexts. This is
followed by a description of data collected in this study. The methods, the model estimation results,
Adjusting ITE’s Trip Generation Handbook for Urban Context 7
and verification are presented. Finally the implications for planning and practice are discussed in the
conclusions.
2 Literature review
The ITE Trip Generation Handbook, and accompanying Informational Report, is the most commonly
referenced and utilized practical guideline for predicting vehicle trip rates during the development pro-
cess. ITE also recommends using an approach developed by JHK and Associates et al. (1996) published
in the Trip Generation Handbook (Institute of Transportation Engineers 2004) that reduces vehicle trip
generation for locations in closer proximity to transit with supportive land uses, i.e., greater density,
higher floor-to-area ratios, and available pedestrian and cycling facilities. This report was published as
a draft and is only presented in the Trip Generation Handbook to provide guidance; it does not provie
reductions based on context. The ITE has also supported Gard’s approach (2007) for adjusting trip rates
for assessing the impacts of transit-oriented developments.
Several studies have already been performed to compare the ITE predicted vehicle trip rates to
observed trip rates. Existing research has examined the differences between observed trip rates and the
ITE predicted rates through comparison of the ITE predicted rates to observations at a particular type
of urban context or development. The studies reviewed compared the ITE predictions to sites in mixed-
use development, TOD, development near transit, urban-core areas, and suburban activity centers. To
evaluate these studies together, we have summarized the results of 13 studies performed between 1987
and 2011 (Cervero 1993, Cervero and Arrington 2008, Colorado/Wyoming ITE Section Technical
Committee–Trip Generation 1987, Dill 2008, Fehr and Peers 2008, Hooper 1989, Jeihani and Camilo
2009, Kimley-Horn and Associates 2009, Lapham 2001) in Table 1. The error reported within the table
indicates the mismatch between observations and the corresponding ITE trip generation estimate. A
positive number means observed vehicle traffic was greater than the ITE estimate, and a negative num-
ber means fewer vehicles were observed than estimated by the ITE method.
The greatest range of error in ITE’s estimation of vehicle trips occurred in central business district/
urban core/downtown areas, followed by mixed-use development. Error occurred both in over- and
under-estimating vehicle traffic for retail and residential uses. Dining, office, and service uses all had
actual vehicle trip rates below the ITE estimated rates.
The automobile mode share is provided in Table 1 for studies that counted person trips and calcu-
lated persons taking a vehicle. The central business district/urban core/downtown area shows the largest
range of automobile mode share. But sites in suburban activity centers and corridors contain a substan-
tial range: automobile mode shares were observed to be as small as 54 percent. Places with substantial
non-automobile mode shares further highlight the inapplicability of the ITE vehicle trip rate estimates
in urban contexts, as the ITE method accounts for vehicle travel and does not address person trip rates.
Many jurisdictions, however, have acknowledged the shortcomings and developed local adjustments to
the ITE estimated vehicle trips in their own transportation planning policies.
To summarize the state of the practice of the ITE rate adjustment, we reviewed 23 jurisdictional
guidelines for local adjustment from around the United States and Canada (Baltimore City Department
of Transportation 2007, Bedford County Department of Planning 2004, Charlotte Department of
Transportation 2006, City of Bellingham 2012, City of Bend 2009, City of Henderson 2009, City of
Los Angeles Department of Transportation 2010, City of Mississauga 2008, City of Pasadena 2005,
City of Rockville 2011, City of Salem 1995, City of San Diego 1998, City of Sedro-Woolley 2004, City
of Vancouver 2010, Georgia Regional Transportation Authority 2002, Harris County, Texas, 1991,
Montgomery Planning 2010, New York City, 2010, San Diego Municipal Code 2003, San Francisco
Planning Department 2002, Southern New Hampshire Planning Commission 2010, Flordia Depart-
8 JOURNAL OF TRANSPORT AND LAND USE 8.1
ment of Community Affairs 2006, Virginia Department of Transportation 2010). This review encom-
passed a wide range of cities from New York City to Bend, Ore., and identified current trends in estimat-
ing trip generation rates and conducting traffic impact assessments. Table 2 summarizes this review. Of
the jurisdictions, 22 identify the ITE Trip Generation Handbook rates and methods as appropriate for
their local contexts if local rates or studies are not available, and the 23rd made no recommendation. Six
jurisdictions provide local vehicle trip generation rates of some sort. Six jurisdictions have methods that
allow for bicycle, pedestrian, or transit adjustments to be applied to trip generation rates from mode
share information.
Table 1: Summary of ITE trip rate error findings of 13 studies.
Automobile
A.M. Peak P.M. Peak
Mode Share
Central business district/urban core/downtown -93% to +1109% -99% to +11% 8% to 100%
Eating/restaurant -93% to -57% -99% to -70% 17% to 57%
Office -80% to -22% -62% to -21% 56% to 95%
Residential -83% to +15% -80% to +11% 14% to 85%
Restaurant -35% -26% 34% to 60%
Retail -17% to +1109%* -22% to +8% 8% to 100%
Services -14% -66%
Shopping +30% +3%
Mixed-use development -109% to +181% -170% to +61%
Mixed -109% to +38% -80% to +61%
Town center -108% to +181% -170% to -35%
Transit-oriented development -90% to +20% -92% to +35% 50% to 96%
Office 50% to 96%
Residential -92% to +35% 53% to 93%
Development near transit -58% to +72% -36% to +51% 28% to 90%
Office 28% to 90%
Residential -58% to +72% -36% to +51% 33% to 82%
Suburban activity centers and corridors -37% to -5% 54% to 98%
Office -37% to -20%
Residential -5%
Shopping 54% to 98%
* This retail shop located in Oakland, California, had an observed a.m. peak trip count of 133 vehicle trips and
an ITE estimated count of 11 vehicle trips.
From the review of 13 studies comparing the ITE vehicle trip estimates to observed behavior and
the review of 23 jurisdictions with trip generation methodology adjustments, it is clear that (1) the ITE
Trip Generation Handbook does not provide accurate or consistent vehicle trip estimates across different
urban contexts; and (2) there is no consensus across the jurisdictions about how to quantify the adjust-
ment to the ITE trip rates for urban context.
Alternative approaches to estimating trip generation or adjusting the rates provided in the ITE Trip
Generation Handbook are available. In the United States, several models and methods have been devel-
oped to address various issues related to estimating vehicle trips or adjustments to the ITE methodolo-
gies. URBEMIS (Nelson\Nygaard Consulting Associates 2005) is a pivot model that regionally adjusts
the ITE rates at single-use sites based on built environment features. INDEX (Hagler Bailly Services and
Adjusting ITE’s Trip Generation Handbook for Urban Context 9
At multi-use developments, methods to better account for internal capture (Bochner et al. 2011)
and better estimate trip generation at mixed-use sites (San Diego Association of Governments 2010)
have been developed. However, the authors of a recent evaluation of available smart growth trip gen-
eration methodologies have acknowledged that “no clear winner emerges among currently available
methods” (Lee et al. 2011). Few of these methods address urban context directly in their formulation.
Beyond North America, similar methodologies to the ITE method exist to estimate vehicle trip
generation rates. The Austrailian-based system “New South Wales Roads and Traffic Authority” (New
South Wales Roads and Traffic Authority 2002) provides a dataset similar to the ITE’s, but when trip
rates for a particular land use are not available for Australia, the ITE handbook is a recommended op-
tion. The United Kingdom and Ireland (Trip Rate Information Computer System 2012) and New
Zealand (New Zealand Trips and Parking Database Bureau 2012) require multi-modal information in
the evaluation of the transportation impacts of every development, consider the location’s area type, and
continuously update data. Urban context has been acknowledged in these places as an important factor
10 JOURNAL OF TRANSPORT AND LAND USE 8.1
Data for this study were collected in 2011 from June through early October at 78 sites in the Portland
region. Data collection events occurred from 5 p.m. to 7 p.m. on Mondays, Tuesdays, Wednesdays,
and Thursdays to better capture evening visitors to bars and restaurants. This time frame includes one
hour of the weekday evening peak hour of the facility (5 p.m. to 6 p.m.) consistent with the ITE Trip
Generation Handbook. Because of the limited number of sites included in our study, the study controlled
for weather conditions by only collecting data on days with no precipitation over the five-month span
of data collection. This limitation in data collection has the potential to introduce bias due to a greater
number of non-automobile trips that might be taken when there is no precipitation. Typically summer-
time data collection is usually avoided in transportation studies because of changes in travel behavior
routines that occur with vacation and children being out of school. On the other hand, restaurant and
bar total overall trip rates may actually be greater during this season. The ITE currently does not account
for weather or seasonal variation in the data it compiles, so it is difficult to compare the impact of our
study limitations with the ITE’s data.
Land-use types (the ITE’s land use code shown in parentheses) included in this study are: (a) high-turn-
over (sit-down) restaurants (LU 932), (b) convenience markets (open 24-hours) without gas stations
(LU 851) and (c) drinking places (LU 925). These land uses were selected because they can be found
Adjusting ITE’s Trip Generation Handbook for Urban Context 11
in all area types including places where vehicle trip overestimation has proven most problematic: urban
infill, mixed-use, and TOD areas.
Establishments were selected based on characteristics of their surrounding built environment. To
ensure representation of the establishments located in a broad spectrum of urban environments found
in the Portland region, a sample frame was developed from a complete census of business establishments
of these land use types from the 2010 referenceUSA database (Infogroup 2011) covering these land-
use types over the Portland metropolitan region. For each business establishment in this sample frame,
the following built environment characteristics were collected using archived spatial information at a
0.8-kilometer (0.5-mile), straight-line buffer around the business location: intersection density, aver-
age block size, percent of dwellings that are single-family detached, percent of retail employment, and
percent of lot building coverage. A k-means clustering analysis of these built environment data over the
sample frame resulted in five classifications of area-types that roughly corresponded to: central business
district neighborhoods; urban core neighborhoods; neighborhood and regional centers; suburban town
centers and corridors; and suburban areas. These area types were only used to segment the sample frame
of establishments and ensure businesses were recruited from each area type. The cluster types were not
used in the subsequent statistical analysis or estimation of adjustment models.
This sample frame segmented by area type was used to recruit business establishments to partici-
pate in the study. Business managers were contacted by mail and follow-up telephone call. The result-
ing sample of 78 establishments is shown in Figure 1. Establishments in more dense and mixed-use
areas were over-sampled to ensure sufficient representation in the dataset, as we hypothesized that these
locations were likely to have fewer vehicle trips. Information on business square footage was collected
directly from business managers where possible, and using regional building information data (RLIS) or
estimating using the building footprint (for single-story establishments) from Google Earth.
Most establishments in the study are regionally owned and operated franchises. Local establish-
ments were more willing to participate in the study than national corporate franchises. As such, most
were under 3,000-square-feet gross floor area and may cater to a different market segment than patrons
of national chains. This is one limitation of the study; on the other hand, these are more typical of the
size and type of establishments that may choose to locate in infill developments.
Visitor intercept surveys were administered to visitors as they left the establishment or site. A five-minute
survey was administered using an electric handheld tablet (Clifton et al. 2012). A shorter survey was
administered for those initially refusing the tablet survey. Key information collected from both survey
instruments included travel mode(s), vehicle occupancy and home location. An average of 24 surveys
was collected at each of the 78 establishments, for a total of 1884 surveys. The overall response rate was
52 percent for all surveys. More detail on sample size is provided in Table 3.
Demographic characteristics of survey respondents are compared to data from the 2010 US Decen-
nial Census and the American Community Survey for the Portland Metropolitan Statistical Area in
Table 4. From this comparison, the respondents of the survey appear to be similar to the area population
characteristics of household income, vehicle ownership and household size.
Table 3: Establishments surveyed by area and land-use type, survey sample size.
High-Turnover Convernience
Area Type (Sit-down) Markets Drinking Place Total
Restaurants (Open 24-hours)
Number of establishments 39 26 13 78
Central business district 12 4 3 19
Urban core neighborhoods 10 5 6 21
Neighborhood and regional centers 6 6 4 16
Suburban town centers 5 7 0 12
Suburban areas 6 4 0 10
Visitor survey sample size
Long survey (N) 309 281 107 697
Short survey (N) 369 710 108 1187
Response rates
Long survey 24% 14% 30% 19%
Short survey 52% 61% 50% 52%
2010 Census/ACS
Variable Survey Observed*
Portland (MSA)
Median household income per year $50,000 - $99,000 $55,618
Average household income per year $50,000 - $99,000 $72,200
Median age 25-34 36
Male respondents 57% 49%
Average # vehicles per household 1.6 1.7
Average # bicycles per household 1.7 NA
Average # transit passes per household 0.5 NA
Average # adults per household 2.2 NA
Percentage of households with children 29% 33%
Average household size 2.5 2.5
*Note: Demographic data if from the long survey only (N = 697).
Adjusting ITE’s Trip Generation Handbook for Urban Context 13
At every entrance to the various business establishments in the sample, persons were counted entering
and exiting during the survey period using direct observations, and the resulting average person counts
by land-use type are shown in Figure 2. Person count information is crucial when expanding methods
for transportation impact analysis and trip generation studies to account for all modes of travel.
Likewise, counts of vehicles exiting the site were also collected when feasible, typically when the site had
parking adjacent to the store entrance. Average vehicle counts by land-use type are shown in Figure 3.
Vehicles were not typically counted at sites located in urban areas when adjacent parking lots were not
present or at sites with one shared parking lot for several establishments. It was beyond the resources
of the study to track or follow people to their vehicles upon leaving the survey establishment solely to
count vehicle trips.
Of the 78 sites in the study, vehicle counts were obtained for 44 sites. To account for the remaining
34 sites, the observed mode share, the observed vehicle occupancy rate, the total entering and exiting
person counts, and the size of the establishment were used to calculate an estimated vehicle count for
every study site 2,3. We then compared the estimated vehicle trip rate to the observed rate for the 44
locations with both a vehicle count and a vehicle estimate using Equation 1. The overall ratio was 1.02
observed vehicles for every 1.00 estimated vehicles, a very slight underestimate on average.
To test a variety of measures associated with urban context, built environment information for each site
was gathered from archived data sources within a buffer area of 0.8-kilometer (0.5-mile) radius (Euclide-
an distance) from each establishment. The measures that were included in this study were selected based
on their prominence in the travel behavior and built environment literature and their relative availability
in locations throughout the United States in order to reproduce this method elsewhere. The built en-
vironment measures used in this study are shown in Table 5 and include: number of transit corridors,
activity density (residential and employment), number of high-frequency transit stops, employment
density, average lot coverage, extent of bicycle facilities, retail and service employment density index,
access to rail transit, and intersection density. The average and range of values for the 78 study sites are
also shown, and our analysis is limited to the range of characteristics of the observations in our sample,
and thus the subsequent findings may not be valid for locations with values exceeding those in Table 5.
Most of these measures are straightforward calculations from archived spatial data and commonly
used in the travel behavior literature. The retail and service employment index is a measure of local access
to retail and service destinations. This index is based on the different retail and service establishments
that accommodate everyday non-work activities, e.g., food or clothing stores, restaurants, laundry ser-
vices, supply stores and bookstores. The region is divided into raster cells 264 ft. x 264 ft. For each raster
cell, the index is computed by calculating the number of retail and service establishments that are located
within a 0.8-kilometer (0.5-mile) radius using a kernel density function, and the values are classified into
categories using a 1-to-5 scale using a Jenks’ natural breaks algorithm. In this analysis, the average index
value across all of the raster cells within a 0.8-kilometer (0.5-mile) radius of the establishment is used.
This Retail and Service Employment Index was developed by Metro and based on disaggregate and
confidential information on individual business establishments. Absent access to these discrete data, we
relied on this index in the development of our own measures. Because the discrete, ordinal nature of the
Retail and Service Employment Index does not capture the variations in the number of businesses across
space, we take an average at the 0.8-kilometer (0.5-mile) buffer area to better represent this relative varia-
tion. If discrete data on business establishments were available, a more simple and direct representation
of the built environment could be computed.
Adjusting ITE’s Trip Generation Handbook for Urban Context 15
These built environment measures are highly correlated (see Table 6) and thus cannot be used reli-
ably together in a statistical model in this disaggregate form. However, the authors do recognize that
these various dimensions of the built environment work together to define urban context and influence
travel choice.
Table 5: Built environment measures and summary statistics.
Number of
stops within Bus Stop layer
0.8 kilometer (RLIS, 2010)
C) Number of High-Frequency Transit Stops 47 0 to 244
(0.5 mile) with and TriMet
headways under schedules (2011)
15 minutes
Employees per
ESRI Business
D) Employment Density 4047 square 21 0.4 to 141
Analyst (2010)
meters (1 acre)
Tax lot and
E) Lot Coverage Percent Building Layers 28% 9% to 67%
(RLIS, 2010)
Bike Route layer
F) Length of Bike Facilities Miles 5.4 0.2 to 11.0
(RLIS, 2010)
Density index
based on the
number of retail
Metro Context
and service
G) Retail and Service Employment Index Tool, Portland 2.1 1.0 to 4.2
establishments
Metro
within 0.8
kilometer (0.5
mile)
Presence of rail
Light-rail Stop
station within
H) Access to Rail ** layer (RLIS, 45% No to Yes
0.8 kilometer
2010)
(0.5 mile)
Intersections
Lines file
I) Intersection Density per 4047 square 0.22 0.01 to 0.56
(TIGER 2009)
meters (1 acre)
* RLIS is the Regional Land Information System, Portland Metro; TriMet is the regional transit agency.
** Binary measure indicating presence of rail within the 0.8-kilometer (0.5-mile) Euclidian buffer.
16 JOURNAL OF TRANSPORT AND LAND USE 8.1
Number
A of Transit
Corridors
B Person Density 0.934
Number of
C High-Frequency 0.941 0.956
Transit Stops
Employment
D 0.933 0.988 0.939
Density
E Lot Coverage 0.749 0.851 0.831 0.819
Length of Bike
F 0.799 0.846 0.804 0.810 0.816
Facilities
Retail and
Service
G 0.781 0.887 0.844 0.837 0.921 0.862
Employment
Index
H Access to Rail 0.571 0.531 0.473 0.539 0.396 0.542 0.471
Intersection
I 0.622 0.727 0.753 0.683 0.899 0.777 0.828 0.301
Density
4 Analysis
The aim here is to explore the relationship between urban context, as captured by a variety of built
environment measures, and vehicle trip generation with a larger goal of developing a consistent and reli-
able method for adjusting the ITE trip generation estimates to control for urban context. To this end,
this section describes our key assumptions, the data analysis and methodological development of these
models.
A critical assumption in this study is that person trip rates for a specific establishment type (land-use
category) and size (gross floor square footage or similar measure) are similar across urban contexts. Thus,
it is the distribution of those person trip rates across various modes of transportation that varies by con-
text. Figure 4 provides an illustrated example. If this hypothesis holds true, it suggests that automobile
and non-automobile trips may be substitutes (person trip rates are constant) rather than complements
(non-automobile trips may be additional trips). If non-automobile trips are complementary (vary across
contexts), the ability to compare the ITE vehicle trip rates with data collected here proves difficult be-
cause the ITE does not collect information on person trip rates or the use of non-automobile modes.
In the case of complementarity, the error between observed and estimated vehicle trip rates cannot be
distinguished from non-automobile trip rates.
Adjusting ITE’s Trip Generation Handbook for Urban Context 17
We test this assumption using our person trip data collected in this study. The average person trip
rate (trips per square foot of gross floor area) from the p.m. peak hour (5 to 6 p.m.) across land-use types
is tested for significant variance across contexts. Statistical analysis of variance (ANOVA) is performed
for: (1) all land uses combined across contexts (pooled data) and (2) specific land-use types across con-
texts (data segmented by establishment type). The null hypothesis (H0) states that average person trip
rates are equal across contexts, and the alternate hypothesis (H1) states that average person trip rates
are not equal across contexts. Hypothesis testing is performed by one-way analysis of variance statistical
means testing at 95 percent confidence. In every case, we fail to reject the null hypothesis4, meaning
the average person trip rates per building area are not significantly different across urban contexts. This
result suggests that our assumption appears to be applicable and person trips do not vary significantly
for establishments of a specific size and type.
4.2 Comparison of study vehicle trips tates with the ITE trip rates
This section details a comparison between vehicle trips based on study observations and the ITE Trip
Generation Handbook estimates of vehicle trips for the establishments included in this study. To do so,
we calculate vehicle trip rates (trips per 1000 sq. ft.) for each establishment in our study using the ob-
served person trip counts and the mode share and average vehicle occupancy at each establishment based
upon the survey data (see section 3.3).
These comparisons of vehicle trips for each of the land-use types used in our study (restaurants,
drinking places and convenience markets) for the weekday peak hour of the facility (5 to 6 p.m.) are
shown in Figure 5, Figure 6, and Figure 7. As shown in Figure 5 and Figure 6, the vehicle trips from this
study are consistently below the ITE rates and the ITE data points for convenience stores and drink-
ing establishments. For high-turnover (sit-down) restaurants, the vehicle trips from this study and the
4
Results from ANOVA tests: (1) pooled data, F(4,73) = 0.62, p = 0.652; (2) convenience stores, F(4,21) = 1.86, p = 0.155,
restaurants F(4,34) = 1.97, p = 0.121, drinking establishments, F(2,10) = 1.98, p = 0.189., restaurants and drinking establish-
ments combined, F(4,47) = 1.07, p = 0.382.
18 JOURNAL OF TRANSPORT AND LAND USE 8.1
estimates of the ITE vehicle trips are similar as shown in Figure 7. Table 7 shows a comparison of these
vehicle trip rates for all three land uses.
The ITE has criteria for adoption of the ITE Trip Generation Handbook methodology for local
use and these are shown in Table 8. All conditions must be met to consider application of the ITE
Trip Generation Handbook data for a local context. If not, the development of a local rate or equation
is recommended (Institute of Transportation Engineers 2004). Based on these criteria and the results
presented in Table 7, we recommend a local adjustment to the ITE Trip Generation Handbook rates
for convenience stores and drinking establishments. However, we do not have sufficient evidence to
show that the ITE rates for high-turnover (sit-down) restaurants are inadequate for the Portland region.
Nonetheless, because the establishment sample size is too small to estimate segmented models for each
land use, we include restaurants to increase the sample size used in our estimation of models to adjust
the ITE’s trip rates.
We hypothesize that the differences between these trip rates are largely due to differences in the
travel modes visitors use to access/egress these sites, which is associated with the urban built environment
characteristics where the establishment is located. As discussed in the previous section, this is supported
by the fact that person trip rates are similar across area types. This points to the need to adjust the ITE
rates for urban context, and the next section presents our approach to doing so.
Figure 5: Convenience market (open 24-hours) (LU 851): Weekday, peak hour of adjacent street traffic, 4-6 p.m.
Adjusting ITE’s Trip Generation Handbook for Urban Context 19
Figure 6: Drinking place (LU 925): weekday, peak hour of adjacent street rraffic, 4-6 p.m.
Figure 7: High-turnover (sit-down) restaurant (LU 932): Weekday, peak hour of adjacent street traffic, 4-6 p.m.
20 JOURNAL OF TRANSPORT AND LAND USE 8.1
Table 8: The ITE criteria for using the ITE trip generation methods and data.
Using the pooled-sample data collected for all establishments for the weekday p.m. peak hour (5 to 6
p.m.), nine separate adjustment models are estimated using ordinary least-squared multivariate regres-
sion to provide adjustments to the ITE’s vehicle trip rates for urban context. The dependent variable for
each of the models is the difference between the vehicle trip rate found in this study and the trip rate
estimated using the ITE’s methods (Institute of Transportation Engineers 2008, Institute of Transporta-
tion Engineers 2004). In most cases in this study, the vehicle trip rate given by the ITE is greater than
that calculated based on our study observations, resulting in a negative value for the adjustment. In each
of the nine models shown in Table 9 (numbered A through I), the independent variables include indica-
tors of the land-use type (drinking establishments are the base case) and one of the built environment
measures listed in Table 5. Despite the statistical caution of using highly correlated independent vari-
ables together, several models were estimated with combinations of these correlated built environment
variables, but they did not yield significant improvement in model fit and were not more telling in terms
of policy than those models employing a single built environment variable. The model specification is
shown in Equation 2 below.
Where,
Adjustment= Vehicle Trip Ratestudy–Vehicle Trip RateITE ;
DummyRestaurant = Indicator if the land-use type is a restaurant;
DummyConvenience= Indicator if the land-use type is a convenience store;
Built EnvironmentA to I= One of the built environment measures (A-I) shown in Table 5.
From the estimation results shown in Table 9, all of these models perform quite well with adjusted
R2 ranging from 0.76 to 0.77. It is obvious from the parameter estimates that the land-use indicators
contribute more to the adjustment than the built environment variables that represent urban context.
This is expected since we had insufficient sample size to estimate separate models for each land-use type,
and one would expect different land uses to have different trip generation rates.
However, once land use is controlled for, significant differences in trip generation rates can be at-
tributed to the various context variables. In all models the built environment measures are negatively
and significantly associated with the adjustment, although with varying levels. The negative coefficient
indicates that the urban context measures are inversely correlated with the adjustment. This means that
locations that have greater density, transit access, bicycle network, lot coverage and pedestrian connectiv-
ity are associated with vehicle trip rates that are lower than those predicted by the ITE’s methodology.
This makes sense given that one would expect greater non-automobile mode shares in locations with
these built environment measures and thus require a reduction in the ITE vehicle trip rates. Interactions
between the built environment and land-use type dummy variables were not significant and, therefore,
were not included.
In models A through C, the built environment variables (number of transit corridors, activity
density, and number of high-frequency bus routes) have greater significance with a p-value ≤ 0.01,
compared to a p-value ≤ 0.05 for the built environment variables in models D through I. Since each
of these built environment variables has a different operational construct (units of measurement), ex-
amining their standardized coefficients is telling. Models A through C also have the built environment
22 JOURNAL OF TRANSPORT AND LAND USE 8.1
variables with the largest standardized coefficients with number of transit corridors having the largest
(-0.153), followed by number of high-frequency bus routes (-0.150) and activity density (-0.149). The
other models (D-I) have similar model fits and significance levels, but there are some distinct differences
from the standardized coefficients. Employment density has the greatest explanatory power of these
remaining built environment variables with a standardized coefficient of -0.143 and the binary variable
indicating access to rail transit within the 0.8-kilometer (0.5-mile) buffer has the lowest with a standard-
ized coefficient of -0.116.
In terms of application, the results of this estimation alone do not provide sufficient direction to
guide the use of these models or for the direction of future research needed to further inform models. To
aid in this and better understand the performance of these models, we verified these model estimation
results using independently collected data. We use the word “verification” instead of “validation” due to
sample sizes required to warrant statistical merit. Because of segmentations across different land uses and
urban contexts, the verification sample would need to be increased fivefold to meet statistical require-
ments for a validation effort. Vehicle count data were collected from 34 additional establishments (10
convenience stores, 12 drinking places, and 12 restaurants) with varying built environment attributes
in April and May of 2012, and the summary statistics for these locations are shown in Table 10. For
these sites, vehicle counts entering and exiting the locations were collected using the same methodol-
ogy outlined above. Using each of the estimated models A through I from Table 9, the adjustment was
calculated. Then this adjustment was applied to the ITE vehicle trip rate for each establishment, and
the new adjusted rate was used to estimate the number of vehicle trips for that location for the evening
weekday peak hour. This model application was compared to the observed vehicle trip data collected for
each of the sites. The results of this verification exercise aggregated by land-use type are shown in Table
11, Table 12 and Table 13.
From these tables, we can see that the models developed and verified here for convenience markets
and drinking places predict vehicle trips better overall than the ITE methods alone. From Table 11, the
mean squared error (MSE) for the estimates developed from data and methods outlined in the ITE’s
Trip Generation Handbook is 1121 for convenience markets, compared to a MSE range of 36-45 for the
context adjustment models developed here. In addition, the ITE overestimated vehicle trips in every
case (N=10) with an average percentage error of 195 percent. From the verification results for drinking
places shown in Table 12, the MSE for the ITE’s estimates is 30, compared to a range of 8 to 24 for the
urban context adjustment models. For 11 of the 12 bars in the verification sample, the ITE estimates of
vehicle trips were greater than the observed verification counts with an average percentage error of 129
percent. However, the verification results for restaurants are less supportive of the context adjustment
approaches developed in this paper. From the results in Table 13, the ITE estimates appear to be com-
mensurate with our adjustment model performance. The ITE MSE is 40, compared to a MSE range of
33 to 45 for the adjustment models. This suggests that another approach may be necessary to adjust the
ITE estimates or that the ITE method may be adequate for application in the evaluation of trip genera-
tion attributed to restaurants across urban contexts.
Adjusting ITE’s Trip Generation Handbook for Urban Context 23
Convenience
Built Environment Measure Adjusted Intercept Restaurant Standardized
Model Market Coefficient
(units) R2 Coefficient Coefficient Coefficient
Coefficient
Equation 2 Coefficients β0 β1 β2 β3 ---
Number of transit corridors
A 0.77 -4.31 * -25.48 *** 7.62 *** -0.09 *** -0.153
(count)
Activity density (residents
B and employees per 4047 0.77 -3.41 -26.19 *** 7.24 *** -0.07 *** -0.149
square meters (1 acre))
Number of high-frequency
C 0.77 -3.62 -26.07 *** 7.19 *** -0.05 *** -0.150
bus routes (count)
Employment density
D (employees per 4047 square 0.76 -4.24 * -26.13 *** 7.15 *** -0.08 ** -0.143
meters (1 acre))
E Lot coverage (%) 0.76 -0.86 -26.60 *** 6.97 ** -0.17 ** -0.131
Length of bike facilities
F 0.76 -0.75 -26.24 *** 7.55 *** -0.79 ** -0.131
(miles)
Retail and service employ-
G 0.76 0.64 -26.04 *** 7.41 *** -3.29 ** -0.141
ment index (count)
H Rail access (binary) 0.76 -5.19 ** -24.31 *** 8.09 *** -3.99 ** -0.116
Intersection density (number
I per 4047 square meters (1 0.76 -2.20 -26.85 *** 6.47 ** -14.35 ** -0.117
acre))
NOTE: N = 78
***p-value ≤ 0.01
** p-value ≤ 0.05
*p-value ≤ 0.1
Table 11: Verification of adjustment models for convenience markets (ITE land use: 851).
Table 12: Verification of adjustment models: Drinking places (ITE land use: 925).
5
Percentage error is defined as the percent ratio of the estimate error (estimate minus observed) to the observed value.
Adjusting ITE’s Trip Generation Handbook for Urban Context 25
Table 13: Verification of adjustment models: High-turnover (sit-down) restaurants (ITE land use: 932).
While this study tested a limited number of land uses in one metropolitan region, it confirms the need
for amendments to the long-term industry standards provided in the ITE’s Trip Generation Handbook.
The results presented here demonstrate that the built environment plays an important role in the trip
generation characteristics of developments and reinforces the need to consider urban context in the esti-
mation of traffic impacts for new development. There is a consistent trend: For all land uses tested here,
vehicle trip rates tend to decrease as the context becomes more urban. This evidence strongly supports
the immediate need for an urban context adjustment to the vehicle trip rates derived from the ITE’s Trip
Generation Handbook for use in the near term.
This research fills this short-term need by providing a means to adjust the vehicle trip rates derived
from the ITE’s Trip Generation Handbook for the urban context of new development. The models esti-
mated and verified in this study are simple, straightforward and consistent. As shown in Table 9, all of
the measures of the built environment performed well in the models, and while there were differences
across models and land uses, the verification results suggest that these built environment characteristics
provide estimates of vehicle trip generation that are as good or better than those provided by the ITE’s
Trip Generation Handbook. They are based on a variety of built environment measures that represent
urban context and have been shown in previous research to influence trip making and mode choice.
These built environment measures are not unique to the study location. They can be easily computed
from readily available archived spatial information from communities in other locations throughout the
United States. Although this work is limited to only a few land-use types, the approach outlined here can
be further verified or validated in other communities around the country and the approach expanded
to include additional land-use types.
One drawback to the approach developed here is the inability to examine the impacts of more than
one built environment characteristic and capture the complexity of the urban environment. As men-
tioned earlier, these built environment measures tend to be highly correlated and thus prohibited from
use together as independent variables in models. One way to mitigate this approach is to define context
through the creation of an index or some other distillation technique such as factor analysis. Although
the creation of an urban context index would solve the econometric problem, it has two barriers to its
26 JOURNAL OF TRANSPORT AND LAND USE 8.1
widespread use. First, it lacks policy sensitivity in interpretation as the individual built environment
measures, such as activity density and pedestrian connectivity, get combined into one. Second, the use
of an index is limited to the area where it is created. Unless an index is created using national-level data,
it would be difficult to transfer it to another urban area or generalize its relation to trip generation. Thus,
the choice between using individual elements of the built environment and a combination is not an easy
one when the goal is to have a sound method that can be applied anywhere.
The various shortcomings to the ITE’s Trip Generation Handbook as a planning reference, includ-
ing insensitivity to urban context, have been well documented in the previous literature. The work here
provides more empirical evidence to support those critiques. Given the need for methodologies backed
by empirical evidence that provide planning support for the automobile as well as non-motorized and
transit modes in urban environments, a complete overhaul to the methods and data used in the Trip
Generation Handbook or the development of an altogether new approach for assessing the transporta-
tion impacts of new development is necessary over the long term. Any new methods should consider
the impacts of motorized and non-motorized modes for all person trips and would require an entirely
different data collection agenda (see Clifton et al. 2013). Specifically, the current approach to methods
and data provided by the ITE would be improved by a movement away from a focus on vehicle trips
and expanded to consider person-trip information and multi-modal travel. Traffic impact analyses can
be important and powerful planning tools, but only if they reflect the multi-modal nature of urban
environments. The analysis should provide a basis for how these person trips are distributed across the
various modes, as a function of site and urban context characteristics.
This study represents a first step in moving this bar forward and advancing national standards.
Data for more land uses and covering a wider range of urban contexts are needed to inform a nationally
relevant methodology. But many communities across the country already have a great deal of informa-
tion from their own local trip generation studies to inform a larger-scale study and validate available
methodologies for regional and urban context variations. The opportunity exists to make these data
more readily available to researchers to help improve practice and create new professional standards that
better reflect the multi-modal nature of our cities.
Acknowledgments
The authors wish to acknowledge Metro, the regional government for the Portland, Oregon, metropoli-
tan area, and the Regional Partners; the Oregon Transportation Research and Education Consortium;
and the Oregon Department of Transportation for their support of this research.
References
Badoe, D. 2000. Transportation land-use interaction: Empirical findings in North America, and their
implications for modeling. Transportation Research. Part A: Policy and Practice 2000: 263.
Baltimore City Department of Transportation. 2007. Procedures and Requirements for Conducting a Traf-
fic Impact Study in Baltimore City. Baltimore, MD: Baltimore City Department of Transportation.
Bedford County Department of Planning. 2004. Bedford County Traffic Impact Study (TIS) Guidelines.
Bedford, VA: Bedford County Department of Planning.
Bochner, B. S., K. Hooper, B. Sperry, and R. Dunphy. 2011. NCHRP Report 684: Enhancing Internal
Trip Capture Estimation for Mixed-Use Developments. Washington, DC: Transportation Research
Board.
Adjusting ITE’s Trip Generation Handbook for Urban Context 27
Cervero, R. 1993. Ridership impacts of transit-focused development in California (UCTC No. 76).
Working paper. Berkely, CA: The University of California Transportation Center.
Cervero, R., and G. B. Arrington. 2008. TCRP Report 128: Effects of TOD on Housing, Parking, and
Travel. Washington, DC: Transportation Research Board.
Cervero, R., and K. Kockelman. 1997. Travel demand and the 3Ds: Density, diversity, and design.
Transportation Research Part D 2 1997(3): 199–219.
Charlotte Department of Transportation. 2006. Land Development Rezoning and Traffic Impact Study
Review Process. Charlotte, NC: Charlotte Department of Transportation.
City of Bellingham. 2012a. Development Guidelines and Improvement Standards: Section 11-2.02 Traffic
Studies: Project Trip Generation and Design-Hour Volumes. Bellingham, WA: City of Bellingham.
City of Bellingham. 2012b. Development Guidelines and Improvement Standards: Section 11-2.13 Traffic
Studies: Techical Requirements of Final Report. Bellingham, WA: City of Bellingham.
City of Bend. 2009. Chapter 10-10, development code, in Bend Code. Bend, OR: City of Bend.
City of Mississauga. 2008. Traffic Impact Study Guidelines. Mississauga, Ont., Canada: City of Missis-
sauga.
City of Pasadena. 2005. Transportation Impact Review Current Practice and Guidelines. Pasadena, CA:
City of Pasadena.
City of Rockville. 2011. Comprehensive Transportation Review. Rockville, MD: City of Rockville.
City of Salem. 1995. Development Bulletin 19: Transportation Impact Analysis. Salem, OR: City of Salem.
City of San Diego. 1998. Traffic Impact Study Manual. San Diego, CA: City of San Diego.
City of Sedro-Woolley. 2004. Traffic Impact Study Guidelines. Sedro-Woolley, WA: City of Sedro-Wool-
ley.
City of Vancouver. 2010. City of Vancouver Traffic Study Requirements. Vancouver, WA: City of Vancou-
ver.
Clifton, K. J., K. M. Currans, and C. D. Muhs. 2012. Contextual Influences on Trip Generation, Project
Report OTREC RR-12-13 prepared for Oregon Transport Research and Education Consortium,
Portland, OR. http://otrec.us/project/407.
Clifton, K. J., K. M. Currans, and C. D. Muhs. 2013. Evolving the Institute for Transportation Engi-
neers’ (ITE) Trip Generation Handbook: A proposal for collecting multimodal, multi-context, estab-
lishment-level data. Paper presented at the 92nd Annual Meeting of the Transportation Research
Board, Washington, D.C.
Colorado/Wyoming ITE Section Technical Committee–Trip Generation. 1987. Trip generation for
mixed-use developments. Institute of Transportation Engineers Journal 57(2): 27–32.
Daisa, J. M., A. Mustafa, M. Mizuta, L. Schwartz, L. Espelet, D. Turlik, G. Bregman, G. 2009. Trip
Generation Rates for Urban Infill Land Uses in California: Phase II Final Report. Sacramento, CA:
California Department of Transportation.
Dill, J. 2003. Transit use and proximity to rail: Results from large employment sites in the San Francisco,
California, Bay Area. Transportation Research Record 2003: 19–24.
Dill, J. 2008. Transit use at transit-oriented developments in Portland, Oregon, area. Transportation
Research Record 2063: 159–167.
Ewing, R., M. Greenwald, M. Zhang, J. Walters, M. Feldman, R. Cervero, L. Frank, and J. Thomas.
2011.Traffic generated by mixed-use developments—Six-region study using consistent built envi-
ronmental measures. Journal of Urban Planning and Development 137(3): 248–261. doi:10.1061/
(ASCE)UP.1943-5444.0000068.
Ewing, R., and R. Cervero. 2010. Travel and the built environment: A meta-analysis. Journal of the
American Planning Association 76(3): 265–294.
28 JOURNAL OF TRANSPORT AND LAND USE 8.1
Ewing, R., and R. Cervero. 2001. Travel and the built environment: A synthesis. Transportation Research
Record 1780 (2001): 87–114.
Fehr and Peers. 2008. Summary of research demonstrating that smart growth reduces traffic. Internal,
referenced in Samdahl, D. Travel Demand Research for Downtown Kent. Kent, WA: Fehr and Peers,
Kent.
Fleet, C., and A. Sosslau. 1976. Trip generation procedures: An improved design for today’s needs. In-
stitute of Transportation Engineering Journal 46(11).
Flordia Department of Community Affairs. 2006. Transportation Concurrency Requirements and Best
Practices: Guidelines for Developing and Maintaining an Effective Transportation Concurrency Manage-
ment System. Tallahassee, FL: Florida Department of Community Affairs.
Gard, J. 2007. Innovative intermodal solutions for urban transportation paper award: Quantifying
transit-oriented develpment’s ability to change travel behavior. Institute of Transportation Engineers
Journal 80(11): 42–46.
Georgia Regional Tranpsortation Authority. 2002. GRTA DRI Review Package Technical Guidelines. At-
lanta, GA: GRTA.
Hagler Bailly Services, Inc. and Criterion Planners/Engineers. 1999. The Transportation and Environ-
mental Impacts of Infill Versus Greenfield Development: A Comparative Case Study Analysis. EPA Pub.
Number 231-R-99-005. Washington, DC: Enviornmental Protection Agency.
Harris County, Texas. 1991. Chapter 5, Article II. In Land Development Code: Traffic Impact Analysis
Guidelines. Harris County, TX: Harris County, TX.
Henderson Department of Public Works. 2009. Traffic Impact Analysis Guidelines. Henderson, NV:
Henderson Department of Public Works.
Hooper, K. G. 1989. NCHRP Report 323: Travel Characteristics at Large-Scale Suburban Activity Centers.
Washington, DC: Transportation Research Board, National Research Council.
Infogroup, Inc. 2011. 2010 reference USA database. U.S. Businesses.
Institute of Transportation Engineers. 2008. Trip Generation 8th Edition: An Informational Report.
Washington, DC: Institute of Transportation Engineers.
Institute of Transportation Engineers. 2004. Trip Generation Handbook: An ITE Recommended Practice,
2nd Ed. Washington, DC: Institute of Transportation Engineers.
Jeihani, M., and R. A. Camilo. 2009. Trip Generation Studies for Special Generators (MD-09-SP808B4J).
Hanover, MD: State Highway Administration, Maryland Department of Transportation.
JHK and Associates, Pacific Rim Resources, and SG Associates.1996. DRAFT Final Report: Accessibil-
ity Measure and Transportation Impact Factor Study. Salem, OR: Oregon Department of Transpor-
taton, Department of Land Conservtion and Development, Transportation and Growth Manage-
ment Program.
Kimley-Horn and Associates, Inc. 2009. Trip Generation Rates for Urban Infill Land Uses in California:
Phase II Final Report. Sacramento, CA: California Department of Transportation.
Lapham, M. 2001. Transit-Oriented Development—Trip Generation and Mode Split in the Portland Met-
ropolitan Region. Portland, OR: Portland State University.
Lee, R., J. Miller, R. Maiss, M. Campbell, K. Shafizadeh, D. Niemeier, and S. Handy. 2011. Evaluation
of Operation and Accuracy of Five Available Smart Growth Trip Generation Methodologies (UCD-ITS-
RR-11-12). Davis, CA: Institute of Transportation Studies, University of California.
Lerner-Lam, E., S. P. Celniker, G. W. Halbert, C. Chellman, and S. Ryan. 1992. Neo-Traditional Neigh-
borhood Design and Its Implications for Traffic Engineering. Institute of Transportation Engineering Jour-
nal 62(1).
Adjusting ITE’s Trip Generation Handbook for Urban Context 29
Los Angeles Department of Transportation. 2010. Traffic Study Policies and Procedures. Los Angeles, CA:
Los Angeles Department of Transportation.
Lund, H. M., R. Cervero, and R. W. Willson. 2004. Travel Characteristics of Transit-Oriented Develop-
ment in California. Statewide Planning Studies FTA Section 5313(b).
Montgomery Planning. 2010. Local Area Transportation Review. Montgomery County, MD: The Mary-
land National Capital Park and Planning Commission.
Nelson\Nygaard Consulting Associates. 2005. Crediting Low-Traffic Developments: Adjusting Site-Level
Vehicle Trip Generation Using URBEMIS. San Francisco, CA: Nelson\Nygaard Consulting Associ-
ates.
New South Wales Roads and Traffic Authority. 2002. Guide to Traffic Generation Developments, Version
2.2. Sydney, Austrailia: Roads and Traffic Authority.
New York City. 2010. City Environmental Quality Review (CEQR) Chapter 16. New York: Mayor’s Of-
fice of Environmental Coordination.
New Zealand Trips and Parking Database Bureau. 2012. URL: www.tdbonline.org.
Samdahl, D. 2010. Travel Demand Research for Downtown Kent. Memorandum submitted to the
City of Kent, WA, June 28, 2010. Kent, WA: Fehr and Peers.
San Diego Association of Governments (SANDAG). 2010. Trip Generation for Smart Growth: Planning
Tools for the San Diego Region. San Diego, CA: SANDAG.
San Diego Municipal Code. 2003. Land Development Code: Trip General Manual. San Diego, CA: City
of San Diego.
San Francisco Planning Department. 2002. Transportation Impact Analysis Guildelines for Environmental
Review. San Francisco, CA: City and County of San Francisco.
Schneider, R. J. 2011. Understanding Sustainable Transportation Choices: Shifting Routine Automobile
Travel to Walking and Bicycling (UCTC-DISS-2011-01). Berkeley, CA: University of California
Transportation Center.
Seskin, S. N., R. Cervero, and J. Zupan. 1996. National Cooperative Highway Research Program (NCHRP)
Report No. 16: Transit and Urban Form. Washington, DC: National Academy Press.
Southern New Hampshire Planning Commission. 2010. Trip Generation Study. Manchester, NH:
Southern New Hampshire Planning Commission.
Trip Rate Information Computer System (TRICS). 2012. Trip Rate Information Computer System Good
Practice Guide. UK and Ireland.
Virginia Department of Transportaiton (VDOT). 2010. Traffic Impact Analysis Regulations Small Area
Plan Amendments: 24VAC30-155-60.D.1: Traffic Impact Statement. Richmond, VA: VDOT.