close

Вход

Забыли?

вход по аккаунту

?

1804-17

код для вставкиСкачать
■ Transportation Research Record 1804
Paper No. 02- 3545
126
Spatial Behavioral Data
Collection and Use in Activity Scheduling Models
Marion Kreitz and Sean T. Doherty
Transport models and their practical applications are becoming increasingly more sophisticated, thus requiring more precise input data, including spatial data. The development, testing, and assessment of a survey
method with which to collect multiday information on activity–travel patterns with a high level of spatial detail and accuracy are described. The
survey consists of a computer-assisted, self-completing weekly household
activity–travel diary survey program combined with an interactive map
for spatial data input and visualization. Compared with traditional paperand-pencil surveys, more accurate spatial data is gathered through this
survey, especially for activity locations, routes, and trip lengths, which a
first trial on a small sample of 30 households clearly demonstrated. In
addition, more details on activity and travel times were collected. Fatigue
effects did not become evident, and the respondent burden was considered to be acceptable by the participants. Additional testing of the
respondents’ map-handling abilities showed that people have different
map-orientation aptitudes that depend on sociodemographic characteristics such as gender, income, and education. This suggests that maps
incorporated in travel surveys should be adapted in different ways
to accommodate different levels of map handling. Future household
travel surveys, especially emerging computerized and Global Positioning
System–supported methods, would appear to benefit from the integration
of a spatial interface both to add detail and to reduce respondent burden.
This is particularly important for activity-based or activity-scheduling
surveys, which increasingly attempt to obtain higher amounts of precise
and detailed information on individual behavior and choices.
these techniques lies in the restricted sample of participants and, for
cell phone positioning, the lack of geographic accuracy.
Another possibility for spatial data collection is through retrospective methods combined with automatic geocoding of the spatial
information in a computer-assisted personal interviewing (CAPI) or
computer-assisted self-interviewing (CASI) survey. For example, in
its Internet-based combined travel–activity and stated preference survey, the Resource Systems Group tested a geocoding module that
allowed entry of activity location by respondents by street or address,
nearest intersection, establishment or business name, and direct map
point-and-click (4).
All entries were geocoded. The same options were also implemented in the Internet-based computerized household activity scheduling elicitor survey developed by Lee et al. (5). Applications of
graphically supported survey instruments with display of locations or
routes showed that the graphical user interface, photos, icons, and
other multimedia elements maintain the respondent’s interest and
increase both accuracy and response rate.
CASI methods offer the additional potential of longer reporting
periods and reduced field costs, especially in light of greater accessibility and use of computerized survey instruments by the population
[e.g., computers and laptop computers, personal organizers, Internet,
and e-mail (6, 7 )]. An opportunity exists to develop and apply a graphical user interface (e.g., a map) that allows for the direct input by the
participant of precise spatial information into the survey system.
In the field of research and planning, more sophisticated travel forecasting models are being applied, thus increasing the need for more
accurate data. This need refers not only to data validity but also to
the precision of spatial (geographical) behavioral information.
Transport modeling—as well as applied transport demand and traffic management—is increasingly in need of precise spatial information, for exact geographic location of activity locations, routes,
parking search areas, and individual activity spaces.
Such data are being collected either in an aggregate fashion, by
induction loops or similar systems, or disaggregately by individual
travel surveys. Travel survey methods can be applied during reporting periods of variable duration, and they differ according to whether
they are simultaneous (online) or retrospective, computer-assisted or
paper-and-pencil (PAPI), and with or without automated positioning.
Simultaneous, computer-assisted automated positioning methods
have already been tested in several studies, by using the Global Positioning System (GPS) (1, 2) or cell phones (3). The disadvantage of
OBJECTIVES
This paper describes the development and testing of a survey method
for the collection of spatial behavioral data within a retrospective
CASI travel survey program. A CASI approach was considered to be
the most suitable because it offers the ability to store large amounts
of spatial data for display and reuse, the ability to incorporate logical
checks, and the opportunity to interact directly with the participant
over a longer period. The primary directive for this development was
the following question: “Is it possible to collect the required spatial
and other information during a longer reporting period (7 days) with
sufficient validity, completeness, and precision along with a reasonable respondent burden?” If so, this method could be considered
for inclusion in a full activity scheduling survey software, thus
incorporating both the activity scheduling and revisions as well as
a spatial–temporal component.
METHODS
M. Kreitz, Institute of Urban and Transportation Planning, University of Technology Aachen, Mies-van-der-Rohe-Strasse 1, D-52074 Aachen, Germany.
S. T. Doherty, Department of Geography and Environmental Studies, Wilfrid
Laurier University, Waterloo, Ontario N2L 3C5, Canada.
Two survey methods were employed to meet the objectives. The first
involved an in-depth computerized activity–travel survey incorporating a geographic information system (GIS) interface to capture
Kreitz and Doherty
spatial information. The second survey involved an investigation
of map-reading abilities. Both were meant to provide a basis for
assessing the future potential of such survey methods.
CHASE-GIS Development
For the purpose of this research, and to compare the survey results
with PAPI and other CASI survey methods, the survey software
CHASE-GIS was developed. CHASE-GIS is a further development
of the CHASE (computerized household activity scheduling elicitor) survey software, developed by Doherty and Miller (8) but augmented with a GIS-based graphical user interface for the collection
of spatial behavior information by Kreitz (9).
CHASE is a Windows-based software designed to trace household
activity–travel patterns over a multiday period and to trace the underlying decision process that led to the observed patterns. This activity
scheduling process is documented by recording the sequence of decisions made by individuals in order to plan, modify, and execute their
final schedule of activities and travel. An initial interview is held to
establish basic household sociodemographic and activity characteristics. This information is reused during the data entry process. In the
course of the reporting week, a weekly schedule similar to a typical
day-planner (calendar) is displayed for each household member,
FIGURE 1
CHASE-GIS activity–travel entry dialogue, activity sheet.
Paper No. 02- 3545
127
leaving any unplanned or unreported time blank. Activities can be
added to this calendar, causing pop-up dialogue boxes to appear for
the entry of activities and related travel data. In addition, the program
records when and by whom such data were entered or modified,
along with supplementary information on the reasons for certain
choices. Previous research demonstrated that this method provides
in-depth detail on observed activity–travel patterns and underlying
scheduling decisions while minimizing respondent burden and fatigue
effects commonly associated with multiday surveys.
During the first application of CHASE-GIS, the focus was on
tracing observed activity–travel patterns over a 7-day period (especially the spatial dimension) rather than on underlying scheduling
decision processes. As such, many of the CHASE software’s decision tracing components were disabled to concentrate testing on the
spatial dimension. This included all questions concerning the planning status of an activity and the reasons why activities were modified. All other parts and functions of CHASE remained activated.
Thus, the resulting CHASE-GIS program essentially combines the
tabular weekly planner view used in CHASE with a new geographic
map interface implemented within the activity–travel entry dialogue.
The map view includes a zoomable GIS-based vector map of the
study region that serves both as a display tool and as an input tool
for spatial data, as shown in Figures 1 and 2. A detailed description
of CHASE-GIS is provided elsewhere (9, 10).
128
Paper No. 02- 3545
FIGURE 2
Transportation Research Record 1804
CHASE-GIS activity–travel entry dialogue, travel sheet.
The survey tool captures information on both in-home (not
detailed) and out-of-home activities, categorized into detailed activity purposes and sorted by categories. As travel diaries usually do,
the survey included questions about the characteristics of the activities performed during the reporting period (e.g., times, type, location,
participants, costs) and the characteristics of travel to the respective
activities (e.g., times, means of transport, costs). Several optional
questions probed travel details further, including information on
multimodal trips and public transport stops used. CHASE-GIS reuses
information that is common to more than one member of a household, such as means of transport, shared activities, and common
activity locations. During an initial face-to-face interview, sociodemographic data are also recorded on the computer along with common activity locations as defined by the respondent and entered by
using the map as entry tool.
In the course of the reporting week to follow, the respondents are
asked to log in at least once a day to report their activities by using
CHASE-GIS. Predefined items can be chosen from drop-down lists
or from the map. If necessary, respondents may define new activities, activity locations and areas, and transport (including public
transport stops and lines). All spatial information can be entered by
clicking on the map. For trips by personal travel modes, the most
likely travel routes are automatically generated and displayed by
using a fastest-path algorithm. The route can then be corrected by
the user through the addition of further reference points along the
route to the map. The route is then redrawn, including the new reference points, and the route length is recalculated. The information
reported is automatically checked for logical consistency. Spatial
information is automatically geocoded and all input is stored in a
database, together with statements of when, in which way, and by
whom any kind of interaction with the program took place. In the
final interview, the data are checked again by the interviewer. A
short interview about the user’s experience with the software also
took place. All data from the initial interview, the diary, and the final
interview are stored in the same database file.
The map view shown in Figures 1 and 2 contains domain information such as area and street characteristics, district and street
names, and an individual (private) transport net with routing information. A public transport network with lines and stops was also
added to it. The map view allows the display of
• Possible and chosen activity locations or areas (origin and
destination highlighted);
• Public transport lines and stops (automatic choice and highlighting of the nearest stops according to the chosen line, possibility
to adjust these); and
Kreitz and Doherty
• Routes (automatically generated fastest-path routes that can be
adjusted by the user; automatic route length calculation).
The first application of CHASE-GIS occurred in the summer of
2000 as part of a two-stage PAPI activity–travel panel survey in the
city of Karlsruhe, Germany, carried out in the course of the Mobiplan project (11). On recruitment for the second panel wave, households were asked whether they would like to participate in the second
wave by computer (on a provided laptop) or with the PAPI method
again. The goal was to survey 20 to 25 households in the second
panel wave with CHASE-GIS instead of PAPI methods. Because
of the restricted time frame of the second wave and the number of
laptops available, this was the maximum number to be surveyed. A
quota sample was drawn depending on the age of the oldest household member and the household size. Households agreeing to the
CASI version were recruited according to the quota sample. The survey attributes for both the PAPI and the CASI surveys were identical, thus allowing the possibility of comparative analysis of the data
gathered with both methods and the experiences and evaluations of
the household members.
Map-Handling Survey
A second survey focusing on an evaluation of the graphical features
of CHASE-GIS was carried out in the spring of 2001. Because some
people find it harder than others to orient themselves on a map, the
purpose of this survey was to identify different types of map-reading
ability of persons with different sociodemographic characteristics.
A total of 178 persons of different sociodemographic backgrounds
were recruited for this map-handling survey. The survey involved
presentation of the same map used in the CHASE-GIS survey. Participants were asked to locate four locally well-known locations on
the map given by the interviewer. The time needed by a person to find
the locations was recorded, as was the geographical accuracy of the
results.
RESULTS
CHASE-GIS Main Study
Between March and August 2000, 30 households were interviewed,
4 of them in a pilot study. Four laptop computers were used in the
field for a 1-week duration in each sampled household. The survey
was intermitted during local holidays. It turned out that 15.7% of the
households that had been given the choice between both methods
were willing to participate in the CASI survey and finally did so.
Recruitment for the CASI version was stopped when 30 participating households, the number of households aspired to be surveyed,
was reached. Statistical tests (binary logistic regression) gave evidence that at the time of recruitment, no significant differences regarding sociodemographic aspects or the mobility (number of trips per
week) in the first panel wave existed between the households that
chose PAPI and those that preferred CASI. The decision to use
CASI, therefore, does not depend on these characteristics, and the
CASI participants can be considered as a representative sample of
the Mobiplan survey participants.
During the computer-assisted initial interview, the same data were
gathered as in the second wave of the Mobiplan (PAPI) study.
Household, person, and vehicle data were already completed with the
Paper No. 02- 3545
129
associated data from the first panel wave. Additional questions covered the regular activities of the household members and their frequencies. Probable activity locations of the reporting period were
defined by using the map. The information was entered into the
CHASE-GIS database so the data could be reused during the reporting week. After the initial interview, the software was demonstrated
to the households by the interviewer, with the aid of an entry example. In addition, a manual was distributed. During the reporting week,
the respondents could call a telephone help-line. The final interview
included a check for complete entries and questions regarding the
respondents’ experiences with and opinions about the survey tool and
methods. The average duration of the initial interview was not much
longer than that held for the PAPI method:
• Six min (for the household and vehicle data),
• Plus 20 min per household member (for the person and activity
data),
• Plus 15 min for the explanation and demonstration of the
CHASE-GIS-software.
The corresponding PAPI survey was completed by 176 households with 276 persons. Excluding pilot surveys, the main study with
CASI included 26 households (40 persons). One of the respondents
was unable to handle the software because of a visual impairment and
instead completed the survey with PAPI methods. Of the remaining
25 households, 21 completed the 7-day diary with the survey software. One household dropped out because of lack of comprehension
of the software, and three others left because of failed data entries
caused by software bugs or computer glitches (mostly caused by running the software on older laptops with capacity constraints); these
also completed the survey with PAPI methods.
The completeness of the data achieved in the CHASE-GIS survey
can be considered to be good. The 21 households provided completed
7-day diaries with all mandatory questions answered for each activity. Problems occurred mostly with questions that required more complex data entry, especially the specification of combined means of
transport and route corrections. Further information was also gathered
with the optional questions, such as those concerning the sequence of
transport modes used, routes, and parking search areas.
With the PAPI method, fatigue effects were evident in the form of
statistically significantly fewer reported trips per week in the second
wave than in the first wave. In contrast, with CHASE-GIS, a slightly
higher number of trips was reported in the second wave, as shown in
Figure 3. Additionally, with CASI, the number of reported trips did
not decrease in the course of the reporting week (Days 1 to 5 compared to Days 6 and 7) as occurred with the PAPI method in both
waves.
The automatic routing led to a more precise statement of trip
lengths. CASI participants generally stated shorter trips than PAPI
participants in both waves. This may be because CASI participants
gave more exact estimations in the first wave, which were confirmed
by the automatic route length calculation in the second wave. Routes
were automatically calculated and displayed by using the fastest
route. These could be adapted by the user in case of private transport.
The trip length was recalculated subsequently. For private motorized
transport, 8 of 25 persons who used this mode during the reporting
week took the opportunity to adapt the routes of some of their trips.
Additionally, 4 of 21 persons using bicycles changed their routes as
well. In the final interview, 70% of the persons using some kind of
private transport during the reporting week stated that the automatic
route suggestion was “mostly” correct, so that they felt no need to
Transportation Research Record 1804
80
70
60
202
508
275
118
262
297
17
264
11
50
106
440
277
40
30
20
10
0
N=
39
274
CG1
MP1
W2 Number of trips per week
Paper No. 02- 3545
W1 Number of trips per week
130
80
70
118
60
75
264
11
106
291
297
50
40
30
20
10
0
N=
37
271
CG2
MP2
Wave 1
Wave 2
Percent
CHASE-GIS
(in Wave 2)
Mobiplan
PAPI
CHASE-GIS
Mobiplan PAPI
25
21.00
19.00
22.00
18.00
50
28.00
24.00
28.00
23.00
75
33.00
29.25
35.50
29.00
FIGURE 3 Number of reported trips per week and per panel wave (W wave; CG CHASE-GIS;
MP Mobiplan).
adapt it. At the same time, 50% of these persons admitted that they
considered the route adaptation to be “complicated.”
Furthermore, the automatic geocoding of locations and routes
provided additional spatial information that could not be achieved
by PAPI data postprocessing, including
• Precise geographical location of activity locations or activity
areas (which helped differentiate activities such as hiking and pleasure driving, both of which intrinsically involve traveling as a pursuit,
not simply as a means to reach an activity location);
• Means of transport used, sequence, and routes;
• Activity spaces for each household member, for time of day,
activity type, and means of transport; and
• Parking search areas and times.
The accuracy of the spatial information gathered with CHASE-GIS
can be measured by comparing with the postsurvey address geocoding
of the PAPI questionnaires. In the second wave, 28% of the trips were
not exactly geocodable because of a lack of a specific address and
number statement. In these cases, only zones (e.g., city center, name
of adjacent cities) were given. Because the geocoding process was
performed by a private company, the remaining effort needed to
manually geocode unfortunately was not recorded. Although this is
not an exact means with which to assess the spatial accuracy of the
software, knowing the number of cases with insufficient address statements and assessing the address reference database’s quality (which
was in this case one of the best available in Germany), these figures
roughly demonstrate the potential of the automatic geocoding facility implemented in CHASE-GIS. Another way to obtain exactly
geocoded activity locations and routes is, of course, to use GPS (12).
Details on activity and travel start and end time, duration, and—in
case of public transport use or being picked up or dropped off—
access, egress, and waiting times were also captured and compared.
Automatic logical checks made sure that temporal statements were
consistent with one another. For the PAPI survey, a comparison
regarding inconsistency in reported travel times was based on the calculation of expected travel times from the average speed of the respective means of transport and the distance covered. Recognizing the
tendency of PAPI respondents to round up to the nearest 5 or 10 min,
only trips with a travel time of more than 10 min were taken into
account. It turned out that 25% of the reported travel times showed a
deviation from the expected value of more than 50% (which might be
because of traffic conditions); 11.5% showed more than a 100% deviation. The latter number indicates that there may be an advantage in
correct travel time reporting from automatic matching with public
transport schedules or similar background data.
Measures of each user’s handling of CHASE-GIS were automatically recorded by the software. It turned out that the survey participants logged into the software an average of 7.3 times per reporting
week but largely did not log in daily. Log-ins predominantly took
place in the afternoon (25.94%) or in the evening (66.04%). Also on
average, the participants needed approximately 22 min per day for
the entries in CHASE-GIS compared to about 10 to 15 min for the
PAPI questionnaire. It was thought that the log-in durations could
have been further reduced if the CASI participants were better trained
and made fewer wrong entries.
In the final interview, most comments from the survey participants were related to the map display and the entry of activity locations. Participants stated that their orientation on the map could be
improved, for example, by adding selected local points of interest
and landmarks to the map or by allowing magnification of the display. They also suggested the provision of an address locator. These
features were eventually implemented in a later version of CHASEGIS and were tested with positive results for handling by a subsample
of the main study’s participants in October 2000. The main study’s
survey design and the survey instrument itself were subject to few
suggestions for improvement.
Overall, the survey participants assessed their experiences with
CHASE-GIS as positive in the final interview. Older households and
Kreitz and Doherty
Paper No. 02- 3545
households without computers tended to be less positive. These
households indicated that CHASE-GIS required more effort than the
PAPI method, although analysis showed that the time required for
data entry was overestimated in most cases. Questioned about their
preferences, 84% of the households stated that they considered
CHASE-GIS to be more interesting and easier to use than PAPI and
that they would prefer CHASE-GIS for another survey of this kind.
They explained that this would apply especially to long reporting
periods following an appropriate period (about 2 days) to become
acquainted with the software.
Map-Handling Survey
An analysis of the accuracy and time required to locate a series of
locations on the CHASE-GIS map interface revealed several tendencies, some of which were statistically significant (for selected
results see Table 1).
The geographic precision with which the locations were to be
detected was at least to the correct block. This was achieved in
43.1% of the recorded attempts.
After opening the map and stating that they knew the location they
had to look for, the participants failed to find the location in only 3%
of the cases. Only 4 of the 178 participants did not find more than one
of the given locations, 3 of them women. All 9 participants needing
the most time to carry out their tasks (more than 1.5 min for each
TABLE 1
Male
Female
Sex
location) were women. No other common characteristics of these
participants were evident.
In further statistical tests, it turned out that women generally needed
more time than men to find a location on the map while not improving the geographical precision. These findings correspond to previous
studies on way-finding (13). The older and better educated a person
was and the higher the income, the faster the person located the given
locations. Those having a car at their disposal needed less time and
achieved more precise results than those with no motorized individual means of transport. People who owned or used computers also
achieved faster and more accurate results, although this may be due
to their general familiarity with the survey instrument in the first place
rather than to their orientation ability on maps. Given these differences in map-handling abilities across socioeconomic groups, maps
could be prepared in different ways for different groups of people,
in accordance with their spatial perception and style of orientation.
Furthermore, special incentives and assistance for these groups could
be taken into account to reduce their burden and, consequently,
improve data quality.
DISCUSSION AND CONCLUSIONS
Results of the CHASE-GIS survey show that respondents are willing to participate in self-completing, computerized activity–travel
diary surveys and consider it very interesting. The CHASE-GIS
Selected Results of Map-Handling Survey
All Locations
Less than 20
25.9
19.0
44.9
Time Needed (sec)
21-40
41-60
14.4
4.8
15.1
7.3
12.2
39.5
60+
5.8
7.6
13.4
% of
All
50.9
49.1
100.0
Less than 20
9.4
17.5
19.3
46.2
Time Needed (sec)
21-40
41-60
8.8
2.7
12.1
3.6
10.3
3.0
9.4
31.3
60+
1.8
8.8
2.7
13.3
% of
All
22.7
42.0
35.3
100.0
Less than 20
0.8
Time Needed (sec)
21-40
41-60
0.2
0.2
60+
0.2
% of
All
1.3
Chi2: 0,009
All Locations
Monthly
Income
(DM)
1800 ≤ 3000
3000 ≤ 5000
5000 +
Chi2: 0,016
All Locations
Education
Secondary
School
Certificate
High School
Graduation
Apprenticeship
or Master
University
16.1
12.6
4.6
6.5
39.7
10.8
6.8
3.0
3.8
24.5
17.6
45.3
9.7
29.3
4.1
11.9
3.0
13.5
34.5
100.0
60+
0.8
12.6
13.4
% of
All
2.8
97.2
100.0
Chi2: not significant on 95 %-level
All Locations
Less than 20
Computer
no
0.6
ownership
yes
44.3
44.9
Chi2: not significant on 95 %-level
NOTE: DM refers to income in deutsche marks.
131
Time Needed (sec)
21-40
41-60
0.3
1.1
11.9
28.4
12.2
29.5
132
Paper No. 02- 3545
survey also resulted in a high level of data quality, especially for spatial and temporal information. Compared with non-computer-assisted
surveys without maps, this survey technique provides
• Spatial behavioral data with a high geographic precision;
• More and consistent details on activity and travel times, including public transport access, egress, transfer, and waiting times;
• More and consistent details on the sequence and combination
of means of transport on a trip, as well as detailed information about
vehicles and routes used;
• Accurate trip lengths; and
• Knowledge about activity chaining, for example, through information on the scheduling status of activities or the reasons for leaving
usual routes.
This information is of considerable use in current transport modeling applications and would be difficult to collect via conventional
methods. Although the respondents initially faced greater demands
in using this method, they became much more comfortable after 3 or
4 days of reporting. However, although several significant tendencies were derived from these findings, they may not be valid for all
population groups and circumstances, given the small sample size
of 21 households that completed the CASI survey.
The experience conducting field work with laptop computers suggests that larger samples would take considerable effort. The effort
for generating an application for an area larger than Karlsruhe, such
as a major metropolitan area, depends on the availability of the
background data. This includes at least the digital availability of the
maps with private and public transport networks and, preferably, an
address locator and detailed points of interest. In Germany, the former data are available all over the country but are provided by a few
private companies. This makes these data extremely expensive,
except when required for research purposes. In the United States,
the TIGER files provided by the U.S. Bureau of the Census would
meet the requirements (but could, perhaps, be improved in content).
With these data available, an enlargement of the study area requires
practically no effort in terms of the survey software; the background
data are simply exchanged.
The effort involved for the recruitment of participants increases for
larger samples and study areas. The participant training and the delivery of the survey instruments do not necessarily have to be undertaken
by an interviewer, which could reduce the costs for manpower. This
depends on the survey method applied. The use of laptops as survey
instruments requires an interviewer for the initial explanation. Other
survey instruments (Internet, personal travel assistant, cell phone)
do not rely on an interviewer for these tasks. Also, more people are
familiar with these instruments than with laptop computers, so they
may find them easier to use. Applying CHASE-GIS on the Internet is
one way to permit increased sample sizes in larger study areas in the
future.
The integration of the spatial features employed in CHASE-GIS
would benefit other innovative survey instruments that incorporate
emerging technological advances. This includes the application of
CHASE on the Internet (5), the incorporation of the survey software
on handheld computerized devices, the use of GPS (14, 15), and the
use of cell phones for survey tasks (3). The CHASE-GIS survey
showed that the implementation of a map improves the completeness
and accuracy of the data gathered. The map-handling survey also
provides insight into the use of maps in survey instruments. Surveys
with automatic positioning (e.g., GPS) may benefit further from an
Transportation Research Record 1804
interface such as the one used in CHASE-GIS, because it would
allow the display of automatically traced spatial data to participants
for interactive updating and completion. This would be particularly
useful at times when or in places where automatic positioning is
disabled or lacks precision.
FUTURE APPLICATIONS
A main future goal concerns how to sensibly implement the spatial
features of CHASE-GIS into the next generation of activity scheduling process surveys. The lessons learned in this initial application
will serve to inform future versions of the programs. The challenge
will be to identify the key attributes of interest and bring them
together into a comprehensive survey instrument with a minimum of
respondent burden. This may include detailed spatiotemporal information on observed activity and travel patterns and information on
the underlying scheduling decision process that leads to the observed
patterns, preferably over a multiday period.
Perhaps more important, the usefulness of such data for improved
model development needs to be assessed. Emerging activity-based
models aim to represent individuals, their activities, and the formation of activity chains depending on their choices, thus requiring
more precise spatial and temporal data than those provided by conventional methods. For example, the activity locations have to be
captured with higher geographical accuracy than at the traffic analysis zone level commonly used in traditional four-step models in
order to model details of activity chains (short walking trips between
several minor activities or similar) within a zone. It seems clear that
methods such as CHASE-GIS would provide the detail required for
the purposes of activity-based models.
More recently, however, activity scheduling process models have
begun to emerge that aim to represent the individual planning process
and, as a result, the sequence of activities over a period (e.g., a day
or a week) (16–18). Their emergence is largely in response to the
need to improve the ability to forecast the more complex responses
to transport demand management strategies and the needs of emerging traffic flow models for more time-dependent trip matrices. Activity scheduling process models require more detailed and different
spatial and temporal data than are available (19). This includes differentiation of activities into routine activities, planned activities
and spontaneous activities, rescheduling behavior, and specification
of parameters such as the spatial and temporal flexibility of activities. These kinds of information have to be captured in a spatial and
temporal long-term perspective and with high precision (20–22).
Because of the possibility of reenabling all the original schedulingrelated modules of CHASE, CHASE-GIS probably is momentarily
the most suitable instrument with which to capture those items of
interest.
Aside from modeling applications, the spatial information provided
by tools such as CHASE-GIS may be of use for spatial analysis of
individual behavior. Understanding of the individual activity spaces
may benefit from exact geographical information about activity locations and types. Additional potentials for analysis are offered by the
visualization of spatiotemporal patterns before data discretization
with a GIS, which may help in the detection of distinctive features
or regularities in behavior. Knowledge about activity spaces in the
context of household characteristics and urban structure may lead to
a deeper insight into the generation of activity and travel demand
and its execution.
Kreitz and Doherty
ACKNOWLEDGMENTS
CHASE-GIS was developed by the Institut für Stadtbauwesen und
Stadtverkehr (ISB), RWTH Aachen, Germany, in cooperation with
S. T. Doherty of Wilfrid Laurier University and PTV AG, Karlsruhe, Germany. The development and first application was enabled
by a grant from the German Ministry of Education and Research, as
part of the funding ISB received for the Mobiplan project. The
authors thank G. Rindsfüser of ISB for his contribution on activityscheduling models to the version of this paper.
REFERENCES
1. Draijer, G., N. Kalfs, and J. Perdok. Global Positioning System as Data
Collection Method for Travel Research. In Transportation Research
Record: Journal of the Transportation Research Board, No. 1719, TRB,
National Research Council, Washington, D.C., 2000, pp. 147–153.
2. Murakami, E., D. P. Wagner, and D. M. Neumeister. Using Global Positioning Systems and Personal Digital Assistants for Personal Travel Surveys in the United States. In Transportation Research Circular E-008:
Transport Surveys: Raising the Standard, TRB, National Research
Council, Washington, D.C., 2000, pp. III-B/1–21.
3. Wermuth, M. Ein GSM-basiertes Verfahren zur Erhebung von Mobilitätsdaten. Strassenverkehrstechnik, Vol. 6, 2001, pp. 269–276.
4. Computer-Based Intelligent Travel Survey System: CASI/Internet Travel
Diaries with Interactive Geocoding. Final Project Report DTRS57-98C-00068. Resource Systems Group, Inc., White River Junction, Vt.,
1999.
5. Lee, M. S., S. T. Doherty, R. Sabetiashraf, and M. McNally. iCHASE:
An Internet Computerized Household Activity Scheduling Elicitor Survey. Presented at 79th Annual Meeting of the Transportation Research
Board, Washington, D.C., 2000.
6. De Leeuw, E., and W. Nicholls II. Technological Innovations in Data
Collection: Acceptance, Data Quality and Costs. Sociological Research
Online, Vol. 1, No. 4, 1996. www.socresonline.org.uk/socresonline/1/4/
leeuw.html. Accessed Sept. 2000.
7. Saris, W. E. Computer-Assisted Interviewing. Sage University, Newbury Park, Calif., 1991.
8. Doherty, S. T., and E. J. Miller. A Computerized Household Activity
Scheduling Survey. Transportation, Vol. 27, No. 1, 2000, pp. 75–97.
9. Kreitz, M. Raumbezogene Daten: Erhebung und Verwendung in Verkehrsmodellen. Arbeitspapier Forschung F6. Institut für Stadtbauwesen,
RWTH, Aachen, Germany, 2000.
10. Kreitz, M. Methods for Collecting Spatial Data in Household Travel
Surveys. Presented at 5th International Conference on Transport Survey
Quality and Innovation, Kruger Park, South Africa, 2001.
Paper No. 02- 3545
133
11. Beckmann, K. J., M. Kreitz, and T. Perian. Mobiplan—Eigene Mobilität
verstehen und planen. Stadt Region Land, Vol. 67, 1999, pp. 57–59.
12. Wolf, J., J. Myers, and C. Arce. Trip Rate Analysis in GPS-Enhanced
Personal Travel Surveys. Presented at 5th International Conference on
Transport Survey Quality and Innovation, Kruger Park, South Africa,
2001.
13. Devlin, A. S., and J. Bernstein. Interactive Way-Finding: Map Style and
Effectiveness. Journal of Environmental Psychology, Vol. 97, No. 17,
1997, pp. 99–110.
14. Doherty, S. T., N. Noël, M. Lee-Gosselin, C. Sirois, and M. Ueno. Moving Beyond Observed Outcomes: Integrating Global Positioning Systems and Interactive Computer-Based Travel Behavior Surveys. In
Transportation Research Circular E-026: Personal Travel: The Long
and Short of It, TRB, National Research Council, Washington, D.C.,
2001, pp. 449– 466.
15. Wolf, J., R. Guensler, S. Washington, and L. Frank. Use of Electronic
Travel Diaries and Vehicle Instrumentation Packages in the Year 2000
Atlanta Regional Household Travel Survey. In Transportation Research
Circular E-026: Personal Travel: The Long and Short of It, TRB,
National Research Council, Washington, D.C., 2001, pp. 413–429.
16. Doherty, S. T., and K. W. Axhausen. The Development of a Unified
Modelling Framework for the Household Activity-Travel Scheduling
Process. In Traffic and Mobility: Simulation-Economics-Environment
(W. Brilon, F. Huber, M. Schreckenberg, and H. Wallentowitz, eds.),
Springer-Verlag, Berlin, 1999, pp. 35–56.
17. Rindsfüser, G., and S. T. Doherty. Konzept, Module und Datenerfordernis für SMART—Simulationsmodell des Aktivitäten-(Re)Planungsprozesses, Tagungsband AMUS 2000, Schriftenreihe SRL–Stadt
Region Land, 69, Institut für Stadtbauwesen und Stadtverkehr, RWTHAachen, Aachen, Germany, 2000.
18. Arentze, T. A., and H. J. P. Timmermans. Albatross: A Learning Based
Transportation Oriented Simulation System. The European Institute of
Retailing and Services Studies, Eindhoven, Netherlands, 2000.
19. Doherty, S. T. Meeting the Data Needs of Activity Scheduling Process
Modelling and Analysis. Presented at 80th Annual Meeting of the
Transportation Research Board, Washington, D.C., 2001.
20. Axhausen, K. W. Data Needs of Activity Scheduling Models. In
Activity-Based Approaches to Travel Analysis (D. F. Ettema and H. J. P.
Timmermans, eds.), Pergamon Press, Oxford, 1997, pp. 229–241.
21. Axhausen, K. W. Can We Ever Obtain the Data We Would Like to Have?
In Theoretical Foundations of Travel Choice Modelling (T. Gärling,
T. Laitila, and K. Westin, eds.), Elsevier, Oxford, England, 1998,
pp. 305–324.
22. Axhausen, K. W., A. Zimmermann, S. Schönfelder, G. Rindsfüser, and
T. Haupt. Observing the Rhythms of Daily Life: A Six-Week Travel
Diary. Transportation (in press).
Publication of this paper sponsored by Committee on Travel Survey Methods.
Документ
Категория
Без категории
Просмотров
2
Размер файла
1 368 Кб
Теги
1804
1/--страниц
Пожаловаться на содержимое документа