Article citation information:
Żochowska, R., Karoń, G., Sobota, A., Janecki,
R. Selected aspects of the methodology of a household interview survey on
an urban agglomeration scale with regard to its services. Scientific Journal of Silesian University of Technology. Series
Transport. 2017, 95, 239-249.
ISSN: 0209-3324. DOI: https://doi.org/10.20858/sjsutst.2017.95.22.
Renata
ŻOCHOWSKA[1], Grzegorz KAROŃ[2], Aleksander SOBOTA[3],
Ryszard JANECKI[4]
SELECTED ASPECTS OF THE
METHODOLOGY OF A HOUSEHOLD INTERVIEW SURVEY ON AN URBAN AGGLOMERATION SCALE
WITH REGARD TO ITS SERVICES
Summary. The article presents the essential
issues and algorithm of the methodology of a four-step transportation model,
which was constructed in order to carrying out a household interview survey.
The results of this research are source data for determining the travel
behaviour of the users of transportation systems, including intelligent
transport systems (ITS). The presented issues regarding the survey methodology
also concern the specifics of the study area, an urban agglomeration area. The
examples particularly relate to an urban agglomeration with the nature of
a conurbation, namely, the Upper Silesian Agglomeration in Poland.
Keywords:
four-step travel demand model; household interview survey; travel behaviour
modelling; urban agglomeration; intelligent transport system (ITS).
1. INTRODUCTION
This article presents the issues and
algorithm for the collection of data during a household interview survey, as part of tasks in the algorithm of
activities during comprehensive traffic and transportation studies (CTTS) on
transportation modelling, along with an evaluation of an ITS.
A
review of the state of the art revealed that discussion relating to household
interview surveys in urban areas, as described in foreign literature, do not
take into account, to a large extent, the determinants of the functioning of
areas such as polycentric agglomerations (conurbations). In the national
literature, there is a lack of comprehensive research; and, while there are
many publications that deal with these issues, they do so in a fragmentary or
superficial manner. Thus, the developed methodology, presented in this paper,
comprising an algorithm for household surveys with regard to transportation
modelling, along with an evaluation of an ITS project (which uses a systematic
approach and a formalization of the key research questions) is an important
scientific contribution.
Currently, the use of ITS services,
in addition to influencing the distribution of traffic flows in the
transportation network (by traffic control subsystems and VMS signs, for
example) may also affect user decisions about certain types of travel. Users’
decisions concerning their travel arrangements may be considered from at least
three aspects:
-
Executive aspect
of the trip: realization vs. retreat
-
Temporal aspect of
the trip: during peak hours vs. during off-peak hours
-
Spatial aspect of
the trip: via the most popular paths (the shortest ones, but also the most
charged by traffic flows) vs. via the least popular paths (the longest ones,
but the less charged by traffic flows).
The construction of a transportation
model requires gathering a wide variety of data, derived from different
sources, i.e., surveys and traffic measurements [2], [3], [5], [9], [13]. The
scope and level of detail of the data collected in a survey generate at least
two outcomes. The first concerns the structure of the four-step transportation
model and the basic scope of the necessary data. The second concerns the scope
of the project (investment project, study, transport development strategy
etc.), especially the assumptions, components and results of the project in
terms of the impact on the transportation system and its users. An algorithm for survey sampling,
included in the algorithm of activities during CTTS on transportation
modelling, along with an evaluation of ITS project, is presented in Fig. 1.
The implementation of ITS services
in the specified functional and utility configuration should contribute to
reducing congestion in the urban transportation network. It is therefore
necessary to gain from the surveys data that will allow for mapping the impact
of ITS services in a specified configuration on users’ decisions about how they
make a journey in terms of the following:
-
Specified time
-
Specified location
-
Specified mode of
travel
-
Specified paths in
the network
The presented transportation model
is a tool that supports analyses of problems associated with the functioning
and development of the transportation system for a specific urban area. This
article addresses the fact that the area is an urban agglomeration (concentration
of cities) and with the nature of the conurbation being studied, namely, the
Upper Silesian Agglomeration in Poland, and its capital city of Katowice. The
presence of an urban agglomeration is specific to the study area. Therefore,
the methodology of a travel behaviour survey on such a scale requires the
appropriate conditions and limitations to be determined, as well as the
assumption of the representativeness of the sample in terms of selected
characteristics [2], [3], [4], [5], [9], [13], [15], [16], [17], [18].
Fig. 1. Algorithm for survey sampling (blocks 5, 5a-5d), which is
included in the algorithm of activities during CTTS on transportation
modelling, along with an evaluation of an ITS Project (in the form of block
scheme)
2. Conditions and
methodological assumptions of Household Interview Survey
When analysing the area on an urban
agglomeration scale, it is necessary to determine its delimitation [2], [3], [4],
[13], [17]. In the proposed methodology, it is assumed that the travel behaviour
surveys are conducted in two zones (see Fig. 2):
-
Zone 0, which is
the internal area, covering municipalities belonging to the urban agglomeration
-
Zone 1, which is
the outer area, covering municipalities that directly impact on the study area
Fig. 2. Delimitation of the urban agglomeration
area (Upper Silesian Agglomeration in Poland) for Zone 0 (a major area of
study) and Zone 1 (direct surroundings).
2.1. The scope of the study
Surveys conducted in households are
complex, providing basic sources of information on many characteristics that
describe the travel behaviour of residents in the study area. In particular,
information and characteristics should at least include:
-
For examined
households:
• The number of traffic analysis
zones (TAZs)
• The number of persons in a
household
• The number of cars in a
household
-
For examined
persons:
• Anthropogenic features and
characteristics of socio-economic activities (e.g., sex, age, occupation,
income level)
• A detailed description of the
trip (e.g., origin and destination, purpose, start and end times, method of
travel with the use of different transportation subsystems, duration,
frequency)
• Assessment of transport
services, travel behaviour and preferences (e.g., concerning public or
individual transport, application of selected ITS services both for public and
for individual transport, accessibility of infrastructure for cycling in terms
of the construction of Bike&Ride parking lots, accessibility to parking
infrastructure in terms of the construction of Park&Ride parking lots)
• Expectations in terms of
improving travel experience by means of ITS services for public and individual
transport
2.2. The basic assumptions
Any methodology of surveys on
households requires:
-
Defining the
general population and random sample
-
Determining the
minimum sample size and formulating assumptions related to the sampling frame
-
Developing the
rules for assessing the accuracy of the results
-
Developing the
principles for the selection and sampling scheme
The proposed methodology assumes
that the study population will comprise residents of the municipalities within
the urban agglomeration (Zone 0), or zone 1, who are aged six years or more,
regardless of their formal place of residence. Such a term also takes into account
persons who do not have an actual address in the study area, but are
temporarily staying within its boundaries and generating traffic. The age
criterion has been formulated in such a way that, when analysing travel
behaviour, one should take into consideration any movement associated with the
need for all children of compulsory school age to get to a primary school. In
the proposed methodology, the study population was divided into smaller subsets
forming strata. Each stratum was then sampled as an independent subpopulation,
comprising:
-
For zone 0:
residents of individual TAZs separated within the municipalities that belong to
the urban agglomeration
-
For zone 1:
inhabitants of an individual municipality belonging to the outer area
The unit of observation is a
household with one or more members. Therefore, the place where the household is
located (and more specifically its address) has been assumed as a basic
sampling unit. Thus, the sampling frame should be an address database
containing the name of the municipality, the name of street and the number of
the building. The frame should unambiguously map the general population and
contain information relating to its subjective, territorial and temporal scope.
The sampling frame has been divided into strata, which correspond to the TAZs
in Zone 0 and individual municipalities in Zone 1. Such a structure ensures the
most reliable and timely data on the socio-economic, economic, demographic and
spatial characteristics. The household size is determined by the number of its
members. The unit of study is a person who is a resident (aged six years or
older) of the study area (municipalities belonging to Zone 0 or Zone 1), while
a single household is treated as a group of units of the study (cluster) from
the point of view of the travel behaviour of the persons belonging to the
household. The interviewer should, therefore, strive to conduct surveys with as
many members of a single household as possible.
2.3. The sampling scheme
The sampling scheme is a concrete
and coherent (in terms of being formal and substantive) procedure for drawing
out sampling items from the general population, such that the condition of
randomness is satisfied. The choice of optimal sampling scheme depends on the
conditions and purpose of the research. To a significant extent, it determines
the way in which specific parameters are estimated, as well as how certain
parameters should be estimated. Household interview surveys require specific
rules for sampling. The drawing out of sampling items requires a cluster
sampling scheme, where each cluster is a single household. The assumptions
concerning this drawing out process are as follows:
-
The sampling frame
must be an address database of municipalities in Zone 0 and Zone 1, which
contains the name of the municipality, the name of the street and the number of
the building. It is usually at the disposal of the state administration.
-
The data contained
in the sampling frame should be current (in the proposed methodology, it was
established up to three years ago) and complete. In addition, the data should
be arranged and contain information about assigning an address to the
appropriate TAZ.
-
The sampling frame
should be divided into individual strata, corresponding to individual TAZs in
Zone 0 and Zone 1.
-
The sample size
(number of respondents) in the individual municipalities in Zone 0 and Zone 1
is fixed on the basis of proportional allocation, according to the relation:
where:
G- the set of all municipalities in
the study area (Zone 0 or 1)
-
The sampling
scheme in the programming phase of the survey (Stage I) includes:
• Determining the initial number
of addresses in relation to each stratum, which should take into account the
surplus for the randomly selected addresses (e.g., 20%) concerning the
situations where there is an absence of response or a refusal to carry out the
interview.
• Drawing out sampling items of a
predetermined number from the sampling frame, which is an address database.
• Grouping randomly selected
addresses within each stratum. The essential criterion for each grouping is the
location of addresses. It is worth combining the addresses located in close
proximity into one set in order to reduce costs and improve the survey’s
outcomes.
• Determining the number of
surveys and identifying specific addresses relevant to the survey for each
interviewer. In the proposed methodology, the number of surveys in a single TAZ
of Zone 0 has been established on the basis of proportional allocation, according
to the relation:
where:
RK(g) - the set of TAZs in
-
The assumption
about the proportional allocation of the number of surveys in the TAZs and the
number of respondents in the municipalities of Zone 0 and Zone 1 satisfies the
requirement concerning the equal probability of selection for all items
constituting the general population.
These assumptions were used in the
methodology for constructing the transportation model when preparing the tender
documentation (including the specification of the essential terms of the
contract as well as the description of the subject of the order) for two areas:
the Upper Silesian Agglomeration (in Poland) and the territory of the central
subregion of the Silesian Voivodeship.
2.4. Representativeness of the
surveys
It is assumed that household
interview surveys are conducted using the representative method, based on a
random sample, which offers the possibility to generalize, within a specified
margin of error, the survey results for all units of the general population.
The concept of representativeness of the sample refers to the specific features
of the study unit, which means that the structure of the random sample must
correspond to the structure of the general population from the point of view of
these features. In other words, a representative sample is a microcosm of the
population due to the selected features, while it is assumed that the
representativeness of the sample essentially depends, to a certain extent, on
two factors:
-
The choice of the
respectively numerous sample
-
The method of
sampling and fulfilling the conditions of homogeneity, randomness and
independence
A sample that is representative of
one particular feature may not be representative in terms of other features.
Accordingly, the problem regarding the proper selection of characteristics, for
which representativeness is expected, seems to be very important. It is worth
remembering that providing a representative sample for multiple features is
simultaneously difficult to implement, while the complexity of the problem
increases with the number of variables. Therefore, when designing the surveys,
which are then used to analyse travel behaviour analyses, the minimum number of
variables, which are closely related to the study variables, is usually
selected in order to assess the representativeness of the sample. In this case,
the choice of variables, on the presumption of their representativeness, is
burdened by a certain degree of subjectivism.
In the case of drawing out large
samples (at least tens of elements), one should expect that the probability of
selecting samples of the structure, which are significantly different from the
structure of the population, is relatively small. This obviously does not
exclude the occurrence of extreme situations, in which surveys are only
subjected to certain groups of respondents (e.g., due to the absence of working
people, interviews only conducted with non-working elderly people), which may
significantly distort the estimation of the real value of the observed feature.
Therefore, to increase the accuracy of surveys, it is assumed that structure of
the sample may differ by no more than 25% from the population structure in
terms of sex and age with regard to inhabitants in each municipality of Zone 0
and Zone 1.
Putting together a sample of
predetermined structure requires the efficient organization, high precision and
constant control of the results of the survey at the various stages of its
implementation. When conducting a survey, the current verification of the
structure of the sample with the above assumptions and the potential correction
are advisable. In the proposed methodology, it is assumed that the control of
the sample structure, at the stage of carrying out the survey, is conducted at
least once a week. This approach also allows for the verification of intensity
in respect of the inflow of surveys and the corresponding adjustment regarding
the choice of the sample items in order to obtain the desired data structure
for the construction of a transportation model.
The result of the draw is a random
sample, which is the basis of statistical inferences of unknown parameter
values of the general population from which the sample has been selected. To
assess the values of the parameter estimates, their estimators are used. They
are the relevant functions (statistics) related to the sample, which correspond
to the sequence of value of a random variable with a specified distribution,
which is called the distribution from the sample. It is assumed that the
applied estimators must at least be:
-
Compatible, which
means that sequences of assessments, which are obtained with their help, are
stochastically convergent with the real value of the estimated parameters, that
is, they are subjected to the law of large numbers. In practice, the effect of
the compatibility of the estimator is obtained in a situation where, with
increasing sample size, the variance of the estimator heads towards zero
-
Unencumbered,
which means that the expected values (mathematical expectations) of estimators
correspond exactly to the real values of the estimated parameters.
To assess the precision of estimates
in household interview surveys, four parameters of the general population (the
characteristics that are the most relevant to the aim of the study) are chosen:
-
The number of
trips made for a certain purpose:
where:
-
The share of trips
made in a certain mode:
where:
-
The average
duration of the trip:
where:
-
The average
mobility rate of respondents:
where:
Among the selected characteristics,
only mobility rate, which is understood as the number of trips made during the
day, is determined at the municipal level. Other characteristics should be
analysed in the individual strata (TAZs).
3. CONCLUSIONS
The results of travel behaviour analyses are necessary
when constructing a four-step transportation model, which reflects the existing
state of the transportation system [12], [14], [19], and assessing the
directions and trends from the perspective of changes in the system. A
transportation model is also a tool for analysing the problems associated with
the functioning and development of the transportation system. This feature
enables a broad spectrum of activities under the conditions of the
implementation of ITS services in urban areas.
Developing the methodology for a household interview
survey on an urban agglomeration scale with regard to ITS services is related
to the following decision problems:
-
Defining the
population and the sample
-
Selecting the
characteristics of the sample unit, which are important in terms of the
representativeness of the survey
-
Defining the units
of observation and the units of study (basic and group units)
-
Determining the
strata and the sampling frame
-
Selecting the
sampling scheme
-
Determining the
process for controlling the implementation of the survey in terms of the
representativeness of the data collected
Moreover, these decision problems are conditioned by
the methodology of delimitation for the study area, by the methodology of
description for the transportation systems, and by the methodology of
identification for traffic flows in transportation systems.
Verification of the assumptions about the described
methodology was made during the preparation of the tender documentation
(including the specification of the essential terms of the contract, as well as
the description of the subject of the order) for two areas: the Upper Silesian
Agglomeration (in Poland) and the territory of the central subregion of the
Silesian Voivodeship.
References
1.
Aaron Agbenyegah Agbo, Wenfeng Li, Lanbo Zheng, Yanwei Zhang, Charles
Atombo 2017. “Optimisation of intermodal freight transport network”. European Transport/Transporti Europei
63(1). ISSN 1825-3997
2.
Boyce David, Huw
Williams. 2015. Forecasting Urban Travel.
Past, Present and Future. Cheltenham, UK; Northampton, MA: Edward Elgar
Publishing. ISBN 978-1-84844-960-2.
3.
Cambridge
Systematics, Inc. 1996. Travel Survey
Manual. Washington, DC: Department of Transportation and Local Government.
4.
Department of Economic
and Social Affairs. 2005. Designing
Household Survey Samples: Practical Guidelines. Studies in Methods. New York, NY: United Nations.
5.
Department of
Transport. 2016. National Travel Survey:
England 2015. Available at: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/551437/national-travel-survey-2015.pdf.
6.
Drisya Manghat, Aswathy KV, Krishnamurthy
Karuppanagounder. 2017. “Automobile level of service criteria for two-line
undivided heterogeneous urban corridors”. European
Transport/Trasporti Europei 63(9). ISSN: 1825-3997.
7.
Erfan Hassannayebi, Seyed Hessameddin Zegordi,
Mohammad Reza Amin-Naseri, Masoud Yaghini. 2017. “Optimizing headways for urban
rail transit services using adaptive particle swarm algorithms”. Public Transport. DOI:10.1007/s12469-016-0147-6.
8.
Jacyna-Gołda
Ilona, Jolanta Żak, Piotr Gołębiowski. 2014. “Models of traffic flow distribution for various scenarios of the
development of proecological transport system”. Archives of Transport 32 (4): 17-28. ISSN 0866-9546. DOI:
10.5604/08669546.1146994.
9.
Kulpa Tomasz,
Andrzej Szarata. 2016. “Analysis of household survey sample size in trip
modelling process.” Transportation
Research Procedia 14: 1753-1761.
10.
Nosal Katarzyna,
Wiesław Starowicz. 2015. “Evaluation of influence of mobility management
instruments implemented in separated areas of the city on the changes in modal
split”. Archives of Transport 35(3):
41-52. ISSN 0866-9546. DOI: 10.5604/08669546.1185186.
11.
Pallav Kumar, Krinshnanunni, Shriniwas Arkatkar,
Gaurang Joshi, 2017. “Examining microscopic and macroscopic traffic flow
parameters at diverging section on multilane urban roads in India”. European Transport/Trasporti Europei
63(2). ISSN: 1825-3997.
12.
Pyza Dariusz.
2011. “Multi-criteria evaluation of transportation systems in supply chains”. Archives of Transport 23(1): 47-65. ISSN
0866-9546.
13.
Richardson Anthony
J., Elizabeth S. Ampt, Arnim H. Meyburg. 1995. Survey Methods for Transport Planning. Melbourne: Eucalyptus Press.
14.
Sierpiński Grzegorz.
2017. “Distance and frequency of travels made with selected means of transport.
A case study for the Upper Silesian conurbation (Poland). In:
G. Sierpiński (ed.). Intelligent
Transport Systems and Travel Behavior. Advances in Intelligent Systems and Computing,
75-85. Cham, Switzerland: Springer International Publishing. ISBN
978-3-319-43990-7. DOI: https://doi.org/10.1007/978-3-319-43991-4_7.
15.
Statistics Canada.
2010. Survey Methods and Practices. Catalogue no. 12-578-X. Statistics
Canada: Ottawa. ISBN 978-1-100-16410-6.
16.
Taylor Michael
A.P., William Young, Marcus Wigan. 1992. “Designing a large-scale travel demand
survey: new challenges and new opportunities.” Transportation Research Part A: Policy and Practice 26A(3):
247-261. ISSN: 0965-8564. DOI: 10.1016/0965-8564(92)90035-6.
17.
Transportation
Research Board. 2007. Metropolitan Travel
Forecasting. Current Practice and Future Direction. Special Report 288.
Washington, DC: Transportation Research Board.
18.
Transportation
Research Board’s Travel Survey Methods Committee. The On-line Travel Survey Manual: A Dynamic Document for Transportation
Professionals. Available at: http://www.travelsurveymanual.org.
19.
Wasiak Mariusz.
2016. “Vehicle selection model with respect to economic order quantity.” Archives of Transport 40(4): 77-85. ISSN
0866-9546. DOI:10.5604/08669546.1225471.
Received 03.02.2017; accepted in revised form 30.04.2017
Scientific Journal of Silesian
University of Technology. Series Transport is licensed under a Creative
Commons Attribution 4.0 International License
[1] Faculty of Transport, The Silesian
University of Technology, Krasińskiego 8 Street, 40-019 Katowice, Poland. E-mail:
renata.zochowska@polsl.pl
[2] Faculty of Transport, The Silesian
University of Technology, Krasińskiego 8 Street, 40-019 Katowice, Poland. E-mail: grzegorz.karon@polsl.pl
[3] Faculty of Transport, The Silesian
University of Technology, Krasińskiego 8 Street, 40-019 Katowice, Poland. Email: aleksander.sobota@polsl.pl
[4] Faculty of Economics, University of
Economics, 1 May 50 Street, 40-287 Katowice, Poland. E-mail: ryszard.janecki@ue.katowice.pl