Decoding Behavioural Norms in School Mobility:
A Structural Equation Modelling Analysis
KORNILIA MARIA KOTOULA1, GEORGE BOTZORIS2
1Centre of Research and Technology Hellas, Hellenic Institute of Transport,
6th km Charilaou - Thermi Road, 57001 Thessaloniki,
ORCID: 0000-0001-6765-7240
GREECE
2Department of Civil Engineer,
Democritus University of Thrace,
Kimmeria Campus, 67100 Xanthi,
ORCID: 0000-0002-8983-8058
GREECE
Abstract: - Travel demand modelling for school travel, remains a subject of limited research. The exploitation
of factors influencing parents in school mode choice and the understanding of the significance parents attribute
to these factors is important, contributing to transport planning and leading to a strategic direction with an
ultimate scope to improve the school transportation system and promote the use of alternative transport modes
for upgrading the living environment and quality of life in general. The current paper examines the
development of a Structural Equation Model (SEM) describing the interrelationships between the factors
influencing parents in the decision-making process and the final mode choice. For that, a questionnaire survey
is conducted for parents of children aged six to eighteen years old. The collected data are analysed through
Exploratory and Confirmatory Factor Analysis. Following, an SEM is developed examining the proposed
authors' conceptual model, basic hypotheses of school travel choice, and direct and indirect correlations of
factors composing parental behaviour.
Key-Words: - Structural Equation Model, mode choice, school trip, school transportation system, alternative
transportation
Received: August 19, 2022. Revised: April 9, 2023. Accepted: April 23, 2023. Published: May 26, 2023.
1 Introduction
School mobility is an integral and important
parameter of social activities as it ensures students'
right to education, while at the same time
contributing to knowledge acquisition, socialization,
and adoption of mobility behavioural patterns, [1].
The design, organization, and general functioning of
a school transportation system, is a research subject
that gained ground within the last decades among
the global scientific community. However, in
Greece, the research on related issues is still in its
early stages, [2].
The overall view of students’ mobility patterns is
a particular scientific subject addressing not only
transport experts and transport planners but also
public health scientists and policymakers.
Nevertheless, school trip modelling remains a
subject of limited research. Investigating the factors
that influence parents in the school mode choice and
understanding the importance they attribute to these
factors, is particularly important for transport
planning and for shaping the appropriate strategic
directions towards an overall improvement of the
school transportation system.
Based on the above, the current paper presents
the development of a methodological framework
that investigates and analyses in-depth, personal
hidden characteristics influencing parents in the
school mode choice process. The research examines
aspects of human behaviour in terms of school trip
completion and reveals the importance that parents
attribute to specific factors which positively or
negatively affect the selection of the transport mode
students use for traveling from their residence to
their school unit and vice versa.
According to the existing literature, a significant
number of researchers have concluded the most
basic factors that influence the school mode choice.
Examples include: student's gender, [3], [4], [5], [6],
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
520
Volume 22, 2023
[7], student's age, [8], [9], [10], parents’ attitude
towards the use of alternative transport modes, [11],
[12], [13], [14], [15], [16], distance, [17], [18], [19],
[20], [21], built environment, [4], [20], [22], [23],
[24], and road safety, [25], [26], [27].
Based on the above-defined factors, a
questionnaire survey was designed by the authors
and used as the main tool for primary research and
data collection. The main stages of the research
include i) a questionnaire survey conducted to
students’ parents; the survey was designed after
having investigated the type of variables involved in
the process of selecting a transport mode, ii)
correlations’ analysis through the application of
Exploratory and Confirmatory Factor Analysis
(EFA, CFA) and iii) Structural Equation Model
(SEM) development, highlighting direct and indirect
interrelations between the independent variables
presenting positive or negative effect on school
mode choice.
2 Methodology
2.1 Questionnaire Survey Conduction
For the primary research and data collection
procedure, a questionnaire was designed based on
an in-depth literature review analysis conducted to
identify the factors affecting parents in the school
mode choice process. The research questionnaire
has a structured character of a clear and predefined
sequence of consecutive questions. It consists of
three sections collecting i) data on various socio-
economic characteristics of participants, ii)
information regarding the factors that seem to
motivate parents in the school trip mode choice
process, iii) information regarding parents’ mobility
patterns and perceptions regarding the use of
specific transport modes.
The sections of the questionnaire are as follows.
The first one includes questions regarding the socio-
economic characteristics of the respondents. The
second part includes questions regarding school
trips completion, while the third part consists of
three subsections: in the first one, eighteen crucial
factors that motivate parents in the mode choice
decision process are given for the level of
significance to be defined. For that purpose, a
typical 5-level Likert scale is used (1 corresponds to
very significant, and 5 corresponds to not significant
at all). Following, in the second sub-section, the role
of the structured environment in which students
travel is examined. Parents are asked to declare their
level of agreement or disagreement in 13 statements
describing the environment that includes the route
students follow from their residence to the school
unit and vice versa. Once again, a 5-level Likert
scale is used for that purpose (1 corresponds to
strongly agree, 5 corresponds to strongly disagree).
The questionnaire is completed in the third sub-
section where fifteen statements related to parents’
travel habits are examined to identify the impact of
their perception on the selection of different school
transport modes.
The survey took place in Thessaloniki city, the
second largest city in Greece, numbering
approximately one million residents and 100,000
students. In total, 512 parents of Primary and High
Public-School students of the Thessaloniki
Metropolitan area participated in the questionnaire
survey that took place from May to June and from
September to November 2019. The minimum
sample size was defined based on the following
method, [28]:
-1
2
α/2
N -1 d
n N 1+ p (1- p) z







(1)
where:
N size of the population, e.g., the total number of
students in the under-study area,
n sample size, that is, the number of individuals
required to respond to achieve the desired level
of accuracy,
p a probability parameter estimating the chance
that the sample contains a specific
characteristic. It is an estimation of the
proportion of people (with a specific
characteristic) falling into the group in which
we are interested within the population. If no
previous experience exists (as in the case of our
survey), then a percentage
p = 50% is considered the worst case, [28],
d margin of error that we could accept or tolerate,
such as say ±5%. The margin of error describes
how close the answer of the sample is to the
true value of the population. It is evident that
the smaller the margin of error is, the closer the
findings of the survey are to reality,
zα/2 parameter related to the confidence level (c),
which measures how certain we can be that the
sample accurately reflects the population within
its margin of error. For c = 90%, za/2 = 1.645,
for c = 95%, za/2 = 1.960, and for c = 98%, za/2 =
2.326 (values of za/2 are derived from the two-
tailed standard normal distribution, [28]).
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
521
Volume 22, 2023
Based on Eq. (1) and for N = 100,000 students, p
= 50%, d = ±5%, and zα/2 = 1.96 (confidence level
95%), we calculated that at least 383 questionnaires
were required to be completed. However, in total
512 were collected and used for the SEM
development. The questionnaires’ completion
followed a two-fold process. In person, interviews
were conducted while also parents were invited to
complete the questionnaire online by using a google
docs format file received in their e-mails.
2.2 Exploratory Factor Analysis Results
The Exploratory Factor Analysis (EFA) was initially
adopted to investigate and identify the factors (latent
variables) that the 49 observed variables
(questionnaire items) may form. Initially, the
variables were tested regarding their correlations
(use of Pearson coefficient). The results showed that
there is a large number of statistically significant
correlations making it possible to group the
observed variables into factors. Due to the high
correlation between the two items representing the
preferred mode of transport (residence to school and
school to residence route), only one was used in the
analysis. Regarding the EFA, the principal axis
factoring method was deployed using the direct
oblimin rotation technique, due to the fact that high
correlations (>0.32) in more than 10% were found
in the factor matrix when the varimax rotation
technique was initially applied, [29].
Out of 49 observed variables, 6 were not
included in the analysis as their weights were found
to be less than 0.05. In more detail, the observed
variables excluded are: student’s gender, lack of
appropriate infrastructure for cyclists, constant use
of the private vehicle may form a student’s
dependency on this mode, private vehicle use
contributes to traffic congestion, car ownership may
be a symbol of prestige and the traffic congestion
does not bother me. Additionally, travel cost and
family income although presented with statistically
acceptable weights were finally excluded from
further analysis, as for a factor’s creation more than
two observed variables are required, [29], while also
these two variables could be inserted separately into
the final SEM as exogenous in order their influence
to be examined. The rotation technique identified 9
factors with eigenvalues greater than 1, accounting
for 61% of the total data variation. Based on the
variable’s conceptual framework, the factors'
labelling followed (Table 1).
2.3 Confirmatory Factor Analysis Results
To investigate whether the observed variables’
attribution to the factors is valid, a Confirmatory
Factor Analysis (CFA) was applied, allowing the
correlation between the latent variables and the
errors of the observed variables under the use of
modification indices. Several are the reasons for
examining the correlation between latent variables
and errors of observed variables, namely:
Identification of model: correlations’ estimation
determines the error variance in the observed
variables (not accounted for latent variables),
leading to a distinction between the measurement
error and the true constructs represented by the
latent variables.
Model fit assessment: the examination of
correlations between latent variables and errors
provides useful insights for the model's
adequacy. If a lack of significant correlations is
noticed, problems with measurement are possible
to occur.
measurement precision understanding;
examining the correlations, it can be well
assessed to what degree the observed variables
capture the latent constructs. The higher
correlations between latent variables and errors
noticed the higher precision in measurement is
assured, suggesting that the observed variables
are reliable indicators of the constructs.
According to the CFA, all observed variables
were found to be statistically significant (p-
values<0.001), indicating and confirming their
significant contribution to the creation of factors.
Figure 1 illustrates the flowchart of interrelations
between the variables.
Covariances between factors connected with
two-way arrows in Figure 1, (e.g., MOTMODE and
ATTCAR) are those found highly correlated when
allowed to covariate. Similarly, covariances
between errors connected with two-way arrows (e1–
e7) reveal a high correlation between the observed
variables (e.g., MOTsaf and MOTdist). Correlations
between errors are only allowed for variables
belonging to the same factor and not for different
ones.
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
522
Volume 22, 2023
Fig. 1: Covariances’ significance between the factors and the errors of the observed variables
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
523
Volume 22, 2023
Table 1. Exploratory Factor Analysis: Factors labelling and description
Questionnaire items
(observed variables)
1
Student’s safety (MOTsafe)
2
Travel time (MOTtime)
3
Parents’ working hours (MOTwork)
4
There is someone to help (MOThelp)
5
Student’s age (MOTage)
6
Student’s convenience (MOTconven)
7
Distance from residence to school unit (MOTdist)
8
Student’s socialization (MOTsocial)
9
Student’s health (MOThealth)
10
Luggage weight (MOTweight)
11
Environmental sensitivities (MOTenviron)
12
I have more quality time with my child during the school trip
(MOTqualtime)
13
Satisfied with the comfort of urban bus services
(MOTbuscomf)
14
Urban bus is a reliable transport mode (MOTrelbus)
15
Satisfied with time reliability with urban bus services
(MOTtimerelbus)
16
I like traveling by urban bus within the city (MOTlikebus)
17
Current transport mode used, School-Residence (TMsr)
18
Preferable transport mode, School-Residence (PRTMsr).
19
Current transport mode used, Residence-School (TMrs)
20
I like driving within the city (ATTlikedrive)
21
I use my car for all trips within the city (ATTusecar)
22
Driving is more comfortable than walking/bicycling
(ATTcomfcar)
23
Owing a car makes my life comfortable (ATTcomfdrv)
24
I would prefer my child walk or drive to school under
different circumstances (ATTwalkbike)
25
Walking/bicycling to school is a good way for my child to be
familiar with the environment (ATTfam)
26
Walking or cycling to school increases students’ physical
activity (ATTphac)
27
Parent's car ownership (MMOTcar)
28
Parent's driving license possession (MMOTlic)
29
There are no parking limitations outside my residence or the
school unit (MMOT park)
30
There are no trails of vandalism in the neighborhood
(NEIGHvand)
31
Residences of the neighborhood are in good condition
(NEIGHcond)
32
The neighborhood the student travels to is safe (NEIGHsaf)
33
Sidewalks have sufficient width (SIDwidth)
34
Sidewalks are clean (SIDclean)
35
Sidewalks are separated from traffic with trees (SIDprot)
36
There are no obstacles on the sidewalks (SIDobst)
37
Crossings are safe (SAFcross)
38
Traffic conditions are not dangerous for students (SAFtraf)
39
It’s unlikely for my child to be harassed by others (SAFhar)
40
It’s unlikely for my child to be injured or abducted by a
stranger (SAFinj)
41
There is adequate lighting in the school trip route (SAFlight)
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
524
Volume 22, 2023
Table 2 presents the results of all covariances
included in the model (variables and errors). All
covariances were found statistically significant
(p<0.05, therefore the null hypothesis for non-
significant covariances can be safely rejected).
Regarding the CFA’s modification indices
values, these were calculated within the permissible
limits of international literature, indicating a good
model’s adaption. To further evaluate the model’s
adequacy, reliability analysis was performed
through Cronbach's alpha. All values were above
0.70 and therefore none of them should be
eliminated indicating the high homogeneity of the
variables and their matching to the relative factor.
Table 2. Regression weights and statistical significance of observed variables and factors
Estimate
Standard
error
Critical
ratio
Level of significance
(p-values)
ROUTESAF
↔
NEIGBSAF
0.349
0.042
8.219
< 0.001
ROUTESAF
↔
MOTCAR
-0.231
0.043
-5.423
< 0.001
ROUTESAF
↔
ATTCAR
0.094
0.029
3.220
0.001
ROUTESAF
↔
MODE
-0.330
0.065
-5.086
< 0.001
MOTMODE
↔
NEIGBSAF
-0.087
0.023
-3.797
< 0.001
MOTMODE
↔
MOTCAR
0.354
0.046
7.756
< 0.001
MOTMODE
↔
ATTCAR
-0.072
0.022
-3.282
0.001
MOTMODE
↔
ATTBUS
-0.061
0.020
-3.144
0.002
ATTCAR
↔
NEIGBSAF
0.103
0.025
4.086
< 0.001
MOTCAR
↔
NEIGBSAF
-0.122
0.034
-3.576
< 0.001
MODE
↔
NEIGBSAF
-0.303
0.059
-5.160
< 0.001
MODE
↔
MOTCAR
0.349
0.076
4.590
< 0.001
MODE
↔
ATTWLKBIK
-0.124
0.049
-2.533
0.011
ATTCAR
↔
MOTCAR
0.126
0.035
3.612
< 0.001
ATTBUS
↔
ATTCAR
0.163
0.031
5.234
< 0.001
ROUTESAF
↔
ATTBUS
0.090
0.022
4.005
< 0.001
MOTMODE
↔
ROUTESAF
-0.215
0.032
-6.643
< 0.001
MOTHEALTH
↔
NEIGBSAF
-0.114
0.029
-3.915
< 0.001
MOTHEALTH
↔
MOTCAR
0.525
0.057
9.229
< 0.001
MOTHEALTH
↔
ATTWLKBIK
0.060
0.020
2.915
0.004
MOTHEALTH
↔
ROUTESAF
-0.215
0.037
-5.764
< 0.001
MOTHEALTH
↔
MOTMODE
0.461
0.050
9.147
< 0.001
e14
↔
e15
0.519
0.042
12.354
< 0.001
e9
↔
e8
0.307
0.035
8.746
< 0.001
e5
↔
e3
0.377
0.054
7.002
< 0.001
e7
↔
e2
0.298
0.039
7.618
< 0.001
e13
↔
e12
0.148
0.033
4.518
< 0.001
e42
↔
e44
-0.087
0.034
-2.565
0.010
e42
↔
e43
0.272
0.044
6.229
< 0.001
e18
↔
e17
0.280
0.043
6.467
< 0.001
e37
↔
e38
0.391
0.047
8.391
< 0.001
e11
↔
e10
0.243
0.034
7.184
< 0.001
e10
↔
e9
0.214
0.033
6.396
< 0.001
e11
↔
e9
0.154
0.027
5.612
< 0.001
e13
↔
e14
0.066
0.019
3.507
< 0.001
e3
↔
e2
0.077
0.030
2.588
0.010
e5
↔
e1
-0.069
0.032
-2.136
0.033
e5
↔
e4
0.165
0.048
3.398
< 0.001
e6
↔
e1
0.107
0.038
2.805
0.005
e7
↔
e4
0.103
0.035
2.948
0.003
e10
↔
e8
0.135
0.030
4.553
< 0.001
e25
↔
e26
0.102
0.039
2.601
0.009
e40
↔
e43
0.258
0.041
6.293
< 0.001
e40
↔
e42
0.114
0.038
3.018
0.003
e4
↔
e3
0.119
0.043
2.752
0.006
e6
↔
e2
0.062
0.030
2.035
0.042
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
525
Volume 22, 2023
2.4 The Conceptual Model and the Research
Hypotheses
Based on EFA and CFA results the conceptual
model was created, which is necessary for the
determination of the correlations between the factors
and the observed variables. The conceptual model
(Figure 2) is structured by eight factors, forming
three main categories of effects on school mode
choice and are related to:
1. The motivation, including objective parameters of
parents’ motivation to select a specific transport
mode (MOTMODE factor), motivation
parameters related to the physical and mental
health of the student (MOTHEALTH factor), and
finally, parameters related to the possibility of
using private vehicle for school trips (MOTCAR
factor).
2. The parents’ mobility patterns, including their
shaped perception and attitude regarding the use
of the private vehicle (ATTCAR factor), the use
of bus (ATTBUS factor), and the use of non-
motorized transport modes, namely walking and
cycling (ATTWALKBIK factor),
3. The built environment safety, including the
neighborhood safety (NEIGBSAF factor) and the
route safety (ROUTSAF factor).
What is highlighted in this point, is that the
MODE factor has also been extracted from the EFA
and confirmed by the CFA, incorporating the
choice/preference of parents’ school transport mode.
This is the factor that forms the core of the
conceptual model, the dependent variable.
Based on the conceptual model, the research
hypotheses (H) were built, the validity of which was
subsequently examined with SEM’s development.
The main hypotheses considered in the present
study are the following:
Η1: Do motivation factors affect directly or
indirectly the school mode choice?
Η2: Do the parents’ shaped travel patterns affect
directly or indirectly the school mode choice?
Η3: Do the built environment safety factors affect
directly or indirectly the school mode choice?
Η4: Do the exogenous factors affect directly or
indirectly the school mode choice?
Given the fact that within the conceptual model,
there are three factors expressing motivation
(MOTMODE, MOTHEALTH, MOTCAR) and
another three expressing the shaped perceptions and
attitudes of parents towards motorized and non-
motorized transport modes (ATTCAR, ATTBUS,
ATTWALKBIK) it will be further examined
whether these factors can create second-order
factors.
Finally, beyond the four basic hypotheses that
were previously posed and will be examined
through SEM, any correlations of the factors
themselves (latent variables) with each other will be
tested.
Fig. 2: The conceptual model
3 Structural Equation Model Results
and Conclusions
Structural Equation Model (SEM) was evaluated
using the maximum likelihood technique, which
attempts to estimate the factor model’s parameters
that are very likely to produce the initial correlation
matrix, assuming that the sample conforms with the
multivariate normal distribution. The values
obtained by the goodness-of-fit indices are the same
as those of CFA, proving an adequate SEM. More
specifically (values in parentheses show the desired
literature values for each indicator, [28], [29]): IFI =
0.91 (≥ 0.90), TLI = 0.90 (≥ 0.90), CFI = 0.91 (≥
0.90), RMSEA = 0.06 (< 0.08) and X2 (CMIN/DF)
= 2.65 (between 1 and 3).
The model is composed of (Figure 3):
i. The three factors expressing parents' motivation
to select the school transport mode
(MOTMODE, MOTCAR, MOTHEALTH
factors).
ii. Three factors expressing the parent's attitude
towards non-motorized modes; walking and
bicycling, and motorized modes; car and bus,
(factors ATTWALKBIK, ATTCAR, ATTBUS),
Mode choice
(mode)
BUILT-ENVIRONMENT SAFETY
PARENTS’ SPAPED TRAVEL HABITS
MOTIVATION
EXOGENOUS FACTORS
Parent’s attitude
regarding the use
of non motorized
transport modes
(walking, bicycling)
Parent’s attitude
regarding the use
of private vehicle
Parent’s attitude
regarding the use
of public transport
Motivation
(based on school
travel
characteristics)
Motivation (based on
elements affecting
students’ physical and
mental health)
Motivation
(related to the
usability of private
vehicle)
Parent’s
professional
status
Parents’
educational
level
Parent’s
age
Parents’
educational
level
Neighbourhood
safety
Route safety
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
526
Volume 22, 2023
iii. Two factors expressing safety (NEIGBSAF,
ROUTSAF factors).
iv. The dependent variable (which as emerged from
the CFA consists of a separate factor) represents
the school mode choice (MODE).
Fig. 3: Graphic depiction of SEM
The developed SEM contains the multi-
dimensional element, known as the second-order
factor since the three factors of parental motivation
(MOTMODE, MOTCAR, and MOTHEALTH)
configure the second-order factor MOT. According
to SEM’s results, all the interrelations between
factors and errors are statistically significant (p
<0.05). The developed SEM is distinguished by a
series of direct and indirect correlations and
interrelations between the factors directing to the
dependent variable which is the transport mode
(MODE).
The main conclusions emerging from SEM’s
development, and more specifically from the
interrelation (denoted as ↔) and the relation
(denoted as ←) effect analysis, are the following
(Table 3):
Three factors appear to have an immediate effect
on the school mode choice (blue arrows of Fig.
3). The factor representing neighborhood safety
(NEIGBSAF), the motivation factor (MOT), and
the motivation factor affecting parents on using
private vehicles (MOTCAR).
The remaining variables indirectly affect the
dependent variable (MODE), through the three
latent variables (MOTMODE, MOTHEALTH,
MOTCAR in green circles) that configure the
second-order factor (MOT), indicating parental
motivation to the mode choice decision. All these
effects derive from the latent variables related to
parents’ attitudes towards the use of private
vehicles (ATTCAR), buses (ATTBUS), and
alternative transport modes such as walking and
bicycle (ATTWALKBIK), as well as the
variables representing the safety of the school
route the student follows (ROUTSAF) and the
safety provided by the neighborhood
(NEIGBSAF) the student moves for reaching the
school unit.
The factor representing students’ mental and
physical health (MOTHEALTH) appears to be
inactive, as no other factor seems to affect it.
However, its contribution to the configuration of
parents’ motivation factor (MOT) is important.
Parents’ shaped attitude towards the use of non-
motorized transport modes (ATTWALKBIK)
has a positive effect (0.303, p=0.0.2) on
motivation (MOT) representing the positive
predisposition of parents for students to walk or
bike to and/or from the school unit. At the same
time, it has a negative effect (-0.261, p=0.004) on
MOTMODE, concluding that this positive
predisposition is significantly reduced when
considering the parameters composing the latent
variable MOTMODE, such as student’s age,
distance to school, student’s comfort, travel time,
etc.
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
527
Volume 22, 2023
Table 3. Interrelation (denoted as ↔) and relation
(denoted as ←) effect analysis of SEM
Estimate
Level of
significance
(p-values)
ATTCAR
↔
ATTBUS
0.303
< 0.001
NEIGBSAF
↔
ROUTESAF
0.447
< 0.001
ATTBUS
↔
ROUTESAF
0.164
< 0.001
NEIGBSAF
↔
ATTCAR
0.147
< 0.001
ATTCAR
↔
ROUTESAF
0.171
< 0.001
MOT
←
NEIGBSAF
-0.460
< 0.001
MOT
←
ATTWALKBIK
0.303
0.002
MOTCAR
←
MOT
0.952
< 0.001
MOTCAR
←
ATTCAR
0.242
< 0.001
MOTCAR
←
ROUTESAF
-0.154
0.012
MODE
←
MOT
-0.263
0.036
MODE
←
NEIGBSAF
-0.761
< 0.001
MODE
←
MOTCAR
0.688
< 0.001
MOTMODE
←
ROUTESAF
-0.152
0.001
MOTMODE
←
ATTCAR
-0.107
0.011
MOTMODE
←
ATTWALKBK
-0.261
0.004
MOTHEALTH
←
MOT
0.871
< 0.001
MOTMODE
←
MOT
0.768
< 0.001
Parents’ shaped attitude towards the use of buses
(ATTBUS) does not appear to have any direct
effect on any of the other variables, but only a
two-way interaction with parents’ shaped attitude
towards the use of private vehicles due to the
comforts offered (ATTCAR) and the feeling of
security provided by the sidewalks and the
school route the student follows (ROUTSAF).
The first two-way relationship (ATTBUS ↔
ATTCAR) presents a high correlation (0.303,
p<0.001), indicating on one hand the
complementary nature of private vehicle and bus
use and on the other the positive attitude of
parents towards the use of motorized transport
modes. The second two-way relationship
(ATTBUS ↔ ROUTSAF) is noticed lower but it
is also statistically significant (0.164, p<0.001).
This relationship can be interpreted from the fact
that the use of the bus for school travel is part of
a more complicated process, as it is combined
with walking (the student has to walk from the
bus stop to the school unit and vice versa, or/and
from the residence to the bus stop and vice
versa). More specifically, this is the first and last
part of the school trip directly linked to the safety
of the route the student follows. Therefore, in
deciding whether the student will travel by bus or
not, the parent has to also consider the factors
affecting the safety levels provided by the route
the student follows.
Additionally, observing the set of two-way
relationships between the latent variables related
to the parents’ shaped attitude towards the use of
private vehicle and the comforts it offers
(ATTCAR) and bus (ATTBUS), and the
variables related to the route safety (ROUTSAF)
and neighborhood safety (NEIGBSAF), it is
clear that these four variables influence each
other. At this point, it should be mentioned that
the two latent variables representing route safety
(ROUTSAF) and neighborhood safety
(NEIGBSAF) were found to have a statistically
significant correlation (0.447, p<0.001). Even
though this correlation presents a high
covariance, a new second-order factor could not
be configured, most probably due to the absence
of one or two additional latent variables that
could contribute to the configuration of such a
second-order factor.
A similar inability to configure a second-order
factor was presented in all three latent variables
related to the parents’ shaped attitude towards
the use of the motorized transport modes bus and
private vehicle (ATTCAR and ATTBUS) and the
non-motorized transport modes bicycle and
walking (ATTWALKBIK)). Although the first
two variables are correlated with statistical
significance (0.303, p < 0.001), the variable
representing parents’ shaped attitude towards the
use of non-motorized transport modes does not
present any correlation both to the attitude
towards motorized modes and the remaining
latent variables of the SEM. This can be
theoretically interpreted by the fact that these
two transport mode options (motorized and non-
motorized) are diametrically opposed to
configuring two different transport users’
categories.
A significant influence on the factor
ATTWALKBIK deriving from at least one of the
two factors composing the neighborhood safety
(NEIGBSAF and ROUTSAF) was not observed,
although this was expected.
Parents’ shaped attitude towards the use of a
private vehicle (ATTCAR) has a positive effect
(0.242, p<0.001) on the factor depicting the
motivation of using a private vehicle
(MOTCAR), which in turn has a positive and
statistically significant effect (0.392, p<0.001) on
the school mode choice (MODE).
At the same time, the parent’s attitude towards
the use of a private vehicle (ATTCAR) has a
negative effect (-0.107, p=0.011) on the factor
consisting of the parameters of parents’
motivation to select a transport mode
(MOTMODE). This negative effect indicates that
the more positive the parents are on using the
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
528
Volume 22, 2023
private vehicle due to the comfort this offers, the
less is the effect of the parameters composing the
factor MOTMODE (student’s age, school
distance, student’s comfort, convenience, school
travel time, etc.). The positive attitude of parents
towards the use of private vehicles leads them to
completely ignore or not sufficiently evaluate the
observed variables composing MOTMODE.
Consequently, parents’ shaped attitude towards
the use of private vehicles negatively affects part
of their motivation in relation to these variables,
which, however, are particularly important in the
configuration of the latent variable MOT
(representing the overall parents’ motivation).
After all, MOTMODE configures to a large
degree MOT based on the comparatively large
positive effect noticed (0.768, p<0.001).
Both the parents’ shaped attitude towards the use
of a private vehicle (ATTCAR) and the factor
related to car possession and the ease of finding a
parking space close to the residence or the school
unit (MOTCAR) appear to contribute negatively
to the whole system that tries to interpret the
school mode choice process. This negative effect
hides that private vehicle ownership and parents'
attitude towards its use can motivate the mode
choice in favor of private vehicle, leading parents
to overlook essential parameters (e.g., distance
from residence to school unit, travel time,
student’s age, etc.). Therefore, it can be well
argued that parents who strongly support the use
of private vehicles in the school mode choice
process tend to ignore the observed variables
composing MOTMODE.
The school route’s safety (ROUTSAF) is found
to have a negative effect (-0.154, p=0.012) on the
factor of parental motivation based on private
vehicle possession and the ease of finding a
parking space close to the residence or the school
unit (MOTCAR). This relationship indicates (and
at the same time confirms the logical sense) that
the greater safety provided by the school route,
the less motivated the parent is to use his vehicle.
Neighborhood safety (NEIGBSAF) seems to be
particularly important, as it affects not only the
motivation of the parent in general but also
directly affects the transport mode choice. Its
negative effect on MODE is mainly based on the
strongest coefficient noticed in the model (-
0.761, p<0.001). Dilapidated or damaged
buildings, traces of vandalism, and the feeling
that the built environment is unsafe seem to
negatively affect the mode choice process. This
means that parents are essentially obliged to
choose the mode that provides the greatest
possible security. Therefore, the lower the safety
levels, the more a parent tends to use a private
vehicle.
Private vehicle appears to play a strong role in
the school mode choice process. Among the
three factors that directly affect the transport
mode (MODE), the factor MOTMODE has a
positive effect (0.688, p<0.001), while the two
others, NEIGBSAF (-0.761, p<0.010) and MOT
(-0.263, p=0.036), negative. Therefore, the more
important car ownership and driving license
possession are considered by parents and the less
insignificant the restrictions of finding a parking
slot close to the residence and the school unit are,
the more parents tend to use their private
vehicles.
4 Discussion for Further Research
The present study presents perspectives for future
additions that could, on one hand, improve the
content and the expected result, while, on the other
hand, providing answers to other research findings.
In this context, the following paragraphs present
additional issues to be explored that may lead to an
extension of the findings of the present study.
The research was conducted in a Greek urban
city, in which policies and interventions that serve
the standards and principles of sustainable urban
mobility have begun to be adopted and implemented
only in recent years and this is the case to a greater
or lesser extent in all other Greek cities. The
complete lack of appropriate infrastructure or even
the inadequacy of existing infrastructure seems to
negatively affect the attitude of parents towards the
adoption of different travel patterns, thus failing to
enhance the use of alternative transport modes. An
integrated walking and bicycle network as well as
the creation of school rings around the school units,
could potentially further reduce the use of private
vehicles for school trips completion. Further
research could hypothetically focus on the existence
of relatively organized infrastructure and examine
the intention of parents to select the use of bicycles
for school trips (found to be completely limited in
the current study), but also to further increase
walking even from the first grades of elementary
school. The present study showed a superiority of
walking (mainly due to the proximity of the school
unit to the student's residence), however in most
cases the choice of a student “walking alone”
mainly concerns older age groups. It would
therefore be of scientific interest to examine
parents’ behaviour, considering the existence of
appropriate infrastructure, as beyond the security
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
529
Volume 22, 2023
provided by such infrastructure, any personal
insecurities of parents that are not primarily related
to infrastructure but with deeper prejudices and
fears, might emerge.
In the current questionnaire survey, parents were
asked to evaluate only the quality of services of
Thessaloniki’s public transport system leaving out
the assessment of school buses provided to students
by Greek prefectures according to the current
legislation. Therefore, the evaluation as formed and
reflected by SEM presents a generally negative
attitude of parents towards the use of buses. Adding
targeted questions to the questionnaire regarding the
use of dedicated school buses would probably lead
to different results.
The research focused mainly on the urban
environment. The incorporation of rural areas would
potentially outline different characteristics of school
trips and highlight different needs and requirements
for improving the school transportation system.
References:
[1] Kotoula, K.M., Botzoris, G., Ayfantopoulou,
and G., Morfoulaki, M., The Existing School
Transportation Framework in Greece –
Barriers and Problems Comparing to other
European Countries, Transportation Research
Procedia, Vol. 24, 2016, pp. 385-392.
[2] Kotoula, K.M., Botzoris, G., Ayfantopoulou,
G., and Profillidis, V., Urban School Travel–
Understanding the Critical Factors Affecting
Parent’s Choices. In Advances in Mobility-as-
a-Service Systems: Proceedings of 5th
Conference on Sustainable Urban Mobility,
pp. 912-922, Cham: Springer International
Publishing, June 2020.
[3] Noland, R., Park H., Von Hagen, L.A., and
Chatman, D., A Mode Choice Analysis of
School Trips in New Jersey, In 89th Annual
Meeting of the Transportation Research
Board, Washington, D.C., January 2010.
[4] Seraj, S., Sidharthan, R., Bhat, C., Pendyala,
R., and Goulias, K.G., Parental Attitudes
Towards Children Walking and Bicycling to
School: A Multivariate Ordered Response
Analysis. In 91st Annual Meeting of the
Transportation Research Board, Washington,
D.C., January 2012.
[5] Yang, Y. and Markowitz, E., Integrating
Parental Attitudes in Research on Children’s
Active School Commuting: Evidence from a
Community School Travel Survey,
Transportation Research Record, Journal of
the Transportation Research Board, Vol.
2318, No. 1, 2012, pp. 116-127.
[6] Larsen, K., Gilliland, J., Hess, P., Tucker, P.,
Irwin, J., and He, M., The Influence of the
Physical Environment and Socio-demographic
Characteristics on Children’s Mode of Travel
to and from School, American Journal of
Public Health, Vol. 99, No. 3, 2009, pp. 520-
526.
[7] Clifton, J., Independent Mobility Among
Teenagers: Exploration of Travel to After-
School Activities. Transportation Research
Record: Journal of the Transportation
Research Board, Vol. 1854, 2010, pp.74-80.
[8] Larsen, K., Buliung, R.N., and Faulkner,
G.E.J., Safety and School Travel: How does
the Environment along the Route Relates to
Safety and Mode Choice, Transportation
Research Record, Journal of the
Transportation Research Board, Vol. 2327,
2013, pp. 9-18.
[9] Panter, J.R., Jones, A.P., Sluijs, E.M.F., and
Griffin, S.J., Attitudes, Social Support and
Environmental Perceptions as Predictors of
Active Commuting in School Children,
Journal Epidemiology and Community
Health, Vol. 64, 2010, pp. 41-48.
[10] Schlossberg, M.J., Greene, P., Phillips, P.,
Johnson, B., and Parker, B., School Trips:
Effects of Urban Form and Distance on Travel
Mode, Journal of the American Planning
Association, Vol. 72, No. 3, 2006, pp. 337-
346.
[11] Leslie, E., Kremer, P., Toumbourou, J.W., and
Williams, J.W., Gender Differences in
Personal, Social and Environmental
Influences on Active Travel to and from
School for Australian Adolescents, Journal of
Science and Medical in Sport, Vol. 13, No. 6,
2010, pp. 597-601.
[12] Loucaides, C.A., Jago, R., and Theophanous,
M., Prevalence and Correlates of Active
Traveling to School Among Adolescents in
Cyprus, Central European Journal of Public
Health, Vol. 18, 2010, pp. 151-156.
[13] Mendoza, J.A., Watson, K., Baranowski, T.,
Nicklas, T.A., Uscanga, D.K., Nguyen, N.,
and Hanfling, M.J., Ethnic minority Children's
Active Commuting to School and Association
with Physical Activity and Pedestrian Safety
Behaviors, Journal of Applied Research on
Children, Vol. 1, No. 1, 2010, pp. 1-23.
[14] Rodriguez, A. and Vogt, C.A., Demographic,
Environmental, Access, and Attitude Factors
that Influence Walking to School by
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
530
Volume 22, 2023
Elementary School-aged Children, Journal of
School Health, Vol. 79, 2009, pp. 255-261.
[15] Yu, C., Zhu, X., Impacts of Residential Self-
selection and Built Environments on
Children's Walking-to-school Behaviors,
Environment and Behavior, Vol. 47, No. 3,
2015, pp. 268-287.
[16] Yu, C. and Zhu, X., From Attitude to Action:
What Shapes Attitude Toward Walking
to/from School and How does it Influence
Actual Behaviors, Preventive Medicine, Vol.
90, 2016, pp. 72-78.
[17] Fyhri, A. and Hjorthol, R., Children’s
Independent Mobility to School, Friends and
Leisure Activities, Journal of Transport
Geography, Vol. 17, No. 5, 2009, pp. 377-
384.
[18] Carver, A., Timperio, A., and Crawford, D.,
Parental Chauffeurs: What Drives their
Transport Choice, Journal of Transport
Geography, Vol. 26, 2013, pp. 72-77.
[19] Curtis, C., Babb, C., and Olaru, D., Built
Environment and Children’s Travel to School,
Transportation Policy, Vol. 42, 2015, pp. 21-
33.
[20] Waygood, E.O.D. and Susilo, Y.O., Walking
to School in Scotland: Do Perceptions of
Neighbourhood Quality Matter, IATSS
Research-Journal, Vol. 38, No. 2, 2015, pp.
125-129.
[21] Manz, W., Bauer, U., Herget, M., and
Scheiner, J., Familienmobilität im Alltag,
Final Report of Project “Determinanten und
Handlungsansätze der Familienmobilität”,
Federal Ministry of Transport and Digital
Infrastructure, 2015, Berlin.
[22] Kerr, J., Rosenberg, D., Sallis, J.F., Saelens,
B.E., Frank, L.D., and Conway, T.L., Active
Commuting to School: Associations with
Environment and Parental Concerns,
Medicine and Science in Sports and Exercise,
Vol. 38, No. 4, 2006, pp. 787-794.
[23] Mitra, R. and Buliung, R.N., Intra-Household
Travel Interactions, the Built Environment
and School Travel Mode Choice: an
Exploration using Spatial Models, In 91st
Annual Meeting of the Transportation
Research Board, Washington, D.C., January
2012.
[24] Basbas, S., Tetou, V., and Politis, I., Ordinal
and Binary Logistic Logit Models for
Examination of Behavioral, Infrastructure and
Perception Factors Influencing Biking, WIT
Transactions on the Built Environment, Vol.
130, 2013, pp. 573-584.
[25] Dellinger, A.M., and Staunton, C.E., Barriers
to Children Walking and Bicycling to School-
United States, Journal of the American
Medical Association, Vol. 51, No. 32, 2002,
701-704.
[26] Martin, S. and Carlson, S., Barriers to children
Walking to or from School – United States.
Journal of the American Medical Association,
Vol. 294, No. 17, 2005, pp. 2160-2161.
[27] McDonald, N.C., Aalborg, A.E., Why parents
drive children to school: implications for Safe
Routes to School Programs, Journal of the
American Planning Association, Vol. 75, No.
3, 2009, pp. 331-342.
[28] Profillidis, V.A. and Botzoris, G.N. Modeling
of Transport Demand, Amsterdam: Elsevier,
2018.
[29] Tabachnick, B.G. and Fidell, L.S. Using
Multivariate Statistics – 5th Edition. Boston:
Pearson Education Inc., 2007.
Contribution of Individual Authors to the
Creation of a Scientific Article (Ghostwriting
Policy)
-Kornilia Maria Kotoula carried out the literature
review (section 1) and conducted the questionnaire
survey (sub-section 2.1).
-Kornilia Maria Kotoula and George Botzoris were
responsible for the factor analysis deployment (sub-
sections 2.2, 2.3, and 2.4), SEM’s development
(section 3), and discussion for further research
(section 4).
Sources of Funding for Research Presented in a
Scientific Article or Scientific Article Itself
No funding was received for conducting this study.
Conflict of Interest
The authors have no conflict of interest to declare.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en
_US
WSEAS TRANSACTIONS on SYSTEMS
DOI: 10.37394/23202.2023.22.52
Kornilia Maria Kotoula, George Botzoris
E-ISSN: 2224-2678
531
Volume 22, 2023