modelling portferry choice in roro freight transportation (2002)
TRANSCRIPT
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
1/14
Modelling port/ferry choice in RoRo freight transportation
John Mangan a,*, Chandra Lalwani b, Bernard Gardner b
a Irish Management Institute, Sandyford Road, Dublin 16, Irelandb Cardiff Business School, Cardiff University, Wales, UK
Abstract
Maritime transport choice is an often neglected area in the transport literature. This paper makes two contributions to that
literature: firstly, it brings together all of the major contributions to date to the port/ferry choice literature (and particularly those of
Brooks, DEste, Matear and Gray, and other authors); secondly, it employs an in-depth, triangulated research methodology which
incorporates both quantitative and qualitative methodologies for an investigation of port/ferry choice in the Ireland/UK and Ire-land/Continental Europe markets. The qualitative methodology employed, construct elicitation, has not previously been employed
in this context. In addition, determinants of choice and the role of other selection variables have been elucidated and two techniques
(input-oriented modelling based on the Aaker and Day model and process-oriented modelling) have been employed to model the
decision making process. 2002 Elsevier Science Ltd. All rights reserved.
Keywords: RoRo freight transport; Port/ferry choice; Transport modelling
1. Introduction
This paper reports the results of a study into Roll-On/
Roll-Off (hereafter referred to as RoRo) port/ferry route
choice in the Ireland/UK and Ireland/Continental Eu-rope markets. This market is first described and the
literature on decision variables in transport choice, and
on modelling transport choice, is then detailed. The
output of a triangulated, three-phase research study is
then reported.
1.1. Research context
The RoRo freight market is of central importance to
the Republic of Irelands Celtic Tiger economy, an
economy that has grown considerably in recent years.
The Republic of Ireland (hereafter referred to as Ire-
land) is an island country geographically located in theNorthwest of Europe with a population of some 3.75
million people. Irelands recent economic success has
resulted from growth in both the manufacturing and
service sectors and is a consequence of, inter alia, a
combination of careful economic planning, investment
in infrastructure, high standards of education and, not
least, EU grant aid. A member of the EEC/EC/EU
since 1973, Ireland was one of the first qualifiers for
European Monetary Union (EMU) and is now a
member of the single currency (Irish Pound IR Euro
1.27). GNP is substantially lower than Gross Domestic
Product (GDP) in Ireland due to the importance ofprofit repatriations by foreign firms and interest pay-
ments on the national debt. Forecast GDP for 1999 was
IR68,299m (GDP per capita was 23,125) and fore-
cast GNP was IR58,369m. In 1999 exports from Ire-
land were forecast at circa IR58.2 billion and imports
at circa IR49.5 billion yielding a balance of trade
surplus of circa IR8.7 billion. According to the
Economic and Social Research Institute (www.esri.le),
Irelands GDP was forecast to increase by 8.8% in
2000.
A feature of the whole island of Ireland is that, since
the opening of the channel tunnel linking Britain with
Continental Europe, Ireland is now the only EU mem-ber country without a landlink to the rest of the EU and
is thus totally dependent on both the air and maritime
transport modes for external access and egress. Ireland
has a large economic dependence on external trade
(which is the fastest growing sector of the economy) and
is in a peripheral location vis-aa-vis the economic centre
of gravity of the EU. Consequently, ports and ferry
services are of special importance to the Irish economy.
Fig. 1 illustrates both the growth and corridor shares
of RoRo freight in the Irish market while Fig. 2 illus-
trates the RoRo routes to and from Ireland. RoRo
International Journal of Transport Management 1 (2002) 1528
www.elsevier.com/locate/traman
* Corresponding author. Tel.: +353-1-2078574; fax: +353-1-2955150.
E-mail address: [email protected] (J. Mangan).
1471-4051/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 1 4 7 1 - 4 0 5 1 ( 0 1 ) 0 0 0 0 3 - 9
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
2/14
freight comprises approximately 19% by volume of all
goods handled at Irish ports. The northern corridor
comprises the ports of Belfast, Larne and Warrenpointin Northern Ireland, the central corridor comprises
Dublin and Dun Laoghaire, both in the Republic of
Ireland, and the southern corridor comprises Cork and
Rosslare, both also in the Republic of Ireland.
It is apparent from Fig. 1 that the northern corridor
has the largest share of the market, albeit a share which
has been declining somewhat in more recent years.
Northern Ireland has a population of approximately 1.6
million and an economy which has not mirrored that of
the Celtic Tiger in terms of growth and prosperity.
Consequently its share of the whole islands RoRo
freight traffic is somewhat disproportionate. Assumingan origin/destination spread of RoRo freight traffic
broadly in proportion to economic output in the Re-
public of Ireland vis-aa-vis Northern Ireland, it is evident
that a significant share of the Republic of Ireland traffic
has been in the past, and to a lesser, but still significant,
extent is still availing of RoRo ferry services to and from
Northern Ireland ports. The usually cited reason for this
scenario is the higher frequency of cheaper sailings with
shorter sailing times from the Northern Ireland ports. It
needs to be stressed of course that there is nothing
wrong with Republic of Ireland shippers using Northern
Ireland ports they are merely making what they per-
ceive to be the optimum choice in a free market. It doeshowever raise the question of whether or not ineffi-
ciencies exist in the whole RoRo freight market as a
result of shippers having to route their traffic through
distant ports in order to avail of what they perceive to be
the best ferry route choice.
It is estimated that approximately 47% of the RoRo
freight traffic in the entire Irish market travels unac-
companied. Furthermore, it is estimated that approxi-
mately only 2% of RoRo freight traffic in the Irish
market travels on direct sailings to Continental Europe,
the remaining traffic (i.e. 98%) either terminating in the
UK or else landbridging the UK enroute to other
destinations.
2. Transport choice variables
Murphy and Hall (1995) reviewed a range of studies
from the 1970s, 1980s and 1990s that evaluated modal
or carrier selection decision making in transport. The
aggregate rankings across all of these studies of the
relevant variables which determined choice are shown in
Table 1. Murphy and Hall add that new variables are
emerging from studies of transport choice in the 1990s
such variables include rate negotiation, service nego-
tiation, carrier response in emergencies, willingness to
improve service quality, and quality of dispatch per-
sonnel. They add however that more definitive conclu-
sions will require further research. Cullinane and Toy
(2000) undertook a content analysis (a technique which
can measure the frequency of different words or morecomplex measures in a series of documents) of 75 articles
contained in a database of literature relevant to the
subject of freight route/mode choice decisions and ob-
tained (using the less abstract methods of content
analysis) the following hierarchy:
1. cost/price/rate;
2. speed;
3. transit time reliability;
4. characteristics of the goods;
5. service.
Cullinane and Toy stress the importance of accu-
rately identifying relevant factors in transport choice
studies. Factors so identified are often used as inputs to
predictive models. The output of these models some-
times differs to the actual practice of decision makers
and this may occur because the wrong input factors were
identified and used in the first instance, hence illustrating
the importance of correctly identifying the relevant
factors.
Hall and Wagner (1996) provided evidence to show
that the key selection criteria for one mode or modal
segment may not be applicable or critical for another
mode or modal segment, hence it is important to at-
tempt to distil the more critical factors for a specific
context. DEste and Meyrick (1992) distinguishedquantitative factors (e.g. frequency, cost), which could
potentially be measured and compared in an objective
manner, and qualitative factors (e.g. marketing, tradi-
tion, etc.) which were more subjective, in transport
choice. Given then the range of pertinent choice factors
it is important when studying transport choice to cap-
ture the impact of all of these factors, including those
qualitative factors which may be difficult to measure
and, as stated, to distil those critical factors which are
specific to the modal segment considered in a given re-
search effort.
Fig. 1. RoRo freight units (import and export) on the three corridors.
(Data from the Central Statistics Office Statistics of Port Traffic
published annually, and from Trade at the Principle Ports published
annually by the Department of Economic Development, Belfast.)
16 J. Mangan et al. / Transport Management 1 (2002) 1528
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
3/14
Fig. 2. RoRo routes to and from Ireland.
J. Mangan et al. / Transport Management 1 (2002) 1528 17
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
4/14
Table 2 lists, in descending order of importance,
factors considered important in port/shipping company
choice which were identified in three studies which
looked at port/shipping company selection in the mari-
time context.
Empirical evidence shows that differences in selection
criteria are often observed among decision makers in
transportation. Murphy et al. (1992) showed that portselection factors are evaluated differently by freight for-
warders, large shippers, small shippers, ferry operators
and port managers. Matear and Gray (1993) showed that
shippers and freight forwarders employ different criteria
in selecting transport services. DEste and Meyrick
(1992) observed differences between freight forwarders
and producers (i.e. consignors) in selection criteria and
they cited several other studies which found that differ-
ences exist between shipper and carrier perceptions of the
selection variables. Murphy et al. (1997) however noted
that while there may be differences between shippers and
carriers in terms of mean importance ratings for different
factors, the relative importance of the different factors
within both groups (which could be illustrated by cor-
relation) may be similar. DEste and Meyrick (1992), in
their study of the AustraliaTasmania RoRo market,
noted that predictably consignors were more inclined to
leave shipping decisions to an agent or haulier.
The context within which freight transport choice
decisions are made has changed considerably in recent
years. This has arisen as a result of improvements in
transport efficiency which have changed considerably
the operational effectiveness of the transport sector, the
evolution of logistics and supply chain management
within which transport services are key links, and in-
creased outsouring of the transport function by firms.
3. Modelling transport choice
Fortunately (or unfortunately!) there are copiousapproaches to modelling freight transport choice.
DEste (1992b) categorised the various modelling ap-
proaches into three broad categories, viz:
Input-oriented models.
Outcome-oriented models.
Process-oriented models.
This paper will follow this broad categorisation
(although it must be stressed that the classification in this
paper of the various reviewed techniques into these three
broad categories is the onus of these authors, not
DEste).
Table 1
Variables affecting freight transportation choice (Murphy and Hall,
1995)
1990s Rankings
1970s 1980s
Reliability 1 1 1
Freight rates 4.5 2 3.5
Transit time 2 3 5.5
Carrier considerations 6 5 2
Shipper market considerations 4.5 5 3.5
Over, short and damaged 3 5 5.5
Table 2
Factors important in port/shipping company choice
Matear and Gray (1993)
Irish sea: all maritime
traffic types
Spencer et al. (1992)
RoRo across the
English Channel
DEste and Meyrick (1992) RoRo
between Australia and Tasmania: ferry
factors
DEste and Meyrick (1992) RoRo
between Australia and Tasmania:
port factors
Punctuality of ferry Service frequency Frequency Proximity to origin
Space availability Convenient schedules Price Port turnaround time
Service frequency Delays, cancellations Transit time Record of strikes
Response to problems On fastest route to
destination
On-time Loading facilities
Value for money rate Space availability Damage Port marketing
Ferry arrival time Fast check-in /disem-
barkation
Commitment Port charges
Ferry departure time Speed of customs Problems Tradition
Sea crossing time On che ape st route t o
destination
Technology
Low freight rate Low tariffs Extra space
Carrier relationship Port vehicle congestion Door-to-door service
Proximity to freight
destination
Pre-booking facilities Flexible contracts
Proximity to freight
origin
Chance for driver break Long contracts
Special rates/discounts Ferry drivers facilities Promotion
Shipper preference Congested roads to port
Standard of these roads
Note: The factors are ranked in descending order of relative importance.
18 J. Mangan et al. / Transport Management 1 (2002) 1528
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
5/14
3.1. Input-oriented models
These models relate to the range and relative impor-
tance of the various factors that influence carrier choice
(the decision making ingredients) but do not give an
insight into the actual decision making process. In the
view however of Pisharodi (1991) there is a need to shift
the focus of transport choice decision modelling from
the determination of factors which influence decisions
(as is the case with input-oriented modelling) to the
determination of the activities actually involved in de-
cision making. Examples of input-oriented modelling
include: calculating importance means (e.g. Hall and
Wagner, 1996), factor analysis (e.g. Stank et al., 1996),
and the Aaker and Day Model(e.g. Brooks, 1985, 1990,
1995, 2000). The latter technique is useful in that it
distinguishes attributes judged to be merely important
from actual determinants of choice: Brooks called the
latter factors salient factors (Fig. 3).
3.2. Outcome-oriented models
These models are concerned with predicting the
outcome of a particular decision situation and, ac-
cording to DEste (1992b), they tend to be broad-brush
and mechanistic in nature, with their success judged by
their predictive power rather than their explanatory
ability. In these models mathematical formulations are
developed which attempt to predict the decision out-
come, at least in an aggregated sense. Examples of
outcome-oriented models include logit modelling (e.g.
Spencer et al., 1992), decision trees (e.g. Coles and
Rowley, 1995), the Analytic Hierarchy Process (e.g.
Lehmusvaara et al., 1999) and Conjoint Analysis (e.g.
Aaker et al., 1995).
Outcome-oriented models are not however without
their critics. DEste (1992b) expressed the view that
outcome-oriented models have done little to improve the
level of understanding of ferry choice in the maritime
context, while Gray (1982) noted that better results
would be achieved if greater effort was devoted to un-
derstanding how shippers and other interested parties
behave, rather than developing more sophisticated
mathematical models.
3.3. Process-oriented models
These models attempt to explain how the various
pertinent factors interact (the decision making ingredi-
ents), and the nature of the environment within which
the interaction occurs, to produce the observed decision.
In essence then such models represent the carrier choice
paradigm, the recipe for decision making. A good ex-
ample of such a model is the process-oriented model
developed by DEste (1992a) to represent ferry choice in
the RoRo freight market between Australia and Tas-
mania; a similar such model was developed by Brooks
(1990) for a different market.
Central to any process-oriented decision model is an
understanding of how the various influences fit together
and interact. DEste (1992a) identified a strategic level of
interaction where there is an arrangement of the various
stages in the evaluation and choice process, and which
can be represented in flowchart form, and a lower level
of interaction of individual choice factors which can beeithercompensatoryornon-compensatory.Compensatory
interaction allows strong performance in one factor to
compensate for poor performance in other factors, thus
usually combining the effects of all decision factors into
a single decision variable (e.g. as is the case in regression
based modelling). Non-compensatory interaction how-
ever stipulates that there are minimum levels of perfor-
mance and that any given factor cannot necessarily be
traded-off against other factors (closely related to these
concepts of compensatory and non-compensatory in-
teraction is the shippers objective which can be either
one ofsatisfyingor maximising).
3.4. Modelling transport choice: the need to take a holistic
view
DEste (1992b) noted that carrier choice cannot be
understood as an isolated decision in time and space. It
encompasses past, current and future implications in
both the transport and wider organisational context.
He recommends that, in carrier choice research, there is
a need to take a much broader perspective that places
the decision in context in space and time, while carrier
choice factors should be understood and not just mea-
sured. In a similar vein, Pedersen and Gray (1998) argueagainst the assumption of context-free determinants of
choice in transport studies and they suggest that the
context within which the choice is made must be con-
sidered in such studies. Brooks (1995) concluded that a
market is not homogeneous in its requirements of car-
riers and that different elements surface as important
both in identifiable geographical markets and customer
groups, and that furthermore choice criteria are a
moving target over time and vary significantly between
segments of the market for a single mode. Brooks
(1995) further suggests that carrier choice decisionFig. 3. The Aaker and Day model (adapted from Brooks, 1985).
J. Mangan et al. / Transport Management 1 (2002) 1528 19
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
6/14
making has shifted from a process involving a short
timeframe, few decision makers, etc. to a process which
is more complex, has a longer timeframe, etc. which in
essence reflects the shift in interfirm relationships from
an adversarial or transaction basis to a more partnership
or relationship-driven basis.
It is evident from this discussion that transport
markets must not be viewed as homogeneous and that a
broader, holistic view needs to be taken of the transport
choice process. It is important to realise when consid-
ering transport choice that it is likely that an elaborate
decision making process isnotengaged upon each time a
consignor or haulier has a shipment to send. Brooks
(1984) showed that the Buygrid Framework, first put
forward in 1967 by Robinson, Faris and Wind in In-
dustrial Buying and Creative Marketing, can be applied
to the purchase of liner shipping services. The buygrid
framework identifies three classes of purchase:
New tasks appropriate for new purchasers in the
market. Modified rebuy when the purchaser is experienced
but not committed to a particular supplier.
Straight rebuy when the choice of supplier is auto-
matic.
DEste (1992a), in his study of the AustraliaTasmania
RoRo trade, observed that in most cases straight rebuys
were made and that only occasionally were shipping
options re-examined (modified rebuy); he found new
tasks to be a minor, but increasing, component of the
overall freight market, and the evidence suggested that
the most thorough comparisons were made for such new
tasks. Whyte (1992) also pointed to the prominence of
source loyalty (i.e. buying a product or service from the
same source from which it was obtained previously in
other words straight rebuys) in freight markets. This
tendency then to engage in objective assessment for only
a minor proportion of shipments is thus a key feature of
the transport choice process.
All of the foregoing discussion in this section points
to the necessity to take a holistic view of the transport
choice process rather than just looking at the actual
nature and structure of the transport choice decision per
se. Such a holistic view considers, inter alia, that trans-
port choice changes over time, is different for different
markets, involves multiple actors, is often encapsulatedwithin partnership based intercompany interaction,
rarely involves new tasks, is dependent upon the supply
chain and logistics strategies of shippers, etc.
4. Methodology
On the basis of the preceding discussion it was
decided to discount output-oriented modelling and to
attempt both input-oriented modelling (calculating
importance means, factor analysis, and using the Aaker
and Day Model) and process-oriented modelling. In
order to generate the requisite data it was decided to
employ a combination of research methods structured
into a three phase, triangulated research effort as
follows:
Phase 1 involved an investigation of RoRo port/ferry
choice in the Irish market in an inductive (phenome-
nological) manner. This involved an analysis of the
Irish RoRo freight market and interviews with (24)
key actors in the Irish RoRo freight market (plus
245 short interviews with truck drivers). This pro-
vided the researchers with an in-depth insight into
various issues concerning the operation and structure
of the market.
Phase 2 investigated RoRo port/ferry choice in a de-
ductive (positivist) manner. This involved: (i) a rigor-
ous sampling exercise which involved eliciting a
representative sample of decision makers engaged in
the Irish RoRo freight trade from 14 separate sam-
pling frames, (ii) development and pilot testing of aquestionnaire (across nine interviews), (iii) personal
interviews (57) with decision makers using a struc-
tured questionnaire, and (iv) survey data analysis us-
ing various input-oriented modelling techniques
(including theAaker and Day model).
Phase 3 involved again investigating RoRo port/ferry
choice in an inductive (phenomenological) manner.
This involved taking the results obtained in Phase 2
and further investigating these (by way of some 19 in-
terviews) using construct elicitation, an interviewing
methodology based on Kellys theory of personal
constructs.
5. Results
5.1. Factor analysis
A principal components factor analysis was per-
formed on the importance ratings obtained from the
survey data in Phase 2 of the research. A VARIMAX
rotation converged in 10 iterations, producing six ag-
gregate factors (the term aggregate factor is used here
to identify factors generated by factor analysis so as to
avoid confusion with individual port/ferry factors). Itwas not possible however to identify or label these six
aggregate factors.
The procedure was repeated with the intention of
extracting three factors and in this instance VARIMAX
converged in six iterations. Table 3 illustrates the ro-
tated factor matrix scores obtained. The three aggre-
gated factors, based on the factor scores obtained have
been labelled:
Factor 1 speed/risk/convenience factor.
Factor 2 cost and convenience factor.
Factor 3 driver factor.
20 J. Mangan et al. / Transport Management 1 (2002) 1528
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
7/14
Factors with loadings in excess of 0.5 (shown in bold
in Table 3) were used to determine aggregate factorlabels McGinnis (1980) and Matear and Gray (1993)
adopted this approach in their factor analyses in freight
transport choice studies.
The first two factors are perhaps unsurprising, but the
third factor, the driver factor, is unique in that it has not
appeared previously in the literature in the context of
factor analysis and ferry choice.
5.2. Identifying salient factors
The 57 respondents in Phase 2 were asked to score
their choice of port/ferry route for a hypothetical ship-
ment. Respondents were then asked to score their nextchoice of port/ferry route in the event of their first choice
not being available. Table 4 illustrates:
(i) the mean importance rating which respondents
gave to different choice factors,(ii) the mean performance ratings given by respon-
dents to 14 pertinent choice factors in their first
choice of a port Performance 1,
(iii) the mean performance ratings given by respon-
dents to 14 pertinent choice factors in their second
choice of a port Performance 2,
(iv) indicates if the ratings are significantly statisti-
cally different, and
(v) indicates if the factors can be considered salient
factors.
The number of respondents who gave performance
ratings to individual choice factors for both first and
second choice ports ranged from a minimum of 16 re-spondents for the factor intermodal/connecting trans-
port links to a maximum of 31 respondents for the
Table 3
Factor analysis
Variable Factor 1 Factor 2 Factor 3
Facilities onboard for drivers ).08 ).16 +.79
Proximity of ports to freight origin/destination +.04 +.15 +.34
Port and ferry chosen is on fastest overall route +.37 ).12 +.42
Preference of consignor/consignee +.15 ).17 ).06
Spaces available on ferry when needed )
.09 )
.28 +.01Intermodal/transport links at ports +.65 ).09 ).34
Port and ferry chosen is on cheapest overall route ).04 +.68 ).14
Speed of getting to/through ports +.65 +.14 +.21
Ferry suitable for unaccompanied/special cargo ).03 +.54 ).12
Delays due to driving bans, tacho, weight, etc. +.39 ).16 +.31
Opportunity for driver rest break ).01 ).09 +.86
Cost of ferry service/discounts ).29 +.82 +.23
Sailing frequency/convenient sailing times +.32 +.50 ).07
Information on sailing options +.85 ).01 ).02
Risk of cancellation/delay +.58 +.38 +.17
Table 4
Mean importance ratings and identifying determinants of choice (the Aaker and Day model)
Factor Mean importance Performance 1 Performance 2 Different ? Salient ?
1. Space available when needed on ferry 4.578 4.4 3.9 Y1 S1
2. Sailing freq/convenient sailing times 4.408 4.1 3.9 N
3. Risk of cancellation/delay 4.340 3.9 3.7 N
4. Port and ferry on fastest overall route 4.160 3.9 3.6 N
5. Proximity of ports to origin/destination 4.083 4.2 3.9 Y2 S2
6. Cost of ferry service/discounts 4.000 3.9 3.4 Y1 S17. Speed of getting to/through ports 3.959 3.7 3.8 N
8. Port/ferry on cheapest overall route 3.776 3.7 3.5 N
9. Ferry suitable for unacc or special cargo 3.558 3.8 3.9 N
10. Delays due to driving bans, tacho, etc. 3.543 3.6 3.4 N
11. Availability of info on sailing options 3.314 3.8 4.0 N
12. Facilities for drivers 3.250 4.2 3.7 Y1 S2
13. Opportunity for driver rest break 3.118 4.0 3.4 Y2 S2
14. Intermodal/connecting transport links 2.093 3.3 3.0 N
Y1 indicates factors significantly different at the 95% confidence level, Y2 significantly different at the 90% confidence level, N no significant
difference. Salient factors, labelled S1, are those with mean importance ratings above 4 and significant performance differences at the 95% confidence
level. Other potentially salient factors, labelled S2, are those with an importance rating above 3 and significant performance differences at least at the
90% confidence level. Owing however to the relatively small sample size in the survey (57 respondents) and the resulting potential for error it is only
possible to accept the S1 factors as these meet the more stringent criteria for salience and are thus more statistically valid.
J. Mangan et al. / Transport Management 1 (2002) 1528 21
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
8/14
factor port and ferry is on the fastest overall route to
destination.
It was obviously necessary to define the statistically
correct criteria for salience. To recap, salient factors are
those which both perform differently and are regarded
as important. The column headed Different? in Table 3
identifies those factors for which there is a difference in
indicated performance ratings between the chosen op-
tions. In this table, Y1 indicates those factors signifi-
cantly different at the 95% confidence level, while Y2
indicates those factors significantly different at the 90%
confidence level. It was decided that Y1 factors only
should be selected (i.e. those significantly different at the
95% confidence level) owing to the relatively small
sample size in this study and the resulting potential for
error. It was also decided that, for the same reason, for
any Y1 factors to be considered salient factors, they
should have a mean importance score greater than 4, or
at least greater than the mean of the sum of the mean
importance ratings (which in this case is 3.73). On thisbasis two salient factors (these are labelled S1 in the
table) are thus identified from the data, namely space
availability on ferries and cost of ferry service/discounts.
Other factors have also been identified as potentially
salient, although not with statistical accuracy from the
perspective of the above criteria. These are labelled S2 in
the table.
5.3. Construct elicitation
The interviews in Phase 3 employed a technique
known as construct elicitation (Note 1) to identify
factors considered important in port/ferry choice (see
Table 5):
(i) Outlines the ranking of mean factor importance
obtained in the Phase 2 research.
(ii) Identifies the two salient factors which emerged
from the Aaker and Day analysis in Phase 2.
(iii) Indicates if any of the Phase 2 factors listed were
also elicited during the Phase 3 construct elicitation
exercise (if the factors emerged in a slightly different
form and/or multiple forms, then these descriptions
are given in brackets).
Twelve of the 15 factors used in the Phase 2 research
were also elicited during construct elicitation in Phase 3.
Three factors, used in the Phase 2 research, were not
however elicited in Phase 3. The three factors were:
Port/ferry on the cheapest overall route (ranked eightmost important in Phase 2).
Availability of information on sailing options
(ranked 11th most important in Phase 2).
Preference of consignor/consignee (ranked 14th most
important in Phase 2).
Three issues are of note with regard to the absence of
these factors from Phase 3. Firstly, none of these three
Table 5
Comparison of port/ferry choice factors elicited during the Phase 2 research and during the Phase 3 research
Factor Mean importance in
Phase 2 research
Salient in Phase 2
research?
Elicited in Phase 3?
1. Space available when needed on ferry 4.578 Yes Yes
2. Sailing freq/convenient sailing times 4.408 Yes (frequency of ferry service; ferry arrival
and departure times; parallel sailing timings
reduce choice; convenient sailing times)
3. Risk of cancellation/delay 4.340 Yes
4. Port and ferry on fastest overall route 4.160 Yes (also : speed of ferry service)
5. Proximity of ports to origin/destination 4.083 Yes (location of traffic origin; location of
traffic destination)
6. Cost of ferry service/discounts 4.000 Yes Yes (bulk discounts from ferry companies;
cost of the ferry service)
7. Speed of getting to/through ports 3.959 Yes (delays at ports; potential for delays
on certain port access/egress routes;
port protests)
8. Port/ferry on cheapest overall route 3.776 No9. Ferry suitable for unacc or special cargo 3.558 Yes (ferry suitable for unacc cargo; ferry
suitable for haz cargo)
10. Delays due to driving bans, tacho etc. 3.543 Yes (tacho restrictions; driving bans; if load
overweight avoid routings over UK; Irish
trucks delayed by UK police)
11. Availability of info on sailing options 3.314 No
12. Facilities for drivers 3.250 Yes (catering; staff attitudes; incentives;
distractions from passengers)
13. Opportunity for driver rest break 3.118 Yes
14. Preference of consignor/consignee 2.413 No
15. Intermodal/connecting transport links 2.093 Yes (options to get units from UK ports to
other UK ports; partner ferry services on the
English Channel)
22 J. Mangan et al. / Transport Management 1 (2002) 1528
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
9/14
factors are determinants of choice as identified by the
Phase 2 Aaker and Day analysis. Secondly, none of the
factors were ranked as being of particular importance in
Phase 2 (the most important of the three was only
ranked eight). Thirdly, the Phase 2 factors were chosen
based on a review of the literature and subsequent pi-
loting of a draft survey instrument. In contrast, the
Phase 3 factors were actually elicited from the respon-
dents. Based on these three issues then it could therefore
be argued that the absence of the three factors from the
Phase 2 results is not a matter of particular concern.Four other factors, not specifically used in Phase 2, were
also elicited in the Phase 3 construct elicitation and these
are shown in Table 6.
5.4. Building a process-oriented model
The combined research results were used to build a
holistic, structural, process-oriented model, similar to
that of, for example, Brooks (1990) or DEste, but ob-
viously modified for the Irish RoRo freight market in
the context of the Phases 2 and 3 results obtained.
Furthermore, it was intended that the model would be
combined with the many qualitative findings which
emerged during the research to yield a rich overview of
RoRo port/ferry choice in the Irish context. The model
developed is shown in Fig. 4.
6. Discussion
The triangulated research methodology which was
selected, developed and employed allowed for the
fullest possible exploration of RoRo port/ferry choice.
Perhaps the greatest affirmation of this approach can
be seen from the presentation of the Phases 2 and 3research results shown in Table 5. The Phase 3 research
results have in a broad sense affirmed the various re-
sults obtained during the Phase 2 research. In addition,
they have also added extra richness to the whole
research effort by giving extra detail to many of the
Phase 2 factors, as well as generating a number of
other factors which did not arise during the Phase 2
research. A good example is the Phase 2 factor delays
due to driving bans, etc. which was subsequently use-
fully disaggregated into a myriad of relevant factors in
Phase 3.
6.1. Implications for the Irish RoRo freight market
During the course of this research it was apparent
that the Irish RoRo freight market is quite sophisticated
from both logistical and operational perspectives. Many
hauliers are providing value adding, total logistics so-
lutions for their clients. Vertical integration, where a
service provider controls more than one element in the
transport chain, is evident in some instances. Indeed this
is an issue of particular note for many of the RoRo ferry
companies should they just operate RoRo ferry ser-
vices or should they provide a door to door integrated
transport solution for their clients? Furthermore, what is
their strategic role in the evolving logistical strategies of
their customers (both hauliers and manufacturers)? In
terms of service provision there appears to be some
demand for new service offerings, particularly more high
speed ferry services, more direct sailings to Continental
Europe, and more sailings scheduled to depart Ireland
later in the evening/night. There was much discussionamong those hauliers and manufacturers who were in-
terviewed during the course of this research concerning
the latter issue of later sailings. Obviously the com-
mercial viability of this and the other demands would
have to be assessed. One issue which was apparent
during the research was that there appears to be widely
fluctuating demand for space on RoRo ferries, largely
driven by the production schedules of manufacturers.
6.2. Implications for transport/logistics theory
The emergent themes running through the literature
review in this research were that:
For proper research into transport choice the re-
searcher should not assume that the market is homo-
geneous and should note that selection criteria vary
according to mode, time, market segment, etc.
That the context within which the decision is made is
of importance and that furthermore, in order to mod-
el decision making, it is often more beneficial to
adopt a broad, holistic view of the port/ferry choice
process rather than look at the nature and structure
of the port/ferry decision per se.
This research undertaking exposed a variety of per-
tinent selection criteria. These were usefully disaggre-gated to ferry issues and overall route issues. The actual
determinants of choice, or as termed by Brooks the
salient factors, in the Irish RoRo freight market were
space availability on ferries and cost of ferry services/
availability of discounts. Legion other factors emerged,
many important for different contexts in the heteroge-
neous Irish RoRo freight market. This should also allow
the definition from the model (and more specifically
from the data behind the model illustrated as the Phases
2 and 3 research results) of benefit segments for the Irish
RoRo freight market.
Table 6
Other factors elicited during the Phase 3 research
1. Ferry company marketing (sales reps visits, etc.)
2. Depends on when are the goods due
3. Dictated by the location of the hauliers depot/configuration of
the hauliers network
4. Security (security of units onboard from competitors and
passengers; risk of hijacking)
J. Mangan et al. / Transport Management 1 (2002) 1528 23
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
10/14
New variables identified in this research include issues
such as security of units onboard ferries from competi-
tors and passengers, a much wider definition of facilities
for drivers, and some quasi-ethical issues such as the risk
of being caught if outside regulatory limits, etc. The
decision model shown in Fig. 4 indicated the following
hierarchy of importance: overall route, followed by the
ferry service, followed by the port. There are obvious
implications in this for port marketing. This also con-
curs with the views of Fleming (1997), Slack (1985) and
DEste and Meyrick (1992) that the port is of lesser
importance in the transport chain. Some differences in
factor importance between hauliers and manufacturers
were observed. What is particularly of note however in
the context of this research is the significant degree to
which the routing decision is delegated by manufactur-
(a)
(b)
Fig. 4. (a) A model of RoRo port/ferry choice in the Irish market. (b) The decision making process. (c) Ferry issues. (d) Overall route issues.
24 J. Mangan et al. / Transport Management 1 (2002) 1528
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
11/14
ers to hauliers, and consequently the reduced relevance
of the formers role with regard to choosing routes.Risk played a significant role in the routing decisions
of hauliers. The many baskets approach, illustrated by
DEste (1992b) who showed that hauliers reduce risk by
spreading their traffic over a number of carriers, was
also evidenced in the Irish context. What is perhaps
notable however is that many hauliers have moved to a
more partnership, as opposed to adversarial, relation-
ship with the ferry companies, and consequently are
more inclined to only consider the port/ferry options
which they are familiar with and to reduce their pa-
tronage to selected, key service providers.
Fig. 4 also shows decision makers only occasionally
revising their choices, on the basis that they are nowmore in partnership with the ferry companies, and are
generally more content with their routing decisions. One
of the few instances where many of the considerations in
the port/ferry choice process were pushed to one side
was with urgent shipments whereby it was revealed that
there is a tendency in the Irish market for these to be
rushed onto the next available ferry from any port. On a
final point, it is important to highlight the central role
played by both drivers and tacit knowledge in the port/
ferry choice decision. Tacit information as opposed to
information sourced through marketing or electronic
(c)
(d)
Fig. 4. (Continued).
J. Mangan et al. / Transport Management 1 (2002) 1528 25
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
12/14
channels played a central role in port/ferry choice.
Drivers were usually the essential conduit of this infor-
mation. Indeed the role of drivers in port/ferry choice
must not be underestimated their legal driving limits
combined with their impressions of onboard service
seem to be given significant weight during the port/ferry
choice process (except in instances where the same ferry
is used all of the time, in which the case the driver does
not obviously have much of a say).
6.3. Implications for research methodology
It was already pointed out above that the triangu-
lated research methodology which was selected, devel-
oped and employed allowed for the fullest possible
exploration of RoRo port/ferry choice. The Phase 3
research added to and enhanced the Phase 2 research
output. As regards choice modelling, and in the context
of the various approaches considered in this research,
certain approaches were usefully applied, particularlytheAaker and Day(input-oriented) modelling approach,
so usefully applied by Brooks, and also process-oriented
modelling, again usefully applied by Brooks and also by
DEste. Various qualitative methodologies were con-
sidered for the Phase 3 research and ultimately construct
elicitation was employed. This inductive methodology
added significant value to the information which was
already gleaned from the Phase 2 research, and was
both enjoyable and rewarding to employ as a research
technique.
7. Conclusion
This paper has made two significant contributions.
Firstly, it has brought together into one study all of the
major contributions to the port/ferry choice literature,
and particularly those of Brooks, DEste, Matear and
Gray, and other authors to date. Secondly, it has em-
ployed an in-depth, triangulated research methodology
which incorporated both quantitative and qualitative
methodologies for the investigation of port/ferry choice.
The qualitative methodology that employed construct
elicitation had never previously been employed in this
context.The output of the combined, three phase research
effort comprised a synthesis and model of RoRo port/
ferry choice in the Irish market. The research has pro-
vided a valuable insight into the operating dynamics of
the Irish RoRo freight market which will be of partic-
ular use to all stakeholders in the sector. Various issues
have been identified and clarified such as required im-
provements in ferry services, the reasons for the fluctu-
ating demand for space on ferries and, not least, the
insignificant impact of port, as opposed to ferry com-
pany, marketing on route choice. The research has made
specific contributions to the transport choice literature.
The actual determinants of port/ferry choice have been
identified (space availability on ferries and cost of ferry
services/availability of discounts) as have some new, or
seldom noted, selection criteria such as security of
freight units, a wider consideration of facilities for
drivers, and the risk of being caught if outside regula-
tory driving limits. Other issues illustrated include the
high degree of delegation of the routing decision by
consignors, the significance of risk, the infrequency of
revision of routing decisions, and the surprisingly cen-
tral role of drivers and tacit knowledge in the port/ferry
choice decision.
As regards research methodology, this research effort
has highlighted the significant benefits which can accrue
from employing a triangulated research methodology.
One particular methodology, construct elicitation, was
employed as an interview technique in the third phase of
the research and was shown to be a particularly re-
vealing interview technique. From the raft of many andvaried techniques which can be used to model transport
choice two techniques, namely the Aaker and Day (in-
put-oriented) modelling approach and process-oriented
modelling, were shown to be particularly illustrative and
effective.
Based on the research described in this paper, a
number of recommendations are suggested for further
research. The role of RoRo ferry companies in the lo-
gistics chain is worthy of further investigation. Should
such ferry companies just operate RoRo ferry services or
should they provide a door to door integrated transport
solution for their customers? Furthermore, what is their
strategic role in the evolving logistical strategies of their
customers? A second area for further research concerns
taking the holistic model developed in this research and
testing it in other markets (e.g. the geographically re-
verse markets of UK/Ireland and Continental Europe/
Ireland). A third area for further research might be to
attempt to explore the link between routing decisions
and manufacturing schedules. From a total costs anal-
ysis perspective it would be useful to explore if sub-op-
timum routing decisions are made in order to comply
with manufacturing schedules.
8. Notes
(1) Construct elicitation is an interviewing method-
ology based on Kellys theory of personal constructs
and was first put forward in his seminal work A theory
of personality: the psychology of personal constructs
(Kelly, 1955). The basis of Kellys technique was for the
researcher to identify and gain a wider knowledge of
what he termed the interviewees personal constructs.
Kelly defines constructs as the way in which some
things are construed as being alike and yet different from
26 J. Mangan et al. / Transport Management 1 (2002) 1528
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
13/14
others. Hussey and Hussey (1997) define constructs as
the qualities which an individual uses to describe and
differentiate between the elements, and they describe
elements as the objects of the interviewees perceptions.
Kelly (1955) defines elements as the things or events
which are abstracted by a construct.
In the context of this study then, the elements would
be the various ports and ferries which the decision ma-
ker could possibly use, while the constructs are the
various qualities which the decision maker uses to de-
scribe and differentiate between these various ports and
ferries in order to make their decision. According to
Stewart and Stewart (1997) the essence of Kellys tech-
nique is that it puts the rigour where it needs to be, that
is, in the methodology rather in the model of personality
(or behaviour) it propounds. In their view Kellys writ-
ings and his techniques were years ahead of their time
when they first emerged. Kellys (1955) emphasis then
was on anticipation rather than on reaction to help the
practicing psychologist who is attempting to gain abetter understanding of a patient and their problems.
Kellys approach suggests that instead of asking the
patient prescribed questions and attempting to slot pa-
tients into boxes, the interviewer should instead allow
the patient to identify their own constructs, what Kelly
refers to as the creative capacity of the living thing to
represent the environment, not merely to respond to it.
This effectively then alters the usual interviewerinter-
viewee balance with the interviewer adopting an ap-
parently less domineering role in the interaction.
Bannister and Fransella (1989) noted that one of the
overriding questions in the field of personality has
concerned the question of what to measure. They note
however that such a difficulty does not arise with Kellys
personal construct theory the unit is clearly stated, it is
the construct. Personal construct theory has, since
Kellys original seminal work in 1955, caught on quite
significantly in the field of psychology. Stewart and
Stewart (1997) detail the application of the technique to
the management field and point out that, especially in
the first few decades after the publication of Kellys
seminal work in 1955, it was confined to use within
psychotherapy and related areas. Personal construct
psychology was first used in a business context in
market research and then in the personnel area. Ex-amples of applications of the technique include: Moy-
nihan (1998) who used it to identify the factors (i.e. the
constructs) which software project managers take into
account when planning new projects; Hisrich and Jan-
kowicz (1990) who used the technique for examining
intuition in venture capital decisions; and Anderson
(1990) who suggested use of the technique for employee
selection.
Kelly went further than just a consideration of the
elicitation of personal constructs. Drawing upon his
training in mathematics, Kelly designed a technique
whereby the mathematical relationships between con-
structs could be obtained this is called the repertory
grid technique. Indeed Kelly, a keen geometer with ex-
perience in navigation and an interest in multi-dimen-
sional geometry, very ably applied mathematics to his
theories not alone did he develop the repertory grid
technique but, according to Bannister and Fransella
(1989), he also went so far as to provide the future user
of the technique with his own method of non-parametric
factor analysis. In addition, Kelly presented his theory
as a geometry of psychological space (see, for example,
Shaw and Gaines, 1992).
References
Aaker, D., Kumar, V., Day, G., 1995. Marketing Research, fifth ed.
Wiley, New York.
Anderson, N., 1990. Repertory grid technique in employee selection.
Personnel Review 19 (3), 916.Bannister, D., Fransella, F., 1989. Inquiring Man: The Psychology of
Personal Constructs, third ed. Routledge, London.
Brooks, M.R., 1984. An alternative theoretical approach to the
evaluation of liner shipping. Part l: Situational factors. Maritime
Policy and Management 11 (1), 3543.
Brooks, M.R., 1985. An alternative theoretical approach to the
evaluation of liner shipping. Part II: Choice criteria. Maritime
Policy and Management 12 (2), 145155.
Brooks, M.R., 1990. Ocean carrier selection criteria in a new
environment. Logistics and Transportation Review 26 (4), 339355.
Brooks, M.R., 1995. Understanding the ocean container market a
seven country study. Maritime Policy and Management 22 (1), 39
49.
Brooks, M.R., 2000. Performance evaluation of carriers by North
American companies. Transport Reviews 20 (2), 205218.
Coles, S., Rowley, J., 1995. Revisiting decision trees. Management
Decision 33 (8), 4650.
Cullinane, K., Toy, N., 2000. Identifying influential attributes in
freight route/mode choice decisions: a content analysis. Transpor-
tation Research E 36, 4153.
DEste, G., 1992a. Carrier selection in a Ro/Ro ferry trade Part 2.
Conceptual framework for the decision process. Maritime Policy
and Management 19 (2), 127138.
DEste, G., 1992b. Frameworks for understanding the behaviour of
purchasers of general cargo shipping services. In: 1992 World
Conference on Transport Research, Lyons, France.
DEste, G., Meyrick, S., 1992. Carrier selection in a Ro/Ro ferry trade
Part 2. Decision factors and attitudes. Maritime Policy and
Management 19 (2), 115126.
Fleming, D., 1997. The meaning of port competition. In: Conference
of the International Association of Maritime Economists (IAME),
City University, London.
Gray, G., 1982. Behavioural approaches to freight transport modal
choice. Transport Reviews 2 (2), 161184.
Hisrich, R., Jankowicz, A., 1990. Intuition in venture capital decisions:
an exploratory study using a new technique. Journal of Business
Venturing 5 (1), 4963.
Hussey, J., Hussey, R., 1997. Business Research: A Practical Guide for
Undergraduate and Postgraduate Students. MacMillan Press,
London.
Lehmusvaara, A., Tuominen, M., Korpela, J., 1999. An integrated
approach to truck carrier selection. International Journal of
Logistics: Research and Applications 2 (1), 520.
J. Mangan et al. / Transport Management 1 (2002) 1528 27
-
7/25/2019 Modelling Portferry Choice in RoRo Freight Transportation (2002)
14/14
Matear, S., Gray, R., 1993. Factors influencing freight service choice
for shippers and freight suppliers. International Journal of Physical
Distribution and Logistics Management 23 (2), 2535.
McGinnis, M., 1980. Shipper attitudes toward freight transport choice:
a factor analytic study. International Journal of Physical Distri-
bution and Materials Management 10 (1), 2534.
Moynihan, T., 1998. How Software Project Managers Characterise
New Projects and Their Settings. Available at http://www.com-
papp.dcu.ie/-tonym/tonym.Murphy, P., Daley, J., Dalenberg, D., 1992. Port selection criteria: an
application of a transportation research framework. Logistics and
Transportation Review 28 (3), 237255.
Murphy, P., Hall, P., 1995. The relative importance of cost and service
in freight transportation choice before and after deregulation: an
update. Transportation Journal Fall 1995, 3038.
Murphy, P., Daley, J., Hall, P., 1997. Carrier selection: do shippers and
carriers agree, or not? Transportation Research E 33 (1), 6772.
Pedersen, E., Gray, R., 1998. The transport selection criteria of
Norwegian exporters. The International Journal of Physical
Distribution and Logistics Management 28 (2), 108120.
Pisharodi, R., 1991. The transport choice decision process, the
potential, methodology and applications of script-theoretic mod-
elling. The International Journal of Physical Distribution and
Logistics Management 21 (5).
Shaw, M., Gaines, B., 1992. Kellys geometry of psychological space
and its significance for cognitive modelling. The New Psychologist
(October), 2331.
Slack, B., 1985. Containerization, inter-port competition, and port
selection. Maritime Policy and Management 12 (4), 293303.
Spencer, A., Anderson, S., Whitcombe, M., 1992. Channel choices: a
study of options, timings and costs of international roadhauliers. Transport Studies Group, University of Westminster,
London.
Stank, T., Daugherty, P., Ellinger, A., 1996. Information exchange,
responsiveness and logistics provider performance. The Interna-
tional Journal of Logistics Management 7 (2), 4357.
Stewart, V., Stewart, A., 1997. Business Applications of Repertory
Grid, originally published in 1981 by McGraw-Hill; the authors
have published an updated electronic copy of the first seven
chapters of the original text on the world wide web at the following
website: http://www.enquirewithin.co.nz/businappof.
Whyte, J., 1992. Characteristics of transport managers which influence
haulier selection. International Journal of Physical Distribution
and Logistics Management 22 (5), 1424.
28 J. Mangan et al. / Transport Management 1 (2002) 1528