TIIw;4 Rn.-..t Vol. I3A. "'" 91-109 0041-1647/79(0401-4091/S02.

0CtI0
@ Papmoa Pms L:d.. 1979. Prialtd ill OIUl Britain

ACCESSIBILITY INDICATORS FOR TRANSPORT PLANNING

J. M. MORRIS, P. L. DUMBLE and M. R. WIOAN

Australian Road Research Board. SIlO Burwood Highway. Vermont South. Victoria. Australia
(Received 6 November 1978)
Abstract-Both perceptual and measurable specifications of accessibility are reviewed and their relevance to
transport planning established. The wide variety of analytical (orms which can be used to quantify different aspects
of accessibility are categorised and grouped by conceptual basis. The different forms of accessibility index are then
related to underlying theories which link consumer demand. evaluation and accessibility.

INTRODllCTJON illustrations are provided by a simple analysis of trans-

Transport planning organizations have historically adop- portation study data for the Victorian City of Ballarat,
ted a view that transport problems and transport solu- and some traffic restraint studies in Coventry (U.K.).
tions can be treated without considering non-transport
aspects of urban life. This view is reftected in the highly POrEN11AL APPLICAnONS
selective nature of traditional transport planning solu- Accessibility has generally been defined as some
tions-solutions which, by and large. set out to improve measure of spatial separation of human activities.
and accommodate an ever-increasing flow of vehicles, Essentially it denotes the ease with which activities may
not necessarily even improving the flow of people be reached from a given location using a particular
(Hillman. Henderson and Whalley 1973. 1976). But per- transportation system. Several broad applications of ac-
ceptions of transport planning objectives have changed cessibility indicators may be identified, including evalua-
substantially in recent years. Attention has shifted from tion of the transport/land-use system, modelling travel
plans catering for a continuation of existing trends. to choice situations, modelling urban development, and
plans which attempt to substantially alter those trends- summarising spatial structure (Wachs 1978). With the
by encouraging the use of public transport and non- exception of the third application, the concept of ac-
motorised methods of travel. and by attempting to in- cessibility is equally applicable in rural and urban con-
tegrate transport with land-use planning. At the same texts. In this paper the main emphasis is laid on the
time much greater emphasis is being given to dis- urban environment.
tributional questions. and evaluation of alternative land-
use/transportation plans is no longer based entirely on System evaluation
efficiency criteria. The focus of transport planning is Accessibility is already important as an evaluation
moving from "vehicular mobility" to "personal mobility" criterion. Evaluation of alternative transport plans is
(Dalvi 1978) and from "traffic congestion" to "ac- best considered in relation to the activities of
cessibility provision" (Wilson 1972). interest to individuals and groups because most daily
These changes have undoubtedly led to a more com- travel owes its existence to the spatial separation
prehensive view of travel. and to questioning of many of activities. Since accessibility is a function of both
assumptions at the base of transport planning (Hensher land-use patterns and the performance of the trans-
1976). But there has also been much confusion over port system, it is a particularly appropriate criterion
terminology and over the precise role that these new for evaluating the service provided by the transport
concepts should be assigned in the planning process system to di1ferent categories of users (Koenig 1977,
(DaIvi 1978). This paper explores some of the various Black and Conroy 1977). A useful feature of accessibility
concepts employed to measure accessibility. A closer indicators is their ability to· generate remedial solutions
integration of both accessibility and mobility con- and to influence the plans being developed, by indicating
siderations in transport planning is needed. although we which areas or groups are currently under-provided.
concentrate here on accessibility. Such solutions may not necessarily involve modifications
There is a critical distinction between the derivation of to the transport system: and in some cases improvements
"objective" indicators of accessibility, and perceived in accessibility may be achieved more eft'ectively by
measures. The first sections of the paper concentrate on reorganising the distribution of activities in space and/or
the areas of relevance of accessibility measures and a time. Accessibility indicators may also be used to moni-
typology of functional forms. A wide variety of tor changes in the urban system. irrespective of whether
measures is drawn from descriptive analyses found such changes are planned or unplanned (Black and Con-
helpful in planning. The selection of indicators ap- roy 1977).
propriate for transport planning is then considerd, with Despite these advantages of accessibility indicators
special reference to the combined inftuence of land-use there is currently some debate on whether accessibility
and transport. Selected theoretical frameworks for con- or mobility should be the objective in transport planning.
sidering accessibility are related to the models of This issue is compounded by the fact that the concept of
behaviour and perception used for forecasting. Empirical mobility has been used rather indiscriminately to refer to
91
 J. M. MORRIs, P. L. DUMBLB and M. R. WIOAN

both the supply side and the demand side of transport transport system that connects them (Wachs 1978). It so
services (Dalvi 1978). For the purpose of the present doing accessibility indicators are important descriptive
study. personal mobility is interpreted to mean the ability measures of urban spatial structure and performance.
of individuals to move from place to place: this depends With these broad applications in mind. let us now turn
principally upon the availability of different modes of to examine the various concepts and measures of ac-
transportation, including walking (see HiHman et aL 1973. cessibility which may be of value in transport planning.
1976). When defined in this sense. mobility is concep-
tually distinct from actual travel: and the argument over DEFINING AND MEAStlJlJNG ACCESSIBIUTY
mobility or accessibility as an objective in transport Accessibility measures are based on the premise that
planning is seen to be a futile exercise. Mobility and space constrains the number of opportunities available.
accessibility together influence an individual's capacity Beyond this point, definitions of the concept differ
to travel in daily life. It is important to recognise. widely. There is considerable variation in the other ele-
however, that perceilJed accessibility and perceilJed ments which may be included. and in how they are
mobility-the real determinants of behaviour-will be at measured and combined. As Gould (1969, 1964) has
variance with "objective" indicators of accessibility and noted, "accessibility •.. is a slippery notion ... one of
mobility. those common terms which everyone uses until faced
with the problem of defining and measuring it".
TralJeJ demand models To some desree. variations in accessibility measures
Accessibility indicators may also be used as input are inevitable since the appropriate definition will depend
variables in modelling travel choice situations. Travel upon the intended application. However. most of the
involves costs in time, money and human effort which confusion stems from fundamental differences of
must be borne directly by the community. Consequently. opinion. There is a basic dilemma in choosing between
accessibility not only influences the distribution of travel "process" indicators (measures of the supply charac-
costs within the community but may also affect levels of teristics of the system and/or individuals) and "outcome"
service use and participation in desired activities. It has indicators (such as actual use and levels of satisfaction).
been suggested that individuals make a set of mutually On the one hand accessibility may be interpreted as a
dependent choices or decisions which are highly depen- property of individuals and space which is independent
dent upon individual household members' perceived ac- of actual trip making and which measures the potential
cessibilities to various opportunities by a given trans- or opportunity to travel to selected activities. Alter·
portation system (Ben Akiva and Lerman 1975). Such natively, it may be held that "proof of access"lies in the
decisions include. for example. where to live. how many use of services and participation in activities. not simply
cars to own. and what trips to make at what times by in the presence of opportunities. Consequently there is a
which modes (Burns and Golob 1976). Accessibility, tendency to want to measure accessibility in terms of
therefore. represents an important element to be con- actual behaviour (Wachs 1978).
sidered in virtually all choice issues relevant to transport This basic conflict gives rise to a range of accessibility
planning. Once again. however, there is a fundamental measures which differ in terms of their behavioural
problem of measuring perceived values. component. And yet this represents only one of many
sources of variation in accessibility indicators. Since
Urban development models there is no concensus on an operational definition of
This third application of accessibility is closely accessibility. it is necessary to develop a broad
related to the second. although it represents a somewhat classification of accessibility measures before any
more longstanding interest held by transport planners. meaningful attempt can be made to evaluate them.
This concerns attempts to model the relationship be-
tween accessibility and urban development (Clark 19SI, A cltusi/icalion 0/ accessibility indicators
Hansen 19S9. Patton and Clark 1970. Davidson 1973, A useful classification of accessibility indicators is
1977, Beggs 1976). Here the focus is not so much on given in rIB. I. This is Jarsely an amalgamation of
modelling individual choices but on modeHing urban previous attempts to classify accessibility measures (In-
form in the aggregate. gram 1971, Briggs and Jones 1973. Wachs 1978). Exam-
ples of specific formulae and references for further dis·
Description cussion are presented for each terminal class shown in
Accessibility indicators provide possibly the most Fig. I.
useful and appropriate means of summarising a great The two principal bases of classification are the
deal of information on the location of households in behavioural dimension mentioned earlier. and a dis-
relation to the distribution of urban activities and the tinction between "relative accessibility" and " integral
accessibility" developed by Ingram (1971). Relative ac-
tThe term relative accessibility is sOmetimes used to descn'be cessibility describes the relation or degree of connection
an integral measure which has been normalised in order to between any two points. whereas integral accessibility
facilitate temporal or spallal comparisons. For instance, the
measure may be normalised to correct for changes over time in describes the relation or -degree of interconnection be·
the number of opportunities (sec Palton and Clark 1970) or for tween a given point and all others within a spatial set of
spatial variations in the competing pressures of demand for the pointst(see Fig. 2.}. Essentia11y. relative accessibility is a
relevant opportunities (Morris 1976). measure of the effort involved in making a trip;· while
 Accessibility indicaton for transport planning 93

II • I I I I I I • • • I I I I

i s i • c w i IS
i
, r:: ~
- --
~
I

J i
r- .J of
I ~
~

.-- J-
~ II?
_II - ,il ~ II ~ _ iJ
lid - J
.8
~

r- II1I
1,1 t---

~
I
I! ~ 'r -
I~· ... r!
'i
I .~
0

r----

-I I-
II i
Jllh Iii I!
• P-
:=

. '- .--fi ~ ~fq

'If B :l
J5
J §
..---- j r- I! '"

~ i~
I IIII!
j~
l- I I 'I t!
r----
e
I...--
~ I- iI ~i IIf ~~
.B
B n
II! ~~
i~'
:s
--
.5 ~
'--

f 1,1
-
:is

J
i--

1
I
'- i~I~ I
I
, !

~E- - - - - - - - - - - -.....
stlOJ.WIGNI

UlI:xJIIcI
_,c
8tIOJ.WlCOO
..
lIfIOo.&110
 94 J. M. MORRIS, P. L. DUMBLE and M. R. WIOAH
Relative Accessibility IntesroJ ACCOI8IbIllty from administrative restrictions on the use of services or
participation in activities (see Oberg 1976). The problem.
then, is to choose the most appropriate form from the
mass of alternatives.
I

,/ CHOOSING APPROPRIATE INDICATORS FOR EVAWATION

It is clearly outside the scope of this paper to prescribe
suitable measures of accessibility for every conceivable
application in transport planning. We coniine our atten-
AI-Cij AI-l...S!l tion here to the broad area of system evaluation, and
n
.... tJ..... hIM ta ............11\ dfnIIiI "8. • Ift tr.... u..10 IIII1Mm1 dlNGt .. tM
MgIOn
give detailed consideration in the next section to the use
cSlI"~ to Ceft!r.ll ........ Diatrlcl ""'''' ~.e.c. eo .. otMr aOftn
of accessibility indicators in modelling travel demand.
Fig. 2. Relative and integral accessibility. However, these aspects are closely related, as are all
potential applications of accessibility indicators. Irres-
integral accessibility is some measure of total travel pective of intended application, the practical value of
opportunities (Oberg 1976). The former undoubtedly accessibility indicators depends upon the extent to which
gives rise to the simplest measures of accessibility, al- they reliect behaviour and perception.
though operational measures of integral accessibility The principal differences in selecting suitable measures
vary considerably in complexity. of accessibility for evaluation rather than for some other
The emergence of a large range of measures of in- purpose are, first, the level of disaggregation of the
tegral accessibility is essentially the result of continu- population and activities, and second, the weight given to
ing attempts to link accessibility with behavioural ease of operation and interpretation of the measure. Four
theories. These attempts have concentrated mainly on general guidelines may be identified to assist in the
three aspects: first, the choice of an appropriate measure selection of accessibility indicators for evaluation:
of impedance to reflect the perceived cost of travel; (I) The indicator should incorporate an element of
second, assumptions about the perceived choice set of spatial separation which is responsive to changes in the
opportunities; and third, the choice of appropriate performance of the transport system.
attractiveness variables to ret1ect the availability of (2) The measure should have sound behavioural foun-
opportunities at destinations to satisfy the particular dations.
wants and desires of travellers. Consideration of the (3) The indicator should be technically feasible and
latter effectively differentiates the "process" indicators operationally simple.
into two groups: those which simply describe the ease of (4) The measure should be easy to interpret, and
traversing space via a given transport system (public or preferably be intelligible to the layman.
private); and those which measure accessibility to selec- These criterial are occasionally in conflict with one
ted activities or opportunities using a given trans- another. Nevertheless all should be considered to some
portation system. degree in the selection procedure.
Although the distinction between "relative" and "in-
tegral" accessibility was originally developed in relation The unit 0/ spatial separation
to "process" indicators, it is equally applicable to The question of the appropriate measure of spatial
measures of actual behaviour (such as trip rates and separation is not independent of the issue of the
travel times) which are in some sense measures of ac- behavioural basis of accessibility measures, but is treated
cessibility. Simple behavioural measures of relative ac- separately here for the sake of convenience. Spatial
cessibility include standardised trip rates between separation may be measured in terms of travel time,
specific areas. Likewise, the trip distnbution pattern in a distance, cost, or some combination of these or other
given region may be used to compute a measure of total characteristics of the transport system. In tum, each of
accessibility. Such measures assume that revealed travel these may be derived in different ways. For instance,
patterns are good indicators of how people value ac- estimates of travel time may be either measures of
cessibility when they choose their destinations (Zakaria perceived travel time, as reported by respondents in
1974). home interviews, or estimates of network travel times
In reality, the range of possible accessibility indicators obtained from shortest path algorithms. Unfortunately,
is almost endless, and only a broad outline is presented systematic errors are associated with every approach,
in Fig. 1. For example, the composite indicators which in and the problem becomes one of choosing the measure
themselves constitute a large family of measures, may be which best suits the problem at hand from the available
modified in a number of ways. These include varying the alternatives.
unit of separation, time of day, mode of travel, measure While a measure of perceived separation is attractive
of attractiveness of opportunities, measure of demand, on behavioural grounds when modelling individual res-
and level of disaggregation. In addition, the "gravity ponses, some form of actual separation is preferable for
type" indicators, as introduced by Hansen (1959), lend evaluative purposes. Moreover, measures (such as time,
themselves to a variety of functional forms of impedance cost and convenience) which monitor network quality
(power, exponential, Gaussian, etc.); and most indicators and performance are more satisfactory than measures of
may be modified to allow for "barrier effects" arising network distance, especially in urban areas. Koenig
 AccessibDity indicators for transport plannina 95
(1977), for example, employs a gener.dised cost function when examining the relationship between accessibility
based on the time, cost and effort involved in travelling and travel behaviour. This is best tackled by stratifying
by different modes. the population into relatively homogeneous groups, and
calculating accessibility for each group separately (see
Behaviouralloundations Turner 1972, Koenig 1977, Black and Conroy 1977,
Behavioural considerations inftuence two major Mitchen and Town 1977)".
choices when selecting appropriate accessibility in- The mode of transport available to individuals is a
dicators for evaluation: first, the choice between "out- particularly vital element in calculating accessibility.
come" and "process" indicators: and, second, the choice Countless studies have highlighted the marked dis-
between indicators of accessibility to the transport sys- crepancy between the number of opportunities which
tem, and indicators of accessibility to opportunities oia may be reached by car within a given time period, as
the transport system. compared with those which may be reached by public
"Outcome" venus "process" indicaton. The concern transport (Wachs and Kumagai 1973), or on foot (Hill-
for a sound bebavioural foundation does not automatic- man et aI. 1973; 1976).t Ac:c:ordingly, the short-run im-
ally imply a preference for "outcome" indicators, since pacts of particular land-use/transportation plans may
planning strictly on the basis of observed behaviour can depend substantially upon the mobility cbaracteristics of
be attacked on many grounds. Observed behaviour is the population. The findings of a Sydney study are a case
simply the response to current circumstances, giving in point: Black and Conroy (1977) found that a dispersed
only a single point on a demand curve of unknown arrangement of workplaces improves accessibility to
shape. In consequence, modelling on the basis of obser- employment for residents of outer suburbs, especially
ved behaviour can be interpreted as tautological: it leads those who have access to private transport (notably men
to self-justification (Vickerman 1974), and existing in- and higher socio-economic status women); while im-
adequacies merely become self-fulfilling propbecies for proved public transport favors women more than men by
the future. Moreover, it requires inordinately heavy data reducing, but Dot eliminating differences in ac:c:essibility.
input and is descriptive rather tban explanatory in the In recognition of the importance of mobility con-
formal sense. siderations, some researchers have proposed composite
The major disadvantage of using measures of actual "mobility" indices, or measures of "access to oppor-
behaviour to evaluate the transport/land-use system is tunities", derived by weighting accessibility indices by
that it is difficult to disentangle the inftuence of choices actual travel behaviour (viz. relative use of dilferent
and constraints. For instance, an increase in the total transportation modes and trip purpose frequencies) (see
time spent travelling may represent an improvement in Wickstrom 1971, Briggs and Jones 1973, Popper and
community wen-being if it is linked to increased levels of Hoel 1976). Such indices, however, are subject to the
participation in desired activities. Alternatively, the in- same criticisms as outcome indicators. Also the indices
crease may denote a worsening situation if it arises apply specificaHy to areal units, and thus do not permit
purely because a given set of activities is harder to reach .detailed consideration of distributional effects. The fact
(see Koenig 1977). Likewise, higher trip generation rates remains, however, that the more satisfactory alternative,
do not necessarily denote increased well-being. A desir- i.e. constructing separate mode-specific accessibility in-
able outcome for both individuals and society may wen dicators depends upon knowledge of actual travel pat-
be one in which activities can be pursued with minimum terns-only in this way can mode-availability be inferred
travel effort, rather than one which involves the largest on a large scale. Consequently an analysis of observed
number of trips. behaviour is a necessary (but by no means sufficient)
While actual bebaviour is in itself an inadequate basis condition for the modelling of accessibility.
for transport planning, there is a critical need to under- Accessibility to transpol1, or to OPPOl1unities'! Since
stand the relationship between supply factors and actual most travel is a means to an end, an accessibility
behaviour. Implicit in the use of "process" indicators in measure which reftects the distribution of activities
modelling and evaluation is the assumption that out- within the city is preferable t9 a measure which simply
comes are in some way affected by them. ·A detailed describes the ease of traversing space via a given trans-
analysis of actual travel patterns gives some indication of port system. There may yet be a place for measures of
the behavioural constraints operating on different groups connectivity of the transport network or measures of
in the population, and also provides a meaningful basis ac:c:essibility to pUblic transport-such measures may be
for classifying the population. As will be shown later, useful in pinpointing glaring deficiencies in the transport
socio-economic, demographic, and mobility characteris- system. But for most of the broader issues tackled in
tics exert a strong inftuence on the demand for travel, and present-day transport planning these measures must be
and consequently it is important to control for these effects rejected on behavioural grounds. Indicators of travel
time, distance or cost fail unless supplemented because
they reftect only one of the components of tbe satis-
tThe dilference between the freedom of movement of the car faction an individual may derive from his travel. Account
owner and the captive public transport user is not entirely a should also be taken of the probable interest of the
matter of speed. Rather it relates to the loss of time at terminals
and junctions and to the channelled nature of the public transport destination reached.
network (which limits the number of opportunities that can be Hence, the range of choice narrows considerably to
conveniently reacbed). the set of "process" indicators which describe ac-
96 J. M. MoRRIS. P. L. DuMBLE and M. R. WIOAN
cessibility to opportunities via the transport system. In stitution: for any given set of spatially distributed activi-
most cases this amounts to a choice between the various ties S located at time I from the origin i, there is an
forms of composite indicators shown in Fig. I; but in equivalent set of activities S'. which if located at the
some cases a simple "relative" accessibility index may origin would give an equal accessibility value (Weibull
be more appropriate. For instance. when services have 1976. Black and Conroy 1977).
administratively defined catchment areas the "choice" of Accessibility-related comparative indices have been
destination is not an issue. and accessibility may be more employed by Flowerdew (1977) to avoid tbis problem.
meaningfully measured by the "effort" involved in These correct the scalar effects between alternatives A
reaching the prescribed activity centre. Simple measures and B, by checking to see if option A is stiD better if the
of proximity to the nearest opportunity may even be costs/times of option B are used in A, and vice versa.
more appropriate for some very loc:aI activities. especi- The package provides both time and generalised cost
ally if the potential destinations are fairly homogeneous. In indices. Flowerdew (1977) comments on the difficulty of
the majority of cases, however, consumer choice prevails, finding the best method of measuring accessibility. noting
and the destinations vary considerably in potential utility. that a weighted average of costs or times (in which the
Accordingly, composite indicators are the most ap- weights are numbers of trips) may increase even when all
propriate since they not only reflect transport conditions costs (times) have decreased. This is because travellers
but also the wealth of choice provided by urban structure may. as a result. make longer trips. Flowerdew (1977)
(Koenig 1977). used Laspeyre and Paasche type indices. The Laspeyre
The choice of appropriate attractiveness variables for indices measured whether it would have been
inclusion in a composite indicator will depend upon the cheaper/more expensive (quicker/slower) to make the
specific activity or group of activities under study. Such journeys in configuration A with the generalised costs of
an indicator should normally include simultaneous con- configuration B instead of A. The Paasche indices were
sideration of supply and demand elements. For example, the same but based on the journeys of configuration B
accessibility to employment not only depends upon the Indices of less than 100 represent improvement. The
number of relevant job opportunities availatile within a forms of the indices used in the Sao Paulo package are:
given area, but also upon the number of persons com-
peting for those job opportunities. This aspect is in- Laspeyre
corporated in the modified gravity index specified by (I T"C,J1 T"C,,) x 100
Weibull (1976). or
The final selection of an appropriate operational form (I T"t"/1 T"t,,) x 100
of accessibility may be governed by technical con- Paasche
siderations of operational simplicity and ease of com-
prehension. There is a distinct trade-off between the or
behavioural relevance and the operational simplicity of (I T"IJI T"I,,) x 100
accessibility indicators. A composite measure which in-
corporates the perceived cost of travel and the level of where suffices a and b refer to the configurations A and
competing demand is the most acceptable on behavioural B being compared, T denotes trips, C generalised cost
grounds, but is undoubtedly the most difficult to apply. and I time. The summations are carried out over the
appropriate population group-e.g. for consumers in
Technical consideralions specific areas using public transport and travelling to any
The selection of an appropriate impedance function is destination.
essentially a technical issue. There is no universally Other researchers have turned to cumulative-oppor-
accepted theoretical basis on which to select the correct tunity indices or accessibility profiles as measures of
function. Moreover, calibration requires heavy data input accessibility (see Fig. 3). The principal disadvantage of a
and there are major difficulties in identifying the "true" graphical measure is that it does not produce a single
value of the separation d~cay exponents (Wilson 1971, value of accessibility which can be used to immediately
Curry 1972. Ewing 1974, Dalvi and Martin 1976). A compare alternative land-use/transportation plans. It
further difficulty may arise in the context of evaluation if does, however. ofter three advantages. First, the value
different separation decay exponents are used for weigbtings of the relative importance of separation and
different population groups. Conroy (1978) argues that attraction are made explicit. Second, the distribution of
this introduces bias into the evaluation process. opportunities with increasing distance from a given loca-
Whitbread (1972) suggests that a further disadvantage tion is apparent and may be compared for different areas,
of gravity-type indicators is that they implicitly weight modes and socio-economic groups. Third.· graphical
one unit of separation as equivalent to one unit of measures enable standards to be more clearly specified
attraction.t This relates to the property of distance sub- (e.g. S opportunities within C units of spatial separation)
in terms which are readily intelligible to the layman
(Whitbread 1972. Briggs and Jones 1973). To some extent .
tThls criticism is related more to the way in which these this third feature also applies to cumulative-opportunity
accessibility indicators have been applied In practice. rather than
to intrinsic features of the indicators, themselves. Vickerman indices of accessibility. but such measures are based on
(1974) represents one of the few attempts to determine the an artificial boundary and there is a problem in deciding
independent influence of attraction on travel behaviour. where to set the limit.
 Accessibility indicators for transport planning

AcceaalbUlty Protlle:
100

_""" 75
'!loG'
~tln so
2S

Cumuloth,o-opportunlty Indox:

-- 01

1
IthcU. . . . . 0'
ckaI~"11on ~omlU.'1n
• •UnaUon zone
ISII
30
omlta Gf _1111
_"lion
IromotlQlftl
ICQI
1
AI- ,,~fl

"c:.no.
"c:.no. -- 1.2.AI-
AI- 30
" c.n.. 3. AI - 200
;0

Gravity-type Index:
110

,Z 40
110 ,2 AI - I SI.'ICQI
I'Wlf_ _ 11'511IfICIII- ~

4 40 2 dP
.', AI- !1I+!1I+i!I +!11
1 4 II 4
- 30 + 10 + 10 + 10
- eo
Fig. 3. Sample calculations of some accessibility indicators.

Nevertheless, the similarities between the various appraisal of the spatial differences, and is based on a
types of composite indicators are more notable than their weighted ranking of life cycle, age group, immigration,
differences (see Weibull 1976). Indeed, Black and Conroy household and public facilities, car ownership, employ-
(1977) have devised an accessibility measure which ment and socio-economic characteristics. The higher the
combines the relative advantages of graphical and score, the greater the disadvantaged nature of the dis-
numerical indicators; specifically, a numerical value or trict.
index of accessibility may be derived by integrating the By comparing the different diagrams in Fig. 4 it
area under the cumulative opportunity curve bounded by becomes evident that a disadvantaged area would suffer
a given spatial separation limit. Unlike other cumulative- heavy restraint under supplementary licensing (requiring
opportunity indicators this index preserves information an extra licence to operate a vehic:le in the central area or
on the distribution of opportunities within the chosen "the railway triangle"). This is not a simple result to
separation band. The index also conforms with the six interpret. H the resident did not own cars (likely, in this
axioms of accessibility postulated by Weibull (1976), and area), then the sharp traffic reduction would be a key
has been shown to give empirical results which agree benefit, but if all the employment in the area was unsuit-
c:Iosely with those produced by a Hansen-type index able for the residents, they would be suffering a large
(Conroy 1978). But the index is still based on an artificial reduction in accessibility to their jobs. Further questions
boundary; and, as presently applied, does not allow for then arise on the degree of balance between residents
variations in demand at the supply points. and jobs in the area, and the average length of journey to
work.
An applied access/bUity indicator The detailed result of matching the different diagrams
The complications in definition and application of provides several illustrations of these distributional
different accessibility indicators should not be allowed to questions. Supplementary licensing and parking produce
confuse the issue: accessibility even 'as a simple relative, a very wide range of effects, and consequently pose
or uncomplicated integral, measure (see Fig. I) is an numerous awkWard distributional questions (W'Jg811 et aI.
effective addition to our assessment armoury. A practical 1974). In both cases the central area is the worst hit, and
example is siven to illustrate this point. FJBUte 4 shows it is interesting to note that this is the area most socially
four different diagrams on a common geographical basis, disadvantaged. It might therefore be argued that the
that of the city of Coventry in the U.K. The results are triangle restraint area (which forms the boundary for the
drawn from work (Wigan el al. 1974) done for the U.K. application of all the policies discussed) Is too large as it
Department of the Environment (1977) Traffic Restraint extends into areas beyond the central business district of
Study, where a wide range of different traffic restraint Coventry (a small area at the bottom of the triangle).
policies were examined using an equilibrium model (in- The accessibility changes for the cordon policy show
cluding elastic travel demand for private and public the lowest generalised costs (i.e. best accessibility) of the
passenger travel and goods transport). policies applied to the railway triangle, and even lower
The key point to be drawn from the diagram is that costs under restraint in the central area than in the
while two of the policies shown produce closely similar unrestrained state. This is a result of greater freedom of
net benefits, the spatial accessibility impacts have very movement for trips solely within the triangle, which
different forms. The diagrams illustrate simple measures therefore escape charging at the cordon.
of total separation (i.e. ICii. as shown in Fig. 1). The The parking costs show cost reductions for a very
social indicators diagram provides a basis for the social , large primary residential area to the north and west of
PARKING CORDON
:i
Benefit Benefit
£1400/hr £s2Oo/hr

GENERALISED GENERALISED
COST CHANGES COST CHANGES
IN ACCESSIBILITY IN ACCESSIBILITY

26-21 mmm 27 -16 IIIIlI

20-16 ~ 16-10 ~
12-8 t<~~:d
--
-~... 10-7 t~~~
!-
6-0 £IIIIll 7-0 £IIIIll §!:

o- 1-31

1-31-1-141
Itr:ttl
l1li
0-1-41

(-51 - 1-91
1<.;>',1
Bl
!
:u
r

THE PATTERN OF SOCIAL

r
!;;
DISADVANTAGE
Benefit
1
£65oo/hr §!:
i'"

f
GENERALISED
COST CHANGES

---
IN ACCESSIBILITY PDINTS AGAINST

56-41
~ 80-65

41-32
rat 35-30

32-23

23-16
.,
ID 12-9

8-0

o
~
~

o
15-7 ~
7-0 IJIlII]
, Fig. 4. Spatial impacts of various transport policics in Coventry (After Wigan et aL 1974).
 Accessibility indicators for transport plonnins 99

the triangle (as a direct consequence the number of trips tive opportunities index, as one dimension of the cate-
rise in this area). This has implications not only for land gory analysis matrices produced (Don 1975); the cell by
use but also for the public transport system which would cell (household) trip production rates varied with the
suBer reciprocal decline in passengers. level of accessibility, but in no consistent manner. Both
It may be concluded that: studies, however, adopted an aggregate approach to ac-
(I) Supplementary licensing produces the least pro- cessibility measurement (i.e. all persons in an origin zone
gressive effect; inducing the greatest accessibility shift in were assigned the same level of accessibility). Ac-
the three central wards (i.e. the triangle), and the least in cessibility, as pointed out earlier, will vary from in-
the peripheral areas to the north, east and west. dividual to individual and zonal aggregation provides an
(2) Parking charges produce the same general patterns inadequate basis for reproducing variations between in-
as supplementary licensing but the range of accessibility dividual circumstances. Individuals, or relatively homo-
shifts is not so large, and in some areas, the charges geneous groups of individuals, should be adopted as the
actually induce traffic. basic unit throughout the modelling process and should
(3) Cordon charging actually produces progressive not be introduced solely at the evaluation stage.
effects, and might therefore be rated more highly as a Several relatively succesful attempts have been made
result. The less advantaged areas retain their mobility to formulate models of travel behaviour based upon the
and are affected least, while the outer areas suffer the principles of micro-economic consumer demand theory,
revenue. particularly in the realms of travel choice (see McFadden
The general social distributional impacts are clearly (1973) for an excellent treatment). Some of these
highlighted by this analysis. The change in emphasis of represent explicit attempts to formulate trip generation
the assessment produced by the extra information pro- sub-models on micro-economic bases. For example
vided by a simple accessibility indicator is substantial, in Koenig (1977), after accepting that the exponential for-
the light of the close economic comparability between mulation of the gravity model was the correct model for
cordon and supplementary licensing. trip distribution, demonstrated that the trip generation
However, it is clear that none of the established rate was a function of accessibility. Bums and Golob
measures of accessibility satisfy all of the requirements (1976) demonstrated how an accessibility measure can be
for transport evaluation. Typically, simple measures fall incorporated into several of the travel choice areas.
down on behavioural grounds, while indicators with Niedercorn and Bechdolt (1969) and Cochrane (1975), in
stronger behavioural foundations are complex and attempts to give the gravity model a micro-economic
difficult to apply in practice. More importantly, even basis (using vastly different assumptions and ap-
though some indicators have a stronger behavioural basis proaches), produced trip generation sub-models which
than others, none are completely acceptable on incorporated what could be considered to be accessibility
behavioural grounds. This is because the established measures. Lastly, Williams (1977) in a rigorous
measures do not explain why increased accessibility theoretical treatment of travel demand models once
should lead to increased trip-making. Since this probably - . deduced the same result by indirect means.
represents the major stumbling block for accessibility Each of these direct or indirect approaches to ac-
indicators, the following section gives detailed con- cessibility and its importance in travel demand modelling
sideration to the theoretical underpinnings of accessibility . can be classified into one of the four approaches identified
indicators. by Koenig (1977):
(a) Common sense (e.g. Hansen 1959)
MJCJlO.ECONOMlC THEORY, TRAVEL DEMAND AND (b) Axiomatic (e.g. Weibull 1976)
ACCISSIBILITY (e) Consumers surplus (e.g. Neuberger 1971)
A perceived change in accessibility either affects travel (d) Behavioural utility (e.g. Koenig 1977).
behaviour directly or alters levels of satisfaction with the Distinctions between the approaches are blurred. parti-
transport/land-use system. Various theories, founded on cularly between the latter two, and often, depending on
models of micro-economic consumer behaviour have the vitial assumptions, lead to the same result. It is not,
been specified to express this implied causal relationship however, the aim of this paper to indulge in a theoretical
mathematically. Empirical results in support of these and abstract discussion on the merits of each of the often
theories are reviewed here together with the essentials of highly mathematical treatments given to some of these
the theories themselves. approaches. Rather the intention is to demonstrate that
Specifically, we concentrate on the trip generation there are accepted theories of micro-economics which
sub-model. This represents an area where it has long suggest that trip generation is likely to be influenced by
been recognised that accessibility (or supply) conditions accessibility. In particular we will examine the attempts
have a genuine inftuence, although previous attempts to of Koenig (1977), Niedercorn and Bechdolt (1969) and
model this relationship have not been particularly Cochrane (1975) to demonstrate this point. These are
encouraging. Daor (1975), for example, concluded that supported here with some empirical results.
the level of accessibility to relevant activities (when Some analyses were carried out on a household travel
measured by a Hansen-type index) produced insufficient survey executed in Ballarat in 1970 as part of the Ballarat
improvement in the trip generation sub-model to warrant Transportation Study (Harris Lange-Voorhees 1971).
inclusion. The Victorian Ministry of Transport (MoT) Ballarat was chosen because the sample size was small
included levels of accessibility, measured by a cumula- enough to be manageable-1284 households containing
 100 J. M. MoRllfS. P. L DlBIBLI! and M. R. WIGAN
3804 persons over the age of S-and the survey included It should be made clear that this derivation focusses
data on all trips made, including walk and bicycle modes. on the individual. Hence the accessibility measure to be
Systematic under-reporting of walk trips is expected to used should be the accessibility as perceived by the
have occurred, as only one mode was recorded for each individual (and therefore dependent on the mode upon
trip. Where two or more modes were used, the access which they rely).
mode (often walking) was eliminated at the trip linkage There is no real reason why accessibility should not
stage. Such conventions of "dominant mode" coding also vary between trip purposes. It is conceptually ape
ignores key information on access modes which is now pealing that both the measure of opportunity for inter-
being realised to be of central importance in mobility and action and the wiUingness of an individual to travel
market segmentation approaches to modal choice. should vary with trip purpose.
Figure S gives practical weight to this first derivation.
This shows graphs of daily home-based trips per person
The approach 0/ Koenig (all modes) vs accessibility to relevant opportunities for a
Koenig demonstrates (on the assumptions briefly out- particular person category for some French cities
lined above) that the net utility (i.e. benefit) Ul, derived (Koenig 1977). The graphs show an increase in the
from a trip is, average for an individual of type k living observed trip generation rate with an increase in ac-
in zone i. given by the expressiont: cessibility, when both are defined and calculated on the
basis of the individual.
I
U," =x log. Al +constant
(I)
It has not been possible to compute similar indices for
Ballarat (along the lines of those in Fig. I), but the effect
where of varying levels of accessibility has been approximated
by subdividing Ballarat into a series of concentric rings.
AI" -- I Sle -,,~t, an ' These were numbered sequentially 1-4, outwards from
I S," "integral" (normalised) Hansen
the CBD. The mean trip generation rates for groups of
index (2) individual were then estimated on the assumption that
accessibility, by all modes, to all activities considered to
and S,· is the number of potential destinations for in- be important trip attractors, declines with distance from
vividuals of type k in zone I. x' is the exponential the centre of Ballarat. In a city such as Ballarat which
parameter associated with the destination choice has a single centre this assumption appears to be valid,
decision for individual k. C1, is the "cost" of travel particularly as Koenig (1977) demonstrates that it holds
between i and J. for the individual of type k. From now for tertiary employment places in the more complex city
on the suffix k will be removed, but it is implicit in all of Marseilles.
formulae. Figure 6(a) shows the mean trip generation rate for
The reasoning behind this hypothesis implies that if various groups of individuals by the number of the
the individual at i were to rank all potential trips in concentric ring in which they reside. For this analysis,
decreasing order of perceived net utility, then he would individuals were grouped according to their employment
travel just enough to ensure that the gross utility derived status (Harris Lange-Voorhees 1971). The result is in-
from making the last trip was exactly offset by the conclusive: the hypothesis that accessibility affects trip
disutility of undertaking it. That is, as well as the usual generation is not fuHy supported; but neither can it be
assumption that the trip rate for an individual at zone I is rejected. as there are probably several other influences at
a function of his socia-economic characteristics, Eit work. Fltst, the method of stratifying the population is
expressed mathematically as: probably not detailed enough to account for factors such
as income, socio-economic status, number of cars
(3) owned, etc., which also vary systematically with distance
from central Ballarat (Harris Lange-Voorhees 1971), and
eqn (I) implies that it is also a function of the average net which have previously been found to be strongly related
utility. hence of accessibUity, to be gained from making a to trip generation rates. Second, the method of ap-
trip. Thus eqn (3) becomes: proximating the accessibility level by distance from cen-
tral Ballarat attributes the same accessibility level to
(4) every individual resident in any particular concentric
ring, contrary to Koenig's assertion that a person's ac-
cessibility is dependent on the mode they choose (or are
forald) to take (Oenig 1977).*
tKeoaig's formulation bas been sJiahdy modified to facilitate FIgUre 6(b) shows that the non-car trip generation rate
easier comparison with other researchers' work.
*WiDlams (1m) would araue that accesiibility is not to be is affected by accessibility as postulated. The falloff with
calculated solely on the basis of chosen mode. His measure is decreasing accessibility is most marked for walk trips,
based on composite generalised cost and is internally consistent whUe tram§ tripS falloff sliPdy. and bus triDs increase
with his travel demand model. If an Individual has (or perceiVes) margmally. t<igure 6(c) snows tne enect ot accessibility
no real alternative mode, then Williams' measure reduces to that
for the chosen mode. summed over all person types for trips stratified by
_ ISince the Ballarat Transportation Study Was carried out in mode. This clearly indicates that car travel is the main
1970 the tram service has been withdrawn. component which weakens the hypothesis.
 Accessibility indicators for transport planning 101

PERSON CATEGORV:

8 9 10
AI-ACCESSIBIUTV TO TERTIARV EMPLOYMENT PLACES

Fig. S. The elfect of accessibility on trip rates in selected French cities.

-
PERSON CATEGORIES: MOOE OF TRAVEL
((a) and (b) ) (el

----
------- .......,
- - u-tGyocl TI11'II

~
A _A'-
7---"$.-=--------
,- "",:,...,
•.• -
-.---.
\
... , -----
..- ......\..- \

-- ---...------
~

" 3 3

"
"----
.... '.
:-- ---"'._---_._- ---.
\
-----..-
-- ~.... :::--'~-..<--,-"
'.-- '\. -'" ~

~ ------
3
RING No.
•
RING No.• RING No. •
(a) All modes (b) Non-Car Modes (e) All Tripe. All Persons

Fia. 6. The elfect or accessibility on trip rates in Ballarat.

An attempt was made to overcome the effect of car which indicates a slight but significant decrease in trip
ownership on the overall trip generation rate by dividing generation with a decrease in the level of accessibility.
the population into licence holding and non-licence holding The category:
groups. This step approximated the range of person types
used by Koenig. Thus Koenig's person category of: "housewives licensed to drive"

"non-working, non-car owning and over 60 years old" is also shown. Socio-economic factors are again likely to
explain the increase in trip generation rate with tlie
decrease in level of accessibility. However, housewives
is approximated by our person category: with driving licences at all levels of accessibility make
significantJy more trips than those without. indicating
"retired and non-licence holding". that the former group is far more "mobile" in the sense
discussed earlier in this paper. The same conclusion can
be drawn for all employment status groups although the
Figure 7 shows that once again our analysis does not
unequivocally support the hypothesis; apin the result is results are not presented here.
inconclusive. However, also plotted in Fig. 7 are the Fundamental allumptioflS underlying Koenig's approach
results for the person category: There are some practical considerations and theoreti-
"housewife, without driving licence" cal assumptions in Koenig's formulation which may limit
 102 J. M. MORRIs, P. L. DUMBLE and M. R. WIGA'I

PeRSON CATEGORIES:

_.. ---
---- -----
- - RoIlrad. Non-oriver

- - - Housewife. Non-Oriver

------ Housewife. Driver

Nol.: Nc>1 OIIOIIGI' ""••"VIIl,ON InRmo No. I

4
RING No.
Fig. 7. Furtber effects of accessibility on trip rates in Ballarat.

the effectiveness of the whole approach when attempting in Koenig's formulation (though shared with current
to incorporate it into a working trip generation sub- travel demand modelling practice) is that all travel is of a
model. simple nature, That is, travel is assumed to be composed
One practical consideration is that of the zoning sys- solely of two-stage journeys:* starting at home, going to
tem. According to micro-economic consumer choice a single destination for a single purpose and then return-
theory the individual perceives that a set of alternatives ing bome. As a large proportion of travel is accounted
is open to him (Henderson and Quandt 1971) and that each for by multi-stage journeys, this assumption is incorrect.
alternative possesses a certain level of utility.t In a work- This may undermine the behavioural veracity of most
ing model the alternative destinations are typically agg- trip generation models in current use, due to the difticulty
regated within a zonal framework. For the model to be in specification of mode and purpose in multi-stage
behaviourally sound it is therefore necessary that the journeys and the mutual inftuence of each stage on
individual perceives the spatial distribution of activities perceived accessibility relevant to preceding and suc-
as this discrete pattern of zones. This is perhaps unlikely ceeding stages.
except for trip purposes such as shopping for high order One deficiency specific to Koenig's model is that the
goods which are available only at a very limited' number theory involved in the formulation does not provide us
of locations. It has been shown that accessibility indices with a behaviourally based functional form. That is,
are sensitive to the zoning system used (Dalvi and Martin while we know that:
1976). Ben Akiva and Lerman (1978) further suggest that
unless the zoning system is carefully designed "the 1i =I(E" V,) = g(E" log. A,) (5)
measure of accessibility will in general be biased".
Another general problem may be caused by the we are left with no clues as to what the function may be,
necessity to construct separate indices for different It would seem that increasing accessibility leads to an
modes. This requires some previous knowledge of the increasing trip rate, ad Infinitum, as eqn (I) suggests that
chosen mode; knowledge which does not become avail- the net utility derived from making any particular trip is
able in the sequential approach to travel demand model- independent of the number of such trips already under-
ling until after the trip distribution (destination choice) taken in the time period under consideration. The
stage. Some commentators have suggested a mode- concept of satiation must somehow be introduced. In
specific approach to trip generation to overcome this micro-economic utility theory this corresponds to the
drawback (Vickerman 1974, Bums and Golob 1976), requirement that marginal utility be a positive, but
given the marked effect of car availability (defined at the decreasing function of the quantity consumed (Hender-
time the decision is made to make, not to make, or to son and Quandt 1971):
delay making, a trip). Figure 7 indicates the strength of
this effect assuming that licence/non-licence holding is a
(6)
proxy for car availability, Approximating assumptions
can also be employed if one wishes to avoid the adoption
of mode-specific trip generation values. Other derivations using the same framework as Koenig
One other important, yet tacitly accepted assumption (and thus containing the same general assumptions) have
been proposed which attempt to incorporate such a
satiation effect.

tit is ollly necessary that the individual be able to rank all the Nledeftom and Bechdolt's approach
perceived alternatives open to him (HendenoD and Quandt 1971, Niedercorn and 8echdolt (1966) adopt the approach of
Samuelson 1971), , lith tili' f' d' 'd aI·th th .
*A journey is detlned as a trip sequence startina and endins at maxun s n~ e u tyo. tn IVI U S WI respect to etr
the bome bue. A mUlti-stage journey involves more than one . travel requtrements subJect to the constraints of the total
intermediate destination. amounts of time and money that individuals are willing
 Accessibility indicators for tI'aIIsport plannlna 103
to spend on travel. This arises in the context of deriving S -AlII
tbe gravity model from micro-economic theory. (2) a trip destribution term, IS,ee-x1/J'
As a first approximation they assume that the net
utility derived by an individual at i from travel, u" is a
function of the number of trips undertaken to eacb which is simply the gravity model as formulated by
destination j, Tilt and the potential for interaction at each Koenia (amongst others) but with travel time rather than
destination. Thus generalised travel cost as the impedance measure.
Equation (12) can be rewritten as eqn (16) by sub-
stituting for ;" as follows:
u, =a~AJ'(TII) (7)

(13)
where a is a constant of proportionality, and A, is the
perceived attractiveness of j for interaction. and
A simpler prObfein statemelil-is obt8iDe&i by iDodifYins -
the constrainl term slightly 10 cover only a time con-
straint (i.e. a travel time budget):

maximise (14)
(8)

subject to _ ~/IISlJe-ul/
(9) ,,=---- (IS)
~S,e-A.I/

where H, is the total time allocated to travel; til is the

travel time from; 10 j; Til is the number of trips from i (16)
to j.
Niedercom and Bechdolt produce a solution assuming
a logarithmic function: Summing over all destinations (i.e. over all J) gives the
total trip generation rate for an individual in zone;
U, :;:10g.T, (10)

where T, is the sum of all trips produced at ; by the T,=H,---- (17a)

individual.t L/IJS,e- Atll
This obeys the first and.second order requirements: or
T,:;: H,·A. (l7b)
(6) L/I,IS, e-AII/

The logarithmic assumption leads to the result Thus the total trip generation rate is an increasing
function of the level of accessibility, although not
HA directly proportional to it as might appear from a first
TII:;:~~ (II)
I, lA, glance at eqn (l7b).
The effect of the accessibility term (.4,) is dampened
by the denominator. Thus if A, increases due to a fall in
where ;; is the average travel time for trips taken by an any or all til'S, the denominator will also increase, but not
individual at i. If one then assumes that an individual's
by as much as A" hence T, will increase at a slower rate
perception of the attractiveness of j for interaction (.4,)
than .4,. Similarly if .4, increases due to a redistribution
is the accessibility of j with respect to I as defined by
Koenia, i.e. S, e-A'ij, then of opportunities in favour of locations closer to i, the
increase in the denominator will be dampened by the I."
which is smaller, hence carries less weight, for the closer .
H. §ie-uli
T.=.....! (12) zones than it is for the more distant zones.
II ;; IS, e-Aij One consequence of Niedercom and Bechdolt's ap-
which is composed of two terms proach is that each individual has a maximum amount of
time (and/or money), which he is willing to devote to
travel. The individual will use his maximum time, except
(I) a trip generation term, (HI;'); and in the unlikely event that he is able to make aU the
interactions he desires in less time. Therefore an im-
provement in the transport system will generally nol
me fuU derivation is not reproduced here: the reader is
referred to the orisinal article (Nledercom and Bechdolt 1969). cause an individual to spend more or less time travelling.
Slight differences have been introduced here to conform with Thus each individual's travel time budget is simply
Koenig's terminology. obtained by observing his travel behaviour. i.e. the
TR-A VoL 13A, No. 2-C
104 J. M. MORRIS, P. L. DUMBLB and M. R. WIOAN

amount of time he wishes to spend travelling equals the counted for by the discrepancy in car usage; i.e. when
amount he actually travelled. time spent travelling as a car passenger (not shown) is
The average time spent travelling daily by Ballarat added to that spent driving, men still spend more time
residents was analysed for different categories of in- travelling than women in all age· groups. A· second fea-
dividuals. In doing so it was possible to establish. ture is the drop in tot81 time spent travelling by women in
amongst other things, which grouping gave the greatest the 4O-5S age bracket 'Ibis may possibly be due to
between-groups variation. The results for all individuals women in this group no longer needing to accompany
are presented below in Table I. their children on trips. They may even send their chil-
A method of stratification which showed a large dr~n on errands as they become old enough to ac-
amount of between-group variation was a combination complish these tasks by Ihemselves. These and other
sex/age grouping. One group (males, between the ages of similar observations rapidly lead one to realise that the
18 and 24 inclusive) exhibited a daily travel time budget travel demand of individuals cannot be considered in
of almost 93 min (43% above the average), while another isolation from their role in the household.
group (males, less than 10 years) exhibited a daily travel Niedercom and Bechdolt's approach, whilst retaining
time budget of only 39 min (40% below average). FIgUre the desirable feature that accessibility be considered on
8 shows the results for aU sex/age groups. Included on an individual basist also manages to dampen down, but
Fig. 8, for interest mainly, are the daily travel times not prevent the ever-increasing trip rate effect of in-
allocated to car driving and walking by residents in the creasing accessibility in Koenig's model.
various sex/age groupings. However any general deficiencies and underlying
The graph for time spent walking is quite similar in assumptions inherent in Koenig's model will still be
shape to that obtained from an analysis of a National present.
Travel Survey (NTS) of the United Kingdom by Daor
and Goodwin (1976). In particular, the small amount of The approach 0/ Cochrane
time spent walking daily by men in the age range 20 to SO The approach of Cochrane (1975) could be considered
years is observable in both Ballarat and NTS results. The almost to be begging the question in relation to his
most obvious difference between the two analyses is the treatment of accessibility and trip generation. His under-
relatively low average amount of time spent walking in· lying assumptions are very similar to Koenig's as
Ballarat; 10.5 min compared to 18. This is partly expressed by eqns (IH5), but Cochrane introduces the
explained by the method of "dominant mode" coding concept of satiation, albeit in a somewhat arbitrary
adopted in Ballarat. manner, by assuming that the demand for trips between i
Some interesting sociological influences on observed and all j by an individual, is related to a factor 0, (which
mobility are apparent in Fig. 8. For instance, the ten- is really a saturation level of trip making) as well as to
dency of men, at all age levels, to spend more time A,.
travelling than women. The difterence is more than &C- Cochrane then derives* the following expressions for
TIJ and T,:

tThat is eqns (l7a) and (l7b) can be disauresated by person

type (incorporatins both personal and household attn'butes),
mode and purpose (although it could be stretchins credibility too
far to suaest that individuals have travel time budsets for each and
travel purpose).
me denvation is not presented here, but it appears in the (19)
original article under the Section: "The unconstrained model"
(Cochrane 1975). Note that his symbols stand for quantities which where 0, can be thought of a saturation trip rate, and K
differ from ours. is a parameter.
Table 1. Mean daily travelling timet per person in Ballarat (1970)

Time Destination purpose Tune

Primary mode (Min) (M"m)

Car Driver 29.3 Home 28.9

Car Passenger 13.2 Work 11.3
Tram 13.6 Employer's Business 1.5
Bus 2.4 SociaI/Remational 9.0
Taxi 0.5 Eat Meal 0.5
Truck Passenger 0.2 Medical/Dental 0.7
Walk 10.5 Personal Business 1.7
School Bus 0.2 Shopping-Convenience 4.2
Other (Bicycle) 5.0 -Comparison 1.5
School 5.5
TOTAL 6S TOTAL 6S
tThe travel time for eacb trip made by each individual was stated (in terms of a beginning time
and an ending time) on his/her travel diary. These stated times are used throughout this section.
 Accessibility indicators for transport planning lOS
ICO ICO

r\
I \
CAACf!lV£A-FB.IAI.E

"\'EIWlE
CAA~ -
20._
- -
I \ - --
-1--"'\-
~,
J
':>.c-- "
A_ - ,""WALK-nwu: / '
10 WALK "Ioii:i:iiJ-{ - -',:::;'7- - - - 10

<10 1(1." IS·" ,,·2< 25·38 _ SS'" _

-
OWAeo
......
Fig. 8. Daily travelling time per person in Ballarat.
AGEGROlJ'

Once again, this trip generation sub-model is best veys, where trip stage and sequence tend to be ignored.
applied to relatively homogeneous groups of individuals The coding conventions themselves can cause the loss of
and can then be mode and purpose specific. An iterative critical data: the choice of a single dominant mode-
Ordinary Least Squares (OLS) procedure would typic- usually omitting the access mode-in a complex journey
ally be used to estimate G. and K. Rough estimates of G. is of special significance. Further information may be
and K were made for the results depicted in Fig. S for ignored at the analysis stage: for example undue aggre-
the City of Nice (1966): gation of purpose codings can result in significant loss of
detail within a trip sequence. Nevertheless information is
G, =4.7 (daily trips per person) retrievable by going back to the basic survey data.

and
K =0.068.

Cochrane's model is perhaps the most useful from an

operational standpoint The model also has the theoreti-
cal nicety that it obeys the requirement of decreasing
marginal utility of consumption. It remains to be seen
how well it performs in practice, although this will
depend on the method adopted for stratifying the popu-
lation into homogeneous groups. The idea of a saturation
trip rate is an intuitively appealing way of overcoming
the major deficiency of Koenig's formulation. However,
the general assumptions underlying current modelling
processes have not really been questioned in the Coch-
rane model, nor in any of the others discussed.

. LINKAGE OF TRIPS AND ACTJ\'rI'IES

The Ballarat data were examined to check on the
I.tI9oI'LOYED All ~ _ _ _ _ _oc:::=
-
=
- IwO·,lago

-Intttt-siago

proportion of multi-stage journeys. r1gW'e 9 shows that

the incidence of multi-stage journeys in Ballarat was up
to SO% for some groups and similar figures have been
found elsewhere.
The simple calculation of the different accesslbUity 1.0 ~.O
indices Is complicated by the inclusion of linked trips.
STATUS DAY NUMBER OF JotJANEYS
There are two distinct problems. The first is the practical
coding of the data at the initial stage of transport sur- Fis. 9. The incidence of journey making in Ballarat.
106 J. M. MORRIS. P. L. DUMBLE Bnd M. R. WIOAN
The second problem is conceptual. and posed by the of consistency questions arising from the underlying
treatment of behaviour: i.e. is travel sequential or simul- base of individual utility in tbe construction of the for-
taneous in nature? Accessibility and mobility are both millism. The concept of accessibility is related most
indicators designed to summarise actual or perceived naturally to a simultaneous view of travel and destination
potential for travel. 'lUId are therefore closely linked with demand and choice, where the combinations of mode
hypotheses of destination choice (and in a similar man- and destination may be seen to define the accessibility to
ner, modal choice). This link is made the more difficult to tbe home base of the journey. This view can be recon-
handle by the need to relate both travel behaviour and ciled with sequential choice models of mode and des-
accessibility concepts to individual utility specifications. tination fairly easily for out-and-back home based jour-
The first level of aggregation poses the problem of neys. but as we have already seen earlier in this section a
differentiating behavioural, preceived, and resource significant fraction of journeys are multi-stage in nature.
(actual) determinants of travel utility and is treated in The following questions may now be posed. Is ac-
detail by Mcintosh and Quarmby (1972) and Wigan cessibility to be attributed to: (1) the home base of a trip
(1971). At the level of zonal aggregation involved in sequence? or (2) eacb successive zone visited?
mobility and accessibility calculations, such differen- In the latter case there are further choices for attnout-
tiation is more relevant to general evaluation issues than ing the accessibility so calculated: either by zone-by-
to individual utility questions. and though more familiar zone recalculation where each zone in the sequence is
to transport analysts, must be treated later in the chain of treated as a "home base" with access to opportunities
analytic:aI procedures. one stage away: or by an accumulation of sucb cal-
The fundamental issue is that of utility specification culations and the total attributed either to the home base
for individuals, and the manner in which this conditions or credited to every zone visted in the sequence. Hanson
and structures the functional forms at a level aggregated (1978), however reports the empirical results of one
enough for practic:aI choice analyses. Williams (1977) has study which demonstrates that ''95% of all stops are
reviewed a family of such necessary consequences in planned before the individual leaves home".
functional forms, showing how both the unrestrictive If a simultaneous decision model is adopted. then all
assumptions of entropy c:aIculation5-which contain no these choices collapse to a cumulative accessibility value
specific utility assumptions or specifications other than allocated only to the calculation, although other varia-
the range of random combinations of choices, but solely tions could be embraced which would then include some
aggregate constraints-and the cumulative choice prob- non-home based relevance. If a sequential model is
abilities from specified utility functions lead to families adopted, the relevant accessibility calculations become
of choice models of very similar form (but with critical further ill-defined. and strongly influenced by tbe precise
underlying constraints inherent within their structure). models adopted whatever the index of accessibility
The cboice of destination and of mode is frequently desired.
assumed to be a (simultaneous) single decision. but in The loss of specific linkage labels on multi-stage jour-
fact represents two separate choice functions whicb may neys in conventional transport models does not neces-
or may not correspond to a single simultaneous choice sarily rule out "correct" accessibility calculations in all
function. The separability of the mUltiple logit function is these cases: the case of sequential models with zone-by-
frequently exploited in this way, to overcome such prob- zone accessibility calculations with no accumulation will
lems as are raised by different behavioural models for give tbe same results, although requiring the recovery of
destination and mode choice. This assumption is most the information that a new unlinked trip is actually to be
important. The development of any utility-based choice treated as "home based" for this purpose. These close
model covering destination choice must place constraints inter-relationships between elastic travel demand, travel
on the evaluation framework. It should also include or and destination cboice hypotheses, accessibility, and the
develop a summary measure of destination opportunities unifying'effects of individual utility theories have the net
(or accessibility) which affects both the level of trip result of further constraining our freedom to chop and
demand and its geographical potential. Williams (1977) change models between different stages of the transport
illustrates this point by a minor observation that "if a trip analysis process. The emergent importance of trip link-
production model is developed from an underlying utility age in this web shows up clearly yet further constraints
structure, then the appropriate measure of net utility, on tbe transport planning process as so often carried out.
which involves level of service variables in the logit This link between modelling analyses and ac:cessibility
approximation, is proportional to the log transform of a assessment binds different activities together through the
modified Hansen measure of accessibility and of similar multi-stage jOl,lmey and through the fundamental links
structure to the index proposed by Koenig (1977)". Pre- between destination choice and the activities at those
cisely the same sort of requirement arises from the destinations which provide the motive for movement
inclusion of accessibility measures in category analysis The most closely related area of special concern is that
procedures (Daor 1975. Dalvi and Martin 1976), wbere of directly representing activity linkages, without the
the link is drawn at the evaluation stage when the resul- intermediary of links between journey stages. Descrip-
tant fUnctions in the models of elastic travel demand tive models of the multi-stage journey and the chained
must be integrated. activity structure involved have been buUt using Markov
The weight of Williams' (1977) synthesis is towards and transition matrix formats (e.g. Wigan and Richards
sequential models of choice, due to the readier resolution 1978). Such descriptive models are inadequate for more
 Accessibility indicaton for transport planning 107
• than pragmatic use, and causal hypotheses are needed to Second. marginally pushing back the limits of the
complete the network of motives and constraints for current models by incorporating accessibility measures
travel behaviour. into the models, and by other refinements such as allow-
ing the intermediate stages of multi-stage journeys to
CONCLtSIONS affect the trip generation 8I!d modal choice decisions.
The limitations to current practice brought out by our
review of accessibility indicators are now summarised. IIEF1!RENCIS
(I) The current travel demand modelling practice of Begs J. J. (1976) Some empirical findings on the relationship
treating trips as separate events sets the limits of our between residential density and accessibility to job oppor·
tunities. Proc. 8/1t ARRB Con/tltnce 8(6), Session 32, 7-IS.
ability to explain behaviour in the following areas: (a) Becker G. S. (1965) A theory of the allocation of time. The
trip generation, whether accessibility is used as a Economic Journal 75(299). 493-S17.
production variable or not, due to the inability of inter- Ben-Akiva M. and Lerman S. (197S) Forecasting models in
mediate stages to influence the decision to make a trip, transportation planning. A paper prepared for presentation at
and (b) modal choice, due to the inability of intermediate the ConI. Population Forecas/inglor Small Areas.
Ben-Akiva M. and Lerman S. (1978) Disaggregate travel and
stages of journeys to influence this decision (Bowyer and mobility choice models and measures of accessibility. Pro-
Tao 1978). ceedings o/Ihe Third In/er. ConI. Behavioural Travel Model-
(2) Household surveys of travel are deficient in the ling. Tanunda. South Australia. Croom Helm, London.
following areas: (a) trip purpose-only one destination Bentley G. A.. Bruce A. and Jones D. R. (1m) Intra-urban
and origin purpose for each trip is recorded, when in fact journeys and activity linkages. Socio-Econ. Plan. Sci.
11,213-220.
more than one activity may be pursued at any particular Black J. (1m) Public inconvenience: Access and Tral1e1 in Swen
destination. including home; (b) the scope of the travel Sydney Suburbs. Urban Research Unit. The Australian
diary-frequently only one day of travel is recorded. National University. Canberra.
which is inadequate to cover the full range of frequent Black J. and Conroy M. (1977) Accessibility measures and the
social evaluation of urban structure. Environmen/ and Plan-
and regular activities pursued by household members, ning A 9. 1013-103L
individually and co-operatively; and (c) the management Brigs D. A. and Jones P. M. (1973) Problems in transportation
of travel within the household-details are not usually planning for the conurbations: the role of accessibility. In-
recorded on decisions regarding, say, the allocation of slilule 0/ British Geographers 1973 Annual Coni.
D. P. and Tao A. (1978) An Appraisal 0/ Trallel Slruc-
cars between licenced drivers in the household, or who Bowyer,lures in Canberra. Australian Road Research Board. Internal
will undertake the travel necessary for the collective Report AIR 289-S.
well-being of the household. Bums L. D. and Golob T. F. (1976) The role of accessibility in
It is possible to reduce the effect of these constraints basic transportation choice behaviour. Transpol'talion 5(2).
by inverting the traditional approach to transport plan- 17S-198.
Clark C. (1951) Urban population densities. 1. Royal Statistical
ning. This is to adopt the view that travel is simply Soc. 54(4).
brought about by the physical separation of people from Cochrane R. A. (1975) A possible economic basis for the gravity
some of the activities in which they desire to parti- lJIodel. J. Trans. Economics and Policy 9. 34-49.
cipate-the activity linkage view of travel (Jones 1976, Conroy M. (1978) Accessibility and the evaluation of land
1978, Bentley et aI. 1977. Hanson 1978). Moving from use/transport plans, paper presented to 91h ARRB Conf.. 21-25
August 1978, Session 25. 1-17.
travel per se to personal and household activity patterns Curry L. (1972) A spatial analysis of gravity Hows. Regional
and aspirations in general, must eventually lead to a Studiu 6. 131-147.
better understanding of individual reactions to transport DaJvi M. Q. (1978) Behavioural modelling, accessibility, mobility
policy. Indeed, it may be found that particular "transport and need: concepts and measurement. Proceedings 0/ 3rd
Inter. Coni. Behavioural TralJel Modelling, Tanunda. South
problems" can best be solved by non-transport or in- Australia. Croom Helm London.
stitutional methods, which allow for the re-arrangement Dalvi M. Q. and Martin K. M. (1976) The measurement of
of institutionally determined travel patterns. accessibility: some preliminary results. Transpol'tation 5( I),
Unfortunately, whilst the data requirements for the 17-42.
activity linkage approach are reasonably clear (Oix 1975), Daor E. (1977) The eslimallon o//ravel demand. Greater London
Council. Department of Planning and Transportation,
the type of models required is far less so. A' startill8 Rescorch Memorandum No. 490.
point is provided by consumer choice models of activity Davidson K. B. (1973) Relallonshlps betwttn land-use and ac-
demand which address the problem of the value to be eelSlbUily. Greater London Council. Department of Planning
placed on travel time savings (Becker 1965, De Serpa and Transportation. Research Memorandum No. 378.
K. B. (1m) Accessibility in transportland:use model-
1971, de Donnea 1971). Heggie (1976) gives some of the Davidsonling and assessment. Environment and Planning A 9, 1401-
necessary conditions with which new travel demand 1416.
models would have to comply. However, we are still De Donnea. F. X. (1971) Consumer behaviour, transport mode
some way from satisfactory working models of activity choice and value of time: Some micro-economic models.
linkage in travel demand. It is therefore suggested that DeRegional and Urban Economics 1(4) 3SS-J82.
Serpa A. C. (1971) A theory of the economics of time. The
research proceed on two fronts: Economic Journal, December. 828-846.
First. attempting to improve our understanding of Dix M. C. (I97S) Applicalion 0/ In-Deplh Inleroiewing TtCh-
travel behaviour by undertaking simple descriptive niques 10 Ihe Sludy 01 Traut! BehalJiour: Some Preliminary
analyses of journey making and behaviour and, if the Results. University of Oxford. Transport Studies Unit. Work-
ing Paper No.9.
data can be obtained, of activity patterns pursued by Don P. A. (I97S) The new models I: trip generation. trip dis-
persons and households. tribution, mode choice. Revised Transpol'tatlon Models lor
108 J. M. MORRIS, P. L. DUMBLE and M. R. WIOAN

Melbourne. The Institute of Engineers, Australia. Victoria SpeclJic Services and Regulations in Sydney. Unpublished •
Division, Transportation and Hishways Branch, 4-1 to 4-38. Ph.D. thesis. Urban Research Unit, The Australian National
Donald O. C. (1977) Health Care Delivery in Sydney, Unpub- University, Canberra.
lished Ph.D. thesis, Urban Researcb Unit, The Austra\ian Mitc:heU C. O. B. and Town S. W. (1977) Accessibility o/llarious
National University, Canberra. social groups to different act;lIities. TRRL Supplementary
Ewins O. O. (1974) Gravity and linear regression models of Report 25g, Transport and Road Research Laboratory. Crow-
spatial interaction: a cautionary note. EcollOmic Geognzphy thome, U.K.
50,8~. Morrill R. L. (1963) The distribution of migration distances.
Flowerdew A. D. J. (1976) An evaluation packll8e for a strategic Papen 01 th, Re,ional Scltnce Alloc/otion II, 75-84.
landuse transportation planning program. In Urban Trans- Morrill R. L. and Earickson R. (1968) Hospital variation and
portation Planning Conference (Edited by P. Bonsall, Q. DaJvi patient travel distances. Inquiry 5, 1-9.
IUId P. J. Hills) 161-176. Abacus Press, United Kingdom. Morris J. M. (1976) Access to community health facilities in
Freestone R. (1975) On urban resource allocation: the dis- Melbourne. POPtr3 of tile Fint Australia-New ZuJand
tn"bution of medical practitioners in Sydney. Geography Bu/- Regional Science Con/ermce, Brisbane. August 1976. 143-173.
letin 7, 14-25. Morris J. M., Dumble, P. L. and Wigan M. R. (1978) Accessibility
Gamer B. J. (1976 The etrect of local government reform on access . indicators for transport planning. Forum Papen. 4th Aus-
to public services: a case study from Denmark, Bristol Essays in tralian Transport Researcb Forum: Perth. May 1978. Director
Geography (Edited by P. Haagelt, M. Chisholm and R. Peel) General of Transport's Office, Perth, W.A.
319-338 Heinemann, London. Neuberger H. (1971) User benellt in the evaluation of transport
Gould P. (1969) Spatial Diffusion. Association of American and land use plans. J. Tran,. Econ. Policy 5(5).52-75.
Geographers Resource Paper No. 17, Commission on College Nledercorn J. H. and Bechdolt B. V. Jr. (1969) An economic
Geography, Washington, D.C. derivation of the "Gravity Law" o( spatial interation. J.
Great Britain Department of the Environment (1977) A Sillily 0/ Regional Sci. 9(2). 273-282.
Some Means of Traffic Restraint. Department of Environment Oberg S. (1976) Method, 0/ Dtscribln, Physical Access to Supply
Research Report 14, Department or Environment, London. Points. Lund Studies in Geography Series B. Human Oeo-
Hansen W. G. (1959). How accessibility shapes land-use. J. Am. grapby No. 43, Department of Geography, The Royal Uni-
IlISt. Planners 15,73-76. versity of Lund, Sweden.
HlUlson S. (1978) Urban travel linkages. Proc. 3rd Inter. ConI. OECD Road Research Programme (1977) Transport Require-
Behallioural Travel Modelling, Tanuda, South Australia. Croom ments for Urbau Communities: Planning for Personal Travel.
Helm, London. F"ma1 Report of Research Group Til. Paris Appendix D, 84-92
Harris Lange Voorhees (1971) Ballarat Transportation Stlllly. (unpublished).
Consultants Report to the Study Commiltee-Chief Planning Pallon T. and Clark N. (1970) Towards an accessibility model for
Engineer, Country Roads Board-Chairman. residential development Analysis 0/ Urban Development
Heggie I. G. (1976) Consumer Cholct and Urban Travel Dtmands (Edited by N. Clark), 266-288. (proceedinss of the Tewksbury
A RtIIlsed Theory. University o( Oxford. Transport Studies Symposium) Special Report No.5, Transport Section. Uni-
Unit. Working Paper No.4. venity of Melbourne.
Henderson J. M. and Quandt R. E. (1971) Micro-Economic Popper R. J. and Hoel L. A. (1976) Mobility evaluation (or urban
Theory: A Mathematical Approach. 2nd Edn. McGraw-Hili, public transportation. Transportotlon Plannins and Tecll-
Kogakushll. noloBY 3, 131-141.
Hensher D. A. (1976). The structure of journeys and nature of Samuelson P. A. (1971) Foundotions 0/ Economic Analysls.
. travel patterns. Environment and Planning A 8, 6S~72. Harvard University Press, Cambridge.
Hillman M, Henderson I. and Whalley A. (1973) Penonal Mobil- Sherman 1.., Barber B. and Kondo W. (1974) Method for evalu-
Ity and Transport Po/icy. Political and Economic Planning, ating metropolitan accessibility. Transportation Reseoreh
London. Record 499, 70-82.
Hillman M.. Henderson I. and Whalley A. (1976) Transport Turner C. O. (1972) A model framework (or transportation and
Rtalities and Plonnin, Policy. Political and Economic Plan- commWlity plan analysis. J. Am. Inst. Plonners 38, 325-330.
ning. London. U. D. P. A. Planners (1975) Alta Servim: Melbourne. Report
Ingram D. R. (1971) The concept of accessibility: a search for an prepared for the Australian Cities Commission by Urban
operational form. Regionol Studies 5. 101-107. Design and PlanDing Associates. Melbourne.
Jones P. M. (1976) Travel as 0 manifestation of actillity choicel: Vickerman R. W. (1974) Accessibility. attraction and potential: a
trip generation re-interpreted. University of Oxford. Transport review of some concepts and their use in determining mobility.
Studies Unit Workins Paper No. 14. Oxford. Enlliranment ond Planning A 6. 67~1.
Jones P. M. (1978) New approaches to understanding travel Wachs M. (Reporting Chairman) (1978) Report on Plenary Ses-
behaviour: the human activity approach. Proc. 3rd Inter. Con/, sion. Workshop H: Behavioural ModeUing, Accessibility,
Behavioural Travel Modelling. Tanunda, South Australia. Mobility and Travel Need. Proc. 3rd Inter. Con/, BehoDio/UQ1
Croom Helm. London. Trallel Modelling, Tanuncla, South Australia. Croom Helm,
Kaiser E. J., Butler W. W., McAllister R. J. and Thibeault R. W. London.
(1972) Some problems with accessibility standards: a com- Wachs M. and Kumagai T. G. (1973) Physical accessibility as a
parison of household preferences to standards (or work. social indicator. Socio-Economk Plannillg ScL 7,437-456.
shopping and school trips. Rtlliew of Regional Studies 3, WeibuU J. W. (1976) An axiomatic approach to the measurement
111-123. of accessibility. Regionol Science ond Urban Economics 6,
King R. (1975) Urban senicel and distributive Justice: a Sydney 351-379.
cose study. Centre for Environmental Studies Working Paper. Whitbread M. (1972) Ellaluation In the planning process-the
University of Melbourne, Melbourne. cose of accessibility. Plannins Methodology Research Unit.
Koenig G. (1977) Les indicateurs d'accessibilites dans les studes University College London., Workins Paper 10.
urbaines: de la theorie a la prac:tique. Rtllue Generale des Wickstrom G. V. (1971) Defining balanced transportation-a
Routes. July 1977 (English translation available). question of opportunity. Traffic (Jlltulerly 15, 337-349.
McFadden D. (1973) Conditional loglt analysis of qualitative Wigan M. R. (1971) Benefit IUsellmlllt lor networlc traffic models
choice behaviour. (Edited by P. Zaremka) Frantien in ond applicotion to raod priclns. TRRL Report LR 417, Trans-
Econometrics, 105-142. Academic Press, New York. port and Road Research Laboratory. Crowthome. U.K.
Mcintosh P. T. and Quarmby D. A. (1972) Generalised costs and Wigan M. R:, Bamford T. J. O. Broughton J., Paulley J. and
the estimation of movement costs and benefits in trans- Lynam M. (1974) Two scenarios for traffic restraint: the traffic
portation planning. Highway Releoreh Record 383, 11-26. social and environmental elfeets. Working paper (unpublished:
Manning I. (1972) An Economic Study of Some Location- available from the lint author at ARRB on request as: Dis-
 Accessibility indicators for transport planning 109
• cussion Paper DN300). Enl1ironmenl and Planning A 9(3), 285-344.
Wigan M. R. and Richards E. (1978) The analYIi.r of journey Wilson A. G. (1971) A family of spatial interaction models, and
It",ctllrt: linkage between different stages of a journey. Aus- associated developments. Enrironment and Planning A3, 1-32.
• tralian Road Research Board Internal Report AIR 300-1. Wilson A. G. (1972) Developing issues in urban transport plan-
Wilbur Smith and Associates and Len T. Frazier and Associates ning. Memorandum in Second Report from the Expenditure
(1969) Mtlbournt Transportation StlUly, Volume I: Su""y. Committee, HCS7, Session 1972-73. Urban Transport Plan-
Melbourne Transportation Committee, Melbourne. ning 2, (HMSO, London).
Williams H. C. W. L. (1977) On the formation of travel demand Zakaria T. (1974) Urban transportation accessibili!}, measures:
models and economic evaluation measures for user beneftt. modifications and uses. Traffic Quarttrly 28, 467-479•

·.