planologisch
studiecentrum TNO
research centre
for physical planning TNO
in Dutch geography and
urban and regional planning
in the seventies:
Geographical curriculla
and methods applied in
"Spatial Research",
Frans Dieletnan
Dree Op
't Veld'\ Jl
I
"'r '
urban and regional planning in the seventies:
Geographical curriculla and methods applied in
"Spatial Research",
Frans Dieleman* and Dree Op
'tVeld
nr ~
Bibliotheek TU Delft
IIIIIIIIII~
I
c:
1995966* Institute far Geagraphical Studies and Urban and Regianal Planning, Free University, Amsterdam.
Delft, April 1981/ Planologisch Studiecentrum TNO
Schoemakerstraat 97 - Postbus 45, 2600 AA Delft Tel. 015-569330
Both authors highly appreciated the help and advice of Emma Bijlsma and dr. Harry Timmermans as well as the comments of prof. dr. W. Heinemeyer on an earl ier draft.
A special thanks to Kitty van Geestand Brigitte Andoetoe for the careful preparation of the lengthy manuscript.
KEYWORDS
Quantitative Methods Human Geography Curriculla
Dutch Quantttative Spatial Research Regional Economies
Urban and Regional Planning Gravity and Entropy Models Correlation and Regression Mathematical Programming Logit /1ode 1 s
Urban and Regional Models Cluster Analysis Autocorrelation Multi-Dimensional Sealing MarKov Chains Categorieal Data Contingency Analysis
l. I NTRODUCTI ON 1 2. QUANTITATIVE TECHNIQUES AND DUTCH GEOGRAPHY 2 3. QUANTITATIVE METHODS IN HUMAN GEOGRAPHY CURRICULA 9 4. QUANTITATIVE METHODS IN DUTCH "SPATIAL RESEARCH": 14
MAIN ACCENTS AND PREVAllING TRENDS DURING THE SEVENTIES.
- DATA COLLECTION 14
- THE ANNUAL NUt~BER OF PUBLI CAT! ONS 16
- MAIN ACCENTS 19
- PREVAllING TRENDS IN THE SEVENTIES 23
5. CONCLUSIONS AND SPECULATIDtlS 26
6. REFERENCES 28
APPENDIX
1. Introduction
In this paper an attempt will be made to give some, sometimes fairly personal, impressions about the development and use of quantitative methods in Dutch geography and urban and regional planning during the seventies. The paper consists of three main parts.
In order to pave the way we will first of all give a number of impressions about the development of human geography in general in Holland during the last ten years.
Next we will describe the gradual development of the training in quantitative methods and research methodology at the five departments of human geography during the seventies. Only limited attention will be paid to the curriculla in related disciplines like regional econo-mics and urban and rural sociology. Studying the curriculla is of course just one possibility to analyse what sort of techniques we re considered to be most relevant for human geography and to arrive at a picture of the changes in emphasis on different topics that occured in the last ten years.
Consequently the third part of the paper intends to throw some light on the use of quantitative methods in -what we called
arbitrarily- "spatial research" or "spatial sciences" in the Nether-lands during the seventies. Unlike the discussion of the developments regarding training facilities, the treatment of this topic is not restricted to the contributions of human geographers only. The main reason to extend our attention to other disciplines is that in applied spatial research, geography does not function independently, but is part of a complex whole of interlinked disciplines showing considerable overlap. Disciplines like regional economics, transportation research, rural and urban sociology and physical planning make contributions to spatial research too, especially where quantitative methods and mathematical model building are concerned. t·10reover this broadening
of scope offers the opportunity of an interesting comparison of
curriculla in human geography and spatial research activities in genera 1 .
-2-2. Quantitative Techniques and Dutch Geography
The introduction of advanced quantitative techniques and social science
researchmethodology into Dutch geogra[Jhy dates back to roughly speaking 1970, as will be shown later on. In the Anglo-American geography the new emphasis on quantitative methods was just one aspect of more general [Jaradigm changes and discussions as described by Johnston (1979). For Dutch human geography this is far less true. It is easy to find Dutch geographers in the late sixties who specify regional geogra[Jhy as the ultimate aim of geography and view geography as an essentially idio-graphic science. Keuning (1969) for example states that:
'Nomothetic thinking ... has attracted many supporters in recent years, partially due to the influence of the strong a[Jplication of mathematical and statistical-mathematical methods in ... for example economics and sociology. However ... it seems to me that human geography ... in its [Jursuit of the individual aspect and the unique aspect ... must nevertheless be viewed as a branch of science which is by definition idiogra[Jhic.' (p. 135).
It can be safely assumed that views like this were [Jrevalent among a significant porti on of Dutch geographers around that time. Nevertheless it would be incorrect to conclude that the introduction of quantitative methods around 1970 is evidence of areaction against a idiographic human geography in the Netherlands. Heinemeyer (1977) [Joints out correctly that the involvement of Dutch geography with a type of
re-search along the more general guidelines bf social science research methodo 1 ogy i s of a much longer standi ng. r~uch of the work long before 1970 was of a more thematical than regional nature, as for example in urban and rural geography. This is substantiated"in publications by Heslinga (1974, 1980). Even Keuning whose explicit view on the nature of human geography we mentioned above was partly involved in research of a much more nomothetic character, like the hierarchy of urban centres in the Netherlands (Hei nemeyer , 1977). Moreover, long before 1980 Dutch geography has established close contacts with urban and regional plan-ning.
-3-A considerable number of human geographers was working in "economisch-technische" and "sociografische" institutions in the Netherlands and received a research training oriented towards these jobs.
The fact remains however that Dutch geography was fairly unaffected for a long time by the turbulent developments in the field in the United States. This is certainly strange because in the international journals these developments were clearly evident, and Dutch geographers had always been largely inspired by ideas from abroad. If Dutch geo-graphers drew any inspiration from work in the United States in the early sixties it was more from the human ecologists and the work of for example Stuart Chapin than from "spatial geography" or "quantita-ti ve geography" (Hei nemeyer, 1977).
Dnly at the end of the sixties a number of staff members was appointed at various university departments of geography, who were given the task of mastering the quantitative methods and the positivist oriented scien-tific methodology. Dur impression is that these appointments met gene-ral consent. Even geographers who were not attracted at all by a quan-titative type of approach to geographical problems tended to feel that at least some Dutch geographers should pay more attention to these matters. Anyway it means one of the steps in a rapid orientation of Dutch geography in the seventies towards Anglo-American developments in the discipline and Anglo-American literature. Dne of the manifes-tations of this re-orientation was a seminar on "Mathematical methods in human geography" in 1971. Hauer (1971) discussed the "new geography", the "new orientation" in the field and advocated abandoning regional geography as the ultimate aim of the discipline. The pursuit of syste-matic priciples, the application of general social science methods of investigation and the use of quantitative techniques were recommanded. At the seminar, a great amount of attention was devo~ed to factor ana-lysis. In Dutch geography this technique was still quite new. It shows how limited at that time the knowledge about quantitative methods still was. It is also indicative of the great interest in especially
quanti-tative methods and less in research methodology that arose in Dutch geography in the seventies, a point which will be illustrated several
-4-more times in the following paragraphs.
It is difficult to answer the question why little attention was paid for a fairly long time to the paradigm changes in Anglo-American geography and the introduction of quantitative techniques that accom-panied these changes. There is no written discussion in Dutch geographic literature on this theme. So one can rely on personal impressions only. Perhaps the following issues played a role:
- until the seventies the influence of French geography, where the regional monograph was quite important, remained significant. Lec-tures on the nature of human geography for example often included a review of the opinions of well-known German and French geographers about concepts and methods which were especially meaningful for regional geography;
- many geographers had a humanities background, and their interest was fairly history oriented. Understandably, they had very little desire to start working with unfamiliar quantitative methods and to devote their attention to general scientific research methodo-logy;
- the computer was introduced in the Netherlands at a relatively late date: it was not until the seventies that the universities had easy access to it. Probably this is one of the reasons why the use of quantitative methods -which af ter all was an important element in the paradigm changes in Anglo-American geography (see for example John-ston, 1979)- did not develop rapidly;
- changes in the way a field of science is practiced are partly a question of generation differences, the new generation replacing the old, as was also stated by Taylor (1976) and Johnston (1979).
Heslinga (1980) also, pointed out the importance of the personnel history of the discipline for a good understanding of the development of Dutch geo-graphy. Until the seventies, Dutch professors were in a position to shape the academie training in human geography to fit their own preferences. This changed in the seventies, when the staffs increased in size and the staff units ('vakgroepen') began to take over many of the professors' responsibili-ties. Moreover, a generation of professors retired in the early seventies, and the arrival of a new generation was one of the factors leading to a
)
-s i
li-different approach to the field of geography.
One of the most striking aspects of Dutch human geography in the period from 1970 to 1980 is the great expansion of the university staffs and the increase in the number of students, making this expansion possible. Heslinga (1980) spoke of a 'miraculous multiplication'. Up until 1950, there were only two universities where students could major in human geography, only a few professors and an extremely small number of staff members. In 1950, 1959 and 1961, at three other universities an insti-tute of human geography was established. Between 1959 and 1979, the num-ber of geography professors and lecturers rose form four to twenty (Heslinga, 1980), and in the past decade the number of staff members tripled or quadrupled. At each of the five universities, the human geo-graphy departments now employ between forty and sixty staff members. Without a doubt, this spectacular growth of the field of human geogra-phy at the universities can be viewed as the driving force behind chan-ges in the actual practice of the discipline and in the curriculum for students. Regional geography and historical geography, with a great amount of interest in the study of the landscape, continued to contri-bute significantly to 'the activities in the discipline. In the seven-ties though - stimulated by the increase in the number of staff members - other specializations like urban and rural geography, economic geo-graphy, geography of underdevelopment, educational geography and urban and regional planning expanded more rapidly.
With the growth of these specializations within Dutch human geography, there was also a clear shift in orientation to the Anglo-American
lite-rature. The Anglo-American approach to geography was incorporated into the Dutch counterpart almost automatically and certainly without any fuss. Students were amply confronted with quantitative methods and empirical-analytical research methodology. In geographic research and education the thematic orientation became more dominant and the impor-tance of regional studies declined. The stages which Johnston (1979) distinguished in the development of Anglo-American geography can, to a certain extent, also be found in Dutch geography in the seventies. For example, in the early seventies attention was devoted to geostatistical methods (Ter Hart, 1968), social physics (Hakkenberg, 1969), factor ana-lysis (Dieleman, 1971; Van der Knaap, 1971) and the rank-size rule
-6-(Deurloo, 1972). Within less than a decade, similar to recent trends in
Anglo-American geography, there were already the first reactions to the
partially empirical-analytical - if not positivist - orientation in Dutch
geography. Kouwenhoven (1979) felt drawn towards such geographers as
Buttimer, Yi-Fu Tuan, Entrikin etc. and chose to follow the lines of
'humanist' geography. Jansen (1980) preferred a phenomenological approach
in the spatial sciences and expressed his opposition to geography with
a logical-positivist tint. 50 within a decade, a re-orientation took
place in Dutch geography which bore many similarities to the changes in
the Anglo-American counterpart in the last th ree decades. The great
ma-jority of human geographers at the Dutch universities now feel closely
involved with the profession as practiced in America and England. More-over, the younger generation of human geographers - and there are a
great many of them - received an academie education with a clear
Anglo-American accent. Nevertheless, we still have the impression that
gene-rally speaking, Dutch human geographers are less intrigued by the use
of quantitative methods, empirical-analytical research methodology and
model-building than their Anglo-American col leagues are or were.
There is a number of reasons for this:
- in urban and economie geography the orientation in the seventies was
mainly towards the Anglo-American practices and the use of
quantita-tive methods expanded rapidly in these fields. In historical
geogra-phy, the study of landscape, the geography of underdevelopment and
educational geography - all of which play an important role in the
Netherlands - this was much less the case. For example, some
geogra-phers continued to devote their attention to regional studies and a
large number of books was written about a wide range of countries. In
the field of historical (urban) geography, Deurloo and Hoekveld (1979)
are an exception to the rule regarding the use of quantitative methods.
Consequently, when we discuss the use of quantitative methods in
spa-tial research in the following sections, we can very well as far as
human geography is concerned, restriet our attention to research in
the Western world, primarily in the fields of urban-and-rural and
economie geography;
in the past decade, Dutch human geography, particularly urban and
urban and regional planning. Many geographers are employed in the field of urban and regional planning. The research done by urban and rural geographers and economic geographers was and is largely centred around problems which are viewed to be significant within the framework of urban and regional planning in the Netherlands. This will probably be even more so in the future, once the new way of financing university research - more dependent on non-university financial sources - be-comes more wide-spread. This is why research into the above-mentioned specializations of geography is rather practically oriented: quantita-tive methods and mathematical models are used and developed if consi-dered necessary, but the majority of the work remains problem orien-ted. Extensive experiments with quantitative methods or extremely theoretical investigations are fairly rare and mainly practiced by staff members appointed to teach quantitative methods;
the last point is again related to the increase in staff concomitant with the expansion of Dutch human geography in the past ten years. Although many of the recently appointed younger staff members have had an Anglo-American oriented education, this was hardly the case with the professors and lecturers appointed during this period. When
the increase in the number of chairs for human geography gave the uni-versities the opportunity to appoint new professors and lecturers, the generation of human geographers who had received a training in the use of quantitative methods was still to young to fill these chairs. Con-sequently, in the large majority of cases, professors and lecturers were appointed with hardly any inclination to apply and develop tech-niques. In spite of the fact that most of them are familiar with the Anglo-American literature, they have less affinity with quantitative techniques and mOdel-building. This is one of the reasons why, in Dutch human geography, there is still a great diversity of opinions regarding the actual practice of the profession. We have the impres-sion that writing a book about a country or a region has remained just as acceptable as conducting research with essentially theoreticalor quantitative aspects, although Kouwenhoven (1979) and Jansen (1980) seem to have a different opinion on this matter. Hopefully the rela-tive tolerance and mutual respect that exists now will not disappear in the debate on the conduct of geographic research that is developing
at the moment.
Just as has been the case with the other specializations of geography, the field of research methods and techniques in human geography has pro-fited from the increased interest in the field and also from the more pronounced Anglo-American orientatiQn of Dutch geography. At a rapid pace and, in our view, with wide-spread approval, the various universi-ty departments of geography have introduced courses in statistics, quan-titative methods and research methodology. In the past decade, there ha~
been a great increase in the number of university staff members who de-vote a large part of their time to quantitative methods, which has led in turn to an increased amount of knowledge on this field. At the de-partments, th ree to five staff members are in a position to teach and do research involving quantitative methods. A number of chairs of geo-graphy has been reserved for this specialization. Added to the staff members who are specialized inphilosophy of science and the history of the profession, this means that at the five human geography departments alone, there is a considerable manpower engaged with the technical, metho-dological and theoretical aspects of human geography.
If one analyses the use of quantitative methods in spatial research in general - as we shall attempt to do later in this paper - one sees that in addition to the work of human geographers, the contribution of rela-ted specializations is considerable. In this respect, special reference must be made to regional economists (particularly with regard to mathe-matical model-building\ the sUb-departments of the Universities of Technology in Eindhoven and Delft, the Research Centre for Physical Planning (PSC)-TNO in Delft and sociologists such as Van Doorn and Van Vught (1978) and Van de Vall (1980).
3. Quantitative methods in human geography curriculla.
If we take the curriculla at the five departments of human geography at the Dutch universities to be the measure of the geographers' knowledge on quantitative methods then, as might be expected, the 1970 knowledge level was still quite low. Table 1 shows when and for which groups of students various quantitative methods were introduced into the curriculla at the five universities. The dates when certain branches of statistics or the philosophy of science were introduced into the curriculla do not give a fully accurate picture of the gradual manner in which the various subjects were usually introduced. Usually, there was a development simi-lar to the one experienced at the Free University. At a given department, a number of staff members get the feeling that !something has to be done' about teaching a certain subject. Since the necessary expert knowledge is not at hand, no lectures are given yet about the subject, but a suitable-looking book is selected to be read by the students, though no questions are asked about it yet at the examinations (Gregory, 'Statis-tical Methods and the Geographer', was sometimes used this way). This situation gradually changes because a staff member acquires the neces-sary knowledge or because a new staff member possessing this knowledge is appointed. The lectures and examinations tend to be more serious and to occupy a place of their own and slowly but surely become an integra-ted facet of the curriculum. Discussions about this with staff members from departments at other universities confirmed this picture of a gra-dual ir.troduction. Consequently, the years given in table 1 must be seen in this light, and not as all toD absolute dates.
Table 1 makes clear that prior to 1965 there were no serious courses in statistics at the geography departments of the Dutch universities. At most of the departments, basis statistics (column 1) were not intro-duced as a separate course until 1970. It is no wonder, in view of the recent start of courses in statistics being given to large groups of geography students in an early stage of their studies, th at the intro-duction of more advanced quantitative methods (column 2) has been even more recent. Moreover, in the initial stage, these courses are of ten given to a few of the graduate students and the subject is dealt with in a superficial manner: the use of the methods is illustrated by taking an example from the available literature, but the technical aspects are
TABLE 1. TllE INHODUCTlON QF STATlSTlCS. QUArmTATlVE METllODS Alm RESEARCH METHOOOlOr,y Hl THE CtJ~RICULLA OF TllE F1VE OUTCH UNIVERSITY OEPAKT"BITS OF HUMAN GEOGRAPHY IN THE EARLY sEvnmES.
1. ~~~~~n~~~-~~:i~~~~~~ons 2. regressien and correlation
analysis Factor analys;s, geo-s tati 5 ti cs, simulation. network analysis. Markov-Chains etc.
Has been included for a Ever since + 1970/1971. fac-~~n~r!~:t!Q ;~~d~~~;~ c~~~; ~~; ~~~~{~~T u~a}o~n~!~~:~
n
in
attended lectures on the sub- groups of graduate students.
jeet at the economics depart- An introductien to matrix ment. Given for the first algebra was provided for time at the human geography them by the department of department itself in 1974.1n econom;cs.
1974, statistics was a150 ;n
-cluded in the first and second year curri cul urn. Ever since + 1959/1960. Rijken van lHst (first volume) and Gregory have been s tudi ed by undergra-duate students. Not in
detail, however.
Since + 1966, extensive lectur1!s on statistics have been ineluded in the curriculum for undergraduate s tudents. Ever since 1971/1972
lectures have been given
on statistics to under-graduate students. Be-fore then Rijken Van
Olst was studied.
al-though no quest;ons were asked about 1 t on examinations. Since the fifties seminars have been held for fi rst year and second
î
year students wherere-< ~~:~f~t~;h~~l~~h and
~ taught. Ever s;nce 196B, there has been a
separa-te course in descriptive
statistics , and ever since 1970 there has a1-so been a course i n i
n-ferential statistics.
Ever since + 1968. they have been dëalt with
su-perficially for
postgra-duate students. Ever since 1969, lectures on
statistics have been given
to undergraduate students.
Since + 1970, lectures in most Of the above-mentioned techniques were included in the curri cul um for some of the th;rd year students. In
rather great detail.
Ever since 1971/1972. the use of these methods in geography has been dea 1 t wi th by a sma 11 qroup of graduate students. No separate lectures on the
subject.
Ever since + 1971, gra-duate studéiits have been in a position to become
familiar with them. No separate lectures.
They have been tauqht
superfieially to
anprox-imately half the graduate s tudents ever si nce + 1·968. and in IJreater
aetail ever since 19701 1971. 3. Use of the computer. Since + 1971 for a nudler-of graduate students. Since + 1970, a num-ber orgraduate stu-dents were given the opnortuni ty to become
familiar with it. From
1974 on. underqraduate students are offered
this oODortunity as
well.
4. Philosophy of sc1ence
Ever sinee + 1970. in addition to-a survey of the history of the qeo-granhy discinline. the ~hilosoohy of science
(De Groot is dealt with) has a1so been included in the curriculum for a group of graduate stu-dents.
5. Research metho-do10gy
Ever since + 1970.
attention has been
devoted to .th; 5 by a group of graduate students soeciali-zin~ in research. In the third study
year, methodology is a150 dealt witl1
in a oractical cour-se involved with
field work. Ever sinee + 191i5, some Ever since + 1966,
attention his been devo- methodologynas
ted to it, although not been dealt with in
in great detail. Sinee the framework of + 1972, it has been the thesis written
aealt with in the first in the finat year. study year. Ever sioee 1970.
undergraduate 5 tu-dents have a1so devoted an increa-sing amount of
at-tention to it.
From 1972 on, a smalt Sinee 1970/1971, atten- Ever since 1970/71 undergraduate as well as graduate students have de-voted a great deal
of attention to it.
qrouo of 9raduate stu- tion was devoted to it dents was briefly intro- starting in the first duced to it. study year, and all the
students devoted a great amount of atten-tion to it throughout
their training.
In 1970, lectures were introduced for graduate students.
Jn 1974 it was introdu-eed for aonrox; ma te Iy ha 1f the '}radua te s tu-dents.
Ever since + lQ6fl, it has been dei1 t wi th on the reading list for underqraduate s tudents. In lQ6fi, seeond year students were I'lOre or
less confronted with i t by means of lectures
on "quanti tati ve me-thods" .
Ever since + 1966, graduate stüdents have had to study a number
of books about i t. In
+ 1972/1973 it was
in-trodueed to the fi rs t year curriculum.
For a long time, a large amount of attention has been devoted to these methods by undergraduate as wel1 as graduate students. Ever since + 1968 extensive atten-tion has been de-voted to this topic in the framework of the third year
re-search training progr!!1'fIIE!.
ti. The tenns undergraduate student and graduate student are used here for the Outch "orekandidaatsstudent" and "doctoraalstudent". These are only the rouqh
equivalents of the words. slnce the dHferences between the Anglo-American From:Dieleman, 1975 uni vers i ty sys tems and the Dutch one, where the "kandi daa ts" taken mi dway i snot
not discussed in detail. For the large majority of Dutch geography stu-dents, the computer (column 3) remained a very vague concept up until the mid-seventies. It is still not something they can utilize to its full extent and use independently.
In Columns 4 and 5 of table 1, facts are also given with respect to the introduction of the philosophy of science and social science research methodology into the human geography curriculla. The introduction of these
subjects was an even more gradual process than in the case of statistics
and quantitative methods. It is obvious that for a long time, courses had been given in the methodology of human geography, and students had been taught to conduct research while writing their theses. However, as is understandable in view of the interest in regional description, these parts of the curriculum partly consisted of dealing with the concepts
and methodology judged as being fitted and proper for regional descrip-tion. Consequently, the philosophy of science was initially not touched upon in these lectures. Nor were themes dealt with such as the empirical
cycle, the operationalization of concepts and the testing of hypotheses.
Gradually, however, these topics were either added to already existing parts of the curriculum, or certain parts were adapted in such a way as to fit into the basic lines of general research methodology. This is why the dates given in columns 4 and 5 of table 1 are only a rough esti-mation of when this development took place. However, with all the due reservations concerning these dates, it can be noted that no real atten-tion of any importance was devoted to social science research methodolo-gy prior to 1970 (with the exception of the University of Amsterdam). Roughly speaking, the teaching of research methodology and quantitative methods at the five human geography departments was in a preliminary stage up until 1975. A number of staff members, of ten a mathematician in conjunction with a geographer, became familiar with the basic material and tried to find a way to integrate it into the human geography curri-culum. Ever since approximately 1975, lectures on the basic principles of descriptive and inferential statistics and research methodology have been fixed and integrated parts of the curriculla of all the five de-partments, as is clear from an inventory drawn up by the 'Working-group of Mathematical Geography' in 1978. Af ter 1975, basic statistics and
-12-the fundamental principles of -12-the quantitative methods came to demand less attention, so that it became possible to devote more attention to the improvement of graduate courses. Consequently, af ter 1975 a wider range of subjects could be discussed and dealt with in greater depth. Partially this was also due to the fact that research had provided a greater amount of experience with the subjects involved (see the follo-wing section).
If we compare the 1978 graduate curriculla of the five human geography departments, then considerable differences in emphasis can be noted (see also Dieleman
&
Hauer, 1980). In the first place, these differen-ces in emphasis pertain to the various specializations which graduate students can choose from. We have the impression that it is mainly the group of the students majoring in the geography of Western industriali-zed countries, economic geography, urban geography etc. - many of whom are to be employed in the field of urban and regional planning in the Netherlands - who are intensively confronted with quantitative analysis. This seems to be much less the case with students majoring in historical geography, political geography, educational geography and geography of underdevelopment. At Nijmegen, Utrecht and Amsterdam (Free University), special chairs in quantitative methods and research methodology make it possible for graduate students to major or minor in these subjects.A second difference in emphasis pertains to whether it is research metho-dology or the techniques which occupy a more central position in the department curriculla. In Nijmegen (at least in 1978), research methodo-logy seems to dominate; a relatively small amount of attention is devo-ted to the quantitative methods, whereas basic analytical tools like correlation analysis, regression analysis and factor analysis occupy a large part of the curriculum. At departments where the techniques are more strongly emphasized than research methodology, the above-mentioned quantitative methods also recieve a large amount of attention. In addi-tion, however, input-output modelling, gravity models, entropy models, Markov models, linear programming, network analysis and so forth also occupy a more or less important position. It is almost self-evident that as a result of this development of the graduate curriculla af ter
~ 1975, the use of the computer has also been introduced; many graduate students learn to work independently with standard programmes such as
S.P.S.S. So by the end of the seventies all the human geography depart-ments were offering a diversified training programme on quantitative methods and research methodology. Nevertheless, a number of shortcomings can be noted. For example, cluster analysis received very little atten-tion in the curriculla, although this technique recently spread like wildfire into the various fields of science and is also quite widely used in human geography research. Multi-dimensional scaling and methods of discrete multivariate analysis (see e.g. Everitt, 1977) were also hardly discussed, if at all, around 1978. Whereas these are precisely the techniques that are suitable for processing low measurement level data. If one compares the human geography curriculla with those of re-lated sciences, one sees th at the latter devoted a greater amount of attention to these subjects.
To a certain degree, there was also a dearth of attention devoted to
spatial interaction models, allocation models, multicriteria-analysis and so forth. In spatial research in the Netherlands these methods play an important role, as will become clear in the following section. They are only very briefly dealt with in human geography curriculla. Human geographers hardly seem attracted to these techniques. Understandably, the curriculla for students majoring in physical planning do devote more attention to these topics, but it is especially the regional eco-nomists who grant them an important place in the educational program. Other methods which are more or less exte~sively dealt with in the human geography curriculum, however, are largely lacking in the regional eco-nomics educational program. If one compares the curriculla of the diffe-rent fields, one gets the impression that up .to a certain extent each field utilizes and emphasizes its own techniques, with very little atten-tion devoted to developments outside the field. It can be assumed, how-ever, that since 1978 more attention is devoted gradually to some of the methods just mentioned.
-14-4. Quantitative methods in Dutch 'spatial research': main accents and pre-vailing trends during the seventies.
Data collection.
In order to acquire an impression of the main accents and trends in the application and development of quantitative methods in Dutch 'spatial research' in the seventies some quantification is cal led for. First of all a yardstick to measure scientific activity is needed. Obviously, an attractive, operational measure concerns the number of publications dea-ling with quantitative methods. As an operationalization of scientific activity, the number of publications clearly has some shortcomings. For instance, it does not take into account the sometimes considerable dif-ferences in scientific scope and quality that exist between publications. The difference between a fundamental handbook and a three page article is neglected. Merely counting the number of publications makes it impos-sible to discriminate between contributions which are the product of a profound analysis and highly relevant for the development of science on the one hand and scientific work of a more superficial nature on the other. Nevertheless, the number of publications was used here, because of its indicative value and ease of measurement. Secondly, it is neces-sary to define the set of titles relevant as exponents of quantitative 'spatial science'. Here, 'spatial science' is thought of as the diver-sified subgroup of social sciences and related applied disciolines, characterized by an essentially locational and, or regional approach
and interest in their research. like human geogra~hy, regional economics
(including spatial econometrics) , urban and rural sociology, trans por-tation sciences, and physical planning. Obviously, this is not exactly a strict and unambiguous scientific definition of the subject and nothing else than a try to write down one part of the global idea that has func-tioned as the main guide to distinguish relevant from irrelevant titles. The other part concerns the 'treatment on an advanced quantitative level' that was used as a 'yardstick' to select within the field of 'spatial science' the subgroup of publications dealing with quantitative methods. The global character of this guidance inevitably induces some subjecti-vity in the selection of relevant titles. But, although it proved to be impossible to ask a number of colleoues how they felt about the (over) completeness of the collected set, we are fairly confident that it meets
the requirements of representativity and reliability that apply here ..
Actually some 457 different titles published throughout the period from 1971 on to 1980 (approximately May) were collected. To arrive at an idea
of the favoured methods, the main research themes in quantitative research, and trends during the period, the collected 457 titles were classified according to three aspects:
- a methodical, - a thematical, and - the time dimension.
The categories of the methodical dimension were developed starting from the basic and generally applicable methods as correlation and regression, analysis of varianee, contingency analysis, factorial analysis, cluster analysis; added to this set were other more specific techniques frequent-ly used to anafrequent-lyse past observations and simulate possible future
condi-tions in 'spatial science', like trend-surface analysis, Markov chains,
Monte Carlo simulation, gravity and entropy modelling, logit-models,
in-terregional economie modelling, etc.
Apart fr om techniques .used for the analysis or anticipation regarding
empirical phenomena, it proved necessary to introduce some specific
decision-aids used in economie and physical planning to provide the
decision-maker with useful information to solve complex problems. The
techniques concerned are cost-benefit-analysis, mul ti-criteria analysis and to some extent mathematical programming. This resulted in the use of 32 specific quantitative methods, that together with a 'textbook' category make up the methodical dimension used forclassificatory purpo-ses (see table 3). For the sake of briev~ty the characteristics of and differences between the various methods are not discussed in detail. The
reader is referred to the abundance of textbooks available on quantita-tive methods.
The set of research topics that make up the thematical dimension was
arbitrarily established by combining the division of activities into
the categories of home, work and services, with the th ree organizational concepts of structure, interaction and system.
Together with three other categories, 'textbook', 'strictly methodical'
-16-in table 3. Obviously the selection of the categories used for the
thema-tical dimension contains a strong element of subjectivity. This applies even more to the ultimate classification of the collected titles accor-ding to the thematical and methodical dimension. We trust though that, given its conscientious preparation, the presented classification of ti-tles is sufficient for the general conclusions concerning main accents and trends drawn in this article.
To obtain suitable results the following practical rules were adopted: - usually each publication was characterized by no more than one
themati-cal category. On the other hand if different techniques were used or dealt with in a publication it was assigned to the respective methodi-cal categories.
- in case of a reader or a colloquium-report each of the contributions was treated as a separate publication.
- identical publications available at different places were introduced in the classifications as only one relevant title. This rule was not applied though in case of identical publications appearing in diffe-rent languages, usually Dutch and English.
Only a selection of the collected and classified publications is inclu-ded in this publication in the list of references. The full list of pu-blications can be found in the appendix.
The annual number of publications.
Turning to the results of the process of collection and classification of titles on 'quantitative spatial research' it is hardly surprising one has to conclude that the number of quantitative studies in Dutch spatial science shows a considerable increase throughout the seventies. (see table 2). The pace of the increase though, is somewhat exceptional. The number of quantitative studies published in 1979 amounts to six times the number applying to 1973, which is not even the year with the smallest number of publications. Besides the important catalyzing influence exer-ted by the 'Vervoersplanologisch Colloquium' (Transportation Planning Research Colloquium) organized annually by a group of active researchers ever since 1974, this swift increase relates to the expansion of a num-ber of existing scientific institutions in terms of manpower and activi-ties as well as to the start of a few new research and training centres.
Table 2 1971 1972 1973 1974 1975 1976 1977 1978 1979
THE ANNUAL NUMBER OF PUBLICATIONS DEALING WITH QUANTI-TATIVE METHODS IN DUTCH SPATlAL SCIENCE IN THE PERIOD
1971 - 1980. number of publications 1980 (until May) 8 25 16 50 62 63 62 52 97 22
In this respect one can think of the swift development of regional eco-nomics as an independent scientific tradition at the Universities of Gro-ningen and Rotterdam and at the Free University in Amsterdam, and the rapid expansion of the spatial research activities at the Netherlands Economie Institute in Rotterdam. Each of the respective institutes esta-blished its own series of Research Memoranda, cal led 'Foundations of Empirical Economie Research' in the case of the NEl. This obviously faci-tated the publication of research results greatly. Clearly, regional eco-nomists at the three universities and the NEl contributed considerably to the rapid increase in the number of quantitative spatial studies pu-blished during the seventies.
The development of groups of staff members specialized in quantitative methods and research methodology, sometimes causing the creation of an independent section on quantitative methods, at the five Dutch geography departments, is another factor behind the rapid growth of quantitative spatial studies. The contribution of academie geography to the number of quantitative studies is relatively small as compared to the abundant pro-duction in regional economics. Partly this reflects a different attitude towards the publication of research results between both groups of scien-tists.
In addition to the activities in regional economics and human geography developments in the field of urban and regional planning were important
---,----~-
-TABlE 3. THE NUMBER OF TIMES QUANTITATIVE METHODS HAVE BEEN DEALT I~ITH IN OUTCH SPATIAl SCIENCE IN THE PERIOD 1971-1980, ClASSIFIED ACCORDING TO TYPE OF METHOO ANO THEMATICAL CATEGORY.
1971 UNTIl 1980
Textbook. ( 0)
Analysis of variance ( 1)
Contingency analysis ( 2)
Correlation and regression ( 3) Di scri ~i nant ana lys i s ( 4)
Factorial analysis ( 5) Cluster analysis ( 6) Multi-dimensional scaling ( 7) Geostatistical analysis ( 8) Graph-theory ( 9) Trend-surface analysis (l0)
Monte Carlo simulation (1I)
Mark.ov Chains ( 12)
Time Series Analysis (13)
Gravity and Entropy Modelling (14)
logit Modelling (l5)
Spatial Choice Models (16)
Spa ti a 1 Pre ference Mode 1 s (17)
Behavioural Models (l8)
Multiregional Economic Models Jt (19)
Location-Allocation Techniques JIt (20)
General Urban and Regional Models .(21)
Multi-region. Demogr. Accounting (22)
Mathematical Programming (23)
Mu1ti-Criteria Analysis (24)
Assignment Techniques (25)
Spatial Autocorrelation (26)
Shift- and -Share Analysis (27)
L inear Structural Models Jt (28)
Canonical Correlation (30)
Geocodi ng Methods (31)
Interdependency Analysis (32)
Cost-Benefit Analysis (33)
Total
(19) Input-output, attractlon. mu1tl-reg; ona 1 growth mode 1 s.
(20) (P-median) (21) (Wilson, Batty-tyne) (28) (Jöresko9's LISREl)
..
u.. ..
~ u..
~ ~ ~ ~ 0 ~ ~..
0 > ~ > c ~ u > ~ c u:;
~ « ~t
«:;
f
~..
u « ~..
~ u..
I
:; u ~ >..
~ ~ u c c ~..
~..
> ~ ~~
g ~ ~ ~ ~ ~ ~ ~ ~ ~..
c..
0 0 c ~ ~ 0 0 ~ ~ ~ 0..
J
~J
~ ~J
~ ~ u..
~ ~ 0 ~ ~ u..
0~
8 i u E ~ 0 ~ ~ ~ ~ 0 ~ u ~ ~ ~ ~ ~ ~ c 1> ~..
~ ~ ~ ';ö ~ ~ ~ 0 c ~ 0 ~ c c c ~€
~ c ';ö 0 ';ö 0 ~ ~ u ~ ~ ~ ~ ~ 0 -1l..
~ ~ u ~ ~ ~ ~ ~ ~ ~..
~..
~ ~ ~..
~ u ~ u ~ ~ ~ 3! ~ 3! !". 8! 8! 0 c c j! c ~ ~ ~ ~ ~ ~ ~ ~ z 12 4 15 2 2 2 1 , 79 F 4 1 1212 29 3 5 5 I 32 15 11 15 15 6 8 58 113 16 24 5.l 2 9 3. 2 1 1 25 32 II 21 66 1 126 46 12 2 1 16 10 3 4 25 49 23 11 10 4 44 ':0 9 .<1 ' 5 111 b" 59 69 614 c cfor the rapid growth of the number of quantitative spatial studies in the
seventies. At the Universities of Technology in Eindhoven and Delft and
at the University of Utrecht sub-departments concerned with research and education in urban and regional planning started and/or expanded rapidly in the seventies, paying considerable attention to quantitative methods; whereas in 1971 the Research Centre for Physical Planning (PSC) came into
existence as part of the Dutch National Organisation for Applied
Scienti-fic Research (TNO). The importance of the latter institution during the seventies is illustrated by the fact that the methodically oriented
perio-dical 'Planning, t1ethodiek en Toepassing', sponsored and co-edited by
PSC-TNO, has rapidly taken over the position of the 'Tijdschrift voor
Economische en Sociale Geografie' as the leading Dutch journal regarding
quantitative methods in spatial science. Main accents.
In table 3 the scientific production in Dutch quantitative spatial re-search in the seventies is summarized. It is obvious that according to
the number of publications gravity and entropy modelling, received by
far the most attention during that decade (113 cases). Other
quantita-tive methods showing high overall frequencies are successively general
correlation and regression (79), mathematical programming (66), logit
modelling (53), multi-criteria analysis (46), multi-regional economic
models (33), cluster analysis (32), general urban and regional models
(32), and factor analysis (29). The leading role of gravity and entropy
modelling is largely the product of two factors. Firstly, gravity and
entropy models can easily be applied to different types of spatial
in-teraction, ranging from migration to the movement of messages. Secondly,
the annual Transportation Planning Research Colloquium, due to its
ca-talyzing influence on the publication of research results in
transpor-tation modelling, accounts for a considerable number
of
publicationsdealing with gravity and entropy-models (for instance):
Le Clerq, 1977; Nijkamp, 1979; Ruygrok, Van Essen and Van der Eems, 1979).
It remains remarkable though, that general correlation and regression, and other quantitative methods applicable to a wider range of spatial
phenomena, like mathematical programming, cluster-analysis and for
-20-from transportation research, gravity or entropy type models were fre-quently used in relation with aggregate residential location processes
(Somermeyer, 1971 ; Drewe, 1972; Klaassen en Drewe, 1974; Drewe and Van der Zouwe, 1979) and the spatial choice of service facilities. The lat-ter group of studies usually concerns shopping studies (for example: Timmermans, 1979; Timmermans and Veldhuisen, 1979; Van der Linde, 1980). In a few cases gravity models were used on data regarding outdoor recre-ation (De Kievit, 1976).
Although logit models (see: Ruygrok and Wieleman, 1975; Verster, Van Leeuwen and De Langen, 1976; Le Clerq, 1979; Van Lierop and Nijkamp, 1979; Ruygrok, 1979; Van der Zwam et al, 1979) do not differ meaningful-ly from gravity and entropy models in terms of their applicability to various types of spatial interaction, they have hardly been discussed in connection with migration analysis and were never applied to model migration flows. This is even more remarkable because the relative fre-quencies of both methods apart from the 'residential location' catego-ry differ only slightly.
Another interesting observation concerns the fact that gravity and en-tropy models as well as logit models have hardly been used or discussed by academie geographers. Publications discussing gravity and entropy, and logit-modelling are usually written by traffic engineers, regional economists and a group of physical planners.
The figures relating to correlation and regression analysis also show an accent on spatial interaction, although they are also fairly of ten discussed in relation to the analysis of spatial structure.
Probably, not only due to their general applicability, but also because of their mathematically simple structure, they are frequently applied, not only in regional economics and transportation research but also in geography and urban and regional planning (for example: Deurloo, 1972; Van Duyn, 1975; Ruygrok and Wieleman, 1975; Bartels and Bertens, 1976; Bartels, 1977; Van der Knaap, 1978).
Because the complex decision problems in economie and physical planning - to whose solustion mathematical programming and multi-criteria analy-sis are used - can hardly be classified according to the thematical ca-tegories, the major part of the publications dealing with these methods is found in the category 'not classifiable'. The remaining
cell-frequen-cies are too low to allow for some reliable conclusions. Inspection of the main authors reveals th at regional economists contribute most to the development and use of both methods (Nijkamp, 1972; Mastenbroek and Pae-linck, 1976; Nijkamp and Rietveld, 1976; Van Delft and Nijkamp, 1977; Paelinck, 1979). Although, with respect to multi-criteria analysis a con-siderable number of studies is signed for by physical planners (Voogd, 1975, 1976, 1980).
As opposed to quantitative methods discussed earl ier, cluster and factor analysis show an accent on spatial structure. In itself this is hardly surprising. A second interesting difference as compared to such methods as gravity and entropy models, logit-modelling, mathematical programming and multi-criteria analysis, concerns the prominenee of geographers in the use of both methods (Dieleman, 1971; Van der Knaap, 1971; Deurloo, 1976; Dieleman, 1978; Van der Knaap, 1978; Masser and Scheurwater, 1978; Van Ginkel, 1979; Op 't Veld, 1979; Op 't Veld and Timmermans, 1980, Timmermans, 1980). With respect to cluster analysis this leading role
has to be shared with physical planners, but apparently part of them are geographers by education.
Multi-regional economie modelling on the other hand, is evidently a strict-ly regional economie affair (Van Wickeren, 1971; Van Leeuwen, Paelinck and Wagenaar, 1976; Ancot and Paelinck, 1980). In the construction of general urban and regional models physical planners occupy a dominant position (Van Est, 1979; Veldhuisen and Kapoen, 1979). In addition to their activities some studies of the Netherlands Economie Institute are also important (Klaassen, 1973; Beumer et al, 1977). Of course, the ma-jority of the studies in general urban and regional modelling applies to the analysis of spatial systems.
Looking at the less frequently appearing quantitative methods (between 10 and 16 cases) it is striking that problems of spatial autocorrela-tion have only received attenautocorrela-tion from regional economists (Bartels and Hordijk, 1977; Hordijk and Nijkamp, 1977; Brandsma and Ketellapper, 1979) whereas a contribution of Dutch geographers lacks so faro The same applies to multidimensional sealing, which apart from regional economists only received attention from physical planners.
(Nijkamp and Voogd, 1978; Van Setten and Voogd, 1979; Nijkamp and Voogd, 1980). Markov chains and shift-and-share analysis on the other hand
re-
-22-present quantitative methods receiving considerable attention from Dutch geographers (see for example: Van der Knaap, 1978; Dietvorst, 1979; Tim-mermans and Op 't Veld, 198~on Markov Chains; and Wever, 1972, Ooster-haven and Van Loon, 1979,on shift-and-share analysis). Assignment tech-niques have been applied of course by traffic engineers mainly (Edelman, 1975; Le Clerq, 1979).
In the case of quantitative methods showing marginal frequencies below 20, the figures for the various thematical categories are too low to arrive at reliable conclusions regarding relations between methods and specific topics. It is therefore impossible to say anything more conclu-sive about the large group of quantitative methods that have hardly ever been discussed throughout th~ whole decade, than that it is strange that most of them have not received more attention. This holds especially for the group of methods pre-eminently suited for the analysis of categori-cal data: analysis of varianee, contingency analysis and linear structu-ral equations. At first glance, one can think of three reasons why a large group of in itself promising quantitative methods have not been applied more:
the majority of the researchers in quantitative spatial science is not familiar with the respective method, because it was not part of the curriculum at the time they attended lectures;
- computer-programs involving the respective methodaredifficult to get; - the respective quantitative method is badly suited for the analysis of
spatial phenomena.
In our opinion the first reason is the most important. As far as geo-graphers are concerned it applies to such methods as for instance contin-gency analysis and time series analysis.
Inspection of the column-totals of table 3 demonstrates that spatial in-teraction received most quantitative attention throughout the seventies. In order to explain this, one might recall one of the factors behind the dominant position of gravity and entropy models: the influence of the
Transportation Planning Research Colloquium. Notwithstanding the increa-sing economie problems in the seventies, only very limited attention was paid to the location process of industry, offices and service activities, whereas the residential location process proved to be a recurrent topic in quantitative analysis.
Prevailing trends in the seventies.
To reveal the main trends regarding quantitative methods in Dutch spatial
science in the seventies we split up this decade into two five-year pe-riods: 1971-1975 and 1976-1980. Inspection of the respective figures gave rise to the conclusion that although in both periods gravity and entropy modelling and correlation and regression are the most popular quantita-tive methods, considerable differences exist between both periods (see table 4).
TABLE 4. THE TWELVE QUANTITATIVE METHODS SHOWING THE HIGHEST OVERALL
FRE-QUENCIES FOR 1976-1980 PERIOD AND THEIR RANK-ORDER FOR BOTH
PE-RIODS, 1976-1980 AND 1971-1975 RESPECTIVELY.
----------------
-Type of method 1976-1980 1971-1975
gravity- and entropy 1 (18%) 1 (19%)
mode 11 i ng correlation and 2 (ll%) 2 (16%) regression logit mode11ing 2 (11% ) 8 ( 4%) mathematical 4 ( 9%) 3 (15%) programming multi -criteri a 5 ( 9%) 6 ( 6%) analysis cluster-analysis 6 ( 6%) 9 ( 3%) factorial analysis 7 ( 5%) 7 ( 5%) multi-regional 8 ( 5%) 5 ( 7%) economie models
general urban and 9 ( 4%) 4 ( 8%)
regional modelling
multi-dimensional 10 ( 4%) 20 0%)
sealing
spatial autocorrelation 11 ( 3%) 12 ( 3%)
Markov Chains 12 ( 2%) 18 (0.0%)
During the seventies notably logit modelling, multi-criteria analysis,
-24-become increasingly popular, whereas the other quantitative methods be-come less important. The enormous increase in studies dealing with logit models and regarding all types of spatial interaction except migration processes, sterns from the wish to incorporate more theory on individual behaviour into the models. Disaggregate multi-nomial logit is considered to be theoretically more attractive than gravity and entropy models be-cause of its firmer underpinning in terms of individual behaviour. The rapid increase in the scientific attention devoted to multi-criteria analysis as an important instrument to solve multi-facetted decision problems in economie and physical planning is partly explained by the in-creased external pressure exerted on decision-makers and politicians, by for instanee the environmentalist movement, asking for a considered-judgement regarding complicated evaluation problems. Without the acti-vities of a few researchers who accepted the challenge to develop the methods asked for, however, multi-criteria analysis would not have re-ceived that much attention as it has now.
The growing attention payed to multi-dimensional sealing is largely re-lated to the swift expansion of multi-criteria analysis. Multi-dimensional sealing offers an opportunity to deal with data of different measurement levels raising ordinal sets to a cardinal level in acontrolled way -a situ-ation frequently occurring in ev-alu-ation studies.
Cluster-analysis evidently has become an important device in Dutch spa-tial research. Markov Chain analysis has become increasingly popular, mainly among geographers and physical planners, to analyse spatial in-teraction patterns for their stability and as a first step in regiona-lization procedures used on flow-data.
As opposed to other quantitative methods also showing the same marginal frequencies for both halves of the decade, the activities concerning ge-neral urban and regional models prove to be clustered around 1976 and gradually fade away towards the end of the decade. So at the start of the eighties the attention devoted to general urban and regional models is at the same low level again as it was at the start of the seventies. If one compares the attention devoted to the thematical dimensions for the 1971-1975 and 1976-1980 period, two meaningful differences can be
observed. The first one refers to the use of general correlation and re-gression. Correlation and regression is far more of ten applied to analyse residential location processes in the second than in the first half. Pre-sumably, this is the result of the increasing value attached to informa-tion regarding migrainforma-tion behaviour in physical planning, because of the increasing influence of migration processes on the future population dis-tribution and the lack of reliable forecasting methods.
The second remarkable difference concerns the incredible increase in the number of quantitative studies dealing with the spatial choice of service activities the second half of the decade shows as compared to the first. This rather sudden change in emphasis was largely caused by the fact that the making of a 'retail development plan' became obligatory within the framework of Dutch Physical Planning law. Consequently municipal and pro-vincial authorities spent a considerable amount of money on the research neccessary to prepare these plans. In addition to this direct influence on physical planning research, this obligation affected academic research indirectly by the resulting demand for know-how regarding the development and planning of shopping centres.
-26-5. Conclusions and speculations.
The preceding paragraphs have illustrated the rapid development of quan-titative analysis in Dutch geography and urban and regional planning in the seventies. The emphasis was very clearly on quantitative methods and the construction of mathematical models. As compared to related social sciences less attention was paid to general research methodology and its translation to specific problems in the discipline. Given the present opinions on this matter it seems reasonable to expect that especially human geographers working in the field of quantitative spatial analysis will turn more to questions of research methodology in the first part of the eighties than they did during the seventies.
Despite the fact that a considerable range of quantitative methods was incorporated in the curriculla, geographers, as far as spatial research was concerned, contributed mainly to the use of quantitative methods like factor analysis, cluster analysis and markov chain analysis. The development and use of mathematical models, within a more deductive re-search design, was left to regional economists and physical planners. We expect though th at in the next few years ties between regional econo-mists and human geographers on the research front will become closer again. The reason for this expectation is that a few geographers have started to work on mathematical model building, while some regional eco-nomists direct their attention more and more to the analysis of empiri-cal data of a low level of measurement and to the discussion of geogra-phical problems without the use of a mathematical framework.
With respect to the popularity of various quantitative techniques ap-plied in Dutch spatial research it will be interesting to see if the increasing interest in the analysis of individual behaviour and motiva-tion will result in a further change in emphasis than the one that can be noticed already in the last few years. If the shift towards the ana-lysis of individual behaviour will persist this might lead to a further relaxation of the emphasis on gravity and entropy models with an increa-sing importance of disaggregated logit-modelling. Quantitative methods suited for the analysis of individual behaviour and or the analysis of data on a low level of measurement, like contingency analysis, lineair structural equations, spatial choice and preference models, multi-dimensional scaling and multi-criteria analysis might become more
ropu-lar in the first part of the eighties.
The question whether the number of publications in spatial research, on or using quantitative methods will continue to grow, is more difficult to answer. Certainly the number of geographers and planners better trained in quantitative methods is still increasing and might be a sti-mulus in this direction. Apart from that the abundance of problems faced in regional economie, transportation and physical planning, to whose solution the use of quantitative methods and models might contribute, could exert a second positive influence. On the other hand a further stagnation of the economy could force the government to cut expenses on research and higher education still more. Another factor of uncertainty relates to the planned reorganization of the financial structure of re-search at the universities and the extent to which this may create shifts in research activities in the different disciplines in the next five years.
