0277-Y53hlX3~0502YY-I 5%03.

00/o
Pergamon Press Ltd

PLURALISM IN THE ORGANISATION OF HEALTH

SERVICES RESEARCH
WARREN KINSTON
Health Services Organisation Research Unit. Brunel Institute of Organisation & Social Studies,
Brunei University, Uxbridge. Middlesex, England

Abstract-The variety of methods used in health services research (HSR) embody categorically different
epistemological assumptions. These are examined in an effort to contribute to a usable framework for
the evaluation of HSR projects. and in the light of a need in the U.K. for adequate institutional
arrangements for the promotion and funding of HSR. Research into organisation has revealed that if
desired values are not explicitly built into the structure. other values may appear unbidden. The ad-
equate institutional base is one which embodies. espouses and funds a meaningful and practical scientific
pluralism. A possible classification is outlined and illustrated and its implications for competition
amongst scientists and the social responsibility of scientists are briefly discussed.

The U.K. has had difficulty in supporting a self-confi- ing and career structure, application of findings and
dent and productive health services research (HSR) so on [lo] can be tackled and an effective HSR sys-
community. This has been partly due to conflicts tem may emerge.
between professional/academic power groups and Of all these other issues, perhaps the most sensitive
partly due to poor communication between and testing of the HSR institution in its understand-
researchers and decision- and policy-makers [I]. If ing of the researcher’s task is its responsibility to
matters are to improve. an adequate institutional evaluate scientific work prior to funding and during
basis for HSR is essential. These were the conclusions or after completion of the research. A meaningful
of a recent one-day symposium on the future of HSR evaluation of research will be based on criteria just as
in the U.K. [Z]. any other evaluation is. If we define scientific research
The aim of this paper is to contribute to a practical, as a particular form of knowledge production [1 l]
that is to say usable and desirable. formulation of the criteria easily fall into two categories, epistemolo-
what any institutional arrangement must be like in gical (is this knowledge?) and methodological (is this a
order that it be adequate for HSR. It has been argued production process?).
that the responsibility for commissioning and long- This paper places the analysis of HSR cdmmission-
term maintenance of HSR should most appropriately ing in the U.K. context, argues for scientific pluralism,
lie with the Government’s health department (DHSS), and offers a possible categorisation of research sys-
with the established research councils for medical and tems. Finally, implications for adequate organisatio-
social science (MRC and SSRC respectively), with nal arrangements are briefly presented. The practical
some new HSR organisation. or with some combi- problems in evaluating the scientific merit and policy
nation of these [J-7]. A more recent suggestion is a relevance of research projects are touched on but not
special Health Authority within the NHS [8]. tackled.
The debate amongst researchers has been fuelled by
three recent social events: first, the recommendation HISTORICAL BACKGROUND .
by the Royal Commission into the National Health
Service [9] that an Institute for Health Services Ai] outhe

Research be set up: second alterations of the DHSS Taylor [12] provides a short lucid historical
in-house research organisation and return of bio- account of the funding of health-related research in
medical research funds to the MRC; third. a major the U.K. The more recent events are summarised
review of DHSS-funded HSR by the DHSS Chief here. In the 196Os, the Government began to back
Scientist. HSR with substantial funds. As HSR grew, the need
The location of a formally set up HSR institution, for an institutional substructure within the DHSS
that is to say an administrative and resource capa- became apparent, and this was developed in the 1970s
bility for commissioning HSR, is only controversial following the Rothschild Report [13]. The report rec-
insofar as there is doubt as to the existence of a ommended the formation of a large complex research
genuine commitment and ability to further HSR. commissioning organiiation headed by a Chief Scien-
From the researcher’s point of view. the fundamental tist within the Department of Health and Social
factor in the commissioning process is an understand- Security (DHSS), and a transfer of 25% of Medical
ing b! the institution of what the researcher is trying Research Council (MRC) funds to this body. The
to do. The commissioning institution must under- DHSS invited Professor Kogan of Brunel University
stand how he is working and why he chooses certain to study its Office of the Chief Scientist (OCS) and
problems and methods. If such an appreciation is monitor activities there. The uneven rise and fall of
lacking. then any institutional base exists in name OCS is described in their two reports [14, 15-J. In
onl!. If an appreciation exists. then the various other 1980. OCS decided to return the transferred funds to
issues. political freedom. funding methods, HSR train- the MRC because the arrangement had been little

199
300 WARREZ KIMTO\

more than a paper exercise. Although the MRC commissioning body becomes captive to scientitic
agreed. in return. to form an HSR board to be given assessors who abjure a pluralist position.
an additional f2 million. editorials in both the British
Medical Jowr~al [ 161 and the Lar~cet [17] questioned SCIEITIFIC PLLR.iLIS\l
the MRC’s capacity to understand the type of
research required. This doubt was exacerbated by the Theoretical debate and informal discussion in the
problems which surrounded the concurrent OCS last two decades appear to endorse the view that
quadrennial reviews of DHSS-funded HSR units, methods in the social sciences are not onI> distinct
which are referred to below. from those in the medical sciences but enormousi\
By 1980. the growth of HSR funding had faltered as varied in themselves [24]. “A research tradition’.
part of the national and world-wide economic reces- writes Galtung [25] “that dogmatically outrules one
sion. In addition government optimism in the use of or more (varieties of approach) only demonstrates its
research, especially social science research, had been own limitations”. HSR is a domain drawing pn a var-
dampened [18]. The effect on HSR and its research iety of fields including clinical medicine. psycholog).
community was serious because the DHSS is essen- sociology. economics. politics. management science.
tially a monopoly promoter and funder of HSR, and, information science. geography. demography. epide-
unlike the MRC. does not tenure its researchers. And, miology, anthropology and others. These different
because HSR is applied and interdisciplinary, many fields have developed norms and values which chan-
researchers were not tenured and could not fit within nel and constrain research work. Even within fields.
the discipline-based University system. In theory, research activities and style of enquiry vary greatly.
funds for HSR might be available from the MRC. It is to be expected therefore that HSR should
SSRC, Universities, Regional Health Authorities, pri- encompass a large variety of research activities and at
vate foundations and elsewhere, but in practice a least several quite distinct scientific outlooks. Ray-
thriving HSR community in the U.K. depended on mond Hainer. who pioneered the application of the
government sponsorship [19]. The part of ttie DHSS behavioural sciences to research and development or-
responsible for HSR sponsorship, the Office of the ganisations in the U.S.A.. concluded that differences
Chief Scientist is therefore the focus of concern and of outlook among scientists were often so great as to
attention in what follows; just as it was recently for prevent adequate communication [26]. Maruyama
HSR researchers faced with an evaluative review of [27,28] has described ways that individuals living
their work. within different world-views misinterpret each other.
He emphasises particularly ‘dimension-reduction’. a
Scientijic merit and accountability form of translation which makes an argument, logical
The OCS evaluations were themselves evaluated by or consistent within one frame, appear blatantly false.
the Brunei team [20] and the comments below are illogical or devious within another; and ‘monopolari-
based on their report [21]. The Chief Scientist is a sation’ in which individuals seek for and base their
DHSS employee responsible for the-health and per- work on ow truth, one theory, or one method and are
sonal social services research budget (E13.4 million in emotionally threatened by heterogeneity. He comes to
1981/1982); and the then current holder interpreted the important conclusion that between some scien-
this responsibility as a duty to uphold ‘scientific ac- tists, communication of basic views may be imposs-
countability’, The research team found that in prac- ible [29].
tice this meant the implicit adoption and explicit In the light of this, it is not surprising to find socio-
advocacy of a particular view of science and scientific logists claiming and demonstrating that criteria of
merit. For example, practical knowledge and scientific scientific value are by no means dominant in the
knowledge were seen as belonging to different and evaluation of research work. Prestige, career open-
distinct universes of discourse; research was regarded ings, reward systems, social values. institutional forms
as a method of legitimating change not stimulating it; and such like contribute to channelling and support-
and disciplinary rather than domain-based thinking ing or suppressing knowledge output C30.331. These
was valued. Most significantly, it was assumed that findings do not contradict evidence that scientific
there existed a unified scientific community with a eminence is determined by the quality of a scientist’s
consensus about the nature of science. If this were SO output [34].
(and Kogan and colleagues were conviced that it was Empirical studies support informed opinion and
not) then there would be little need for explicit criteria sociological exposC. Hudson. for example, has ident-
to aid evaluation. ified two distinct kinds of thinkers. convergers and
OCS operated as if there were strict rules for scien- divergers C35.361. The former prefer to work on
tists to follow which would eliminate error. If there is manageable. well-defined problems which lead to a
no guarantee of truth, and many, like Polanyi [22], single best or right answer. They look for value-free
might accept this as an axiom, then the arrogation of and technical solutions. and argue issues similarly.
the decision as to what is or is not a scientific prob- The divergers prefer ill-defined problems with many
lem or a scientific method becomes of profound social solutions. They are oriented to whole systems. per-
significance with overt political and ethical dimen- ceive in value-laden terms and tend to the pejorative
sions. The validity, indeed the reality. of the OCS and the personal in their argumentative and concep-
evaluations depended upon the OCS convincing or tual style. The concern in this paper is not primarily
coercing others (researchers or decision- and policy- with the personality or cognitive style of scientists but
makers) to accept the evaluation as true. Henkel and with the research that needs to be done and its evalu-
Kogan [23] describe the painful breakdown of a peer- ation. Personality is implicated both because it is
review and funding-evaluation process when the likely that certain problems will appeal more to one
 Pluralism in the organisation of health services research 30 I

type than to another [37]. and because Hudson’s ments which embody multiple viewpoints and
work suggests that members of one group will not approaches remains. One solution would be to have
naturally or easily appreciate the preocGupations, multiple funding arrangements. If. for example. cvcr)
arguments and style of presentation of the other. major University had a Department of Health Service
Mitroff’s studies of the Apollo moon scientists [38] Studies with tenured research-oriented staff then the
provide specific evidence that this argument holds in chances for a broad variety of studies to emerge are
at least some fields in the natural sciences [39]. He greater than if a single Institution run by a powerful
uncovered several categories of scientists correspond- Director is set up.
ing to different ways of thinking and working. His In the U.K. where the dominant funding body.
scientists’ perceptions of each other’s competence and OCS, is headed by one person, that person’s commit-
scientific performance were highly emotionally ment to pluralism becomes the crucial variable.
charged and often dramatically contradictory [40]. Assuming such OCS commitment, the problem with
heavy governmental centralisation of funding power
is operationalising pluralism in a way that seems fair
It seems unlikely that emotions should run more to the researchers involved and that meets criteria of
sedately in HSR. and they do not. The U.K. sym- public accountability. In addition, because HSR is
posium referred to destructive conflicts of interest applied. research must be policy-relevant and must be
between individuals working at different points of the understandable and potentially usable by decision-
spectrum in health-related research [4l]. This spec- and policy-makers.
trum was seen as extending from biomedical through The variety of HSR is enormous and any summar!
clinical to HSR. and the example of conflict provided of its extent or content would be inevitably incom-
was the advocacy of the randomised controlled trial plete. The main issue, as regards scientific pluralism.
as a general model for HSR evaluation. One conse- however, is not the subject of research but the way the
quence of such conflicts has been a struggle for pos- subject is approached. A sole commissioning Insti-
session of the key legitimating terms, scierzcc~ and tution (or whatever body oversees a system with
rrsearch. whose meanings tend to be either widened many commissioning institutions) needs to ensure
or narrowed as outlook. or perhaps self-interest. dic- that the principal research methods and perspectives
tates. are available for use when necessary. The varieties of
HSR has been developed and financially supported research approach are limited and in the remainder of
in the U.K. to help solve pressing practical problems the paper’ I will provide a simple model of the
by using the methods of science [42]. It has been research process and then a five category classifi-
indicated above that the work of science might best cation of research approaches which might be useful
be described as ‘knowledge production’. so it should for commissioning and evaluation. I do not claim this
be noted here that problems can often be solved, or is the only categorisation which might be useful. but
solve themselves. without new knowledge: and that rather that it is a possible and testable model.
much knowledge production is not geared to practical
problems, In other words. science and practical prob-
lem-solving need not be purposefully linked. This MODELLING RESEARCH
holds in practice: Acheson for example. has pointed
out that much high quality HSR does not appear in Model of the process
the conventional scientific literature [43]. HSR involves the use of disciplined inquiry aimed
Lindblom and Cohen [44] surveyed a wide variety at solving practical problems in health services. An\
of activities which both produce knowledge and con- model of the research process, and hence any critical
tribute to social problem-solving. many outside the evaluation of the research must therefore include the
academic scene and many only very indirectly related practical problem. The form of the inquiry and its
to problems. and suggested that the most apt label for relation to the identified problem are the obvious sub-
this work is ‘professional social inquiry’. Kogan er al. jects for peer scientific scrutiny. More peripheral and
[35] took a more restricted view. and for their de- more political is the ‘issue of whether the problem
scrIption of HSR research used Cronbach and ought to be investigated. With this in mind. the
Suppe’s phrase “disciplined inquiry” [46]. The U.K. research process may be schematised as shown in
medical press looked on this expression with approval Fig. I.
[47. 481.
The ditficulties in using social science for social
problem-solving were exa&ined in the U.S. in the late WORLD containing PROBLEMS
1960s [49-511. It was noted that discipline-oriented
studies and their institutions were not generally re-
t
sponsive to social problems. and provided a hostile
REPRESENTATIONAL ( CONCEPTUAL 1
environment for social inquir>. Recent examination
[52] of funded problem-oriented research in the U.S. SCHEME

(including HSR, re\ealed several categories of activity

that fell outside the conventional definitions of Data + Theory - Conclurlon
research and de\,elopment (R & D) work [53].
t
r’/lrrtr/i.\r!f ir? HSR
Irfll’~c,/~~c’~ir,r~!~
EPI STEMOL~O~CAL SCHEME
If the preceding arguments for pluralism are 1
accepted. then the challenge of implementing arrange- Fig. 1. Model of the research process.
302 WARRES KISSTOS

The real world containing decision- and policy- built upon the above general model of the HSR task.
makers with their problems. including some they may Any useful classification would be expected to impli-
be unaware of. is the object of inquiry. The cate simultaneously and coherently all three com-
researchers attempt to model a problem qualitatively ponents of that model: the world. the representation.
and often quantitatively. so as to present information the assumptions. In other words, it should be possible
on its nature to those who have the responsibility to to provide simple explicit formulations for each cate-
take action on it. The method by which they develop gory of the classification and these should distinguish
a representation or conceptualisation of the problem different types of problems. different uses of data and
involves the use of data and theory [54]. Theory theory, and different methods and criteria for their
either leads to the selection of data, or assists in its assessment (see Fig. 2). If this could be achieved. then
processing so as to generate a conclusion which may the first step towards implementing a piuralist
be the desired solution or point towards one. commissioning institution would have been made.
All research activities are founded on some episte- Ideally the classification should encompass all
mological scheme which is not uncommonly left im- forms of systematic knowledge production. Such :L
plicit by researchers as they construct and use theory scheme would both distinguish and link scientitic
and data. The unselfconsciously-held philosophical evaluation which focuses predominantly on method
assumptions determine what counts as knowledge. with practical evaluation which focuses on useful ap-
what can be assumed to be true and how method is to plication.
be rigorously developed and applied to the particular As epistemological assumptions are the foundation
problem. Peer review, therefore. is largely about how and justification of knowledge. and underpin the per-
well epistemological assumptions have been trans- ception and appreciation of problems as well as pro-
lated into methods and adhered to in procedures. viding legitimacy to method, they offer a promising
This model is very similar to the model for the starting point for the development of a classification.
production of common knowledge used to solve every- This approach differs from classifications which start
day problems by the man-in-the-street. Einstein [55] with what researchers do, but it should end up b!
described “the whole of science.. as.. nothing more distinguishing usefully between various research stra-
than a refinement of everyday thinking”; and Bro- tegies and activities.
nowski [56] noted that in his efforts to discover, The initial task of the epistemological categories
explain and control the parameters of his world, the would be to make simple distinctions in the way truth
ordinary man generalises, theorises and tests hunches is assigned within the representational order. In par-
in much the same way as the scientist. One of the ticular the relations and status of data and theory in
important differences between the two lies in the the research process must be articulated. Then the
degree and refinement of expl’ication demanded of the implications of each position for the nature of realit!
scientist. In everyday life, not only epistemological as perceived by the scientist working in that mode
assumptions but also the method and representatio- should be clarified.
nal scheme tend to be left unstated. Nevertheless, Particular methods and ways of assessing methods
when a decision-maker is asked to account for his should flow from these initial formulations, and the
actions. he is usually expected to describe br construct appropriate way that scientific knowledge develops
his understanding (model/representation) of the prob- should be revealed. Each epistemological mode is to
lematic situation and show how the information be defined so as to be discrete and self-sustaining.
(data) at his disposal and his ideas (theories) led him allowing adherents to spend a lifetime believing in
to certain conclusions on which his action was based. and generating knowledge based on its assumptions
Describing the research process in a form easily (though some may switch among the modes). As a
accessible to those who must use the results may con- result criticisms from without a mode should be well-
tribute to improving communication and implemen- developed and characteristic. Aside from such
tation of HSR findings C57.583. expected criticisms, which aim to deny validity or
even prevent the public articulation of key epistemo-
Modrl o/.(1 Usefit/ c’kJs.sqication logical assumptions, there will also be inherent
The next step is to create a classification of HSR dangers from over-dependence on each mode or the
application of the mode indiscriminately to all prob-
lems.
Each mode should be related specifically to real
world problems, that is to say, to a particular type of
Eplstemological
knowledge output. It should also be possible to Ident-
ify conditions, such as nature of problem or type of
researcher, for which that mode is most suitable and
conditions under which it is inappropriate.

A REQUISITE CLASSIFICATIOY

Jaques and colleagues have suggested that activities

generated by human purposes naturally stratify hier-
archically into five levels [59]. Beer [60,61] has inde-
pendently postulated that systems which can ‘survive’
Fig. 1. Model of’ ;I wdul clussificntion. Each cell should can be seen to consist of five levels. each of which is a
contain :I distinct explicit formulation. viable system nested within a metasystem. Although
 Pluralism in the organisation of health services research 303

Table 1. Labelling levels of epistemology ferences lay in the fact that in the U.S.. R = finite popula-
tion of patients. in the U.K., R = finite amount of money.
Data obtained by others was used LO show that predictions
of the model held: and this confirmation was used to jus-
tify further predictions [74].
Within U.K. HSR, two substantial fields of DHSS-
funded research of this sort are notable. Operations
research studies problems which relate to systems of
activities and has developed methods for simulating
and modelling needed technical and programmatic
his findings are. like those of Jaques, derived from changes. Data collection is subsidiary to the primary
study of organisation and management, Beer is equally task of mathematical description of the system under
adamant that the five-tier hierarchic model is gener- study. Areas of investigation have included work-load
ally applicable to systems which can be subjectively allocation, planning. information transfer, waiting-
defined and are expressions of human purpose. Prob- lists, supplies and servicing arrangements [7S-773. In
lem-oriented HSR is of this nature, and hence an the DHSS, the Operations Research Division (ORS)
analvsis along Jaques’s or Beer’s lines would appear is administratively distinct from OCS and has its own
feasible. A literature search revealed that a five cate- comparatively small budget of about El m. This sep-
gory hierarchical structure of epistemological levels aration reflects an unwillingness to let OCS handle
had already been developed. This classification this form of research [78].
emerged from an attempt to design inquiry systems The creation of models of appropriate organisation
which regarded science as part of the reality of the in response to requests from the NHS or Social Ser-
social system which it wished to investigate. Singer vice Depts. has been pursued at BIOSS (Institute of
[62-641. his pupil. Churchman C65.663, and his pupil, Organisation and Social Studies, Brunel University)
Mitroff [67-691 have articulated features of this struc- in the Health Services and Social Services Research
ture over the past half century [70], and their classifi- Units [79-81]. The models, images of how things
cation shows a variety of features which are con- ought to be, are built up from elemental notions of
gruent with the abstract principles of the five level aim, function, task, authority, accountability and so
scheme [71]. If the scheme permits the large variety of on, and developed through intensive field-work; but
formulations required of it in the previous section, the use of systematic data collection to construct the
then it can be adjudged useful at least. Whereas in the model or judge its adequacy in practice is explicitly
previous section we have referred to epistemological rejected [82].
mode or category. from now on we will refer to epis- System I products (formal deductive systems) such
temological level or system (Table I). It is to be as statistical theory and economic theory, not devel-
emphasised that the Roman numerals I-V assigned to oped within HSR, are used in the practice of HSR at
the levels are not to be t’aken to signify that one level later levels.
is more complex or more valuable than another.
An application of Churchman’s classification to the S>asrem II: truth is empirical
model of the research process described above is, as it The researcher at this level places his faith in the
turns out, not difficult and the results are briefly sum- facts of simple observation and the widespread freely
marised in Table 2. Table 2 contains formulations obtained agreement from others about them. This
which. it is expected. will be easily recognised by level depends on system I assumptions, for example.
researchers. Within the confines and purpose of this the idea of a fact is taken for granted.
paper. the content cannot and need not be elaborated.
My early research concerned the mapping of the visual
To ease study of Table 2, the classification is illus-
field on to the thalamus and was built on and with facts
trated below using research of my own at each of the [83]. The study commenced with an acceptance of the
levels taken from various fields: and reference is made existence of neurons, action potentials, maps of the cat
to various HSR workers in the U.K. to demonstrate brain, on-off stimulation in visual fields and so on. as these
that pluralism is a present reality [72]. were all part of an agreed network of factual propositions
based on earlier observations. The new observations sug-
gested an alteration of previous networks of facts (theories)
The researcher’s trust lies with ideas which are self- concerning naso-temporal overlap and representation in
evident and can be manipulated by rules which are certain nuclei. and indicated the need for further specific
observations. The published conclusions resulted from
axiomatic to form patterns which mode1 the real
carrying out the same simple experiment a number of
world. The real world of data is too confusing to be
times. and this experiment could be easily described in
direct]! approached. except through such models. enough detail to allow replication by others.

1 usrd this approach for examining the economic rationale In HSR, facts and factual propositions (inductive
for organising doctors [7?]. The basic idea was that a generalisations) are frequently a desirable output. Sur-
number of doctors (PI compete for a finite resource (RI
veys of geriatric care, gerieral practice, drug prescrib-
which the\ use up in very small amounts (r). Using simple
ing habits, obstetric management. patient attitudes.
mathematics. it could be shown that when P>20. the loss
IO a doctor expendtng I’ becomes very small. A variety of and other topics such as those carried out under the
consequences were then deduced: depredation of R. direction of Dr Ann Cartwright at the Institute for
attempts to increase R. and regulatlon of medical practice. Social Studies in Medical Care (London) are prime
The model was applied to the U.S. and U.K. health care examples of system II research [84-861.
systems which were treated as identical. The apparent dif- The DHSS commissions factual reports from the
304 WARREN KISSTON

Government’s Office of Population Censuses and Sur- spread agreement. a system II approach to the un-
vey and from academic researchers on an enormous known has been devised which creates facts to aid
variety of topics. In addition, its Statistics and planning or political decisions. The Delphi method.
Research Division provides an annual tabulation of for example. manipulates the judgements of experts to
HSR-salient statistics (population, finance. manpower, give them .fact’ status and so offers a social tech-
hospitals. patients. morbidity, dental services, phar- nology for handling the future [SK W].
maceutical services. child care. psychiatric care, etc.)
for use by interested researchers [S-J]. Such statistics S~5fern III: mth is .s~~~lf/wric~
may be combined to develop indices of notions like The researcher at this level believes in the integra-
health status’ or ‘hospital complexity’. tion and interdependence of data and theory and
Because facts carry much automatic influence and draws on and attempts to reconcile both system I and
can be created through freely-obtained and wide- system 11 approaches.

Table 2. Levels (systems) of epistemology

The model or Models unc”*er Elemenrar) formal

Iheor) derives the data thaw can truth lead\ h!
from elemenury he fitted ,” the deduct,“” lo B
formal lollen model Theory network ofe\er-
mathemaucall and data are e\pandlng more
truths It reveal, ,eoarate general lormal
or cmhodw a pr”p”,itIon\ and
lundamental tr”,hs
C”d”r,,IS
ctr”cI”raI lealure
or realIt\,

All proposlllons Data jusufier the Elementar> rmplrloal

ciln he reduced to modelftheory and ]udgemenri ITJW data.
ample is separate from tt observa!,ons.
ob,ervatlons. sensat,“ns~ lead by
Theory IS risky
Vahdlty comes lnductlon lo a network
and dependent on
lrom wldcrpread orever-eypandmg.
data
freely “blatned Increaslngl~ more
.lgreement gene~dl network
or factual propos,r,on\

Data derives lrom Theory and data Multiple or requenual

lheory and v,ce are mseparahle. YestIng olalternatl~e
YC~CP. lhence there each dependlog h>porhenes
wll he muluple on and ~nwract~og
“ICW, or reahty with Ihe other
 Pluralism in the organisation of health services research 305

To determine the most useful intervention in patients suf- vention. Assessment of interventions required the collec-
fering from chronic lung disease. I found myself adopting a tion of data pertinent to the various theories, A ran-
complex research design [90]. Patients with dyspnoea from domised controlled trial of four modes of intervention
chronic obstructive airways disease have widely varying (routine physiological assessment. psychoanalytic therapy,
degrees of disability and distress despite having similar expressive psychotherapy by psychoanalysts. supportive
physiological limitations. It is not known why this is so. therapy from a nurse) was implemented. Expressive ther-
but psychological and social factors are believed to be apy by psychoanalysts was found to be significantly su-
involved. However there was no single model or net of perior than the routine assessment in reducing psychiatric
facts on which I could depend. Instead. it seemed necessary morbidity, and supportive therapy by the nurse was found
to construct and select from a variety of models of psycho- to be significantly inferior. However, further models of
social causation. based partly on available facts (II) and intervention, more refined models of the most successful
current theories 41). and suggest different modes of inter- intervention mode. or slighly different versions of the least

Table 1--cor~rirr~ed

Sullahlc ,I: Ahuccd when:

,I I Well-
understood and
well-defined
prohlcm

Model I\ wcr-c\lended 01
over-elahoraw

131 Researcher
undewandr tlw
prohlcm

Eroesnlve re,,i,nce ,II Wcll- Ill-structured problem II made

“0 general \truclured to looh well-struclurcd
agreemen, problem Hard data is hmlted and
Loss or extreme confusmg.
(?I strong
poss!hilltxs Consensus on data is lackmg
consensual
ageement as lo
It* “B,“X

(41 Researcher has

a ‘feel’ for data

, I, Ill-structured Smgle clear lormulatmo or

problem hu$ a simple expnment IS avadable
plan ol the system Overall pl;m IS unavadable or
IS availtlble and a ObJectIves are confused
sectmn can be Concern wrh non-leaslble or
defined and tnwal alternatives. Researcher IS
locussed on blassed

(?I Oh~ecrwea are

clear

(31 Researcher
takes a balanced
and unhlased \‘I~w

Loss or contact ,I I Ill-structured An opltmal soIu11on IS avadable

wth specdics and problem whose
Problem IS well-structured and
speafic issues true natwe IS 8”
ccl”sens”s costs
doubt and suh,ect
Lack of concern
10 antense debate
for purposes or
the actor\ (21 Opposmg
ohwclner

(31 E\pert,
dlsa$ree and no
other tool IS
avadahle

141 Researcher
capable of
I”,“,l,\.Z
reason,ng

Shoas little (1,Ill-structured Stmpie alternatwes must he

problem exphcitl? declded
concerned utth
Feasihdlt? IS less important
the future and
than other cr,ter,a e,e, cer,am,>
306 WARREXKthSTOY

successful mode immediatei! suggest themselves for testing direction and local autonomy” [lo]. Health care for
in ‘mother stud!. Such a stud) \\ould be cobtly. compli- the aged receives a similar handling when .Armstrong
cated and sttll leave open a range of further options for [ 1021 analyses the policy problems: these are found 111
comparison. typical level IV fashion to be based on a “cognitive
Much HSR seems to call for this approach because contradiction” between gerontology ( = life md
so many issues implicate multiple relevant perspec- natural death) and geriatrics ( = illness and pathologl-
tives or call for comparison of a range of alternatives. cal death).
The large Social Medicine and Health Services System IV assumptions also pertneate some forms
.Research Unit at St Thomas’s Hospital and Medical of sociological investigation and analysis. The
School (London) directed by Professor Walter Hol- research sociologist may collect or emphasise data
land carries out a wide variety of studies which typify which supports one theory and ignore or downpla\
the level III approach. Their Annual-Report for 1980 other data or alternative theories. Critical theor;’
[91] reveals that one research group is estimating the explicitly aims at exposing assumptions through dia-
rate of respiratory illness in different home environ- lectical disagreement: and the desired result is the
ments; the next is assessing the effects of the avail- uncovering of hidden values and generation of con-
ability of school milk: another investigates the reac- flict [103]. Merton’s [ 1041 advice to the social theor-
tion of schoolchildren to education about the role of ist to consider manifest and latent functions also
smoking and diet on health by creating two. groups appears to be rooted in System IV assumptions. The
out of a single longitudinally-studied cohort, a fourth latent function (or dysfunction) is. he implies. the
compares the medical outcome of ditferent ward antithesis of.its manifest function. Davis has genera-
designs: and so on. lised this approach to all interesting social theories:
The randomised control trial so strongly advocated great theories are no more than the direct opposite of
for wider use in HSR by Cochrane [92] and Dollery the assumptions of the audience. and their truth or
[93] and the cost-benefit analyses of economists falsity is irrelevant [105].
[94.95] are prime examples of methods based on sys- Workers trained in the British tradition find the
tem III assumptions. above ideas and practices, based on the notion that
data is irrelevant or hopelessly biassed. disturbing and
distasteful. Professor William Paton. addressing the
The researcher believes that any issue has two Academic Assembly of the British Postgraduate
antagonistically opposed sides. each supported by Medical Federation, refers to the canonisation of con-
System III assumptions. These are dialectically inter- flict and to sociological geese who offer “no more
related in that each side implies the other, and can be than speculative theories put in the form of asser-
synthesised to provide a broader model of the prob- tions” [106]. Professor George Brown. the noted
lem. sociologist, whose work appears to be based on level
Within psychoanalysis. narcissism is generally agreed to be III assumptions [107, 1081 describes this disconnex-
a complex and confusing subject [96.97]: and it has .ion of theory from data, which latter he equates with
become the focus of intense conflictual debate. Two major research, as a “puzzling feature of contemporar)
opposing theories have emerged in recent decades which sociology” [ 1091.
define and describe narcissism and narcissistic phenomena From the point of view of many sociologists the
quite differently. Research papers follow one line and ex- epistemic principle of agreement is something of a
clude the other, and in debates any clinical material (data)
sham. because those with power are the ones who
presented to support one theory is routinely reinterpreted
to support the other. My approach here was to describe
determine which facts will be agreed to. The exposure
the situation In detail and defend the validity of each view- and discomforting of power elites is a historic socio-
point [9X]. Then I demonstrated that the two viewpoints logical mission; and this was reaffirmed by Philip
were dialectically related and that they could be under- Abrams, late Professor of Sociology at Durham
stood in the light of a particular universal childhood ex- speaking to the Centenary Meeting of the British As-
perience [99]. An awareness of this dialectical interplay sociation [ 110. Ill]. He distinguished between “argu-
and its synthesis allowed exploration of implications for mentative knowledge” and “authoritative knowledge”
clinical treatment [ IOO]. and claimed that the role of sociology was to provide
Health services offer much potential for opposing the fuel for political debate. to make politics more
interpretations of data and the range of contlictual sensitive and strenuous not to settle issues or to pro-
polarized issues is great: public vs private care; plura- vide certain answers.
listic vs monistic system: professionalism vs bureauc-
racy: central control vs local autonomy; individual- System V: truth is prugmutit
ism vs collectivism: preventive vs curative; care vs The researcher’s faith is placed in both data and
cure: health vs illness; quality vs equity; political- theory only insofar as they are part of conclusions
incremental vs rational-managerial decision; com- which enable action in the social world. otherwise they
munity vs hospital: acute vs chronic. are irrelevant. The researcher is oriented towards the
In HSR. this is typically the field of policy-studies. future and explicitly incorporates an ethical dimen-
For example Rudolf Klein. Professor of Social Policy sion in his findings.
at Bath, not uncommonly commences his regular
When a family comes for help with its difliculties. often a
articles in the B~irisl~ Mctlictr/ Journul by stating the member with emotional or behavioural disturbance. its
particular dialectic he intends to discuss. A recent problems can be tackled with family therapy. which i\
paper was entitled “The strategy behind the Jenkin based on the principle that there needs to be a change in
non-strategy” and the opening sentence posed the the way family members interact. Many practltloners
dilemma of “ho\+ to strike ;I balance between central specify in detatl the final results if treatment ih to be
 Pluralism in the organisation of health servxes research 307

deemed successful. and then gear therapy lo achieving their origins, may be dismissed as unsubstantiated or
these goals [ 112.I Ii], My colleagues and J believe that simplistic.
such an approach does not always do justice to values of
autonomy and dignity. We have described a quite different
approach to the family in which the therapist only specifies
the direction of therapy. and leaves the family members DISCUSSION OF THE MODEL
open to choose their own new way of functioning
[I I?. I IS]. Working with such general objectives demands An important source of the current conflicts within
a particular therapeutic approach. The therapist must HSR in the U.K. has been identified. This is the
become. temporarily. part of the family system so as to unrealistic belief of those holding the purse strings
explicate the hidden and conflictual value and meaning that the scientific community is unified rather than
systems within the family. He must create the possibility of disparate. The attempt to impose inappropriate con-
a neh future for the family based on a different under- sensus on dissenting researchers has inevitably pro-
standing (i.e. new knowledge) of the past and the present.
voked destructive conflict. If consensus could be
but the restrictions on change imposed by the actual
reached on the nature of differences between
nature of the family members in their particular social and
cultural context must also be recognised. researchers and their methods, a beginning will have
been made in grappling with tile enormous problems
System V assumptions call for the researcher to of assessing the value and quality of research work.
participate in. and create knowledge through the defi- Sub-groups of researchers, themselves, if unencum-
nition of (social) reality. Such an approach seems par- bered by fundamental differences in their vision.
ticularly relevant to innovations which have to be im- might be able to develop useful criteria and tech-
plemented both centrally and peripherally within or- niques for evaluating each other’s work. and conflict
ganisations. involve alteration in the social relation- might then be manageable or even harnessed con-
ships within the system and impinge on other social structively.
systems [ 116. 1171.Argyris [I IS] argues that the aim Five such sub-groups embodying five categorically
of research of this type must be to generate valid different problem-solving research approaches based
information which helps the client make informed on non-commensurable images of reality have been
and responsible choices and develops his irlrerrwl com- identified and their work has been described and illus-
mirmem to rhose choices. Given the contestability and trated (mainly in Table 2). The distinctions have been
contentiousness of social things and the lack of recognised by considering the activities of practising
awareness that reality needs to be defined by those in scientists, including the author, and noting repetitive
positions of power. it is not surprising that the pro- annihilating criticisms between scientists working in
cess and findings are sometimes met with incompre- different ways [131]. Before outlining the implications
hension. of this analysis for the national organisation of HSR.
Within HSR. the principal figure operating in this some questions require examination. First, is any par-
fashion has been Professor Elliott Jaques, who has ticular level a preferable research approach for any
directed the Health Services Organisation Research particular health service problem? Or, put more ex-
Unit at Brunei University. His output is founded on tremely, should the label ‘science’ be reserved or re-
the use of field-work combined with a stringent and stricted to certain levels’? Second, how is this classifi-
. repeated process of conceptualisation and formula- cation related to current descriptive categorisations of
tion to develop valid usable social concepts research?
[ 119. 1201. However. this has not always been under-
stood [l’l. 1221.
Trevor Gambling. Professor of Accountancy at Choice 0J’ system (or level)
Birmingham University, like Jaques, has taken a The usual arguments for restricting the labels
whole systems view and has inevitably expanded ac- ‘science’ or ‘research’ to selected levels lose signifi-
countancy to a theory of society. He argues that cance because they tend to reflect the typical asser-
accounting practice is about magically boosting social tions or criticisms applicable to each level and be part
morale and that its data can be manipulated to look of a desire for a unified scientific community. All the
hard. B\ his own report. he is misunderstood; he finds systems described lead to knowledge and may con-
academic barriers to publication and his research tribute to problem-solving. They can do so scientifi-
projects seeking to determine the changing nature of cally insofar as a sound systematic process of inquiry
accounting in practice tend to fail through disinterest is used. The prime purpose in making distinctions
or active opposition [ 123. 1241. In operational between what is and is not science is to reduce confu-
iesearch in the U.S.A.. both Churchman [125] and sion and maintain identity [132], but if the social
Ackoff [ 126. 1271 have endeavoured to introduce a result is schism and excommunication within the
similar pragmatic and total system approach, appar- scientific community then another solution may be
ent]! also Mith little success [128-l 301. desired. The alternative pluralistic approach need not
The problem of researcher detachment and ethical blur distinctions or threaten identity and can welcome
commitment within a turbulent emotional and value- complexity. uncertainty and disagreement. Tolerance
laden field is significantly increased by his intimate and understanding of the validity of differences could
connection with action and change. Because the lead to fruitful collaboration.
research process at System V challenges assumptions. The classification makes it clear that much of the
generates comple\it!,. requires actors to explicate System II approach so dominant in biological
their value conflicts. and evokes uncertainty in the sciences, and the System III approach so important in
context of immediate decisions. it is easily perceived clinical research, is directly applicable to HSR. How-
as threatening: and the findings. disconnected from ever, Systems II and III like the other levels are only
308 WARREN KINSTOX

fitted for certain problems, so those who say the bio- them. it has tried to capture the phrase and the funds
medical approach is not always applicable are also which accompany it. However. evaluation can also be
correct. brought about by exposure of assumptions using S>s-
Particular issues may require research work at tern IV: evaluation is a continuous process at V: and
several levels. For example, very well-structured prob- at I and II evaluation is inherent in framing and
lems are most easily handled by Systems I and II. answering questions. System III evaluation is insuti-
while ill-structured problems demand a System III, IV cient, inappropriate or impossible when promptness is
or V approach. The attempt to define a clear problem preferable to precision. when the need is not for the
from an unclear issue is characteristic of System III. It best solution but for any feasible solution. or when
breaks down when the situation is so complex and there is a ‘one-shot’ or total-svstem intervention.
confusing that it is not eveh clear whether there is a Similarly, the phrase .action research’. possibly
problem. Substantial research using System IV and V originally referring to System V work by Lewin [ I371
must then be conducted to reveal more about its has been applied to work at all levels. .4lmost im-
nature. mediately. Chein ef al. [I381 discribed diagnostic.
The epistemic principle of agreement [133, 1341, so participant. empirical and experimental versions
strong in System II and also powerful in Systems I which seemed to operate with a variety of assump-
and III. becomes replaced by the value of disagree- tions.
ment in Systems IV and V. Examining the assump- (ii) Dichoromous classjfication. A large variety of
tions underlying agreement or disagreement as a human concerns have generated philosophical
scientific criterion will be immediately recognised as a schools. and though most have an epistemological
System IV inquiry. Disagieement and the challenge of branch few have more than a peripheral or indirect
assumptions behind agreement seem to be associated relation to research work. As a consequence. Susman
with integration of values into the research process and Evered [ I391 in their search for some philosophi-
and a loss of confidence in the supremacy of data cal legitimation for their conception of action
[ 13.51. At levels I to III, value issues either appear research could enlist the support of praxis. hermeneu-
before research commences (built into the design and tics, existentialism, pragmaticism. pragmatism. pro-
data selection decisions) or afterwards (in the inter- cess philosophies and phenomenology. Research has
pretation of the results). At level V, the research pro- its own concerns and requires articulated assumptions
cess engages the values of the decision-/policy-maker which serve its particular purposes, not such a hodge-
as a substantive part of the inquiry and evaluation of podge of philosophies at second-hand.
the process of inquiry [i36]. Although scientists are reluctant. not improperly. to
Problems may need to be tackled from multiple enter another field on its own grounds, they should
angles and at different levels. For example resource not be averse to asserting their own assumptions and
allocation issues often involve level III and level IV guiding principles. At times they are keen to do so.
inquiry; examination of potential health service devel- This enthusiastic ability is most evident in the face of
opments usually requires both a model of the devel- some violation of or threat to deeply held assump-
opment (I) and information on its likely use (II). tions: or when a desire exists to expose the inad-
Because many problems are posed as ill-structured equacy of the assumptions of colleagues. Most classi-
issues (‘impact of unemployment on health care’), they fications generated by researchers in this way and
not only allow redefinition as questions in a variety of naturally used by them are dichotomous. This should
disciplines but also at all levels. Any researcher will not now be surprising, because exposing assumptions
attempt to redefine problems so as to fit the System is usually performed via level IV inquiry which typi-
he prefers to work in and in the light of his training, cally generates opposing viewpoints [ 1401.
institutional pressures, likelihood of funding, available The reader will be familiar with the controversies
assistance, personal interest and so on. The choice of which advocate some method or epistemological base
System is then usually vigorously defended. for research: hard vs soft data; quantitative vs quali-
tative methods: positivistic vs non-positivistic science:
objective vs subjective evaluation: biassed vs
Currently used classifications of scientific research unbiassed reporting; explanation vs understanding:
(i) Descriptive labels. The current approaches to the prediction vs making things happen: detachment vs
controversial topic of research method classification engagement; thinking vs feeling; reductionist vs holis-
are either purely philosophical and distant from tic. As indicated in Table 2. the main output of such
research practice, or descriptive and easily used by an approach is a raising of awareness. not a provision
scientists and others but confusing. Most labels of solutions. Pushed too far. dialectical inquiry and
rapidly lose meaning on close scrutiny. Certain debate become acrimonious. excessively repetitive and
research groups, for example, have tried to capture time-wasting; and such inquiry is never a guide to
labels like ‘problem-oriented’ or ‘hypothesis-testing’, action.
but this paper started from the well-accepted premise When there is a truly dialectical debate on an im-
that all HSR has these characteristics. ‘Experiments’ portant issue whose nature is still unclear, compre-
(i.e. trial and test procedures) are not restricted to one hension within some larger scheme is urgently
System and the mundane and sometimes pejorative required [141]. In some cases it is possible to soften
label ‘descriptive research’ is easy to apply to forms of the polarisation and reduce. if not resolve, intense
research at all levels. conflict by describing the dialectic as a continuum.
‘Evaluation’ is a particularly important label. that is to say by moving inquiry from level IV to level
Because System III research is essentially about com- III, For example. Cherns [ 1421 faced with the conflict
paring hypotheses, i.e. about attributing value to between discipline-oriented approaches and problem-
 Pluralism in the organisation of health services research 3OY

oriented approaches converts it to a four category IflsritutiorIal he

continuum.
No single independently funded HSR-lnstrtute
The preferable approach is to remove the very
engaged directly in research of its own choosing
ground of conflict by creating a higher-level synthesis
would be likely to encompass the total range of HSR
which accepts the dialectic and puts it into perspec-
methods, unless it were very large. An unusual Direc-
tive. This may be aided through level V inquiry: by
tor might manage to promote them with 4&50 staff.
regarding the key accepted notions in the dialectic as
but a board of Directors and over 100 staff sounds
confusing and complex objects for deeper study. it
more realistic. This starts to feel like a domain-based
may be possible to develop new meanings in the con-
University, and the creation of such an institution
text of a commitment to devise a practical means of
would become a political issue.
proceeding. This essay has attempted such an inquiry.
For purely commissioning purposes. there 1s no
reason why an HSR Council could not provide funds
for the total range of HSR. Similarly. in principle. the
I\STITL’TIOKAL IMPLIC‘ATIOYS HSR panel of the MRC might be able to. The difticutt
and challenging task would be to ensure that
This paper has elements of a level I approach to the
members of these bodies. especially the Chairmen.
problem of evaluating scientific research insofar as it
understood and valued scientific pluralism.
uses a model developed by others. Alternatively. it
might be regarded as a level III inquiry insofar as it is
based on the author’s own experience meshed in with Any arrangements for funding should ensure the
current theories in the philosophy of science. The existence of research at all epistemological levels. Thus
underlying spirit is however level V. means that all scientists should not be in open compe-
The level V approach opposes artificial isolation of tition against each other. Competition among scien-
systems which in practice interact. Knowledge gener- tists has more relevance to work within systems than
ated at all levels interacts and both well-structured between systems. For each system. there must be a
and il!-structured health service problems demand certain minimum resource of competent practitioners
attention. As a result. a variety of disciplines and and posts to ensure ongoing development of methods.
whole fields of research have been swept into this to provide adequate training and career structures
inquiry into HSR and made bed-fellows [ 1431. and to act as a competitive spu; to each other.
Another level V feature is that it is written in the Society, not science, will suffer if any approach is neg-
imperative rather than the indicative. The paper at I lected and so society should be concerned with the
or III would say ‘This is u IWJ or even ‘This is rhr balance between systems. A special NHS Health
brsr kr~~rl WUJ to classify HSR. leaving the prescrip- Authority with responsibility for HSR deserves con-
tion implicit or covert. At level V, it says “This classi- sideration as a method for providing the necessary
fication has been devised specifically to assist in or- public accountability of any HSR board or council.
ganising and evaluating HSR: r&e it as the way until
something better comes to mind. or stay in the cur- Research eaaluatiorl
rent difficulties’. This proposition is offered because it Peer review mechanisms focused on method are
is testable. believed to be 55-60”” right on the Milosz essential. Method must be understood to stand on
scale [ 1441 and capable of further substantial elabor- two legs, one lodged in the specifics of the problem to
ation in the service of practical organisation of HSR be solved and the other in the image of reality
(cf. [71]). adopted. It is necessary therefore that the peer share
Such a command with its potentially profound both the epistemological base. as well as being con-
effects on society and the lives of researchers must be versant with the domain of the problem.
,justified ethically. and not just by the need to reduce There is a variety of ways, applicable to all levels.
uncertaint! or to produce an elegant or practical by which the scientific legitimacy of knowledge pro-
model: otherlvise it simply reflects the authoritarian duction might be decided. A preliminary and contest-
attitude so objected to in the OCS review. The ethical able list might include careful reference to other
presuppositions in this case are the right for individ- knowledge sources. especially within the scientific
ual scientists to work in the system they find most world: public description of the methods used in
natural. the importance of reducing destructive con- obtaining the knowledge: publication of the knowl-
flict among scientists. and society’s need for its prob- edge for use by other researchers: systematic develop-
lems to be tackled with all the scientific means avail- ment of knowledge; an active quest to improve
able. methods; and acceptance of peer assessment as de-
The general rule for biological and social processes scribed above.
is heterogenisation [ l45]. A homogenous scientific Less relevant to assessment would be any appeal to
communit) is unable to provide for work on the full norms such as faith in the moral virtue of rationalit!
range of health-related social problems which HSR non-rationality, emotional neutrality/commitment.
must tackle. houever comfortable it may feel for scepticism/dogmatism and the like. Appeal to one or
those included within it. If. in addition. that hom- other side of these norms is no substitute for a
ogenous communit! speaks in the indicative. HSR is detailed exposition of assumptions and intentions
in danger of covert authoritarian rule. and their translation into useful research work [146].
If the command stitched into this paper is heeded. The awareness and use of such assumptions is a per-
Lvhat might it impl! in terms of institutional base. sonal matter and adherence to them a qualit! which
funding and research evaluation’? The comments defines a scientist’s life. Indeed. it could be argued
which follow simpl! aim to open the debate. that “only the scientist (himselfl can tell,. whether he
310 WARREN KIMTO\

is or is not being scientific at any given moment”

[ 1471.
II. Koean M.. Korman N. and Henkel41. Gor~,r~rrrrr~rr’\
Cokissionirly of Rrsetrrch; .A Cuw .Stm/~,. Brunei
University Department of Government. U\bridge.
The aim of this paper has been to suggest the
1980.
nature of a usable framework within which HSR
15. Henkel M. and Kogan M. 7lrr DHSS Fwded
commissioning and evaluation. both thorny problems, Resrurch Ullits: The Pro~~rss o/’ Rcr~cw. Brunei Unr-
could be meaningfully developed: meaningfully, that versity. Department of Government. U\bridge. 1981.
is to say. to the practising researcher, to the public I 6. Leading article. Health services research. 51. hrell. J.
and its decision- and policy-makers. and to the wider 282. 845-846. 1981.
scientific community. 17. Editorial. Commissioning research. Lrwtv i. 33 l-3 13.
Evaluators from the scientific community focus 1981.
explicitly on method and implicitly on epistemologi- 18. For example. the Minister for Educatron. havrng
reduced the finance available to the Social Science
cal foundations: decision- and policy-makers focus
Research Council. set up an inquiry (again under
on types of problems and research output: and
Rothchild) to establish whether it was necessary at .tII.
researchers operate with data and theory using (Lord Rothschild Enquiry ijtto the Soeicll S~~ir~rcc
methods to solve problems. A classification of Rrsrurck Courtcil. HMSO Cmnd X554. 19X2.)
research is offered that can bring these three groups 19. The Government has never accepted the full weight of
and their foci together in a way that fosters mutual this responsibility, for example. in terms of training
understanding and constructive cooperation. and career protection. Holland W. W. Five years after
It is argued that society. by funding HSR, provides and a bit. In &Iuttrrs of‘~f4~omrut (Edited by McLach-
the wherewithal for scientists to offer a level of needed lan G.). Nuffield Provincial Hospital Trust. Oxford.
practical usefulness. It is therefore incumbent on the 1981.
20. Henkel M. and Kogan M. op. cit.
scientific community to ensure that all available
II. Their observations and conclusions are supported by
methods will be used and appropriately evaluated. my own experience of an OCS evaluation.
This requires an acceptance by scientists that HSR 22. Polanyi M. Logic and psychology. /Ial. P.s.I~~/Io/. 23.
inherently demands fundamentally different styles of 27-43. 1968.
inquiry based on differing images of reality. Such 23. Henkel M. and Kogan M. op. cit.
styles and images constitute the identity of individuals 24. Greer S. The Loyic of‘Socru/ Inquiry. Aldine. Chicago.
and groups, hence institutional mechanisms are essen- 1969.
tial to ensure that society does not fall into the grip of 25. Galtung .I. Tlrrorx rrntl lLfrt/wd.s111 Socitrl Resrurch

scientific homogeneity. or become racked by fruitless (Revised Edition), p. 15. Columbra University Press.
New York., 1969.
turmoil.
26. Hainer R. M. Rationalism. pragmatism and existen-
tialism: perceived but undiscovered multi-cultural
REFERENCES
problems: In The Rrsrurch Societ! (Edited by Glatt
This clumsy phrase includes administrative and pro- E. and Shellev M. W. ). D. 7. Gordon & Breach. New
fessronal staff within the NH’S, civil servants within York. 1968. .
central and local Government departments, lay 27. Maruyama M. Awareness and unawareness of misun-
members of health authorities and community health derstandings. Mrtltods 13. 255-275. 1967.
councils. and politicians. 28. Maruyama M. Basic elements in misunderstandings.
Taylor D. Hdth Rrseurchin E/ry/und: A Topic fi)r Diulecticrc 17. 78-92. 99-l IO, 1963.
Dchtrtr. Office of Health Economics. London. 198 I. 29. Maruyama M. Paradigmatology and its application
Acheson R. M. Towards a health services research to cross-disciplinary. cross-professional and cross-cul-
board. Lrmet ii. 1307, 1980. tural communication. World hthropoloy~. Mouton.
Cohen R. H. L. After Rothschild.. Luncrt ii. ll31- Paris, 1973.
1132. 19x0. 30. Crane D. The gatekeepers of science: some factors
Dollery C. T. The Ead of (111Age o/ Optimism: Medid affecting the selection of articles for scientific journals.
Srierfce it1 Retrospect und Prospect (Rock Carling Am. Social. 2. 195-201. 1967.
Mow~gruj~h). Nullield Provincial Hospitals Trust. 31. Merton R. K. The Matthew effect in science: the re-
London. 1978. ward and communications systems of science are con-
Kay A. W. Rrseurch it1 Medicine: Problems und Pros- sidered. Scirrlce 159. 5683. 1968.
pects (Rock Curling Monogruph). Nutfield Provicial 32 Mulkay M. J. Sociology of the scientific research
Hospitals Trust. London, 1977. community. In Science. Tec/~wlo~g.v ud Societ?
McLnchlan G. (Ed.) Ficr Years A/&r: A Rroiew of’ (Edited by Price E. de S.). Sage. London, 1977.
Heulth Cure Re.sr&h Munuyemen~ after Rothschild. 33. Mulkay M. J. Scierlcr urd the Socio/o~g~ of’ Knowledge.
Oxford Unrversitv Press. London, 1978. Allen & Unwin. London. 1979.
Maxwell R. The- organisation and management of 34. Cole J. R. and Cole S. Sociul Strutrficutrorl irl Sciewe.
research in health services. Paper presented to University of Chicago Press. Chicago. 1973.
seminar on Health Services Research. Brunel Univer- 35. Hudson L. Contrury Iuruc/irrtrtiorts. Schocken Books.
sity. Uxbridge. Middx.. 198 I. New York. 1966.
Royal Commission on the National Health Service. 36. Hudson L. Frcrtttes of Mired; Ahilitr. Prrcrptioh urrtl
Report. HMSO Cmnd. 7615, London, 1979. S+Prrceptio,t 11)the Arts ud Scienws. Norton. New
Holland W. W. The oreanisation and funding of York, 1968.
research into public hezdth. J. puhl. H/t/t Po; 2, 37. Kuhn T. S. (1963) The essential tension: tradition and
354360, 19x1. innovation in scientific research. In Scrrntijic Cretrti-
Merton R. K. Tl~e Soc.io/o<r~ of Scierlce. Chicago Unt- vit;, Its Recoyrritiorr trrrd Deorlopmrr~t(Edited by
vcrsity Press. Chicago. 1973. Taylor C. W. and Barron F.). Wiley. New York, 1963.
Taylor D. op. cit. pp. 6-9. 38. Mitroff 1. 1. Tltr Srrhjrcrrre Sit/c of Scirhcr. Elsevier.
Rothschild Report. Tlrr Oryctrri.sutio)r clrtd MUIIU~L’- Amsterdam. 1974.
 Pluralism in the organisation of health services research 311

39. The 42 scientists interviewed included general geol- 64. Singer E. A. Jr (1959) E.sperietw olzd Re/lrctiorl
ogists (7). geophysicists (4). geochemists ( 18).chemists (Edited by Churchman C. W.). University of Pensyl-
(5). physicists (4). astronomers. (4). engineers (2). vania Press, Philadelphia. 1959.
40. Mitroff I. 1. op. cit. pp. 9496. 65. Churchman C. W. Theorv a/ Erperimerztul /n/erenw.
41. Taylor D. op. cit. p. 19. Macmillan, New York, 1948.
42. Rothschild Report da. eif. 66. Churchman C. W. The Desigtl of Inquiring S~wrm~.
43. Acheson R. M. Must health services research be Basic Books, New York, 1971.
funded bv a commissioner’? Luricer i. 788-789. 19X I. 67. MitrofT I. I. Systems. inquiry and the meanings of
44. Lindblom C. E. and Cohen D. K. Liscrhle Ktlowledge: falsification. Philos. Sci. 40, 255-276, 1973.
Sociul Sciewe and Sociul Problem-So/oirt(/. Yale Uni- 68. MitrotT I. I. Epistemology as General Systems
versity Press. New Haven. 1979. Theory: An approach to the conceptualisation or
4s. Kogan M. Korman N. and Henkel M. op. cit. complex decision-making experiments. Phi/as. Sac.
46. Cronbach L. J. and Suppes P. (Eds) Reseurch jar Sci. 3, 1 17-134. 1973.
~omorrow's Sd~ools~ Disciplined Enquiry .liw Edu- 69. Mitroff 1. 1. op. cit. [39].
curior~. Macmillan. New York. 1969. 70. From this point the paper draws heavily on and de-
47. Editorial op. cit. velops the work of these authors without specific
48. Leading article op. cir. reference.
49. National Academy of Sciences. The Brhovioura/ 71. This is not the place to analyse the validity of the five
Sciences and The Federul Go~:ernmem. National level model, or the points of contact between the
Research Council, Washington. DC. 1968. theoretical model and its application. However, it
50. National Science Board. Kuowledgr and A&m: Im- may be noted, for example, that the order in which
proriry the Nation’s Use qf the Sociul Sciettce~. the various inquiring systems are placed by Church-
National Science Foundation. Washington. DC. man, though apparently intuitive, fits the abstract
1969. hierarchic order suggested by Jaques and Beer. Other
51. National Academy of Sciences and Social Research links include: maximum uncertainty at levels I and V,
Council. The Behuaiourul und Sociul Sciences: Outlook with minimum uncertainty at level III; the move from
artd Needs. Prentice-Hall. Englewood Cliffs, NJ. concrete and specific contact with the world to more
1969. general and indirect contact; the intuitive creation or
52. National Research Council. The Federal lrtwstment irt elements and meaning at level I; flexible use of rules
Knowledge of Social Problems, Vol. 1. (Study Project at level II; extrapolation at level III; the caesura
Report). National Academy of Sciences. Washington, between levels I-III and IV-V: instabilitv at level IV:
DC. 1978. completion of the universe of discourse at level V
53. “Of the seven knowledge production and application 72. I use my own work, in HSR and other fields, because
categories. three-research, demonstrations for policy 1 have some intimate and detailed knowledge of the
formulations. and development of (social) materials- mental processes and experiences connected with it
fall within the traditional definition of R & D. Thus, and am not uncomfortable with the epistemological
approximately 5980 million can be called “Social R & propriety or limitations of each approach. My refer-
D”. The four remaining categories-program evalu- ence to other HSR researchers is essentially impres-
ation. general purpose statistics. demonstrations for sionistic A forma1 research study of researchers to
policy implementation. and dissemination-totalling test and. if need be modify or elaborate, this classifi-
$832 million. fall outside the usual definition of R & cation would be feasible as the issues are alive and
D but fulfil legitimate knowledge producfion or appli- meaningful for them. It could sustain a tradition of
cation functions”. (National Research Council. op. cit. research into HSR organisation which the DHSS
p. 91. 1978). Table A-4 of the same document (p. 112) plans to discontinue.
breaks down funding in fiscal 1976 for categories of 73. Kinston W. The organisation of doctors to serve
HSR by the above types of activity. whole populations given finite resources. In National
54 Data and theory are potentially confusing terms and Health Care Spstems in Non-Growth Economies
the brief definitions here are contestable: Data is that (Edited by Svensson P. Cl.), Suppl. 28 of ScatId. J. Sot.
which purports to correspond directly with the real Med.. pp. 117-131, 1981.
world i.e. ‘things’. 7Iteorj. is that which purports to 74 Kinston W. Resource consumption and future organ-
correspond with our expectations of the real vvorld i.e. isation of medical work in the National Health Ser-
‘propositions’. vice. Sot. Sci. Med. 16, 1619-1626. 1982.
S?. Einstein A. Out qfM>, Later Years. Longmans. Tor- 75 Luck G. M.. Luckman J.. Smith B. W. and Stringer A.
onto. 1950. Patients. Hosoitals and Ooerational Research. Tavis-
56. Bronowski J. The Commo~t Sense of Science. Heine- tack, London; 1971. ’
mann. London. I95 1. 76. Barber B. (Ed.) Selected Papers on Operational
57. Rvan A. The Philosoph!~ of the Social Scierlces. Mac- Research in the Health Seruices. Operational Research
millan. London. 1970. Society, Birmingham, 1976.
58. Strong P. M. and McPherson K. Natural science and 77. Boldy D. and Clayden D. (1979) Operational research
medicine: social science and medicine: some meth- projects in health and welfare services in the United
odological controversies. Sot. Sri. Med. 16, 643-657. Kingdom and Ireland. J. Opl RES. Sot. 30, 505-511.
1982. 78. It is noteworthy that ORS was not invited to the
Jaques E.. Gibson R. 0. and Isaac D. J. Levels of Health Research Symposium referred to at the begin-
-th.srructiou irl Logic arid Humarl Acfiort. Heinemann. ning of the paper (see [23).
London. 1978. 79. Rowbottom R.. Balle J.. Cang S.. Dixon M.. Jaques
Beer S. Brain of’the Firm. Allen Lane. London. 1972 E.. Packwood T. and Tolliday H. Hospital Orgarlis-
(2nd Edition. 1981). ation. Heinemann. London, 1973.
Beer S. Fhr Hearr of Enterprise. Wiley. New York, 80. Rowbottom R. W.. Hey A. and Billis D. Social Ser-
1979, cites Departments: Developing Patterns qf Work arld
Smger E. A. Jr Mind a.5 Behariow. R. G. Adams & Organisation. Heinemann. London. 1974.
Co.. Columbus. 1924. 81. Billis D., Bromley Cr., Hey A. and Rowbottom R. Or-
Singer E. A. Jr Ori The Cortrerlted Liti>. Henry Holt & ganising Social Serrices Departmems. Heinemann.
Co.. New York. 1936. London. 1980.
312 WARRES KISSTO\

82. Rowbottom R., W. Social rl!~ul~sis. p. 134. Heine- to/s. lY6&hS. Cambridge Unilrrsit! Press. London.
mann. London. 1977. 1970.
X3. Kmston W.. Vadas M. A. and Bishop P. 0. Multiple 108. Brown G. W. and Harris T. Socral Orry~rl\ of De-
projection of the visual field to the medial portion of pressiorl. Tavistock. London. 197X.
the dorsal lateral geniculate nucleus and the adjacent 109. Brown G. Some thoughts on grounded theor!. SO~,IIJ-
nuclei of the thalmus of the cat. J camp. ,Veuro/. 136. log!, 7. l-16. 1973.
295-315. 1969. 110. Abrams P. Sociolog): could there be another 150
84. Dunnell K. and Cartwright A. .Medicine Takers, Pre- years’? Paper presented at 150th Anm\ersar!. Brltlsh
scribers and Hoarders. Routledge & Kegan Paul. Lon- Association for the Advancement of Science. York.
don. 1972. 1981.
85. Cartwright A. Thr Dignit! of Labour? A Stud! 1 I I. Abrams P. The uses of British sociolog) 1X31-IYSI.
of‘ Childbeariw and I&tctiorl. Tavistock. London. Paper presented at 150th Anniversary. British ,ASSOCI-
1979. ation for the Advancement of Science. York. 19x1.
86. Cartwright A. and Anderson R. Grrlrrrrl Prclcrice 112. Weakland J. H.. Fish R.. Watzlawick P. and Bodin .4.
Revisited: A Second Sfud! cf Parieuts urtd Their Doc- M. Brief therapy: focused problem resolution. F~IIJI.
tors. Tavistock. London, 198 I. Proc. 13. 141-168. 1974.
X7. Department of Health and Social Security. Herr/t/l umf 113. Epstein N. B. and Bishop D. S. Problem-centred sks-
Per.sorla/ Sociul Sercjces Stutisrics fbr Enyhd (with terns therapy of the farnil!. J. .Lftrrr. FUIU. T/IN. I.
Samrtrry Tub/e ,ftir Greut Brirairi) I Y7n. HMSO, Lon- 18-25. 1981.
don. 1980. 114. Kinston W. and Bentovim A. Brief focal farnIl> thrr-
xx. McLaughlin C. P. and Sheldon A. Forecasting new apy when the child is the referred patient. II. Method-
technologies, In Tile .MunoG+jnen[ of Hetrlth Cure ology and results. J. Child PY~~c/IO/. P\l~chrar. 19.
(Edited by Abernathy W. J.. Sheldon A. and Prahalad 119-144. 1978.
C. K.). Ballinger, Cambridge. MA, 1974. 115. Kinston W. and Bentovim A. Constructing a focal
X9. Linstone H. A. and Turoff M. (Eds) Tl~r Delphi formulation and hypothesis in family therapy. .-llr.sr. J.
.Mrrhod: Techiques cold Applicutiorm Addison-Wes- Furm Ther. 4. 37-50. 1982.
ley. Reading MA. 1975. II6 Pressman J. and Wildavsky A. /rtlplr,,le~lrrrrio~~. Un!-
90. Denford J.. Guz A., Heslop A.. Kinston W. er ul. versity of California Press. California. 1973.
Breathlessness and psychiatric morbidity in chronic 117. Parlett M. and Hamilton D. Evaluation as illumina-
bronchitis and emphysema: A study of psychothera- tion: a new approach to the study of innovatory pro-
peutlc managemen;. Psyhol. Med. (In press). grammes. In BeJo& tile Naruhers Gumr (Edited b>
91. Social Medicine and Health Services Research Unit. Hamilton D.. Jenkms D.. King C, McDonald D. and
.411~11l~llReport (I April 1979-31 March 1980). St Parlett M). Macmillan. London. 1977.
Thomas’s Hospital and Medical School. London, 118 Argyris C. Itltercrntiou Thor!, and .Lfrthod. Reading.
1980. Mass: Addison-Wesley. Reading, MA, 1973.
92. Cochrane A. Efictirejless and Efficierlcp: Random Re- II9 Jaques E. A Gemmd Theory of Bareaacrtrc~. Heine-
flecriom on Health Srrcices. Nuffield Provincial Hos- mann, London, 1976.
pitals Trust. London. 1972. 120.Jaques E. (Ed.) H&r/l Services. Heinemann. London.
93. Dollery C. T. op. cit. 1978.
94. Mishan E. J. Cost-Berwfir Aflalysis. Allen & Unwin, 121. Whittington C. and Bellaby P. The reasons for hier-
London, 1971. archy in social services departments: a critique of
95. Williams A. The cost-benefit approach. Br. med. Bull. Elliott Jaques and his associates. Sot. RcL.. 27.
30, 25’. 1974. 5 13-539, 1979.
96. Lichtenstem H. The role of narcissism in the mer- 122. Draper P. and Smart T. Social science and health
gence and mamtenance of a primary identity. fret. J. policy in the U.K. Im. J. Hlrh Sew. 4. 453-470. 1974.
P\xcho-AuulFsis 45, 49-56. 1964. 123. Gambling T. Socieral .Iccoautirq. Allen & Unwin.
97. Pulver S. Narcissism. The term and the concept. J. London. 1974.
,401. p\,Kltour7ul. A\% 18, 3 19-341, 1970. 124. Gambling T. Brxorul t/w Conrerlrio!ls o/ .4ccoamirlq.
9x. Kinston W. A theoretical and technical approach to Macmillan. London. 1978.
narcissistic disturbance. Iut. J. Psycho-Anulysis 61, 125. Churchman C. W. Prediction md Optimal Decr.\~or~:
3X3-394. 1980. Philo.sophicul 1.ssue.s o/ a Soewe of Valwh. Prentice-
YY. Kinston W. An intrapsychic developmental schema Hall. Engiewood Cliffs. NJ. 1961.
for narctsscstic disturbance. Irlt. Rec..Psycho-Analysis 126. AckoR R. L. A Corxent of Corporutr Phr~~w. Wilev.
9, 253-161. 1982. New York. 1970.
100. Kinston W. The positive therapeutic reaction. Paper 127. Ackoff R. L. Redesiyrling rhr Future. Wile!. New
presented to the British Association of Psychothera- York. 1974.
plsts, London. 1982. 128 Ackoff R. L. Optimization + objectivity = Opt Out.
101. Klein R. The strategy behind the Jenkin non-strategy. Eur. J. Opl. Rrs. 1, l-7, 1977.
Br. met/. J. 282. 10X9-1091. 1981. 129 Ackoff R. L. The future of operational research is
102. Armstrong D. Pathological life and death: medical past. J. Opl. Res. Sot. 30, 93-104. 1979.
spatialisation and geriatrics. Sot. Sci. Med. 15A. 130. Ackoff R. L. Resurrecting the future of operational
753-257. 19x I. research. J. Opl. RVS. Sot. 30. IXY~IYY. IY7Y.
IO?. Horkheimer M. Crihd Theory: Selected Essuys. 131. Rapoport noted that most researchers seem to wish to
Herder, New York. 1972. deny validity to many forms of useful research (Rapo-
104. Merton R. K. Sociul Theory cold Sociul Structure. Free port R. N. Three dilemmas of actlon research. H~rrn.
Press, New York. 1957. Rel. 23, 49Y-513, 1970). The documentation of devas-
105. Davis M. S. That’s Interesting! Towards a phenome- tating condemnations of work at a different spistemo-
nology of sociology and a s&iology of phknomeno- logical level to the critic’s own has not been Included
logy. Philos. Sot. Sci. I. 309-344, 197 I. in this paper. It has been assumed that most readers
106. PLyion W. (1979) Ends. means and achievement in have been on the g.iving and or recevlng end at some
medical research. Lrrtlcrr ii. 5 12-5 16. 1979. time. Table 2 provides some indication of the sorts of
107. Wing J. K. and Brown G. f/isrirutio/ialisn7 urld Schizo- criticisms that tend to be repeatedly made.
pltret~itr: A Cowpurtrriw Study of’ The Mrnrul Hospi- 132 Kinston W. Modelling measurement In the social
 Pluralism in the organisation of health services research 313

(and natural) sciences. Document No. 3281. BIOS. 143. Brown has come to a similar conclusion in a recent
Brunel University. Usbridge. Middlesex. 1981. paper. Brown V. A. From sickness to health: an
133 Glass B. The ethical basis of science. Scierwc 150, altered focus for health-care research. Sec. Ser. Med.
12541261. 1965. 15A, 195-201, 1981.
134 Ziman J. M. Public Krmcledyr: ,411 E.s.say Concerning 144. “When someone is honestly 553~ right, that’s very
Sociul Dirmwsiorn of Scicwcr. Cambridge University good and there’s no use wrangling. And if someone is
Press. New York. 196X. 60”/;, right, it’s wonderful, it’s great luck, and let him
13: Oquist P. The epistemology of action research. Acfa. thank God. But what’s to be said about 75Y;? Wise
Social. 21, 1433163. 1978. people say this is suspicious. Well and what about
136 Argyris C. op. cir. IOOU,;,right? Whoever says he’s lOO”/, right is a fanatic,
137 Lewin K. Frontiers in group dynamics. 1. Concept, a thug and the worst kind of rascal”. An Old Jew of
method and reality in social sciences: social equilibria Galicia. Quoted in Milosz C. The Cuptiue Mind.
and social change. 2. Channels of group life: social Seeker & Warburg, New York, 1953.
planning and action research. Hum. Rel. 1, 5-41, 145. Waddington C. H. Towards a Theoreticul Biology. 3
143-153. 1947. vol. Aldine, New York, 1969, 1970, 1971.
138. Chein I.. Cook S. and Harding J. The field of action 146. Much criticism of social scientific method seems not
research. Am. Ps~chol. 3, 43-50, 1948. so much concerned with the nature of knowledge as
139. Susman G. I. and Evered R. D. An assessment of the with implicit assumptions about the nature of human
scientific merits of action research. Admin. Sci. Q. 23, beings or social life (Lessnoff M. The Strucrurr of
582-603, 1978. Social Science. Allen & Unwin. London. 19741. It is
140 Researchers often appeal to philosophers for justifica- irrelevant however to argue that a method which
tion in a level IV way i.e. using those that support treats people as objects is wrong, if, for the purpose of
their contentions. However, because this paper is not the study, it is right. This is because all research
written in such a spirit, and I have no expertise in the under-dimensions reality. Such a method may, how-
discipline, it seemed best to eschew philosophical .ever, be judged unethical or politically unwise. The
references altogether. current state of the ethical and social regulation of
141 Blackburn T. R. Sensuous-intellectual complementar- science contributes to these confusions.
ity in science. Science 172, 1003-1007, 1971. 147. Cowan T. A. Paradoxes of science administration,
142 Cherns A. B. Social science research and its diffusion. Science 177, 964-966, 1972.
Hum. Rel. 22, 210-218, 1969.