British JournulofSociul P s ~ ~ ~ h o b ~ . v ( 120.

249-258
981), Printed in Griwr Brituin 249

The development of social psychology: Consensus, theory and methodology

in the British Journal of Social and Clinical Psychology
Jonathan Potter

Three aspects ofthe development ofsocial psychology were examined by studyingchanges in the
British Journal ofsocial and Clinical Psychology. An analysis ofthe use of‘justificatory statements’
indicated some decrease in consensus among social psychologists. Data suggested that social
psychologists made more use oftheories in the 1970s and the significance ofthis change is discussed.
An analysis of methodology indicated that laboratory experimentation increasingly predominates,
with little sign ofany increase in methodological pluralism. On a more general level, the poor
integration of theory and data in the literature concerning the development of social psychology is
commented upon.

The 1970s were a decade when many social psychologists not only practised social
psychology but also took part in a lot ofdiscussion concerning that practice. They argued
about the way the discipline is developing, what it is, how it could be improved and
sometimes how this development might be impeded. The discussion of these often
interrelated questions is represented in a variety of forms in the literature of social
psychology. Taken together these comprise an important body of knowledge and
conjecture concerning the discipline. However, each of these areas involves, in some more
or less explicit fashion, questions about the development ofsocial psychology as a science,
and on this point they embody some serious shortcomings. Before going on to discuss the
investigation that is the subject of this paper each of these areas of literature will be briefly
discussed, attempting to make these shortcomings more explicit.
(I) Citation analyses ofauthors and departments and analyses ofthe contents ofjournals
have become relatively commonplace (Higbee & Wells, 1972; Garfield, 1976; Rushton &
Endler, 1979; are typical examples). This type ofarticle is often found in the Personality
and Social Psychology Bulletin and the ‘house’journals: American Psychologist and the
Bulletin of The British Psychological Society. Exactly how such articles function in the .
scientific field is not clear. It is difficult to see how this type of publication could usefully
inform, say, editorial policy when they present no articulated theory of scientific quality.
In spite of this certain of these authors are confident that this type of citation analysis does
provide ‘an objective assessment of the quality of work ofboth individual faculty members
and departments’ (Rushton & Endler, 1979). Their justification of this, however, takes no
account of the wide-ranging contradictory literature summarized by Edge (1979). When
presented in their usual form of league tables and trends in research areas this work seems
perhaps better suited to provoking rivalry and one-upmanship than assisting constructive
policy decisions. Such data must be parasitic on the implicit understandings scientists
have of their practice and these understandings have been seriously undermined in the last
few years (see Collins, 1974; Bloor, 1976; Latour & Woolgar, 1979; Mulkay, 1979). This
situation is clearly in conflict with the often avowed desire ofsocial psychologists to
improve on the beliefs and concepts of ‘the man on the Clapham omnibus’, or in this case
the social psychologist’s own folk knowledge.
(2) The literature ofthe social psychology ofthe social psychology experiment has
mushroomed since the end of the 1960s (e.g. Rosenthal & Rosnow, 1969). However the
issues that were originally substantive for social psychology have gradually given way to an
0144-6665/81/040249-10 $02.00/0 @I981 T h e British Psychological Society
250 Jonarhan Potrer
‘academic’ interest in the way experimental subjects construe their role in that particular
situation (Innes, 1980).Some insights have been incorporated into experimental
procedures, although it is not clear how far it is sufficient -at least for publication -
simply to show some awareness of these problems. These studies have provoked little
theoretical debate about the status of such processes in relation to the enterprise of social
psychology. Two recent works have, none the less, tried to extrapolate from research on
the social psychology of the experiment to the practice of social psychology. Barber (1979)
sees this research as picking out the biases in social psychological experimentation which,
as they are made visible, can be removed. Silverman (1977),on the other hand, suggests
that these studies show that we should put less stress on laboratory experimentation and
take up more observational studies, or perhaps attempt to do more field experimentation.
This leaves a problem ofdeciding between two competing ways of interpreting the same
evidence. A multitude of factors would ofcourse be relevant to deciding between these
two interpretations, but there is a particular need for an explicit model of the social
aspects of scientific practice to give meaning to such evidence and to enable decisions to
be made in an informed manner.
(3) There has been a considerable amount of conceptual and metatheoretical discussion
since the 1960s. Such works as Harrb & Secord (1972) explored the philosophical
presuppositions of social psychology. Moscovici (1972)bemoaned the misplacement of
theory by social psychologists. Others have concentrated on the ‘crisis’, or lack of it (Elms,
1975; Westland, 1978).Clearly these are not straightforward empirical questions; however,
their answers often carry certain empirical assumptions about the development and state
ofsocial psychology. For example, Moscovici’s argument rests in part on an empirical
assumption that social psychology is actually lacking in theories ofa certain sort. This
kind of empirical claim is not easy to test because of the complex circularities involved in
using certain scientific procedures to investigate other scientific procedures, as Weimer
(1974) has well illustrated. However, such testing will be dependent on some theory of
what theories are and how they function.
(4) More recently work has been done which attempts to stand back and look at social
psychology in its wider disciplinary context (e.g. Israel, 1979; Lubeck, 1979).Just one
example will be discussed. Stroebe (1980)has come close to the approach which is here
advocated towards metatheoretical issues. He attempted to integrate social psychological
insights with theories from the philosophy and sociology of science, and hence to clarify
the complex interplay oftheory, data and scientific community which contributes to the
development of social psychology. Unfortunately he restricted his philosophical usage to
Popper and drew upon only one perspective on the sociology of science, the American
tradition of Merton (1957)and Hagstrom (1965).Thus he drew upon work which is based
on the assumption that science is a uniquely rational activity on which social processes
impinge only in a supporting role, or occasionally to produce false scientific beliefs. This
restricted choice of Stroebe’s seems hard to justify, since it is exactly the role ofsocial
processes in science he has questioned. In fact it allows him to ignore the opposed
conclusions that he would be led to by an alternative Kuhnian analysis. Stroebe argues
(following Popper, 1963)that progress is dependent on the existence ofa number of
competing subgroups of scientists. Thus as social psychology in Stroebe’s view lacks these
subgroups it has only been slow to develop. However if we substitute Kuhn’s theory -at
least in its early, strong version (Kuhn, 1962)-for Popper’s we would expect progress to
be impeded by competing subgroups, for it would only be during the consensus of‘normal
science’ that there would be broad agreement as to what should be counted as progress.
And for the early Kuhn progress is gauged only by what the participants recognize as
progress.
 The development ofsocial psychology 251
Hopefully this discussion has illustrated some of the main shortcomings of the four
research areas overviewed: that data have been inadequately integrated with the growing
body of theory concerning the role of social processes in science; and that, conversely,
metatheoretical discussion has oAen rested on dubious empirical generalizations.

The developmentof social psychology

The study described in this paper was concerned with the nature of social processes in the
discipline of social psychology. It attempted to avoid the pitfalls outlined above by
addressing theories ofscientific activity, and claims about the nature ofsocial psychology,
with data from a detailed content analysis of the British JournalofSocial and Clinical
Psychology. Three main issues were confronted:
(1) Questions concerning the degree of consensus in social psychology were examined
with the aid ofsome theoretical suggestions made by Spiegel-Rosing (1977). The aim was
to assess whether social psychology is becoming more or less consensual, through an
analysis of ‘justificatory statements’ in social psychology articles.
(2) Various changes in the use of theory in the articles between the 1960s and the 1970s
were examined and interpreted with the aid of a strongly social model of the role of social
processes in the development ofscientific disciplines (Gilbert, 1977).
(3) Claims, both stipulative and empirical, about changes in methodology in social
psychology were examined by an analysis of the kinds of methodology employed in the
articles. Data were used to clarify a debate about the contribution of field experimentation
to the so-called ‘new look’ in social psychology (Fried el al., 1973; Market al., 1976). In
addition the centrality of laboratory experimentation in American and British,research
was examined and the question of methodological pluralism was discussed.

Method
All social psychology articles in the British Journal of Social and Clinical Psychology (BJSCP)
between 1962 (when the journal was first published) and 1978 were content analysed, giving a total of
406 articles. Three groups ofcategories were used in the analysis: ‘justifications’, ‘theory’ and
‘methodology’. The categories that make up these groups were arrived at by a process ofcoding and
then recoding a large number ofarticles to obtain a satisfactory fit between the categories and the
material. Articles of a clinical nature were excluded by using a criterion of reference to therapy,
psychiatry or hospitalization. Short articles and reply articles were accepted for the analysis whilst
book reviews and editorials were excluded. Journal issues were analysed randomly with respect to
decades, to eliminate any progressive coding bias. Four volumes (89 articles) were recoded by
another rater to give a check on reliability.

Generation and definition oJcategories

(I) Justifications. The category system for the analysis of ‘justificatorystatements’ or
‘self-legitimating statements’ followed that of Spiegel-Rosing (1977). It focused on statements in the
text of the articles which could be interpreted as explicit claims of the relevance or importance of the
articles. These statements were generally found in the introduction, although the discussion sections
were also analysed in this way. The category of‘theoretical relevance’, for example, included all
those articles which explicitly stated that they were about theory: testing it, corroborating it,
demonstrating inconsistencies with it, and so on; for example ‘X’s theory can be usefully applied to
behaviour Y’.‘Social and practical relevance’ was a category which covered statements ofthe kind:
‘this could have consequences for the sorts of negotiation that takes place between unions and
management’. ‘Hitherto neglected’ covered statements like ‘few studies have attumpted to relate X
and Y’.As much as possible, the category labels were meant to be self-explanatory. The data were
analysed to show the percentage of articles in each decade using each justificatory statement, and the
percentage of articles using no justificatory statements. Justifications were coded by occurrence, thus
allowing several different statements to be coded for any particular article.
252 Jonathan Potter
(2) Use oftheory. Using the suggestionsof Madsen (1970), Hesse (1974) and others four features were
taken as basic to any theory: (a) it would have some structure of laws or implications; (6)these laws
or implications would be in some way general or universal; (c) it would involve hypothetical or
unobservable terms (usually making reference to a model ofsome sort); and (4 specific hypotheses
may be generated from it. Producing a criteria1 definition ofa term like theory is extremely dificult
(Sellars, 1961) and this present formulation is deliberately broader than that of Deutsch & Krauss
(1965),which is more restrictively positivistic.
The categories were defined as follows:
I . Generalperspective: papers explicitly subsumed under the terms behaviourism, gestalt theory,
personal construct theory, Piagetianism, psychoanalysis, role theory, social learning theory,
symbolic interactionism.
2. Aside theory: reference was made to a theory or theories not directly used in the article.
3. Explicit theory: the term ‘theory’ was used and it satisfied the criterion.
4. Nominaltheory: the term ‘theory’ was used and it did not satisfy the criterion.
5 . Tacit theory: the term theory was not used but the criterion was satisfied.
6 . 7 and 8. These categories were as 3,4 and 5 but with explicit reference made to testing the
theory.
9. Hypothesis: the term ‘hypothesis’ was used.
10. Hypothesis test: the term hypothesis test was used.
I 1. Hypothesisfiom theory- test: the term ‘hypothesis test’ was used where the hypothesis was
explicitly derived from a theory (3.4 or 5).
Each category was ‘all or none’ - irrespective of the number of occurrences in a category, they can
only score 1 or 0 for each article. Not all the categories were mutually exclusive: 6.7 and 8 were
always a subset of 3.4 and 5 respectively; category 10 was a subset ofcategory 9; and category I1 was
a subset of 10 and, in turn, 9. Category 2 was, however, exclusive ofcategories 3 , 4 and 5.

f3) Methodological changes. Following Canter (in preparation) distinctions were made between four
research strategies and three forms ofdata. The four strategies were as follows:
( a ) Experimental simulations. These take place in artificial settings, and involve manipulation of
levels of variables. Data analysis concentrates on differences between groups.
( h ) Field experiments. These take place in natural settings, and involve either manipulations or
naturally occurring effects.
(c) Relational studies. These measure relationships between variables, rather than differences, and
may take place in‘either natural or artificial settings.
(d)Case studies. These focus on a particular existingcontext, and have a descriptive emphasis.
These four research strategies were subdivided according to the form in which the data were
anal ysed.
( a ) Loosely structured. Where the researcher assigns the categories ofanalysis afler collection of
data.
(b) Structured. Where the researcher assigns the categories ofanalysis before collection ofdata.
(c) Performances. Where the data generated are taken to be the degree of behaviour of interest per se
(e.g. illness frequencies where the study is concerned with the frequency ofillness itself).
Clearly in much social psychological research some combination of forms ofanalysis is used. For
example social psychology experiments may involve some sort ofcombination ofstructured and
performance data. In such cases each is scored 0.5, so that any one article can only score 1.0 for
methodology. Further explication of this scheme for characterizing data is given in Canter (in
preparation).

Results
T h e total number of articles analysed was 406 of which 129 were from the 1960s (volumes
I to 8) and 267 were from the 1970s (volumes 9 t o 17). The results are presented in the form
of percentage occurrences in each category per decade -see Tables 1 to 4.

Discussion
(1) Consensus in socialpsychology
Spiegel-Rosing (l977), in a comparative analysis of the use ofjustificatory statements in
different scientific disciplines, suggested that the figures for the absence ofjustificatory
 ?‘he dm4ipmenr ofsocial psychology 253
statements reflect, in part, differences in consensus, i.e. she claimed that the frequency of
the absence ofjustificatory statements might be used as a measure of how closely the
participants share a set of presuppositions about their scientific activities. Her argument
was that, if there is general agreement about the major problems, theories, methods and
directions of development, then statements of legitimation -which make reference to
such things -will be felt to be redundant, and therefore less utilized. To back up this
suggestion she analysed justificatory statements in various disciplines. She found that in
human genetics -a ‘hard’ experimental science -24 per cent of the articles made no
self-legitimating statements, whilst in anthropology the equivalent figure was 16 per cent
and in the social study ofscience it was 8 per cent. Thus she argued that as the disciplines
became increasingly social and historical consensus was reduced and there was more need
of explicit legitimation.

Table 1. Percentage occurrence ofjustificatory

statements
1960s 1970s
(OI’) (‘/n)

Theoretical rclevance 12.9 20.2

Cross-fertilization 0.0 1.1
Relevance to future research 2.9 13.5
Methodological relevance 18.7 13.9
Social or practical relevance 10.1 10.5
Critique or rectify 12.2 19.5
Solve contradictions, inconsistencies 8.6 8.2
Examine, explain or overview 6.5 3.4
Hitherto neglected 9.3 11.6
Persistence as an issue 3.6 4.5
Further evidence or understanding 15.1 18.7
Importance of object investigated 5.0 4.1
Specific case ofbig importance 2.2 2.2
Newness or originality 0-7 1.1
Not classifiable 00 0.7
No justification given 25.2 12.7
Inter-rater reliability = 81 per cent.

If we compare the quantity of articles nof usingjustificatory statements in the BJSCP

with Spiegel-Rosing’s figures, the percentage is 25.2 per cent in the 1960s but drops to 12-7
per cent in the 1970s (see Table I). Thus it would appear that the articles published in the
BJSCP have moved closer to the low consensus end of Spiegel-Rosing’s continuum in the
1970s. If the BJSCP is considered to be a journal representative ofsocial psychology, this
would imply that there has been a reduction in the consensual state of the discipline in the
last two decades.
Unfortunately there are difficulties that must be overcome before this figure can be
accepted as a genuine measure of consensus. Firstly, consensus is not clearly distinguished
from ‘hardness’ in Spiegel-Rosing’s analysis, i.e. it could be that the differences in
justifications found were due to some other difference between ‘hard’ and ‘soft’science
rather than anything specific to consensus. Secondly, Spiegel-Rosing herself notes that
differences in justifications could be partly due to differences in quality ofthe articles-
better articles giving more explicit justification. Thus the figures for the BJSCP might be
partly due to an improvement in the quality of the published articles in the 1970s, perhaps
due to more stringent editing.
254 Jonaihan P o ~ a r
(2) Theory in social psychology
Moscovici (1972), and Kiesler & Lucke (1976), among otheis, have suggested that social
psychology has developed through the accumulation of large amounts ofdata, but with
very little construction and testing of theories. They attacked what they claimed was a
lack ofproper scientific theory in social psychology; to Moscovici it had only
‘protoscientificconceptualizations’,not genuine theories.

Table 2. Percentage occurrence of ‘theory’ and

‘hypothesis’

I. General perspective 8.6 6.4

2. Aside theory 7.2 9.4
3. Explicit theory 9.3 21.3
4. Nominal theory 4.3 3.4
5. Tacit theory 4.3 5.2
6. Explicit theory-test I .4 5.2
7. Nominal theory-test I .4 0.4
8. Actual t heory-test 0.0 1.9
9. Hypothesis 32,4 35.2
10. Hypothesis-test 24.5 25.1
11. Hypothesis from t heory-test 3.6 7.5
12. Categories 3 + 5 13.6 26.5
13. Categories 6 + 8 + I I 5.0 14.6
Inter-rater reliability= 83 per cent.

The data presented in Table 2 appear to support the claim that social psychologists are
rarely concerned with theoretical problems and problems of theory. Of the 406 articles
analysed an average of one in five used theory in a way that satisfied the criterion
(categories 3 and 5 averaged across decades) and an average of one in 10 articles claimed to
be testing a theory, or a hypothesis derived from a theory (categories 6,8 and 11 averaged
across decades). Thus it seems fair to describe the majority ofthe articles in the BJSCP as
atheoretical and, in so far as this journal is representative, this will support Moscovici’s
and Kiesler & Lucke’s claim that social psychology lacks theory.
However if the change in these categories is analysed across decades, a !arge difference
becomes visible: the use of theory has almost doubled between the 60s and the 70s - 13.6
to 26.5 per cent (categories 3 and 5) -and theory testing has almost tripled -5.0 to 14.6
per cent (categories 6,8 and 11). The lack of change in the categories of ‘hypothesis’and
‘hypothesistest’- 32.4 to 35.2 per cent and 24.5 to 25.1 per cent respectively -seems to
indicate that the change is specific to the use of theory and testing, rather than to a broad
shift in the nature of the articles. There is thus strong evidence that articles were becoming
increasingly theoretically orientated during the 1970s.
Other evidence however suggests that the situation may be rather more complicated
than this conclusion implies. Gilbert (1977) has suggested that one ofthe main functions of
journal articles is to persuade other scientists that the findings and ciaims within them
ought to be ratified as knowledge. Therefore scientific reports will normally use various
sorts of persuasive resources: for example ratified methods, novel findings, authoritative
references, clear style, prestigious institutional affiliation and established reputation.
Without effective persuasion, Gilbert argued, scientists risk that their findings will not
 The developmenl of social psychology 25 5
become transformed into ‘scientific knowledge’; instead they may be lost with the vast
majority of articles which receive one or no citation (Cole & Cole, 1973).
If Gilbert’s argument is correct it would be plausible to suggest that theory has become
an element in the persuasive resources that psychologists normally accepted during the
1970s. This could be because Moscovici’s and like views became more generally accepted;
or it might be that social psychologists see an increased use of theory as a step towards
gaining more scientific status. The increase in the justificatory category ‘theoretical
relevance’ from 12.9 to 20.2 per cent (Table 1) seems to support this suggestion. Ifthis
were the case it might be expected that, as psychologists utilized this resource, there would
be an increase in the categories of theory and theory testing- both where theory is
actually used and where it is only nominally used.
The lack ofchange in the ‘nominal theory’ category seems to contradict this suggestion.
However some other evidence supports it. The criterion used for deciding between
nominal and actual use of theories was very weak, and many of the articles which met it as
‘theoretical’ appeared to use little more than small-scale empirical generalizations. The
figures for changing theory use may thus hide a more complex change, where there is both
increased substantial use of theory and an increase in marginally theoretical studies, being
‘packaged’ to display theoretical significance. The category of ‘nominal theory’ use may
well have been too insensitive to register the differences between these two sorts of studies.
Ofcourse it would need further analysis, probably with a subtler technique than content
analysis, to establish this with more certainty.

(3) Methodological changes in social psychology

(a)Field experimentation. The data from this study serve to throw some light onto a
debate concerning field experimentation and the so-called ‘new look’ in social
psychology. Fried et al. (1973) have argued that, despite Smith’s (1972) claim that there is a
‘new look’ in social psychology, emphasizing the study of phenomena in the field as well
as in the laboratory, the evidence does little to indicate its existence. Whilst they found a
moderate increase (rho = + 0.64) in the proportion of field experiments in the Journal of
Abnormal and Social Psychology (JASP), which later became the Journal oJPersonality
and Social Psychology (JPSP), there was a moderate decrease (rho = - 0.54) in the Journal
uJSocial Psychology (JSP). They also noted that in the period studied - 1961 to 1970 -
the proportion of field studies was very small (96 out of 2880 articles or roughly 3.3
per cent).
In contrast Mark et al. (1976), analysing a more recent sample - 1961 to 1974 -and a
new journal (the Journal ofApplied Social Psychology- JApSP), have claimed that there
has been an increase in field experimentation, although this increase was highest in the
area ofapplied research. If Mark et d ’ s figures are averaged together the number of field
experiments can be seen to increase from 2.2 per cent in the 1960s (JASP/JPSP and JSP) to
7.5 per cent in the 1970s (JPSF and JSP). For the JApSP they mentioned that the range for
the years studied (1971 to 1974) was between 17 and 33 per cent.
The data from the present study do not reflect the increase in field experimentation
claimed by Mark el al. In the 1960s field experiments accounted for 4.3 per cent of articles
in the BJSCP, and this figure fell to 1.9 per cent in the 1970s. Thus ifthe three journals
concentrating on basic research -the JPSP, JSP and BJSCP -are taken together, they
support Fried el al.’sconclusion rather than Mark et al.’s. Much of the disagreement
voiced by Mark el al. seems to stem from their inclusion ofdata from the applied journal,
and these data seem to reflect the relevance of field experimentation to applied research,
rather than any ‘new look’ in social psychology.
256 Jonathan Potter
(b)Centrality of laboratory experimentation. The number of laboratory experiments
published in the BJSCP rose to 50 per cent in the 1970s from a 1960s figure of 33-7 per cent
(Table 3). This is similar to the increase seen in the American journals JASPIJPSP and
JSP. If Fried et al.’s data are combined with those of Levine (1978), we arrive at a 1960s
figure of 53.1 per cent for laboratory experiments, which increases to 64 per cent in the
-
1970s (JPSP in 1973 only and JSP in 1972173 only Levine, 1976). In contrast, the
percentage of relational studies in the BJSCP dropped from 53.8 to 41.1 per cent. Taking
the nearly identical category of‘survey research’ in Fried et al. and Levine’s data, a similar
drop from 41.9 to 22.5 per cent may be seen between decades. Thus, in both the British
and the American journals, there has been an increase in the proportion of experimental
simulations, and a correspondingdecline in relational studies. The American journals are,
however, the more solidly experimental.
Table 3. Percentage occurrence of different
research strategies
1960s 1970s
(%I (%)

Experimental simulations 33.7 50.0

Field experiments 4.3 I *9
Relational studies 53.8 41.1
Case studies 0.0 0.0
Inter-rater reliability for methodology categories = 91 per
cent.

(c)Methodological pluralism. The recent social psychology textbooks by Tajfel & Fraser
(1978) and Eiser (1980) have suggested that methodological pluralism is the appropriate
course for the healthy development ofsocial psychology. Tajfel & Fraser in fact suggested
that some sort ofcyclical movement from the laboratory to the field and then back would
be ideal. In practice however, as was noted above, there seems to have been a move
towards a predomination of laboratory experimentation over other methods. The number
of field experiments in the BJSCP was small, with little sign of increase, while there were
no case studies recorded in the entire period studied. Both of these types of research would
be an important part of a methodologically pluralistic social psychology.
Another methodological style rarely found in the BJSCP was that advocated by, for
example, Brenner et al. (1978) and Marsh et al. (1978). This style of methodology would
appear in the category ‘loose structure’, which refers to those studies in which the analytic
categories are generated after the researcher has collected data; for example when the
participant’s accounts are analysed in terms of a repertoire of social rules. In the BJSCP
only 2.1 per cent of the articles came into this category in the 1960s and this fell to 1.3 per
cent in the 1970s (Table 4). This again suggests that social psychology has not yet
developed any substantial methodological pluralism.
Table 4. Percentage occurrence
ofdifferent data forms
1960s 1970s
(%) (%)

Loose structure 2.1 1.3

Structured 83.4 86.1
Performances 6.3 4.8
 The development qfsocial psychology 251
Conclusion
Apart from the specific points made in the discussion, one more general conclusion may
be drawn from the study. Data produced by analyses such as the present one are extremely
difficultto interpret in the absence ofboth a theory of reading scientific texts, and ofthe
relationship between scientific texts and scientific activity. Content analytic methods by
themselves are unlikely to be sufficient for the formulation ofsuch theories, because
theories of this sort are presupposed in any content analysis ofscientifictexts.
Some steps towards the formulation of theories of this kind have been undertaken. For
example Gilbert & Mulkay (1980) have compared a group of scientists’ informal accounts
of their research with the accounts found in their published papers using a detailed
analysis of their scientific language. They found striking and systematic dissimilarities
between the two accounts, for example in the logical and chronological priority of theory
and experimental data. The authors suggest that these differences indicate that the
scientistsare using two separate accounting systems to characterize their activity,
corresponding to two concepts of scientific knowledge. The formal account follows
traditional empiricism, whilst the informal account is intimately bound up with social
and particularistic events in the process ofarriving at data and evaluating theory. This
notion of accounting repertoires has also been used in an examination of the application
of social psychology (Potter, in preparation). Here the role of a stereotypical accounting
system in recharacterizingsocial intervention as a product ofsocial theory is discusied. It
is thus suggested that a more fine grain and systematicanalysis ofscientifictexts and
discourse is likely to lead to advances in the social psychology ofscientific knowledge.
However these analyses are, at present, only preliminary, and if we are to understand the
development of social psychology more fully, we need more of the methods available to
social psychology to be turned back on social psychology itself.
Acknowledgements
I am very grateful t o Margie Wetherell, Peter Stringer and Mike Mulkay for making helpful
and constructive comments on an earlier draft ofthis paper, and to two anonymous reviewers.

References
Barber, T. X. (1978). Pirfalls in Human Research: 10 Pivotal Points. New York: Pergamon.
Bloor. D. (1976). Knowledge and Social Imagery. London: Routledge & Kegan Paul.
Brenner, M., Marsh, P. & Brenner. M. (eds) (1978). The Social Context of Method. London: Croom Helm.
Canter, D. Strategies and Tacticsfor Socialand dehavioural Research. Chichester: Wiley (in preparation).
Cole, J. R. & Cole. S.(1973).SocialStrati/icatron in Science. Chicago and London: University ofChicago Press.
Collins, H. M. (1974). The TEA set: Tacit knowledge and scientific networks. Science Studies, 4,165-185.
Deutsch, M. & Krauss, R. M.(1965). Theories in SocialPsychology. Glencoe, IL Free Press.
Edge, D. (1979). Quantitative measures ofcommunication in science: A critical review. History ofscience. 17,
102-134.
Eiser, J. R. (1980). Cognitive Social Psychology: A Guidebook to Theory and Research. London: McGraw-Hill.
Elms, A. C. (1975). The crisis ofconfidence in social psychology. American Psychologist. 30,967-976.
Fried, S . B., Gumpper, D. C. & Allen, J. C. (1973). Ten years ofsocial psychology: I s there a growing
commitment to field research? American Psychologisr. 28,155-156.
Garfield. E. (1976). Social Science Citation Index Clusters. Current Contents, 27.5-11.
Gilbert, G. N. (1977). Referencing as persuasion. SocialStudies ofscience. 7.113-122.
Gilbert. G . N. & Mulkay. M. (IY80). Contexts ofscientific discourses: Social accounting in experimental papers.
In K. D. Knorr, R. Krohn & R . Whitley (eds), The Social Process ofScienti/ic Investigation. Dordrecht: Reidel.
Hagstrom, W. D. (1965). TheScientific Community. New York: Basic Books.
Harrt, R.& Secord, P. F. (1972). The Explanation ofSocialEehaviour. Oxford: Blackwell.
Hesse, M. (1974). The Structure ofScienti$c Inference. London: Macmillan.
Higbee. K. L.& Wells, M. G. (1972). Some research trends in social psychology during the 1960s. American
Psychologist. 27;963-966.
Innes, J. M. (1980). Fashions in social psychology. In R. Gilmour & S. W. Duck (eds), The Development ofsocial
Psychology. London: Academic Press.
258 Jonathan Potter

Israel, J. (1979). From level ofaspiration to dissonance. In A. R. Buss (ed.), Psychology in Social Context. New
York: Irvington.
Kiesler, C. A. & Lucke, J. (1976). Some metatheoretical issues in social psychology. In L. H. Strickland, F. E.
Aboud & K. J. Gergen (eds),Social Psychology in Transition. New York: Plenum.
Kuhn, T. S. (1962). The Structure ofScientiJic Revolutions. Chicago and London: University ofChicago Press.
Latour, B. & Woolgar, S. (1979). Laboratory Lisp: The SocialConstruction ofScient$c Facts. London and
Beverly Hills: Sage.
Levine. N. (1976). On the metaphysics ofsocial psychology: A critical view. Human Relations. 29,385400.
Lubeck, 1. (1979). Aggression research. In A. R. Buss (4. ).
Psychology in SocialContext. New York: Irvington.
Madsen. K. B.(1970). The languagesofscience. Theoryand Decision, 1,138-154.
Mark, M. M., Cook, T. D. &Diamond, S. S. (1976). Fourteen yearsofsocial psychology: A growingcornmitment
to field experimentation. Personality and SocialPsychology Bulletin, 2,154-157.
Marsh, P., Rosser, E.& H a d , R. (1978). The Rules ofDisorder. London: Routledw & Kegan Paul.
Merton, R. K. (1957). Social Theoryand SocialStructure. New York: Free Press.
Moscovici, S.(1972). Society and theory in social psychology. In J. Israel & H. Tajfel (eds), The Context ofSocial
Psychology:A Critical Assessment. London: Academic Press.
Mulkay. M. (1979). Science andthe Sociology ofKnowledge. London: Allen & Unwin.
Popper, K. R. (1963). Conjectures and Refutations. London: Routledge & Kegan Paul.
Potter, J. A. Nothing so practical as agood theory: The problematic application ofsocial psychology. In P.
Stringer (ed.), Confionting Social Issues. Applications ofSocial Psychology. vol. 1. London: Academic Press
(in preparation).
Rosenthal, R. & Rosnow. R. L.(eds) (1969). Arri/act in Behavioural Research. New York Academic Press.
Rushton, J. P. & Endler, N. S. (1979). More to-do about citation counts in British psychology. Bulletin ofThe
British Psychological Society, 32.107-109.
Sellars, W. (1961). The language oftheories. In H. Feigl & M. Grover (eds), Current Issues in the Philosophy
ofScience. New York: Holt, Rinehart & Winston.
Silverman, 1. (1977). The Human Subject in the Psychological Laboratory. New York: Pergarnon.
Smith, M. B. (1972). Is experimental social psychology advancing?JournalofExperimentalSociulPsychology, 8,
86-96.
Spiegel-Rosing. 1. (1977). Science studies: Bibliometric and content analysis. SocialStudies ofScience. 7,97413.
Stroebe, W. (1980). Process loss in social psychology: Failure to exploit? In R. Gilmour & S.W. Duck (eds), The
Development ofSocial Psychology. London: Academic Press.
Tajfel, H. & Frazer, C. (eds)(1978). IntroducinK Social Psychology. London: Penguin.
Weimer, W. B. (1974). The history of psychology and its retrieval from historiography: 1. The problematic nature
of history. Science Studies. 4,235-258.
Westland, G. (1978). The Current Crises in Psychology. London: Heinemann.
Received 7 July 1980

Requests for reprints should be addressed to J. Potter, Department ofSociology, University of York.
Heslington, York YO1 5DD, UK.