do sociological theories grow weber

8/12/2019 Do Sociological Theories Grow Weber

http://slidepdf.com/reader/full/do-sociological-theories-grow-weber 1/33

Do Sociological Theories Grow?Author(s): David G. Wagner and Joseph BergerSource: American Journal of Sociology, Vol. 90, No. 4 (Jan., 1985), pp. 697-728Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/2779516 .

Accessed: 26/02/2014 20:18

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .

http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

.

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to

American Journal of Sociology.

http://www.jstor.org

This content downloaded from 128.123.35.41 on Wed, 26 Feb 2014 20:18:57 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=ucpress

http://www.jstor.org/stable/2779516?origin=JSTOR-pdf






http://www.jstor.org/stable/2779516?origin=JSTOR-pdf

http://www.jstor.org/action/showPublisher?publisherCode=ucpress



American Journal of Sociology

Theoretical change may very well be progressive in the former; it may not

be in the latter, for observation is assumed to be inherently more value

laden and subject matter more reactive in the social sciences than in thenatural sciences (see, e.g., Gouldner 1970; Winch 1958).

Interestingly, despite the strong differences of opinion about the possi-

bility of theoretical growth, sociologists generally agree about the reality.The consensus seems to be that little or no growth has occurred. Cer-

tainly, none of the sources we have cited sees sociological knowledge asprogressive to any significant degree. Many others concur (e.g., Bergner1981; Alexander 1982).

The fact that those who question even the possibility of theoretical

growth see little of it is, of course, not surprising. What is much moreintriguing is that those for whom theoretical growth is both possible and

desirable see it as such a rare phenomenon. Extolling the virtues of acumulative sociological science is exceedingly difficult with so little evi-

dence for its existence. It is as if half the points at issue have beenconceded before the debate has even been joined.

But should those points be conceded? Is there really so little growth?We think not. In fact, we believe there has been both extensive andimportant growth of theoretical knowledge in sociology.

Why then is there so little evidence? There are several reasons, all

resulting from assumptions sociologists mistakenly make regarding

theoretical work. First, they tend to treat all theoretical activity as having

the same character and intent. Actually, there are several different typesof theoretical activity: some involve growth; some do not. To see evidenceof theoretical growth, therefore, one must first adequately distinguishthese kinds of activity and the characteristics and intentions they involve.

Second, sociologists often assume that all theoretical growth is empir-

ical, that it involves primarily the relation between a theory and anincreasingly supportive body of relevant observations. In fact, theoretical

growth also involves the relation between one theory and another (e.g.,one of greater scope or precision). Theoretical context as well as empiricalcontext is vital to the search for evidence of theoretical development.

Finally, even when sociologists incorporate theoretical context in their

assessment of theoretical work, they tend to treat all theoretical contextsas identical. The assumption again is mistaken. Theories may be relatedto one another in several different ways. Furthermore, each way em-

bodies a distinct form of theoreticalgrowth. Thus, to understandfully thecharacter and extent of theoretical progressin sociology, one must appro-priately conceptualize and distinguish theoretical contexts and their re-

spective kinds of growth.In short, to see evidence of theoretical growth in sociology one must

698







Sociological Theories

first know where to look and what to look for. What is growing?In whatways?

Our thesis, then, is that growth of theoretical knowledge in sociology isnot rare but simply hidden by mistaken assumptions about theoreticalwork. We attempt to correct those assumptions and build a more satisfac-tory scheme for the analysis of theoretical growth. We begin with a

discussion of kinds of theoretical activity. For each activity we provideanswers to two kinds of questions. First, to what extent does theoreticalgrowth occur in that type of work? To what extent can it occur? Second,is theoretical context (i.e., the linkage between theories) represented inthat type of work? Can it be? Our answers to these questions help us

identify one of these activities-one that involves the development oftheoretical research programs-as most appropriate for the analysis of

theoretical growth. Then, through the analysis of actual cases in the two

following sections, we explicate the ways in which theoretical growth is

occurringin such programs. Finally, we consider brieflyhow these ideascan be used to promote the growth of theoretical knowledge in other

branches of sociology.

TYPES OF THEORETICALACTIVITYTheory in sociology has come to include many kinds of sociological

work, from commentaries on the classics to causal modeling. WhatParsons (1954) and Zetterberg (1965), for example, mean by theory areclearly different things.

Failure to distinguish such different types of theoretical enterprises, webelieve, seriously hinders any attempt to identify and understand theorygrowth in sociology; it simply becomes too easy to confuse one kind ofactivity with another. This is particularly evident in sociological use ofKuhn's ([1962] 1970) term paradigm. For example, both Ritzer (1975)and Leinhardt (1977) use the term. Ritzer is concerned with large, amor-phous frameworks for theoretical activity (e.g., the social factsparadigm), whereas Leinhardt is concerned with a much smaller, morehighly focused theoretical enterprise (the social networks paradigm).3 Itis highly unlikely that these enterprises involve theory growth to the sameextent or in the same way. Thus, various types of theory and theoretical

3 Eckberg and Hill (1979) argue that such usages of Kuhn'sconcept are altogethermisdirected.We agree basically; n sociology the paradigmconcepthas been stretchedalmost beyond recognition. Part of our effort here, therefore, is directed toward theidentificationof more specific and more appropriate onceptsfor the analysisof theoryand theory growth in sociology.

699








work must be distinguished if we wish to observe and understandtheoret-ical growth in sociology.

Orienting Strategies

Consider the very large proportion of sociological theory that is in theform of metatheory. It is discussion about theory: about what concepts itshould include, about how those concepts should be linked, and abouthow theory should be studied. Similar to Kuhn's paradigms, theories ofthis sort provide guidelines or strategies for understanding social phenom-ena and suggest the proper orientation of the theorist to these phenomena;

they are orienting strategies. Textbooks in theory frequently focus onorienting strategies such as functionalism, exchange, or ethnomethodol-ogy (see, e.g., Turner 1982).

Development of an orienting strategy may take any of several forms. Itmay include, for example, the development of ontological and epis-temological arguments (often metaphorically stated) concerning the sub-ject matter of sociology, the nature of social reality, and the values andgoals of sociological inquiry. It may also involve the articulation of theconceptual foundations employed in the description and analysis of social

phenomena. Finally, it is likely to incorporate the formulation of direc-tives for the selection of theoretical problems for investigation and for theconstruction and evaluation of proposed problem solutions.

For example, in the conflict orienting strategy it is common to assertthat the history of all hitherto existing society is the history of class

struggles (Marx [1848] 1977). Such a statement is primarilymetatheoret-ical in character. It suggests that class struggles should be regarded as aninherent feature of social reality, that these struggles should be regardedas important in explaining social phenomena, and that therefore such

struggles constitute an important part of the subject matter of sociology.

Similarly, the explication of the AGIL scheme in Parsons and Bales(1953)constitutes part of the conceptual scheme underlying the functionalorienting strategy. In identifying adaptation, goal attainment, integra-tion, and management of latent problems as survivalproblems or sys-tem requisites, Parsons and Bales suggest that all analyses of socialsystems should include specifications of social structuresthat meet theseneeds of the system.

Homans's ([1961] 1974) presentation of an exchange orienting strategyincludes theoretical directives that are explicitly reductionist. Thus rea-sonable solutions to sociological problems should be based ultimately on

psychological principles. Attempted solutions are considered successful tothe extent that they can be derived strictly from psychological principles.

A very large proportion of theoretical activity in sociology is meta-

700








theoretical. For several reasons, however, we do not find much growth atthis level. First, orienting strategies are exceptionally stable, sometimes

even rigid, in structure. Consider the functional strategy: its conceptualframework, its image of social reality, and its directives for the solution of

sociological problems have changed little from Malinowski (1944) andRadcliffe-Brown (1952) to Parsons and Merton (1967, 1968). Similarly,the ontological and epistemological underpinnings of the exchange strat-egy as described by Homans and Blau (1964) differ only slightly fromthose proposed earlier by Durkheim ([1893] 1964) and Mauss ([1925]1967).

Perhaps more important, one orienting strategy is seldom, if ever,

replaced by another. Established strategies may become more or lessdominant over time, but it is rare for a strategy to disappear entirely.

Certainlythe widely heralded decline in functionalist theorizing (see, e.g.,Gouldner 1970) has not eliminated the strategy as a tool for sociologicalanalysis. Nor has any other strategy achieved the disciplinary supremacy

functionalism was assumed to have. Furthermore, comparatively new

strategies such as ethnomethodology seldom replaceolder strategies;more

often they add to the list of metatheoretical options that are available in

sociological analysis. Even when attempts are made to generate new

strategies which adopt important elements of older strategies, the re-placements are often strongly resisted (see, e.g., Denzin's [1969] attemptto link elements of symbolic interactionism and ethnomethodology in asingle strategy and Zimmerman and Wieder's [1970] somewhat indignantresponse). In short, once a strategy is entrenched, it is extremely difficultto replace. Polarization and politicization of the debate are more likelyresults.

Part of the reason for the rigidity both within and among strategies isthat the differences between strategiesare, in general, fundamental. Basi-cally, the claims of an orienting strategy are directives; they are state-ments about values (e. g., the value of function as a conceptual tool), not

statements about facts (e.g., the specific function performed by a particu-lar institutional structure). Such prescriptive arguments are largelynonempirical, and conflicts between them are often unresolvable. For

example, no test could ever demonstratethat the AGIL scheme is or is notempirically true. The claim that goal attainment involves establishingpriorities among system goals and mobilizing resources to attain them

cannot be evaluated empirically in any way. It is true by definition.Similarly, one does not demonstrate the empirical truth or falsity of re-ductionist theoretical directives; one employs reductionist directives indemonstrating the truth or falsity of other ideas. Marx's statement fromThe Communist Manifesto is, of course, often treated as a matter ofcontingent fact. Presumably, one can test whether or not history is rife

701








model of the modern world system, Blau and Duncan's (1967) analysis ofthe American occupational structure, and Cook and Emerson's (1978)

theory of power in exchange networks-all of these are examples of unittheories.

Unit theories are stated sometimes formally, sometimes discursively. Ineither case the basic structure includes a set of concepts (at times sug-gested by an orienting strategy) and a set of assertions relating thoseconcepts to each other in an account of some sociological phenomenon.4Thus, Merton's (1968) account of social structure and anomie includesamong its basic arguments the idea that structuralinconsistency betweensocietal goals and the distribution of opportunities to acquire the legiti-

mate means of achieving those goals results in deviant behavior. Mertonthen uses a set of concepts to identify several types of deviant behavior,each defined in terms of the particular kind of inconsistency betweengoals and means.

Unit theoretical claims are empirically testable, directly or indirectly,and conflicts between them are frequently resolvable through appeal tofact or reason. Thus, while orienting strategiesprescribehow to constructand evaluate theories, unit theories are themselves the particulartheoret-ical constructions to be evaluated.

The bulk of activity at the level of unit theories involves empiricaltesting. To the extent that such testing provides support for our unittheories, growth or progress may be said to have occurred. Perhaps theclearest account of the manner in which growth may result from theempirical testing of unit theories is provided in Stinchcombe's chapter

The Logic of Scientific Inference in ConstructingSocial Theories (1968,pp. 15-56). Basically, (1) the more numerous the empirical consequencesof a theory that are supported by observations, (2) the greater the varietyof consequences supported, and (3) the more frequently those conse-quences are contradictory to the consequences of the most likely alterna-tive unit theories, the more highly developed is the unit theory.Growth, then, is a direct consequence of increasing empirical support.

Activity at the level of unit theories seems much more appropriate tothe discussion of theory development. Empirical support-the relation ofa theory to data-is an essential feature of all scientific knowledge, andincreasingempirical support in the senses outlined by Stinchcombe seemsto be a prominent characteristic of much that we call growth.

However, something is missing. To talkabout theories changing,growing, developing implies, we believe, that one must talk about theo-

ries changing, growing, developing with respect to other theories.

' See Gibbs(1972),Hage (1972),and Cohen(1980) orusefuldiscussionsof the specificfeaturesthat are (or shouldbe) exhibited in the structuresof unit theories.

703








Change, growth, and development are all relational terms. A theory thatis more precise is more precise than another theory. A theory that is

greater in scope is so relative to another theory. A theory that hasgreater empirical support has been compared with another theory.

Even an otherwise isolated theory is compared with the theory that thephenomenon it attempts to explain is statistically random.

Furthermore, many theoretical comparisons involve closely relatedtheories. A more precise or better supported theory often shares much ofits structure with its less precise or less well supported alternative. It isintended as a modification or adjustment of the less satisfactory theory,not as a wholesale replacement for its basic account of sociological phe-

nomena.These problems do not invalidate what we have learned about unit

theories from the theory construction literature. Rather, they requirethatwe extend the concerns that are dealt with in that literature. Even thoughwe can begin to talk seriously about theoretical progress in terms of unittheoreticalactivity, we cannot generate a full and accuratepicture of thatprogress at the unit theoretical level. Unit theoretical activity focusesalmost exclusively on theory-data linkages. To fully describe and under-stand theory growth it is necessary to look also at the relations between

theories, the theory-theory linkages.

Theoretical Research Programs

Theoretical activity in sociology occurs at yet a third level, one whichfocuses on the context of interrelated theories within which unit theoret-ical work occurs. Consider, for example, a unit theory that has recentlybeen assessed empirically. Suppose that the outcome of the assessment isnegative: the evidence does not support the theory. Is the theory thereforescrapped as worthless? More than likely it is not. Instead, some attempt ismade to identify and correct errors in the formulation of the theory. Iferrors can be found, a revised formulation (technically a new, but closelyrelated, unit theory) is then likely to be constructed and tested. If theassessment outcome is again negative, further revision and testing mayfollow. This sort of programmatic activity may continue indefinitely,although at some point the negative evidence may outweigh the worth offurther investigation, and the theory may be evaluated more or less per-

manently as false.Suppose that the outcome of the assessment is positive: the evidencedoes support the theory. Does theoretical activity cease? Again, morethan likely it does not. Instead, attempts are made to improve the theory,perhaps by making its predictions more exact or by broadening its scopeof application. The improved version (once more a closely related, but

704








technically new, unit theory) may then be tested and further revised.Here also such programmatic activity may continue indefinitely.

The collection of theories that emerges from this sort of activity consti-tutes a theoretical research program. Basically, a theoretical researchprogram is a set of interrelated theories, together with research relevant

to evaluating them. A program may also include a body of applied orintervention research that is grounded in the theories in the program.5Much theoretical activity in sociology occurs at the program level. Itincludes, for example, work on sentiment networks, justice and equity,status attainment, mobility, mental illness, and bargaining. In all cases, a

series of interrelated theories has emerged, along with research testing

(and sometimes applying ideas from) those theories.The program notion has become quite prominent in the philosophy of

science through the work of Imre Lakatos (1968, 1970, 1978)on scientificresearchprogrammes. Lakatos and his students have identifiedand ana-

lyzed programmatic theoretical activity in a number of scientific disci-plines, most notably in the physical sciences (see Howson 1976) and in

economics (see Latsis 1976). There has not, however, been much discus-sion of such activity in sociology. We have adapted the idea to describework on expectation states processes (see Berger, Conner, and Fisek1974; Berger, Wagner, and Zelditch 1985); aside from this material therehas been little systematic study of programmatic theoretical activity in

sociology.While we clearly are indebted to Lakatos for the program notion, our

use of it is not intended as a strict application of his ideas. Rather,working from his notion, we have adapted and modified it in identifyingspecific, ongoing cases of growth within sociology. The claims we makeabout different types of programsare developed most directly from inves-tigations of those cases.

Our primary claim is that, if we study the interrelations among unittheories in a theoretical research program, we see that the change,growth, or development among those theories is a multifaceted activity.That is, growth takes several different forms that are manifested in differ-ent types of theoretical relations between unit theories. If we are to under-stand the different forms of growth that occur within the context oftheoretical research programs, we must be able to distinguish the types ofrelations that may develop among unit theories in such programs.

What are the different types of relations among theories in a program?

5 By relevant we mean that the outcomeof researchactivity potentiallyconfirms,disconfirms, or suggests avenues of refinementfor theories in the program. By

grounded we mean that the appliedresearch s basedon, guidedby, or makesuse ofthe theories in the program.

705








TYPES OF THEORETICALRELATIONSAND TYPES OF GROWTH

We have identified five relations that may occur among theories in a

program; each represents a type of theoretical growth. Three of them-theory elaboration, theory proliferation, and theory competition-we de-scribe as basic. The other two-theory variation and theory integra-tion-we describe as special.

The three basic relations may occur in any of a large variety of theoret-ical circumstances and at most stages in the development of a program. In

addition, each may serve as the foundation on which an entire programcan be built, establishing a theoretical context within which other pat-terns of growth (represented by other relations) may occur.

The two special relations occur only in very specific theoretical circum-

stances and at certain (usually more advanced) stages in the developmentof a program. These relations usually appear only in the context of estab-lished patterns of growth founded on the basic relations. In such contexts

the special relations help us choose from among differing theoretical argu-ments and unify our knowledge. Thus, they provide important (and oftendramatic) evidence of progress toward the explanatory goals of theoreticalactivity.

Clearly, both kinds of relations are essential for theoretical growth.The distinction between basic and special relations consequently does not

signify any difference in the value or importanceof particular patterns of

growth; it reflects only differences in the kinds of circumstances in which

each pattern is likely to occur.We consider the basic relations first. Then, in the next section, we look

more closely at the ways in which entire programs may develop from the

basic relations and at how the special relations contribute in very impor-tant ways to that growth. In particular, we explicate three different ways

in which theoretical integration may be accomplished, depending on thekind of relation that exists between the integrated theories.

With these ideas one can begin to make sense of theoretical activity in

sociology from the standpoint of theory growth; that is, one can specifywhat it is that is growing and in what ways it is growing. To demonstrate

this claim we apply these ideas to a number of cases of programmatictheoretical activity in sociology. These cases meet several criteria. First,they are programsof active and ongoing concern in sociology. Artificiallyconstructed cases and cases from other disciplines may have value, but

the most informative cases are those that represent actual contemporarysociological work. Second, each program represents one of the majortypes of theoretical research programs we identify. There may be more

types and there almost certainly are more of one type than another.

Finally, the programs are all ones about which we have intimate know-

706








ledge that has been important in analyzing their development. As a con-sequence, all our cases come from within sociological social psychology,

though we firmly believe that our ideas apply equally well to otherbranches of sociology.

The Basic Relations

Theory elaboration.-Sometimes a new theory is used to make an older

theory more general or more specific. The new theory, T(2), has a struc-ture very similar to that of its predecessor, T(1). It addresses a similar

sociological problem or is applied to a similar base of empirical observa-

tions. T(2) is, however, in some sense more comprehensive, more precise,more rigorous, or has greater empirical support than T(1), from which it

was generated. Thus, the predictions of T(2) say more (i.e., they are

more comprehensive or precise) or fit better (they are better supported

empirically) than the predictions of T(1). Within their common explana-tory domain, the predictions of T(1) and T(2) may conflict only over a

small part-if any-of that domain.This sort of relation, involving increases in the scope, rigor, precision,

or empirical adequacy of a theory, is theory elaboration. Examplesabound. The graph-theoretical formalization of Heider's (1944, 1946)principles of cognitive organization in Cartwright and Harary (1956) is aform of elaboration. So too is the specification in Stinchcombe (1963) ofsome of the empirical consequences of Davis and Moore's (1945) func-tional theory of stratification. Dahrendorf's (1959) account of class con-flict through imperatively coordinated associations elaborates, at least insome ways, Marx's earlier account (see, e.g., Marx [1848] 1977).

Elaboration is the type of relation most sociologists consider represen-

tative of growth or development. Certainly the refinement of knowledgein established areas is common in sociology. Most often the elaboration isprimarily empirical in nature (i.e., each new theory is designed to fit an

established data base more closely). More direct theoretical elaborations(e.g., formalizations and changes in scope or precision) are perhaps some-what less common.6

Theory proliferation.-Consider now a situation in which ideas from

one theory are used to generate a theory concernedwith a new or different

sociological problem or data base. Again, the new theory, T(2), is similar

in structure to T(1). In this case, however, the predictions of the twotheories generallydo not overlap, because they apply to differentexplana-

6 See Fararo (1973) for an impressive sampling of theoretical elaborations throughformalization.

707








tory domains. In a sense, here too the predictions of T(2) say more or fit

better than those of T(1)-but only with respect to the newly considered

problem area. They may have little or nothing to say about the originalproblem area, where the predictions of T(1) remain appropriate.

We call this sort of relation theory proliferation. In this case T(2) ex-

pands the range of application of ideas about social phenomena beyondthe original domain of T(1). Theory proliferation is perhaps less commonin sociology than is theory elaboration. Still, there are many appropriate

examples. Proliferation describes Hannan and Freeman's (1977) applica-tion of ideas from population ecology (see esp. Hawley 1950;Levins 1968)to organizational phenomena. Proliferation also describes the relation

between Heider's (1944, 1946) theory of interpersonal sentiment and twosubsequent formulations. Heider's theory focused on the balance of one'ssentiments toward other actors. Festinger (1957) broadened the balance

principle to consider dissonance among interrelated cognitions. Secordand Backman (1961) extended the range of the balance principle to en-compass relations among various aspects of one's self-conception.

Although proliferation is seldom discussed systematically (either insociology or in the philosophy of science), the theoretical expansiveness itrepresents is a critical feature of theory development. For knowledge to

grow, we must both (1) refine our accounts of established problems and(2) reach out to account for new and different problems. Through prolif-

eration, knowledge can spread into such new domains.Theory competition.-In still other cases a new theory is generated in

an attempt to capture at least some of the explanatory domain of another

theory. In these situations the theoretical structure of T(2) is essentiallydissimilar to that of T(1). Here T(1) and T(2) are interrelatedtheoreticallyby virtue of their similarities in problemfocus or data base. In fact, T(2) is

often generated specifically in response to perceived inadequacies in T(1)'saccount of that problem or data base. The new theory, T(2), claims to

account for the same (ora similar) problem more satisfactorily. The majordifferences in theoretical structure, coupled with the similarities in focus

or data base, may lead to relatively large sets of conflicting predictionsover some portion of the explanatory domain of each theory. If the struc-

ture of T(2) is entirely different from that of T(1) and the explanatorydomains are identical, the predictions of the two theories could conceiv-ably conflict at every point. Both theories may claim to say more or to fit

better.This sort of relation, theory competition, is especially important in

sociology. It is the activity through which theory comparison and choiceare most frequently made. Proposing an entirely new or different set of

theoretical tools for dealing with an established problem, T(2) is intended

708








as a replacement for T(1).7 Proponents of each theory claim the other iswrong. Eventually one or the other theory may win out in the conflictand

become the basic vehicle for further development. A third theory mayalso emerge that is more successful than either of the original competitorsand therefore replaces both as a vehicle for further development.

Theory competition describes quite well the interrelation betweenScheff's (1966, 1974) labeling theory and Gove's (1970, 1975) psycho-physiological theory of mental illness. It also occurs in the interrelationbetween Davis and Moore (1945) and critics of their functional view of

stratification (see, e.g., Tumin 1953).8 Competition appears in the debateover the relative importance of power elites and veto groups in political

decision making. It is fundamental to the conflict between genetic andenvironmental accounts of the bases of intelligence.

Competition is a particularly complex form of theoretical growth. Thestructures of the competing theories may be based on the directives ofdifferent orienting strategies. Consequently, conflicts may not be limitedto differences in predictions. They may arise with respect to the relevanceor interpretation of evidence, the meaning or value of concepts used in the

theories, or the importance of accounting for some specific feature of thecommon explanatory domain. It should not be surprising, therefore, that

theory competition is a slow and arduous means of developing theoreticalknowledge.

TYPES OF THEORETICALRESEARCHPROGRAMS

Isolated instances of theory elaboration, proliferation, and competitioncontribute to the growth and development of theoretical knowledge. Inaddition, each of these kinds of growth may occur in series or in combina-tion with other kinds.

Consider, for example, theory elaboration. A theory that has beenelaborated may undergo still more elaboration; it still may be possible to

I Replacement is also involved in theory elaboration, but in a much more congenialsense. There T(2) may replace T(1), but it retains much of the character of the earliertheory. If the most refined analysis is not always needed, T(1) may even remain aviable theory. In physics, Newtonian mechanics remains viable despite the superiorityof relativity theory, because the predictions of the former are satisfactory for mostpurposes (i.e., for any analysis not involving appreciable fractions of the speed oflight). Note that replacement is absent from theory proliferation altogether.8 This example demonstrates that a single theory may generate development in severaldifferent ways. Stinchcombe (1963) elaborates Davis and Moore (1945); Tumin (1953)competes with it.

709








increase its scope, precision, rigor, or empirical adequacy. Each new

elaboration says more or fits better than the previous one.

Consider also theory proliferation. Proliferationin one explanatorydo-main may reveal key concepts and principles that would be of value

elsewhere, leading to proliferationacrossa broadrange of domains. Each

new proliferation applies the core ideas to a new area.

Finally, consider theory competition. A competing theory is generated

often in response to presumed inadequacies in an earlier theory'saccount

of a particular problem. Additional competitors may also be generated

that address the same problem or similar ones. Each new competitor

attempts to resolve presumed inadequacies in the previous accounts. In

this way the earlier theories affect the content and structureof the latercompeting theories that address a particular, substantive problem.

Further, each form of growth may also occur in combination with the

others. For example, each of a numberof proliferantsmay grow addition-

ally through elaboration of its explanatory accounts.

This sort of collective and systematic effort to develop theoretical

knowledge creates the much larger unit of theoretical activity we have

called a theoreticalresearch program. Each elaboration, proliferation,or competition constitutes a step in the development of the program;

together they define the nature of the interrelationsamong the theories inthe program. In short, the particular pattern of elaborations, prolifera-

tions, and competitions in a theoretical research program constitutes the

program'sanatomy.

Using these concepts we can isolate and describe three basic types of

theoretical research programs, defined according to which relation is the

basic mode of development in the anatomy of the program. Programsin

which the basic mode of development is theory elaboration we designate

as linear, those in which theory proliferation is basic we designate as

branching,and those in which theory competition is basic we designate as

competing.

Two Linear Programs:Conflict Spiral and Deterrence

There are many examples of linear programs;one of the most interesting

concerns the role of threats in bargaining relationships. The program is

based originally on the work of Thibaut and Kelley (1959) on interper-

sonal behavior of all sorts. Their account, built primarilyon principlesofsocial exchange, focused on interdependenceas a basis for relationshipsof

power and conflict among actors. The behavior of A depends directly on

that of B, and vice versa. For example, if A is bargaining with B (say,

over a union contract or an arms treaty), the likelihood that A will

710








threaten B depends on B, not just on A. So too the consequence of A'sthreat depends on B, not just on A.

In 1960 Deutsch and Krauss proposed a theory that makes use of thisinterdependence to explain the behavioral outcomes of threats in bargain-

ing relationships. Unlike many exchange-based theories, theirs assumes

that the explanatory mechanism is nonrational. Basically, threats consti-

tute a form of impression management specifically concernedwith savingface (i.e., avoiding the appearance of weakness). Thus, if A has some

punitive capability (e.g., warheads), A will be tempted to use that capa-

bility to threaten B; using punitive capabilities to threaten maintains an

impression of strength for A. However, A's usage of punitive capabilities

also causes a loss of face for B. To restoreface, therefore,B respondswitha threat; not to do so would give the appearance of powerlessness. Thus,the availability of punitive capability prompts its use against the other;

further, threat is met with counterthreat, creating a spiral of conflict. The

more one party threatens, the more the other party threatens, the conflictin the bargaining relationship escalating with each threat.9

A few years later Shomer, Davis, and Kelley (1966)revised the Deutsch

and Krauss formulation, arguingthat threat should be distinguishedfrom

harm (i.e., punishment or actual damage). Actual harm, they proposed,

always generates the conflict spiral. Threat alone, however, might be

used under some circumstances as a means of communication throughwhich the parties could coordinate their behavior. For example, a union's

warning to management that it is prepared to call a strike might lead toconcessions from management sufficient to avoid the strike. This sort ofresult would be expected when the threatened party anticipates less lossof face from conceding at this point than would result from having toconcede once harm had actually been inflicted. Under such circum-

stances, no spiral of conflict would occur. Shomer et al.'s research sup-ported these revisions.10

Interestingly, starting from similar underlying exchange premises, a

somewhat different account of the use of threats in bargaining hasemerged. Unlike the conflict spiral theories, deterrence theories assumethat parties in the bargaining behave rationally (i.e., in their own self-interest). The greater A's punitive capability, the more B will fear retalia-

tion in response to B's use of such capability. Hence, it is in B's interest toavoid actions that might provoke A into using that capability (e.g., ac-

9 Extensive testing (see, e.g., Deutsch and Krauss 1962)provided supportfor thesearguments.10 Additional theory and research along this path of development is reported inDeutsch(1973).

711








tions such as the use of B's own capability). Increases in A's punitive

capabilities constitute potential threats to B which deter B's use of puni-

tive capabilities.The deterrence process is described very loosely in Schelling (1960,

1966). A much more specific account of part of the process has been

developed by Tedeschi and his associates. Horai and Tedeschi (1969)

proposed a mechanism involving subjective expected utilities for govern-

ing one's response to threats and punishment. The subjective expected

utility of a threat is a function of the magnitude of the threat multiplied by

its credibility (i.e., its probability of occurrence). These dimensions of

subjective expected utility then determine compliance behavior. Spe-

cifically, the greater the magnitude and credibility of A's threat, thegreaterB's compliance with A's demands. The consequence of the use of

threat is therefore often compliance not counterthreat. The use of threat

deters others instead of provoking them, and no spiral of conflict results.

The development of Tedeschi's theory has been primarily empirical.

Each new study has specified an additional set of structural and situa-

tional conditions (e.g., attractiveness, prestige, esteem, or status of the

threatening party) that affect compliance.Tedeschi, Bonoma, and Schlenker (1972)consolidated these findings in

a revised version of the theory. Basically, each structural or situational

condition is a parameterused in calculating the subjective expected utility

of compliance with another party's demands. The higher the utility, the

greater the deterrent effect of A's threats.11

Recently, Bacharach and Lawler (1981) proposed a way to resolve the

discrepancies between deterrence theory (including much of the Tedeschi

work) and conflict spiral theory. Their reconstructiondemonstrates that

the theories deal with the explanatory domain they share using virtually

identical coreconceptual structures. Both are concerned with the puni-

tive capabilities of bargaining parties and with the use of threat and

punishment tactics by those parties. The relationships among these con-

cepts, however, differ somewhat in the two theories. According to deter-

rence theory, each party's attempt to increase punitive capability is met

by the other party with a corresponding increase. A spiral of conflict is

avoided, nevertheless, because each such increase reduces the likelihood

that the other will use that capability. In contrast, conflict spiral theory

postulates that increases in punitive capability increase the likelihood

both of threats to use that capability and of actual attempts to punish theother. Because each increase in punitive capability and in use of threat

and punishment tactics by one party is met with a correspondingincrease

by the other party, the result is an escalating spiral of conflict.

11See Morgan(1977) for additional work on deterrenceprinciples.

712








Although the conceptual structures are the same, the mechanisms bywhich the theories operate are clearly different. Bacharach and Lawler

reconcile these differences by conditionalizing the operations of the twotheories. Deterrence is assumed to be applicable in bargaining situationswhere the stakes are relatively low. As the stakes increase, the likelihoodthat tactics of threat and punishment will prompt counteruse increases.The conflict spiral predictions then become more appropriate. Further-

more, it is impossible to return to the deterrence conditions without re-

structuring the bargaining situation. Bacharach and Lawler then go be-

yond the conditionalization of deterrence and conflict spiral theory to

develop more specific propositions concerned with the effects of punitive

tactics by one party on the concession behavior of the other party. Theirtheory both consolidates and extends the earlier theories of bargainingbehavior.

Both conflict spiral theory and deterrence theory constitute linear

theoretical research programs. The basic means of development for eachis theory elaboration. Consider first the conflict spiral program. Deutschand Krauss (1960) elaborates Thibaut and Kelley (1959); it refines the

latter's account of bargaining relationships, using a face-saving mecha-

nism to generate specific predictions regardingthe consequences of usingthreats in such situations. Deutsch and Krauss says more than doesThibaut and Kelley in this context. Shomer et al. (1966) is an elaborationof Deutsch and Krauss, distinguishing threat from harm conceptually andmaking specific predictions regarding the latter (most of which are sup-ported). Thus, Shomer et al. both says more and fits better than doesDeutsch and Krauss.

Consider next the deterrence program. Horai and Tedeschi (1969)elaborates Thibaut and Kelley in much the same way as does Deutschand Krauss (1960). In this case, however, the refining mechanism in-volves subjective expected utilities. Some of the predictionsmade are alsodifferent, with deterrence commonly the outcome rather than conflictspiral. Nevertheless, relative to Thibaut and Kelley, deterrence theorysays more about bargaining behavior. The consolidation of previous re-search in Tedeschi et al. (1972) is, of course, also an elaboration (this timeof Horai and Tedeschi). It says somewhat more and fits considerablybetter.

Note that the relation between the theories in the deterrence programand those in the conflict spiral program is not one of theory elaboration.Tedeschi's work especially is intended as a challenge to part of the ac-count of bargaining behavior proposed by Deutsch and Krauss. Becausethe theories focus on similar explanatory domains, they are not prolifer-ants either. In a sense the two theories are competitors, but, unlike com-petitors, these two theories are closely related conceptually, having both

713








emerged as elaborations of the same earlier theory. Consequently, therelation between them is a special one not covered by our typology of

basic relations.Note also that the Bacharach and Lawler theory elaborates both deter-

rence and conflict spiral arguments. Not only does it say more than either

version does individually, it does so more compactly. We have identifiedanother special relation between theories that is not covered in our basictypology. We will discuss the character of both of these special relationsbelow.

A Branching Program: Expectation StatesOur example of a branching program comes from work on expectationstates processes. This program is concerned with the processes by whichindividuals come to develop expectations (i.e., stable anticipations offuture behavior) on the basis of general societal and situationally specificinformation they have about themselves and others with whom theyinteract. Theories in the program also deal with the consequences of thedevelopment of such expectations for the actual behavior of individualswhen they interact with others.

The expectation states program originated from a concern with theemergence and maintenance of inequalities in power and prestige amongmembers of discussion groups who were initially equal in status (as in

Bales-type groups). In such situations, systematic and consistent differ-

ences in each member's past performance with respect to the group task

generate expectations for future performance. Once these expectationsemerge, later reactions to an actor's performanceare based on the expec-tations for that group member rather than on actual performance. The

process becomes circular, with power and prestige behaviors reinforcing

expectations, and expectations reinforcing power and prestige behaviors.A major branch of the program, status characteristics theory, has come

to deal with task situations in which actors are already unequal in status

(see, e.g., Strodtbeck's studies of mock juries). In these situations a

status/expectation process is posited; informationthat is culturallyasso-

ciated with the status differences in such situations is used by actors in the

formation of overall expectations for the situations. Power and prestigebehavior is then a direct function of the actor'sexpectations relative to the

other (i.e., their expectation advantage ). Thus, if an actor sees himselfas having higher status than another (e.g., is a male in a professional

occupation working with a semiskilled female), he will come to expect to

perform more capably than he expects the other person to perform. Be-

cause he expects (and is expected to) perform more capably, his positionin the power and prestige hierarchy of the group will be higher; his

714








contributions will be evaluated more highly, his opinions deferredto, andso on. Once again, the process is circular.

Berger, Cohen, and Zelditch (1966), who developed the first statuscharacteristicstheory, considered only the effect of information about onestatus difference. There the status difference generates expectations andpower and prestige behavior. A later version of the theory expanded it toconsider information about multiple statuses operating at the same timeand about statuses of different kinds (Bergerand Fisek 1969, 1974). Thisversion assumed that, in forming expectations, actors combine informa-tion from all the statuses they and others possess that have become salientin their immediate situation. Consequently, a person with inconsistent

statuses (e.g., a black female doctor) is likely to develop expectations thatare an average of all the statuses ratherthan expectations that are basedonly on a subset of the status information (with some statuses eliminatedas irrelevant). A considerable amount of research supports the combin-ing argument. An alternative formulation by Freese and Cohen (1973),however, assumes that eliminating occurs under at least some circum-stances and also receives empirical support.

The most recent version of the theory (Berger, Fisek, Norman, andZelditch 1977), which refines the basic combining argument, considers

status information about more kinds of actors, not all of whom are in-teracting at the same time. The 1977 version also restates the statuscharacteristics ideas in graph-theoretical terms. This formalization per-mits the derivation of more precise predictions about behavior and makesfeasible the detailed analysis of a much larger class of status situations.

A second branch of the expectation states program deals with the man-ner in which actors come to develop expectations for reward allocationsand with how those expectations affect their assessments of the justice orinjustice of actual reward allocations. Status value theory proposes thatreward expectations are based on the activation of referential structures.Basically, referential structures are generalized beliefs about how thedistribution of rewards is related to statuses in society at large. Thegeneral belief that men are paid more than women constitutes such areferential structure.12 Given that this sort of societal and cultural infor-mation is activated (i.e., becomes significant) in the immediate situationof action, people develop expectations for themselves and others in thesituation in terms of what they assume people like them (i.e., those withsimilar

statuses) obtain in general in society. Further, if a referentialstructurehas certain properties(that are specified in the theory), what is

12 A referentialstructureneed not reflectthe actual state of affairs in society. Mostpeoplebelievethat highlyeducatedpersonsaremorehighlyrewarded hanless highlyeducatedones, even though this is often not the case.

715








on the societal level determines what ought to be (in the moral sense) inthe immediate situation of action. Finally, if the reward allocation an

actor actually receives matches these reward expectations, the immediateaction situation is assessed as just; otherwise, it is assessed as unjust.

Most of these ideas are developed in some detail by Berger, Zelditch,Anderson, and Cohen (1968, 1972). The same authors also consider possi-ble responses to injustice (Anderson, Berger, Zelditch, and Cohen 1969).A later statement in Cook (1975) considers more specifically the impactthat responsibility for the unjust allocation and control over future alloca-tion have on responses to injustice. Basically, the more responsibility theactor feels and the more control the actor has over future allocations, the

more he or she is likely to attempt to correct the injustice.Status value theory describes a referential structure/rewardexpecta-

tion process. The process is parallel to the status/expectation process,except that here the expectations concern rewards rather than task perfor-mance capabilities. Thus, information from referential structuresis usedin generating reward expectations; these expectations then determine re-ward allocation (or reallocation) behavior, as well as the assessment ofsuch allocations when they are controlled by an external source.13

A recent formulation by Berger, Fisek, Norman, and Wagner (1983)makes use of this parallel character to construct a new theory that unites

parts of the expectation states accounts of status characteristicsphenom-ena and reward expectation phenomena. The new theory incorporatesconsideration of both expectations for task performanceand expectationsfor reward allocation, specifying structural conditions under which these

two types of expectations become interdependent. As a consequence,Berger et al. are able to incorporate some of the concerns of status value

theory in the graph-theoretical structure of the latest version of status

characteristicstheory. Thus it becomes possible to talk about the effect ofreward expectations on task expectations and the effect of task expecta-tions on reward expectations.

The expectation states program is a branching program. Each branch

represents the application of the basic underlying principles of the pro-gram to a new explanatory domain (e.g., status influence in task groups,assessments of justice or injustice), with little overlap in those domains.In other words, each branch represents a theory proliferation.

Theory elaboration is also quite evident in this particularcase through

its role in the development of each branch of the program. For example,Berger, Fisek, Norman, and Zelditch (1977) is an elaboration of Berger

13 More detaileddiscussionof these and otherbranchesof the expectationstates pro-grammay be found in Berger,Conner,and Fisek (1974)and in Berger, Wagner,andZelditch (1985).

716








and Fisek (1974), which is in turn an elaboration of Berger, Cohen, andZelditch (1966). Proliferation, however, is the more basic relation, for it

defines the interrelation between theories in the different branches of theprogram.14

Again, as in the discussion of linear programs, two relations are notadequately captured in our basic typology. First, the Freese and Cohen(1973) theory is definitely neither an elaborant nor a proliferantof Bergerand Fisek (1969, 1974). Instead, it competes with Berger and Fisek inmuch the same way that Horai and Tedeschi (1969) competes withDeutsch and Krauss (1960). In each pair two theories are closely inter-related conceptually. In each case, both theories are elaborated from the

same earlier theory. Thus, we have another example of one of our specialrelations.

Second, the relationship between the Berger, Fisek, Norman, and

Wagner (1983) theory and its predecessorsalso is not adequately coveredin our basic typology. This special relation seems to share many of its

basic properties with the special relation that the Bacharach and Lawlertheory has with its predecessors in the bargaining programs. In each case

we have a later theory that seems to unite what were previously distincttheories. We now have two examples of each of our special relations. As

we have suggested, later in our analysis we will look more closely at thesespecial relations.

A Competing Program: Distributive Justice

Before elaborating on the special relations, however, we turn to a pro-gram based on competition. The program we have chosen focuses onissues of distributive justice and injustice. Here there are at least two verydifferent accounts of justice phenomena. One has already been in-troduced; it is the status value theory discussed in the previous section.The other approach, equity theory, has a very different conceptual struc-ture.

Equity theories assume that assessments of justice and injustice arebased on direct comparisons of individual actors' inputs and outcomes.For example, drawing on Homans ([1961] 1974), Adams (1965) proposedthat the determination of justice and injustice involves a comparison ofactors' ratios of inputs (i.e., effort expended, or investments )to out-

14 Note that branchingprogramsexist in which there is either little or very inevenelaborationof the differentbranches. This is the case, for example, in the balancetheory program oundedon the workof Heider(1946, 1958).The proliferant heoriesof Newcomb(1953),Festinger 1957),andSecordandBackman(1961)differmarkedlyin the extent of theirelaboration.

717








comes (compensation for effort, or rewards ).Thus, justice is repre-

sented as in equation (1) below.

A's reward/A's investment = B's reward/B's investment. (1)

Injustice then is representedas an inequality in the ratios (in either direc-

tion). All comparisons are local comparisons;that is, actorsA and B are

both in the immediate situation (e.g., both may be teachers in the same

public school). A revision of this theory by Walster, Bersheid, and Wals-

ter (1973) suggested a more adequate specification of the justice equation

that eliminates potential problems with negative inputs.

The status value theory of Berger, Zelditch, Anderson, and Cohen

(1968, 1972)was developed both to provide an account of the distributivejustice process and to deal with several theoretical problems that the

authors claimed existed in the equity formulations of that process. For

example, they argued that some very importantjustice situations (such as

those involving collective, rather than individual, injustice and those

involving other, rather than self-, injustice) cannot be adequately repre-

sented if one uses the equity formulations.The status value account of the justice process that is designed to

resolve these problems proposes that assessments of justice or injustice

are based on comparisons of actual with expected rewards in the im-mediate situation. As previously noted, these expected rewards are based

on activated referentialstructures(culturalbeliefs concerningrewarddis-

tribution in the larger collectivity). Thus, the status value theory predicts

justice when

actual reward = expected reward. (2)

Otherwise, it predicts injustice. If the actual rewardis less than expected,

the actor feels underrewarded; if the actual reward is greater than ex-

pected, the actor feels overrewarded.

Equity theory and status value theory make conflictingpredictionsin a

number of instances, particularly when collective injustice may be in-

volved. Other differences arise when distinctions between overreward

and underreward or between self-injustice and other-injustice are in-

volved. Although the predictions of the two theories do not conflict at

every point, the differences are important and widespread.

Recently, Jasso (1978) developed a new approach to justice issues that

incorporateselements of both equity theory and status value theory.Her

theory is a direct response to the weaknesses she claims exist in both

earlier theories. For example, she criticizes the equity approach for its

inability consistently to distinguish overreward from underrewardand to

handle collective injustice (whether involving overreward or underre-

ward) at all. She also suggests a major weakness in status value theory.

718








Specifically, she points out that the equation used there results in justiceevaluations that are stated in terms of units of the particular reward

commodity rather than in units of justice. The latter is necessary, sheargues, if a comprehensive theory of distributive justice is to be devel-oped.

The theory Jasso creates is shaped in part by the inadequacies she seesin these earlier formulations. Thus, she rearranges the equity and statusvalue equations as a step in her development of a justice evaluation

function, which can be applied to any socially distributedgood. In equitytheory, for example,

justice evaluation =A's reward/A's investment - B's reward/B's investment,

whereas, for status value theory,

justice evaluation = actual reward - just reward. (4)

Jasso's generalized justice evaluation function is then specified as

justice evaluation =

ln (actual amount of good/just amount of good).

That is, an evaluation of the justice of a reward distributionis a function

of the natural logarithm of the ratio of an actor's actual rewards to the

rewards he or she assesses as just. Alternatively, the evaluation may beseen as the difference between the logarithms of the actual amount andthe just amount, a difference mathematically equivalent to the logarithmof the ratio. Jasso supposes that equity theorists would be more likely toprefer the logarithm of the ratio, whereas status value theorists wouldprefer the difference between two logarithms. However, because theforms are mathematically equivalent, the justice evaluation function has

adequately captured critical features of both theories.Jasso makes no attempt to cover all the explanatorydomain of the two

earlier theories. For example, equity theorists are frequently concernedwith predicting specific means of redressingan injustice. Jasso makes nopredictions in this respect. Similarly, status value theory makes predic-tions regarding the assessment of situations where the injustice involvesonly others. Jasso specifically excludes this issue from her conceptualiza-tion.

Jasso's theory does make somepredictions not made by either theequity or the status value theory. For example, the use of a logarithm

results in a weighting that makes underreward more keenly felt thanoverreward, a phenomenon all justice theorists had assumed occurred,but one that had not been incorporated in their theories.

Jasso elaborated her argument considerably in 1980. Among other

719








things, she separates evaluations involving qualitygoods (e.g., beauty)

from those involving quantity goods (e.g., salary) and considers the

application of the justice evaluation function to more than one good at atime. She also develops ajustice evaluation distribution within a social or

conceptual aggregate by assuming consensus about the goods valued

within the aggregate. A variety of empirically testable hypotheses are

then suggested, which relate properties of the distribution to particular

social conditions (e.g., to rates of crime and mental illness).

In the distributive justice program, Adams (1965) and Berger, Zel-

ditch, Anderson, and Cohen (1968, 1972)presentclearlydifferenttheoret-

ical structures; the two theories therefore compete. In fact, status value

theory is in part a response to theoretical problems its authors believedexisted in equity theory (just as Jasso's theory is a response to both status

value theory and equity theory).

As might be expected, theory elaboration occurs frequently in this

programas well, although theory competition is the defining characteris-

tic. It may be reasonable to think of some competing programs as com-

posed of two (or more) linear programs involving distinct conceptual

schemes that are related to one another through competition. A compet-

ing programis probably the most complex form of theoreticalgrowth, for

its brings two or more different theoretical arguments to bear on a prob-

lem or phenomenon at the same time.

In this case as in the two preceding subsections, we see that a develop-

ment in the program is not adequately covered by our typology of three

basic theory relations. The relation between Jasso's theory and the earlier

equity and status value theories is neither theory proliferationnor theory

competition. Instead, Jasso's theory elaborates both earlier theories, uni-

fying them in the process. This is similar to the Bacharach and Lawler

(1981) case and the Berger,Fisek, Norman, and Wagner (1983) case

identified earlier. In all three cases the new theory unifies considerationof

issues dealt with separately in the earlier theories. In one sense or an-

other, the new theories go beyond the theories they unify. However, the

degree to which they unify and amplify is different depending on the kind

of program within which they occur. We shall look specifically at this

phenomenon in our analysis of special relations below.

The Special Relations

As we have seen, not all growth in research programs takes the form of

elaboration, proliferation, or competition. There are at least two special

theoretical relations that appear crucial in describing theoretical growth.

Neither of these relations is likely to appear in the absence of established

research programs; in fact, by their very meanings, they depend on the

720








basic relations which generate such programs. Nevertheless, they repre-sent very important forms of theoretical growth. We now turn our atten-

tion to them.Theory variation.-In some situations the construction of an elabo-rant, proliferant, or competitor reveals slightly different ways in which

the theory's account may be specified. Under these circumstances the

original theory is likely to generate a new theory that incorporatesone of

those slightly different conceptualizations. Ordinarily the theoretical

structureand the focus of explanation of the new theory, T(2), are almostidentical to those of the earlier theory, T(1). The theories differ only in

that they incorporate slightly different working mechanisms. Conse-

quently, they differ only on a very limited set of grounds and in such away that the theories make directly conflicting predictions in that limited

area. Unlike elaboration, however, neither theory initially says more orfits better than the other; the predictions of both can be equally com-

prehensive or equally supported by previous data.

This special relation, termed theory variation, is hard to justify (oreven to see) from the point of view of outsiders (i.e., those not intimatelyfamiliar with the program). Often much of the structure of T(2) is leftimplicit because it is so similar to T(1). Furthermore, the changes in

structure can be quite subtle. Those not thoroughly familiar with thetheories may fail to see any important distinction between them. The

grounds of difference, at least to the outsider, may simply be too small tobe noticed. Finally, precisely because the conceptual grounds of differ-ence are small, a theoretical resolution of a conflict in predictions (giventhat it is amenable to empirical investigation) is likely to occur relativelyrapidly. Variant theories can appear and disappear so quickly they arenot even noticed.

Theory variation, as we have described it, captures the nature of the

relation between Deutsch and Krauss's (1960) and Horai and Tedeschi's(1969) theories of bargaining behavior. They have common theoreticalground in the exchange principles of Thibaut and Kelley (1959). Theymake differential predictions over a limited portion of their shared do-main, depending on the choice of a nonrational or rational mechanism fordealing with reactions to threats. If the mechanism is face saving, theconsequence should be a spiral of conflict; if the mechanism is subjectiveexpected utility, the consequence should be deterrence. Variation also

clearly representsthe relation between the combining argument in Bergerand Fisek (1969, 1974) and the eliminating argumentin Freese and Cohen(1973). Both rely on the basic theoretical structure of status characteristicstheory. In multicharacteristic status situations involving status inconsis-

tencies, however, the two make somewhat different predictions.Although theory variation often appears to be a minor, even a trivial,

721








form of growth, it is actually quite important, particularlyin the refining

of theoretical structures. By constructing variants the theorist is able to

make close comparisons of subtle differences in his or her thinking. Reso-lution of those differences can occur relatively quickly (as is seldom the

case with theory competition). Furthermore, resolution can occur in at

least two different ways. First, one variant may come to be seen as

superior in all cases. Second, as Bacharach and Lawler (1981) have

shown in bargaining theory (and as still may be the case with status

characteristics theory), each variant may come to be seen as superior

under specified conditions. Conditionalization is just as likely as replace-

ment. Both increase our knowledge.

Theory integration.-In a sense, this sort of relation is the opposite oftheory variation. Sometimes two different theories suggest ways in which

the problems they deal with can be treated as similar or related sociolog-

ical problems. Under these circumstances, a new theory may emerge that

incorporatesmuch of the conceptualization of both theories. In such cases

at least three theories, which may or may not have similar theoretical

structures, are involved. One theory, T(3), consolidates many of the ideas

found in T(1) and T(2) in a single formulation, usually suggesting interre-lationships between those ideas. Generally speaking, predictions of T(1)

and T(2) are subsumed in the structure of the new T(3), although it isunlikely that all will be subsumed (especially if the earlier theories are

competitors). Also, the new theory usually generates additional predic-

tions not made by either earlier theory.

With this relation, theory integration, T(3) says more or fits better than

either T(1) or T(2) individually. This is not to say that T(3) is a mere

conjunction of the other two theories (a trivial accomplishmentlogically).

Instead, T(3) is a distinct theoretical structurewhose form depends on the

nature of the relationship between the integrated theories. If T(1) and

T(2) are variant theories, T(3) is likely to involve the specification ofconditions for the application of each variant; T(1) is seen to apply under

some circumstances, T(2) under others. This is what has occurred, for

example, in Bacharach and Lawler's (1981)use of both conflict spiraland

deterrence principles in their theory. The latter applies when stakes are

low, the former when stakes are high. If T(1) and T(2) are proliferants,

integration is likely to entail the identification of propertieswhich permit

the interrelation of disparate phenomena. Such is the case with the inte-

grationin Berger, Fisek, Norman, and Wagner (1983) of status and

reward phenomena on the basis of the interdependence between task

expectations and reward expectations. Finally, if T(1) and T(2) are com-

petitors, integration is likely to require a new theoretical language (with

new concepts) that enables the selection and incorporation of theoretical

principles from each competitor in a coherent new formulation. Jasso's

722








(1978)theory of distributive justice representsan integrationof the equityand status value approaches to justice phenomena in precisely this way.

Although her justice evaluation function captures some of the ideas inequity theory and in status value theory, it does not corresponddirectly tothe form of justice evaluation proposed by either. Moreover, there are nodirect parallels to her justice evaluation distribution (or to the hypoth-eses relating to it) in either earlier theory.

Theory integration often is the major advance that is the clearest evi-dence of progress, because it both unifies and deepens our knowledge.Thus, although integration (especially of competitors)is ratherrare, whensuccessful its impact may be quite dramatic. Successful integration, how-

ever, depends at the very least on the priordevelopment of an establishedresearch program. Significant elaboration, proliferation, competition,and variation are likely to be necessary for the theorist to know whichprinciples are important enough to include in a single theoretical struc-ture. In short, theoretical impact probably requires extensive prepara-tion.

SUMMARYAND CONCLUSIONS

Is there theoretical growth in sociology? Yes-and there is quite a bit ofit. Unfortunately, much of that growth is obscured by our rather limited(and limiting) view of the growth process. To clearly observe, under-stand, and characterize the growth of our knowledge, we have arguedthat it is necessary to

1. distinguish among types of theoretical activities, especially regard-ing the degree to which they do or do not involve growth;

2. consider the relationships that exist among theories, not just therelationship between a theory and the data used to evaluate it; and

3. identify the kinds of relationships that may occur among theories,and therefore the different kinds of growth that may occur.

We have accomplished the first of these objectives by differentiatingbetween orienting strategies, unit theories, and theoretical research pro-grams. Strategies, we have argued, are primarily metatheoretical incharacter. Their manifold, basic, and generally nonempirical differencesmake them rather rigid; whatever changes do occur within or amongstrategies are not likely to be progressive. Hence, a focus on orientingstrategies is inappropriate for the analysis of theoretical growth in sociol-

ogy.A focus on unit theories is considerably more satisfactory. Differenceshere can ordinarily be resolved logically or empirically. Furthermore,much of what we think of as growth of knowledge is representedby theincreases in empirical support that a unit theory may generate;empiricalconfirmation of our theoretical ideas is a form of progress.

723








Empirical confirmation is nevertheless not the only form of progress.

Others include increases in scope, precision, range, and so on. All of these

forms require consideration of the linkages between theories, not just thelinkage between theory and data. Such considerationis only possible with

a focus on theoretical research programs, which are sets of interrelatedunit theories together with the research that supports and the research

that applies them.We have used that focus in analyzing the theoretical activity in several

programs of current interest in sociology: deterrence and conflict spiral,

expectation states, and distributive justice. Our analysis suggests thattheoretical growth is a multifaceted process; we can make sense of the

theoretical growth that occurs in sociology only if we pay attention to itsmultifaceted character.

More specifically, our analysis indicates at least five kinds of relations

among theories that occur in theoretical research programs. Three of

these relations-theory elaboration, theory proliferation, and theory

competition-may serve as the basic mode of development for entirelarge-scale programs. The anatomies of such programscorrespond to thebasic mode of growth. In linear programs the basic mode is elaboration;in branching programs it is proliferation; in competing programs it is

competition.Two other relations-theory variation and theory integration-play

important roles in theoretical growth; however, they are more specialized

in their impact. Neither is ordinarily a basic mode of development for

entire programs. Instead, each occurs within the context of established

programs based on other kinds of relations. Integration, perhaps the

rarest and most dramatic form of theoretical growth, depends especially

on the existence of a well-developed program based on other relations.

Such development is probably necessaryto determine which concepts and

principles in the integrated theories are important enough and promising

enough to be included in a single theoretical formulation.

Perhaps the most surprising and unexpected result of our analysis is

that theory integration may, in fact, occur in any one of three different

ways, depending on the nature of the relation between the integratedtheories. If T(1) and T(2) are variants, integration is likely to involve the

specification of conditions for the application of each variant. If T(1) andT(2) are proliferants, integration is likely to involve identification of prop-

erties which permit the interrelation of disparate phenomena. Finally, ifT(1) and T(2) are competitors, integration is likely to involve the creation

of a new theoretical language that enables the theorist to select and incor-

porate principles from the competing theories in a coherent new theory.Our understanding of the tripartite character of theory integration

emerged only in the analysis of specific concrete cases of theoretical

724








growth in sociology. The analysis of actual cases, we believe, is critical to

the understanding of theoretical growth. With Kuhn, we believe that a

discipline needs exemplars. Each of the cases we have analyzed providesan exemplar of theoretical growth in sociology. Any furtherdevelopment

of our understandingof theoretical growth in sociology must be anchored

in the analysis of such exemplars.

Thus, whatever one's substantive interest in the particular cases con-

sidered, they (and probably many others that can be identified) merit

attention, careful analysis, and study. These cases provide us with valu-

able information about the nature of current theoretical growth in our

field. They also provide us with information that we can use in realizing

theoretical growth in the different branches of sociology.

REFERENCES

Adams, J. S. 1965. Inequityn Social Exchange. Pp. 267-99 in Advances n Experi-mental Social Psychology, vol. 2. Edited by L. Berkowitz. New York:AcademicPress.

Alexander,J. C. 1982. TheoreticalLogic in Sociology. Vol. 1, Positivism, Presupposi-tions, and CurrentControversies.Berkeley and Los Angeles:Universityof Califor-

nia Press.Anderson, B., J. Berger, M. Zelditch, Jr., and B. P. Cohen. 1969. Reactionsto

Inequity. Acta Sociologica 12:1-11.Bacharach, S. B., and E. J. Lawler. 1981. Bargaining: Power, Tactics, and Out-

comes. San Francisco:Jossey-Bass.Bales, R. F. 1950. Interaction Process Analysis. Reading, Mass.: Addison-Wesley.Berger, J., B. P. Cohen, and M. Zelditch, Jr. 1966. Status Characteristicsand

Expectation States. Pp. 29-46 in Sociological Theories in Progress, vol. 1.Edited by J. Berger,M. Zelditch,Jr., and B. Anderson.Boston:Houghton-Mifflin.

Berger, J., T. L. Conner,and M. H. Fisek, eds. 1974. Expectation States Theory:ATheoreticalResearchProgram. Cambridge,Mass.: Winthrop.

Berger, J., and M. H. Fisek. 1969. The Structureof the Extended Theory of StatusCharacteristics nd Expectation States. Paper presentedat the 15th annual meet-ings of the West Coast Conferenceon SmallGroups Research, Seattle, Wash.

. 1974. AGeneralizationof the StatusCharacteristics nd ExpectationStatesTheory. Pp. 163-205 in Berger, Connor, and Fisek 1974.

Berger, J., M. H. Fisek, R. Z. Norman, and M. Zelditch, Jr., eds. 1977. StatusCharacteristicsand Social Interaction.New York: Elsevier.

Berger,J., M. H. Fisek, R. Z. Norman, and D. G. Wagner. 1983. TheFormationofRewardExpectations n Status Situations. Pp. 126-68 in Equity Theory:Psycho-logical and Sociological Perspectives, edited by D. Messick and K. Cook. NewYork:Praeger.

Berger,J., D. G. Wagner, and M. Zelditch,Jr. 1985. Expectation tates Theory:TheStatus of a ResearchProgram. To appear in Status, Rewards, and Influence,edited by J. Bergerand M. Zelditch, Jr. In press. San Francisco:Jossey-Bass.

Berger,J., M. Zelditch, Jr., B. Anderson,and B. P. Cohen. 1968. StructuralAspectsof DistributiveJustice:A Status Value Formulation. Technical Report no. 28.Stanford,Calif.: StanfordUniversity, Laboratory or Social Research.

. 1972. StructuralAspects of DistributiveJustice: A Status Value Formula-

725








tion. Pp. 119-46 in Sociological Theories in Progress, vol. 2. Edited by J. Berger,M. Zelditch, Jr., and B. Anderson. Boston: Houghton-Mifflin.

Bergner,J. T. 1981. The Originsof Formalism n Social Science. Chicago:Universityof Chicago Press.Bernstein, R. J. 1978. The Restructuringof Social and Political Theory.Philadelphia:

University of Pennsylvania Press.Blau, P. M. 1964. Exchange and Power in Social Life. New York: Wiley.Blau, P. M., and 0. D. Duncan. 1967. The American Occupational Structure. New

York: Free Press.Burgess, R. L., and R. L. Akers. 1966. A Differential Association-Reinforcement

Theory of Criminal Behavior. Social Problems 14:128-47.Cartwright, D. P., and F. Harary. 1956. Structural Balance: A Generalization of

Heider's Theory. Psychological Review: 63:277-93.Cohen, B. P. 1980. Developing Sociological Knowledge. Englewood Cliffs, N.J.:

Prentice-Hall.Cook, K. S. 1975. Expectations, Evaluations and Equity. American Sociological

Review 40:372-88.Cook, K. S., and R. M. Emerson. 1978. Power, Equity, and Co-mmitment in Ex-

change Networks. American Sociological Review 43:721-39.Dahrendorf, R. 1959. Class and Class Conflict in Industrial Society. Stanford, Calif.:

Stanford University Press.Davis, K., and W. E. Moore. 1945. Some Principles of Stratification. American

Sociological Review 10:242-49.Denzin, N. K. 1969. Symbolic Interactionism and Ethnomethodology. American

SociologicalReview 34:922-34.

Deutsch, M. 1973. The Resolution of Conflict. New Haven, Conn.: Yale UniversityPress.

Deutsch, M., and R. M. Krauss. 1960. The Effect of Threat on Interpersonal Bar-gaining. Journal of Abnormaland Social Psychology61:181-89.

. 1962. Studies of Interpersonal Bargaining. Journal of Conflict Resolution6:52-76.

Durkheim, E. (1893) 1964. The Division of Labor in Society. New York: Free Press.Eckberg, D. L., and L. Hill, Jr. 1979. The Paradigm Concept in Sociology: A Critical

Review. American Sociological Review 44:925-37.Fararo,T. J. 1973.MathematicalSociology:An Introductionto Fundamentals.New

York: Wiley.

Festinger, L. 1957. A Theory of Cognitive Dissonance. Evanston, Ill.: Row, Peterson.Freese, L. 1980. The Problem of Cumulative Knowledge. Pp. 13-70 in Theoretical

Methods in Sociology, edited by L. Freese. Pittsburgh: University of PittsburghPress.

Freese, L., and B. P. Cohen. 1973. Eliminating Status Generalization. Sociometry36:177-93.

Gibbs, J. 1972. Sociological Theory Construction. Hinsdale, Ill.: Dryden.Gouldner, A. W. 1970. The Coming Crisis in Western Sociology. New York: Avon.Gove, W. R. 1970. Societal Reaction as an Explanation of Mental Illness: An Evalua-

tion. AmericanSociological Review 35:873-84.. 1975. Labelling and Mental Illness. Pp. 35-81 in The Labelling of Deviance,

edited by W. R. Gove. New York: Wiley.Hage, J. 1972. Techniquesand Problemsof Theory Construction n Sociology. New

York: Wiley.Hannan, M. T., and J. Freeman. 1977. The Population Ecology of Organizations.

AmericanJournalof Sociology 82:929-64.Hawley, A. H. 1950. HumanEcology:A TheoryofCommunityStructure.New York:

Ronald.

726








Heider, F. 1944. Social Perception and Phenomenal Causality. Psychological Re-view 51:358-74.

. 1946. Attitudes and Cognitive Organization. Journal of Psychology 21:107-12.

. 1958. The Psychology of InterpersonalRelations. New York:Wiley.Homans, G. C. (1961) 1974. Social Behavior: Its Elementary Forms. Rev. ed. New

York: Harcourt.Horai, J., and J. T. Tedeschi. 1969. Effects of Credibility and Magnitude of Punish-

ment on Compliance to Threats. Journal of Personality and Social Psychology12: 164-69.

Howson, C., ed. 1976. Method and Appraisal n the Physical Sciences. New York:Cambridge University Press.

Jasso, G. 1978. On the Justice of Earnings: A New Specification of the JusticeEvaluation Function. American Journal of Sociology 83:1398-1419.

. 1980. A New Theory of Distributive Justice. American Sociological Review45:3-32.

Kuhn, T. S. (1962) 1970. The Structure of Scientific Revolutions. 2d ed. Chicago:University of Chicago Press.

Lakatos, I. 1968. Criticism and the Methodology of Scientific Research Programmes.Proceedings of the AristotelianSociety 69:149-86.

. 1978. The Methodologyof Scientific ResearchProgrammes, dited by J. Wor-rall and G. Currie. New York: Cambridge University Press.

Lakatos, I., and A. Musgrave. 1970. Criticism and the Growth of Knowledge. NewYork: Cambridge University Press.

Latsis, S. J., ed. 1976. Method and Appraisal in Economics. New York: Cambridge

University Press.Leinhardt, S., ed. 1977. Social Networks: A Developing Paradigm. New York: Aca-

demic Press.Levins, R. 1968. Evolution in Changing Environments. Princeton, N.J.: Princeton

University Press.Malinowski, B. 1944. A Scientific Theory of Culture. Chapel Hill: University of North

Carolina Press.Mauss, M. (1925) 1967. The Gift. New York: Norton.Marx, K. (1848) 1977. The Communist Manifesto. Pp. 221-47 in Karl Marx: Se-

lected Writings, edited by D. McLellan. New York: Oxford University Press.Merton, R. K. 1967. On Theoretical Sociology. New York: Free Press.

. 1968. Social Structure and Anomie. Pp. 185-214 in Social Theory andSocial Structure. Enlarged ed. New York: Free Press.

Morgan, M. P. 1977. Deterrence: A Conceptual Analysis. Beverly Hills, Calif.: Sage.Newcomb, T. 1953. An Approach to the Study of Communicative Acts. Psychologi-

cal Review 60:393-404.Parsons, T. 1954. The Prospects of Sociological Theory. Pp. 348-69 in Essays in

Sociological Theory. Rev. ed. New York: Free Press.Parsons, T., and R. F. Bales. 1953. The Dimensions of Action-Space. Pp. 63-110 in

Working Papers in the Theory of Action, edited by T. Parsons, R. F. Bales, andE. A. Shils. Glencoe, Ill.: Free Press.

Radcliffe-Brown, A. R. 1952. Structure and Function in Primitive Society. Glencoe,

Ill.: Free Press.Ritzer, G. 1975. Sociology: A Multiple Paradigm Science. Boston: Allyn & Bacon.

Scheff, T. J. 1966. Being Mentally Ill. Chicago: Aldine.. 1974. The Labeling Theory of Mental Illness. American Sociological Re-

view 39:444-52.

Schelling, T. C. 1960. The Strategy of Conflict. New York: Oxford University Press.. 1966. Arms and Influence. New Haven, Conn.: Yale University Press.

727








Secord, P. F., and C. Backman. 1961. Personality Theory and the Problem of Stabil-ity and Change in Individual Behavior: An Interactive Approach. Psychological

Review 68:21-32.Shomer, R. W., A. H. Davis, and H. H. Kelley. 1966. Threats and the Developmentof Coordination. ournal of Personality and Social Psychology 4:119-26.

Skinner, B. F. 1953. Science and Human Behavior. New York: Macmillan.Stinchcombe, A. L. 1963. Some Empirical Consequences of the Davis-Moore Theory

of Stratification. AmericanSociological Review 28:805-8.. 1968. Constructing Social Theories. New York: Harcourt.

Strodtbeck, F. L., R. M. James, and C. Hawkins. 1957. Social Status in Jury Delib-erations. American Sociological Review 22:713-19.

Tedeschi, J. T., T. V. Bonoma, and B. R. Schlenker. 1972. Influence, Decision, and

Compliance. Pp. 346-418 in The Social Influence Processes, edited byJ. T. Tedes-chi. Chicago: Aldine.

Thibaut, J. W., and H. H. Kelley. 1959. The Social Psychology of Groups. New York:

Wiley.Tumin, M. M. 1953. Some Principles of Stratification: A Critical Analysis. Ameri-

can Sociological Review 18:387-93.Turner, J. H. 1982. The Structure of Sociological Theory. 3d ed. Homewood, Ill.:

Dorsey.

Wallerstein, . 1974.TheModernWorldSystem.Vol. 1, CapitalistAgriculture nd theOrigins of the European WorldEconomy in the Sixteenth Century.New York:Academic Press.

Walster, E., E. Berscheid, and G. W. Walster. 1973. New Directions in EquityResearch. Journal of Personality and Social Psychology 5:151-76.

Winch, P. 1958. The Idea of a Social Science. London: Routledge & Kegan Paul.Zetterberg, H. 1965. On Theory and Verification in Sociology. 3d ed. Totowa, N.J.:

Bedminster.Zimmerman, D. H., and D. L. Wieder. 1970. Ethnomethodology and the Problem of

Order: Comment on Denzin. Pp. 285-98 in Understanding Everyday Life, edited

by J. Douglas. Chicago: Aldine.

728

do sociological theories grow weber

Documents