Quarterly Journal of Speech The forum

This article was downloaded by: [National Communication Association] On: 10 July 2015, At: 12:23 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place, London, SW1P 1WG

Quarterly Journal of Speech Publication details, including instructions for authors and subscription information: http://nca.tandfonline.com/loi/rqjs20

The forum a

b

c

Donald G. Ellis , Dean E. Hewes , Chaim Perelman & John W. Ray

d

a

Michigan State University ,

b

University of Wisconsin , Madison

c

University of Brussels ,

d

Montana College of Mineral Science and Technology , Published online: 05 Jun 2009.

To cite this article: Donald G. Ellis , Dean E. Hewes , Chaim Perelman & John W. Ray (1980) The forum, Quarterly Journal of Speech, 66:4, 437-449, DOI: 10.1080/00335638009383540 To link to this article: http://dx.doi.org/10.1080/00335638009383540

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://nca.tandfonline.com/ page/terms-and-conditions

T H E QUARTERLY JOURNAL OF SPEECH,

66 (1980), 437-49.

THE FORUM ELLIS ON HEWES N a recent issue of this journal Dean Hewes argued that past research conducted under the theoretical auspices of the Interact System Model (ISM) suffered from a poor fit between theory and methodology.1 Hewes maintained that the methods for developing coding schemes, analyzing temporal sequences, and the lack of subsumptive explanations has hindered the analysis of sequential interaction. Hewes's essay suggests a variety of problems, and, even though I often found myself in substantial argreement with him, I shall argue that Hewes failed to characterize the ISM correctly with respect to (1) the importance of individual differences when analyzing sequential interaction and (2) the proper explanation for interaction sequences.

Downloaded by [National Communication Association] at 12:23 10 July 2015

I

INDIVIDUAL DIFFERENCES

Hewes argues that individual difference information has not been included in ISM research. Such information is ignored for three reasons. The first is based on the assumption that patterns of interaction result from people adjusting to their fellows—not persistent psychological states. When two people are communicating each uses the linguistic behavior of the other as data about how to respond appropriately. In a conversation the only access A has to B's psychological state is through B's expressed communicative behavior. The ISM does not deny the influence of 1 Dean E. Hewes, "The Sequential Analysis of Social Interaction," Quarterly Journal of Speech, 65 (1979), 56-73.

internalized mental states; it does, however, require the researcher to consider these states secondary in importance to externalized patterns of communicative behavior. The performance of a communicative behavior is assumed to constrain the possibilities of what can follow. The resultant pattern of constraints is of interest to the ISM researcher. In this way the researcher emphasizes how interactants structure their conversations rather than why they communicated as they did. Numerous successful lines of research seek to identify interaction patterns at the expense of individual differences information. Scheflen, for example, refers to some communicative events as programmatic.2 These events are programmatic because they are composed of recurring and specific communicative behaviors that are recognizable and meaningful to the participants. All participants in a particular speech community, regardless of individual differences, can recognize and take part in the event. Research that explicates the formal structure of socially appropriate speech is clearly unconcerned with individual differences information. Nofsinger's explication of the demand ticket,3 Ervin-Tripp's work with directives,4 Labov's rules for ritual insults,5 2 Albert E. Scheflen, "Behavioral Programs in Human Communication," in General Systems Theory and Psychiatry, ed. William Gray, Frederick J. Duhl, and Nicholas D. Rizzo (Boston: Little, Brown, 1969), pp. 209-28. 3 Robert E. Nofsinger, Jr., "The Demand Ticket: A Conversational Device for Getting the Floor," Speech Monographs, 42 (1975), 1-9. 4 Susan Ervin-Tripp, "Is Sybil There? The Structure of Some American English Directives," Language in Society, 5 (1976), 25-66. 5 William Labov, "Rules for Ritual Insults,"


458 QUARTERLY JOURNAL OF SPEECH

DECEMBER 1980

variables; it suggests theoretical orientations that increase the salience of the interrelationship among immediate contextual dimensions, especially the communicative behaviors which comprise the activity of the context. This is precisely the intent of ISM research. The third reason for directing attention away from individual difference information emerges from the structuralism that is implied by the ISM.9 Structural analyses are concerned with the relationships among objects or parts rather than the objects themselves. "Structures" result from these relationships and define the work under consideration.10 The specific purpose of ISM research is to "focus on the interrelationships of behaviors over time, regardless of who manifests those behaviors."11 This results in a representation of communication data as patterned interaction where frequencies, types, variations, properties, categories, etc. are evident. This research direction is rooted in structuralism, where the goal is to reconstruct an object (e.g., communicative setting, literary work, myth) in such a way as to make manifest the object's rules and formal properties. The ISM researcher approaches some communicative setting with a scheme for classifying and organizing the interaction. The interaction is then recomposed in order to make certain functions appear and discover the regular conin Studies in Social Interaction, ed. David straints that are responsible for stability. Sudnow (New York: Free Press, 1972), pp.

and Sacks, Schegloff, and Jefferson's accounts of turn-taking and opening sequences6 all emphasize the verbal machinery of communication and not the perceptions and interpretations of individuals. Moreover, the Palo Alto group has met with considerable therapeutic success by assuming that individual differences are not the origins of relationship behavior but result from the active process of communicating.7 A second reason for deemphasizing individual differences is expressed by Jones when he warns against attribution error.8 His research indicates that the role of individual dispositional states when determining behavior is minimal; people underestimate the role of the immediate context when determining behavior. Subjects in experiments too often make the attribution that the underlying disposition of an individual is directly conveyed in behavior and these dispositions "explain" behavior. Jones argues that the presumption of a natural actor-act unit that represents a correspondent disposition is neither a necessary nor inevitable inference. Hewes's recommendation that we increase the amount of individual differences information could lead to an increase in misplaced attributions to individual dispositional states. The line of research represented by Jones suggests a movement away from HSM

120-69. 6 Harvey Sacks, Emanuel A. Schegloff, and Gail Jefferson, "A Simplest Systematics for the Organization of Turn-taking for Conversation," Language, 50 (1974), 696-735; Emanuel A. Schegloff, "Sequencing in Conversational Openings," American Anthropologist, 70 (1968), 1075-95. 7 Carol Wilder, "The Palo Alto Group: Difficulties and Directions of the Interactional View for Human Communication Research," Human Communication Research, 5 (Winter 1979), 17186. 8 Edward E. Jones, "The Rocky Road From Acts to Dispositions," American Psychologist, 34 (1979), 107-17.

9 Warnick provides an excellent explanation of structuralism and how it relates to other perspectives on language and meaning. See Barbara Warnick, "Structuralism vs. Phenomenology: Implications for Rhetorical Criticism," Quarterly Journal of Speech, 65 (1979), 250-61. 10 See Richard Hornby, Script Into Performance: A Structuralist View of Play Production (Austin: Univ. of Texas Press, 1977). 11 Leonard C. Hawes, "Development and Application of an Interview Coding System," Central States Speech Journal, 23 (1972), 93.

QUARTERLY JOURNAL OF SPEECH


The important point here is that the structure and function of communicative settings is obtainable from the surface characteristics of social life. That is, language, customs, cultural expressions, etc. are not simply the after effect of individual differences; they are the substance itself. , EXPLANATION

Criticizing the ISM for poor subsumptive power is akin to criticizing an apple for not being an orange. The ISM, and the research it spawned, is not concerned with subsuming empirical regularities under general propositions; in fact, the ISM was intended as an alternative to this type of explanation and theory construction. Fisher and Hawes12 are explicit throughout their article that the ISM is based on grounded theory; that is, they are concerned with the process of generating concepts and identifying relationships rather than verifying a general proposition by noting an empirical instance. But the adequacy of subsumptive models of explanation is not my concern here; this matter and attendant arguments are available elsewhere.13 Instead, I would like to discuss the form of explanation implied by the ISM and explain how it differs from the subsumptive models advocated by Hewes. Hewes does not make an ontological commitment to the inherent stochasticity of interaction processes. Inter12 B. Aubrey Fisher and Leonard C. Hawes, "An Interact System Model: Generating a Grounded Theory of Small Groups," Quarterly Journal of Speech, 57 (1971), 444-53. 13 See, for example, Leonard Krimerman, ed., The Nature and Scope of Social Science: A Critical Anthology (New York: AppletonCentury-Crofts, 1969); Baruch A. Brody, ed., Readings in the Philosophy of Science (Englewood Cliffs, N.J.: Prentice-Hall, 1970); Frederick Suppe, ed., The Structure of Scientific Theories (Urbana: Univ. of Illinois Press, 1974).

439 ELLIS

action processes are not "effectively" stochastic at this point in time, but are inherently stochastic. This means that interaction processes are governed by their probabilistic structure and, as Anderson writes, the "useful properties appropriate to the time sequence are not the deterministic mean value functions but the probability structure itself."u This has important implications because the appropriate explanation for some communication sequences is not attainable by subsuming the sequences under a general proposition. Stochastic processes are not governed by the logic of the conditional. Hewes's choice of explanatory models is based on if-then statements where some antecedent condition is, at the very least, sufficient to bring about some consequent state.15 Stochastic processes involve no such relationship. If relations among elements of a communication process are, in fact, stochastic, then the movement from one state to another is dependent on previous sequential patterning. Stochastic processes generate an array of alternatives that can be probabilistically induced from historically derived frequency distributions. Predictability (forecasting correctly or incorrectly) is secondary to establishing probabilities among event possibilities. It is important to underscore the predictive limits of stochastic processes. I will argue below that prediction plays a minor role in the type of explanations most suitable for stochastic processes and the ISM. Chatfield makes the case clearly when he writes that a series is "stochastic in that the future is only 14 Thomas W. Anderson, The Statistical Analysis of Time Series (New York: Wiley, 1971), p. 164. Emphasis added. 15 The Hempel and Oppenheim article cited in Hewes's n. 85 (p. 69) is clear on this issue. See Carl G. Hempel and Paul Oppenheim, "Studies in the Logic of Explantion," Philosophy of Science, 15 (1948), 135-75.



partly determined by past values. For stochastic series exact predictions are impossible and must be replaced by the idea that future values have a probability distribution which is conditioned by a knowledge of past values."16 A proper explanation is one which removes puzzlement by taking what we do not know and placing it in the context of what we do know.17 This known context does not have to be a more general network of propositions. Moreover, the indigenous fit between the type of explanation and the nature of the phenomenon to be explained is crucial to an appropriate explanation. In other words, some human behavior is best "explained" by reference to attitudes, beliefs, or values; other behavior might require reference to physiological operations. Strong cultural norms may make some behavior sensible, but the idiosyncracies of a particular relationship are necessary in another case. The upshot is that there must be congruence between the structure of the explanation and the substantive qualities of the phenomenon. How, then, are interaction sequences explained according to the ISM and stochastic processes? In the first place, the appropriate explanation of interaction processes should not involve vertical subsumption but horizontal assimilation. We can explain communicative events in perfectly good sense (remove puzzlement) by referring to any number of temporally prior conditions. The communicative event in question is explained via some relationship between the event in question and historical conditions. The explanation results when some historical regularity can be induced from the 16 C. Chatfield, The Analysis of Time Series: Theory and Practice (New York: Wiley, 1975), pp. 6-7. 17 Michael Scriven, "Explanations, Predictions, and Laws," in Brody, pp. 88-104.

DECEMBER 1980

event in question and there is an understandable relationship between the two. This relationship can take many forms.18 The important point is that the explanation brings out the continuity between successive events and emphasizes the dependence (though not very predictable) of the later on the earlier. This type of explanation is found in a variety of fields, e.g., history, anthropology, embryology, and some parts of developmental psychology. The interaction process characterizing some communicative systems, then, would take the following explanatory structure in the form of a conditional probability: (1)

Pr [S, (t) | x t ]

That is, the probability (Pr) of the interaction process being in some state (S^ at some point in time (t) is dependent on certain occurrences at a previous time. The vector X t denotes the characteristics of these antecedent conditions. There can be any number of states (i) in the system or characteristics of a designated time period (Xt). Three steps are required to develop this model of explanation. First, the researcher identifies the state space. In other words, what are the various states or what communicative events can occur in the identified process. The state space corresponds to the category system used in ISM research. These are the permissible communicative events and can be as simple as sound/silence19 or involve 18 For example, some common relationships between events are reflexive, nonreflexive, symmetrical, asymmetrical, nonsymmetrical, intransitive, transitive, etc. For further discussion, see Hubert G. Alexander, Language and Thinking (New York: Van Nostrand, 1967), esp. Ch. 7. 19 Joseph N. Cappella, "Modeling Interpersonal Communication Systems as a Pair of Machines Coupled Through Feedback," in Explorations in Interpersonal Communication, ed. Gerald R. Miller (Beverly Hills: Sage, 1976), pp. 59-85.

441



a larger number of possibilities.20 The second step is determining the components of X t . The issue is what relevant historical (antecedent) conditions, if any, lead to changes in the probability of being in a particular state. The third step is to establish the form of the relationship between the antecedent conditions and an event state. Most ISM research is perfectly consistent with this model of explanation. True, there are undoubtedly numerous permutations on the model which have not been explored; nevertheless, most of the cited ISM research does not deviate from the practice of identifying system states, observing their development through time, and using these developmental processes to explain the relationships among the system states. Although this type of explanation indicates the continuity between prior events and subsequent results, it does not pretend to great predictive prowess. This is because prior events are not taken in conjunction with universal laws to establish a sufficient condition for a subsequent event. Prediction is simply not central to the ISM.21 Scriven argues that explanation and prediction are nonsymmetrical and espouses a position compatible with the one advocated here when he writes, "to predict we need a correlation between present events and future ones—to explain, be20 Ellis and Fisher used twelve states of decision making and Hawes and Foley had thirteen for interview interaction. See Donald G. Ellis and B. Aubrey Fisher, "Phases of Conflict in Small Group Development: A Markov Analysis," Human Communication Research, 1 (Spring 1975), 195-212; and Leonard C. Hawes and Joseph M. Foley, "A Markov Analysis of Interview Communication," Speech Monographs, 40 (1973), 208-19. 21 Fisher, the foremost proponent of the ISM, makes this point in the original ISM article and in his most recent statement. See B. Aubrey Fisher, Perspectives On Human Communication (New York: Macmillan, 1978), pp. 213-15.

ELLIS

tween present ones and past ones."22 Scriven portrays Darwin as the paradigm of the explanatory but nonpredictive scientist; the evolutionary model of explanation differs from the subsumptive model such that prediction and explanation are not structurally identical. Explanation, in the evolutionary sense, occurs when an earlier event stands in special relation to a later event; predictions are replaced by transition probabilities. In other words, rather than forcasting a future event within an error range, a process is assumed to contain an array of potential system states, all of which have a short-range transition probability. As interaction continues over a longer period of time, the transition probabilities can change. Consider, for example, what we do when we interrupt a conversation half-way through. We ask the conversants how the conversation started and what was said earlier so as to indicate the events which give sense to the present context. We never imply that the present was predictable from the past or else there is no need to continue the conversation. There are, however, a range of states a conversation can take and the probabilistic nature of the movement from state to state is responsible for the emergent structure of interaction. The evolutionary model of explanation advocated here denies Hewes's claim that positing generative machanisms is the key to explaining sequential interaction. Although Hewes does not use the word "cause," he claims that generative mechanisms "produce" regularities in interaction, thereby elevating generative mechanisms to the status of general propositions which subsume empirical instances. 22 Michael Scriven, "Explanation and Prediction as Non-Symmetrical," in Krimerman, p. 121.



This satisfies Hewes's requirement for explanation by tiers. The identification of generative mechanisms is problematic. An inference is required about when a generative mechanism is responsible for producing an interaction sequence. This is complicated by two attendant problems: First, the same generative mechanism can initiate more than one interaction sequence, and, second, several interaction sequences could result from the same generative mechanism. The search for generative mechanisms deflects attention from the more useful concept of morphogenesis. The ability to change is a fundamental theme of any communicative system. And rather than working to establish the general principles (generative mechanisms) which presumably give rise to interaction sequences, it seems more useful to emphasize the communication system's capacity for change and deviation. The values of transition probabilities vary for a variety of complex reasons that often have little to do with generative mechanisms. For example, change in transition probabilities and directions may result from (1) the complexity of organization in the communicative context,23 (2) some facility or rule unique to the communication system, (3) environmental change, (4) the quality and quantity of information available to the participants in the interaction, or (5) the capacity for change in the communicative system. In brief, appropriate explanations of interaction sequences result from local contextual matters and short-term historical conditions. In this brief essay I have tried to de23 For a discussion of organized complexity see B. Aubrey Fisher, Thomas W. Glover, and Donald G. Ellis, "The Nature o£ Complex Communication Systems," Communication Monographs, 44 (1977), 231-40.

DECEMBER 1980

fend a perspective on how social interaction should be studied. My intent has been to take issue with certain suggestions offered by Hewes on how to examine and explain sequential interaction behavior. I believe, however, that Hewes's essay represents something more important than methodological suggestions for the study of communication. It represents an attempt to generate explanations by bringing an organized conceptual framework to a given phenomenon. Stating the relationship between assumptions about the nature of communication and methodological procedures is what is particularly important about Hewes's effort. Nevertheless, I maintain that as communication scholars continue to articulate the fit between the nature of human communication and the structure of our theories, the role of individual differences and subsumptive models of explanations will subside. DONALD G. ELLIS

Michigan State University HEWES ON ELLIS In his critique of my article, Ellis has offered an intriguing set of alternative positions; however, judging from his critique, our field has moved toward consensus over a number of issues. Ellis and I, representatives of divergent perspectives on sequential analysis, seem to have reached agreement on several points—the value of alternative conceptions of time, the utility of multifunctional coding, and the applicability of semi-Markov models to interaction processes. Two areas of apparent disagreement remain. Should individual difference information be incorporated into the state space that governs the representation of interactional processes? How should investigators undertake the


443 HEWES


explanation of sequential structure? I shall attempt to demonstrate that these areas of disagreement result either from simple misunderstandings or identifiable value differences, or both.

munication theorists or their methods as the most legitimate, then I must disagree. I fear that, rather than being interpreted as a spirited defense of the ISM tradition, Ellis' critique could be taken as an attack on my efforts to inTHE PURPOSE OF MY ARTICLE crease the methodological and concepEllis and I are in disagreement over tual flexibility of sequential analysis. the purpose of my article. He apparently My subsequent remarks are addressed believes that it was intended as an in- to this more problematic interpretation. dictment of the Interact Sysem Model.1 INDIVIDUAL DIFFERENCES Thus, he observes that "In a recent issue of this journal Dean Hewes argued that Ellis defends what he claims to be the past research conducted under the ISM tradition of ignoring individual theoretical auspices of the Interact difference information in the sequential System Model (ISM) suffered from a analysis of social interaction.3 Much of poor fit between theory and methodol- what he argues appears to be based on ogy." I did suggest that the assumptions a misconception of my position. The implicit in ISM methodologies have led remainder is based on a real difference researchers to ignore potentially in- in values between us. Let me begin by teresting phenomena. I did not attempt clarifying my position on individual to document a disjuncture between differences in sequential analysis. theory and method within the ISM "'Individual difference variables" are camp. Rather, my purpose was to in- simply those variables that permit recrease the flexibility of approaches to searchers to differentiate among insocial interaction research by increasing dividuals.4 I avoided the words "trait" the number of tools in the methodologi- or "personality variables" because these cal arsenal. As I stated, "I am not advocating methodological solutions to 3 Fisher and Hawes, pp. 444-53; Leonard C. essentially theoretical problems. . . . Hawes, "Development and Application of an Interview Coding System," Central States I am merely proposing a set of options Speech Journal, 23 (1972), 92-99; also see 2 among which a theorist might choose." Leonard C. Hawes and Joseph M. Foley, "A Markov Analysis of Interview Communication," Thus, I believe that Ellis has miscast the Speech Monographs, 40 (1973), 208-19; Leonard C. Hawes and Joseph M. Foley, "Group Deciintent of my article. sioning: Testing a Finite Stochastic Model," in If ISM researchers ask questions for Explorations in Interpersonal Communication, ed. Gerald R. Miller (Beverly Hills: Sage, 1976), which their techniques are appropriate pp. 237-54; Donald G. Ellis, "Relational Conand if the answers will advance com- trol in Two Group Systems," Communication 46 (1979), 153-66. munication theory, I have no quibbles Monographs, 4 "[DJifference/individual: any psychological with them. On the other hand, if ISM character, quality, or trait . . . by which an may be distinguished from others." researchers seek to pass off their ap- individual See Horace B. and Ava Champney English, A proach as the most useful one for com- Comprehensive Dictionary of Psychological and 1 See B. Aubrey Fisher and Leonard C. Hawes, "An Interact System Model: Generating a Grounded Theory of Small Groups," Quarterly Journal of Speech, 57 (1971), 444-53. 2 Dean E. Hewes, "The Sequential Analysis of Social Interaction," Quarterly Journal of Speech, 65 (1979), 57.

Psychoanalytical Terms (New York: Longmans, Green, 1958), p. 152. Lillian G. Portenier, "Individual Differences," in The Encyclopedia of Psychology, ed. Philip Lawrence Harriman (New York: Philosophical Library, 1946), pp. 249-62, treats individual differences as characteristics which differentiate people, including environmentally and genetically induced differences.



would have unduly restricted the range of individual differences that could be investigated. Instead, my treatment emphasized the categorical interpretation of individual differences for I referred to simple identity,5 "doctor" and "patient" roles,6 roles generally,7 subgroup membership, or membership in other aggregate entities (teams, families, etc.).8 Thus, if there are N individuals in a group, each of whom can be discriminated on some basis and each of whom has a repertoire of K categories, then incorporating individual difference information into a sequential analysis simply means constructing an NK-by-NK transition matrix. For instance, if one is coding family triads (N = 3) with a two category system (K = 2, friendly/ hostile), the transition matrix for this interaction is six-by-six (husbandhostile, husband-friendly, wife-hostile, etc.). Of course, this can and has been extended to larger groups and greater numbers of categories.9 This description of my use of individual differences should clarify many of the points of disagreement between Ellis and me. For instance, he argues that "patterns of interaction result from people adjusting to their fellows—not persistent psychological states." Later, he contends that research on attribu5 Hewes, p. 64, n. 54. By "simple identity" I mean that no attribute is assigned to the person other than identity. Thus, "persons" are treated as dummy variables in the same way that "subjects" are treated in a repeated measures design. 6 Ibid., p. 62. 7 Ibid., p. 64. 8 Ibid., p. 64, n. 53. 9 For instance, Dean E. Hewes, Sally Planalp, and Michael Streibel, "Analyzing Social Interaction: Some Excruciating Models and Exhilarating Results," in Communication Yearbook IV, ed. Dan I. Nimmo (New Brunswick, N.J.: Transaction-I.C.A., 1980), pp. 123-41; Dean E. Hewes, Sally Planalp, and Michael Streibel, "Methods for the Analysis of Mutual Influence in Small Group Decision-Making," paper presented at the Central States Speech Convention, 1980; Hewes, p. 64.

DECEMBER 1980

tional bias indicates that people overestimate the role of individual predispositions in determining others' behavior. Through I agree in part with both statements, they do not affect my position. As his arguments indicate, Ellis has construed my use of "individual differences" to mean "stable predispositions" or "traits." My use of "individual differences" in no way commits investigators to the study of "stable psychological states." Nowhere in my article is there any commitment to the study of persistent personality states in interaction analysis. In fact, they are never mentioned. Such states are certainly one form of individual difference that can be encompassed within the framework I suggested, but they do not exhaust its applications. Ellis and I are not at loggerheads at all, except that I would leave the door open for the use of stable predispositions in interaction analysis.10 Ellis does raise one other issue on individual differences that requires comment. The ISM tradition, as evidenced by Ellis' discussion and the extant research, holds that individual differences, even when empirically significant, can be safely ignored. Thus Ellis found such differences in his study of problemsolving groups, as did Hawes and Foley;11 yet, these researchers ignored 10 Since my system for employing individual differences involves the use of a transition from an individual to another individual, the regularities captured by it are not simply the stable effects of personality traits, but rather the interaction effects of the speaker's predispositions with the context provided by the listener's predispositions. Such interaction effects add considerable explained variance to models containing only simple individual difference information. See Norman S. Endler, "The Role of Person-by-Situation Interactions in Personality Theory," in The Structuring of Experience, ed. Ina C. Uzgiris and Fredrick Weizman (New York: Plenum Press, 1977), pp. 343-69. 11 Ellis, "Relational Control"; Hawes and Foley, "Group Decisioning," see particularly Table 2 (p. 245) for individual differences in



445 HEWES

these differences in their descriptions of interaction patterns. Why? A colleague pointed out that they did so because they believed that for theoretical purposes they could ignore any data they wished. In contrast, I hold that in this area the researcher's interests must be subservient to the empirical world. Here lies a real difference in value between Ellis and me. Briefly, I believe that when one makes a theoretical or methodological assumption and it is easily tested, one should test it or its consequences. To ignore individual difference information is clearly an assumption—one which ought to be tested for three reasons:

one goal of sequential analysis is to provide as complete a description of interaction as possible, investigators should not throw away sources of systematic variation. Hewes, Planalp, and Streibel demonstrated the potential severity of this loss in a pilot study.12 In a four-person problem-solving group there was a statistically significant loss of systematic variation when individual difference information was ignored (x2 = 273.5, df = 95, p