[Page 1]

PROCESS VARIABLES FOR PSYCHOTHERAPY RESEARCH [1]

Eugene T. Gendlin, Ph.D.

Wisconsin Psychiatric Institute

University of Wisconsin

Five Characteristics of Research Variables:

In this paper I would like to propose some characteristics of good variables in psychotherapy research. By "variables" I mean what we pick out to observe, measure, and study. A research can be excellent, yet the whole of it may be pointless if the variables are not chosen and defined well.

1) The first characteristic I propose is that research obtain its variables from practice. Academic science was, for a long time, afraid of the vagueness of the raw observations of clinical and human concerns. And many practitioners feared equally the objectification of science. We had two opposing views which came to the same thing: the "hardheaded" view that science should ignore human concerns, and the "soft-headed" view that human concerns are best off when untouched by science.

The growing research into psychotherapy (or counseling) is changing this situation. The researcher into psychotherapy cannot be a person who knows and feels little of what is significant in practice. He must be deeply engaged in direct naturalistic observation of practice, so that he can formulate significant hypotheses and variables. At our stage of science, the selection of variables is all important. Nor is this kind of observation at all "unscientific." One must not confuse the aim of science with its

[Page 2]

starting point. Science aims at precise statements of operational measurement. On the other hand, science starts with naturalistic observation to select and define variables. Research refines observation from practice. It analyzes, defines, and devises many steps of operational definition. Yet, no matter how many defined steps intervene, basically the variables should be the same as those which would be talked about in a case conference, or used in thinking about an hour of practice. When at least some of the variables are basically the same in research as in practice, it then becomes possible to relate many other measures to the psychotherapy area. One needs at least some practice variables to settle significant issues by research, and to obtain research findings that can have a bearing on practice.

2) A second characteristic: We need to study variables of the actual doing of psychotherapy. When a number of professional people call what they do "psychotherapy," we know how to measure their personalities and those of their clients. We even know, to some extent, how to measure changes in the clients. But we do not yet know how to measure or define what these professionals do which makes it "psychotherapy." We can compare the clients to a "control group: which does not receive therapy. But, when such a control group is compared to the "experimental group," we have no way of defining or knowing whether the experimental subjects have received the "treatment." Perhaps half of them did not receive "psychotherapy" any more than the controls did! Nor can we provide the same "treatment" when we replicate the study. Measures of antecedents and change cannot be associated with psychotherapy unless we also define and measure the psychotherapy itself in some way.

[Page 3]

Gradually, this direction of research effort is emerging: The study of the psychotherapy behaviors themselves (and their differential effects). It is a new science, a science of "how." It involves a new type of research variable: aspects of in-therapy behavior, differential effects of different things which may be done during the therapy hour.

3) A third characteristic is that the variables concern change. We do not want to know only how the patient is and was, nor even just what he turns into. We would like variables of the change process itself. What factors constitute the changing?

4) A fourth characteristic of research variables of this kind is that they are objective measures relevant to the essentially subjective process of psychotherapy.

In practice we are not afraid of the subjective. Our practice depends upon subjective events in ourselves and in the client. Research must not drastically distort what it wants to study before it even begins. If we want to study an essentially subjective phenomenon, we cannot begin by pretending that it isn't subjective. We have to face the problem of building a bridge from human subjectivity to objective measurement. For this we can employ theoretical concepts that refer to subjective experiencing. We can use these concepts to guide us to relevant choices of observable factors. And there are observable factors which are subjectively relevant. Otherwise it would be mysterious how we ever sense anything about another person. No matter how intuitive it seems, what we know about another person involves what we observe of him. Our concepts about subjective experiencing help us to select and define the significant observable variables.

[Page 4]

Concepts about subjective experiencing are, of course, a theory. (Gendlin, 1962) Theory always has the function of helping to select observable variables. We may love theories for their own sake, and for their life significance, but from the research point of view theory is an heuristic aid to help select variables and form hypotheses. Theory does not consist of, but leads to scientific statements of precisely defined observable variables and their tested associations.

The special type of theory involved employs concepts which refer to subjective, concrete, privately felt events. With such theory we can select and formulate categories of observable events which are likely to be important because they are those by which we think we recognize significant subjective events.

5) A fifth characteristic which I want to propose for effective research variables is that they are process variables rather than content variables.

Psychotherapy process studies began some ten or fifteen years ago, but although called "process" they were really verbal content variables. Did the individual in therapy most often talk about himself, or others? about his past or his present life? about his sexual conflicts or his self-perceptions? How did these contents change over therapy.

More recently we are finding that verbal behavior can be analyzed not only in terms of content—the "what" is talked about—but one can also analyze how an individual expresses himself. Psychotherapy, it is widely recognized, does not consist mainly of conceptual content. It is a feeling process, a subjective change process. Only a little of it is defined and put into the content of words. Alexander (1946) calls this feeling [Page 5] process emotional learning. I call it "experiencing." (Gendlin, 1961, 1962; Gendlin & Zimring, 1955) Experiencing is the concretely ongoing stream of experience. Conceptual contents, emotions defined as just this or just that, these are only some few abstracted aspects of experiencing, which should not be reified and taken to be units or packages inside an individual. Rather, experiencing is "pre-conceptual," a stream of felt meaning sensed in a bodily way. It does not come in the forms of personality content units.

Effective psychotherapy, I propose, will become definable not in terms of contents, but in terms of behavior variables of the manner of experiential process, and the role which this experiential process plays during the therapy hours.

To sum up the five proposed characteristics of therapy research variables:

• (1) they come from practice and are the same as would be discussed in practice.
• (2) they concern what happens during psychotherapy—what client and therapist do.
• (3) they are variables of ongoing change.
• (4) they are observable factors selected as relevant to the subjective processes of therapy.
• (5) they are content-free, that is to say, they are variables of the manner of experiential process.

Illustrations of this type of variable in client behavior:

Let me illustrate the difference between such process variables and content variables: My illustration begins with a negative finding. In [Page 6] one study (Seeman, 1954) cases were rated according to whether the client focused chiefly on his problems or whether he focused more on his relationship with the therapist. It was found that relationship focus was not associated with success. It seemed to make no difference whether the client focused on the relationship with the therapist, or not. The intention had been to measure relationship engagement. It seemed likely, however, that what had actually been measured was the client's verbal content—did he often talk about the relationship or only rarely?

The new type of process variable might be illustrated by a further study (Gendlin, Jenney, & Shlien, 1960), which built upon and repeated the first and added the following two scales:

• 1) How important to the client is the relationship as a source of new experience? Example: (the client says) "I've never been able to let go and feel dependent as I do now" or "This is the first time I've ever really gotten angry at someone."
• 2) To what extent does the client express feelings, and to what extent does he rather talk about them? This scale differentiates direct expression from report about one's feelings, regardless of whether the feeling is past or present.

These two process-oriented scales significantly predicted therapist outcome ratings, while the content scales again showed that it didn't matter whether the client's verbal content was often or rarely about problems or about the relationship.

In a different study, van der Veen (1962) applied one of these new scales ("direct expression" vs. "talking about") to tape recordings of psychotics in therapy, again finding it associated with outcome.

I am citing these studies here not to give findings, but to illustrate a type of research variable. Content-free variables of the manner of [Page 7] experiencing and expressive behavior. This example may illustrate a "content" and a "process" way of observably defining the variable "relationship engagement."

A sequence of studies have employed a scale developed by Carl Rogers. (Rogers, 1958, 1959, 1960) Rogers incorporated concepts of the process of experiencing into his work, and used these and other concepts to formulate a "Process Scale." To give an idea of the scale, I will cite a few lines from four of its variables: "experiencing," "personal constructs," "self," problems."

On the low end of the scale:

The individual is very remote from his experiencing and unable to draw upon it or symbolize its implicit meaning.

Personal constructs are extremely rigid, unrecognized as constructs, and thought of as external facts. The individual is unaware that he has construed experience as having a particular meaning.

There is. . .expression about self as an object and about self-related experience as objects.

Problems are perceived as external to self.

These descriptions from the low end of the scale all indicate a manner of activity which probably involves little ongoing change. In contrast, the high end of the scale describes the manner of in-therapy behavior thought to involve the most ongoing change. Note that these scale variables fit the five characteristics I discussed: They are descriptions of practice-relevant in-therapy behavior hypothesized to constitute ongoing change, involving subjective processes but behaviorally described, and concerning the manner, not the content of behavior.

At the high end of the scale:

His experiencing is used as a referent to which he can turn again and again for more meaning. . . .experiencing [is] a [Page 8] referent which can be symbolized and checked or rechecked for its further meanings. . . .

There is a dissolving of significant personal constructs in a vivid experiencing of a feeling which runs counter to the constructs. There is the realization that many personal constructs which have seemed to be solid guides are only ways of construing a moment of experiencing.

The individual is living some aspect of his problem in his experiencing. The problem is not "an object in itself."

The self exists in the experiencing of feeling. At any given moment, the self is the experiencing.

Walker, Rablen, and Rogers (1960), Tomlinson and Hart (1962),
Tomlinson (1962a, 1962b), Gendlin et al (1962), found that these high scale descriptions apply significantly more often to the therapy behavior of success cases. Psychotherapy seems more effective when more of the behaviors described higher on the scale occur. These behaviors (so far as they have been defined) characterize the "motor" of constructive personality change. When they occur, change is occurring.

In somewhat similar studies, Halloway [1a] and Cahoon (1962) employed other variables of how the individual employs his experiencing as measured from tape recordings, and also found that these variables have predictive value.

Studies by Mathieu (1961), Tomlinson (1962a), and Truax (1962b) applied similar variables to tape recorded psychotherapy with schizophrenics.

Illustrations of this type of variable in therapist behavior:

My illustrations of this type of variable so far have concerned ongoing change in terms of client behavior. Let me now describe some variables of

[Page 9]

therapist behavior, which fit the five characteristics I am discussing. Rogers (1957) hypothesized that psychotherapy depends upon the therapist's congruence or "genuineness" as a person, his "empathy" and his "unconditional positive regard" for the client. These variables have been measured (as applied to tape recordings) with rating scales by Halkides (1958), Truax (1962a, b, c, d, e, f), and Spotts (1962). To give just one illustration of the specific behavior variables involved, here are stages 4, 7, and 8 of a nine stage "therapist genuineness" scale (Gendlin & Geist, 1962):

Stage 4: "continuous artificial style": The therapist is involved in the interaction and is not blocked by or tangled up in difficulties concerning it. On the other hand, he does not respond spontaneously—his response is "controlled by" his style.

Stage 7. "experiential quality": The therapist's expressions are "continuous" with his present reactions and are the direct expressive outcome of these reactions. The content may be about himself or the other person. If he responds to the other person's meanings, what he says expresses his own inward experiencing of these meanings.

Stage 8: "intimate quality": The therapist intends to, and does, reveal more of his own inner processes and feelings than would be necessary just for a full responding to the other person. He is expressing himself intimately, or adding aspects of intimacy by reporting his inward steps (such as in trying to understand).

These therapist variables attempt to define a manner of interaction held likely to involve ongoing personality change in the client. These scale definitions illustrate the five characteristics of variables: They concern practice-relevant behaviors in an ongoing interaction in which personality change occurs; they employ concepts of subjective therapist [Page 10] processes with which to define observable details; they are content-free, concerning the manner in which the therapist speaks, not what he says.

Subjective Process and Behavioral Specificity:

In the last few years we have been moving toward new scales in which each scale point is defined by specific behavior definitions sufficiently precise so that the rater can give a simple "yes it is on the tape" or "no, it isn't here." As yet we are not that definite. There is still some degree of estimating, although raters usually have adequate reliability. As yet, when the raters are interviewed, we find that the ratings have been made not only on the basis of the scale descriptions of behaviors, but also by use of the general concept or impression. But, equally often, the raters report more specific behaviors which they have picked out and decided to classify under this or that scale definition. Our next, more refined version of the scale then includes these behavior definitions which were first formulated by the raters. Rogers' one original scale has led to five more specific scales. There is still far to go, but we are approaching the creation of significant variables that will be quite precise definitions of behavior.

Psychotherapy research is only just beginning. As it advances, it will differentiate and define many specific variables. Such definitions can give us a language, and, as Roger Brown (1961) has shown in psycholinguistics, our ability to recognize given aspects of observation is greatly increased when we have language, or what he calls "codability." It will help greatly when many facets of behavior are defined and given names, so that our language may guide us to notice them. What at first seems purely subjective and intuitive can point to definable observable differences which we can name and notice.

[Page 11]

A Unifying Theory:

I have given separate examples of client and therapist behavior variables. The client's verbalization is one thing and therapist's verbalization is another. But, concretely in psychotherapy, the client's inward process and the ongoing interaction are not two separate things.

The human being is a process of interaction between a body and an environment. Is the food I have just eaten part of the environment or part of the body? The oxygen from the environment which is now in my bloodcells—is it now the body or the environment? A body without the interacting environment dies in a few moments. Experiencing is an inward sensing of body life. We feel pain when something from the environment cuts our skin. We feel fear when something from the environment seems about to cut our skin. We feel—in one inwardly felt datum—the whole complexity of the organization of our body in its present moment of interaction. This complexity is organized by our body structure and all its modifications from learning and conditioning. Every bit of interaction is symbolic and meaningful in terms of the many ways in which the whole organization is affected. Our words and the abstractions of our logic can only name and outline this or that aspect of this organic complexity, the ongoing life process which we feel inwardly. And so, a person "is" not really these and those abstract personality contents. There are no such conceptual content packages in him. There is only the organically ongoing life process which he feels, refers to, and symbolizes, and which is affected every moment by the responses of the environment—especially the interpersonal environment.

For certain purposes, content entities are helpful variables. They are not helpful variables for the study of personality change.

[Page 12]

Because of the nature of language we must speak in terms of abstract entities. Language consists of abstract entities. If I am to communicate intimately in a life-sharing way with you, I must use such abstractions as "I'm angry." But don't take that to mean that a package uniformly constituted of anger is "in" me. As I use the symbols of a language, and as I refer such symbols to my inward experiencing, I will find a whole complex texture, not packages like hostility or other such artificial entities, "emotions" or "contents." For example, as I refer directly to my experiencing, I may find that my anger seems now to be more a hurt. Then I find, perhaps, that I am not so much hurt by what happened as really afraid. As I refer to my experiencing I may symbolize a complex texture of felt meanings. Perhaps I am afraid because I feel helpless. But I don't want to cope with what I'm up against, I may feel a moment later. I am really neither angry, nor hurt, nor helpless, nor wishing I didn't have to cope. The texture of what I am is already changing and different as I am coping with it in talking to you, and as I find myself with the energy to cope, and the sense of being something active, no longer something cowering. Psychotherapy is not a fact-finding process. It is a change process. Persons are not made up of packaged conceptual contents, they are interaction processes, physiological life processes, inwardly sensed and crudely symbolized linguistically. The act of speaking is much more important than what I say: The very act of saying it to you changes what I am talking about.

The varieties of psychotherapy today are still divided according to the kind of language they use, and according to the kind of content packages which they assume to be in people. If you hold that the most important contents in people are primarily sexual and developmental you are a Freudian. If you say that the most basic contents are attitudes [Page 13] toward the self you are client-centered, and if you see them as power strivings you are an Adlerian. If you see the content packages as avoidances of life and responsibility you are an existentialist. But, I would like to assert, none of these packages exist, except as concepts, as linguistic pointers which can be applied to the stream of felt preconceptual experiencing. Thus the patient can oblige the therapist and use whatever language the therapist likes. Patient and therapist can refer any language directly to the patient's experiencing, or fail to do so. Thus all our orientations have about the same incidence of success in psychotherapy. We still divide orientations along accidental, rather than essential lines. We have currently more than one hundred so-called orientations to psychotherapy. If they all work about equally well, then it is likely that their differences are not the important factors! They differ on the concepts and language they employ. They do not differ, I believe, on the experiential and behavior variables which effect the felt process of personality change.

A unifying theory concerning this experiential interaction process can move past the content differences which divide the field, and can lead to research variables of the manner of ongoing process.

[Page 14]