SOME INTERPERSONAL CONDITIONS
OF EFFECTIVE PSYCHOTHERAPY
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
The author wishes to express his appreciation to
Dr. Sidney M. Jourard, Chairman of the supervisory eemittee,
for his imaginative and unstinting help in defining the problem
area and stimulating the research. The other members of the
supervisory comuttee, Dr. John R. Barry, Dr. George R. Bartlett,
Dr. Hugh C. Davis, and Dr. Marvin E. Shaw were of invaluable
aid in guiding and refining the proposed research. The author
also wishes to thank his wife, Margery, who gave so much of her
interest and energy to this project.
TABZL OF CCONTITS
AC:O.IED3 E TS. ... . . . .
TABLE OF CGEIT. . . . . . .
LIST OF TABLE AND FIGURES . . . .
I. IINRODUCTION . . . . . . . . . 1
Training and experience.
Relation of yrovioua work to Fresent study.
Measurement of verbal behavior.
II. PROCEDURE.E . . . . . . . . . . 21
Maxisna physical presence.
Intermediate physical presence.
lnir nu physical presence.
III. FJSULTS . . . . . . . . . . . 29
IV. ZL:P.VrY L'D CC:ICLUSIONS . . . . . . .. 40
rl2LIOr'.nIY . . ...... . ..... . . . 43
Pr DICE. . . . . . .. . . . . . . . 47
rIOC:AHICf.L EC. . .... . .. . . . . . . . 55
LIST OF TaBLES AID FIGURES
1 Frequency of Verbal Response Category by Group ....... 29
1 Summary of Analysis of Variance for Three
Degrees of Experimenter Physical Presence
and Three Conditions of Response .................... 31
Speculation and research abound on the questions of what
characteristics of the therapist aake for successful psychotherapy
and what procedures constitute the therapeutic process. Shafer and
Lazarus (1952, p. 306) attempt to define the basic function of the
therapist by writing that, "The therapeutic situation must revolve
about and be dependent upon what the patient has to say. The
success of the treatment depends upon understanding the patient, and
understanding cannot be accomplished unless the patient talks. The
technique of getting the patient to talk and to continue to talk must
be the real core of treatment." The present study uses this statement
as a premise and attempts to go on from that point to an inquiry into
how this talking behavior is facilitated.
This research attempts partially to answer the questions
posed by Krasner (1955). In his attempt to outline the major aspects
of psychotherapy amenable to controlled experimentation he asks, "What
aspect of the therapist's behavior oan be isolated and used as a stim-
ulus in the control of the patient's behavior? In eeaence, the question
is, what are the basic irreducible elements of an interpersonal situation
which must exict in order to term the situation a psychotherapeutic one?"
To these questions Krasner responds by writing,
First of all, we must start with the fact that the therapist
is physically present with the patient. In all psychothera-
pies, the therapist, if only by his presence in the same room,
indicated that he is interested in the patient end pays at-
tention to the behavior of the patient. It sounds like the
simplest and most axitmatic thing to say about psychotherapy,
but it would seen reasonable to assume that the factor of the
therapist listening, paving attention, showing some interest,
is a basic and indispensable variable of the therapeutic
situation. The therapist focuses more attention on those
aspects of the patient's verbal emissions which his particular
orientation calls for, but in any case he displays a general-
ised form of behavior cues which we may label as "attention."
These behavior cues may consist of looking at the patient's
face, smiling, writing, nodding of the head, picking up or
putting down a pen, saying "ca hum" or, most common of all,
using overt speech. It would certainly seen that these cues
have some effect on the patient's behavior, more specifically
his rate of verbalizing and the types of material he verbalizes.
It is to this question of the influence of the gross physical presence
of the therapist, that the present study is addressed. We propose to
introduce degrees of physical presence of an interviewer as an inde-
pendent variable, to determine whether it has any effect on measurable
aspects of a subject's talking behavior. It is felt that this is of
importance because this supposition of therapist presence and attention
constitutes an underlying, and often unexpressed premise of most of the
research already done on the therapist, his personality, and his modes
of functioning in the therapy situation. To some of this earlier lit-
erature we must now turn.
A number of studies bear on the personality of the therapist in
relation to therapeutic competence. They attempt to demonstrate that per-
sonality factors affect the proficiency of the psychotherapist. Whitehorn
and Bets (1954), used patient improvement rate as the criterion for rating
competence. These ratings were based on therapy with schizophrenic
patients. All therapists achieving sixty-eight percent or better
improvement rates were designated "A" therapists while those not
attaining this level of improvement were designated "B" therapists.
When a detailed analysis of case records was performed it was found
that "A" therapists tended to select personality oriented rather
than psychopathology oriented goals in treatment, that they had
greater facility in grasping personal memnings and motivations of
the patient's behavior, and that they expressed attitudes freely on
problems and set limits. When tie Strong Vocational Interest In-
ventory was administered it was found that therapists in the "A"
group showed interest in the categories of Lawyer and CPA, while "B"
group therapists had interests like those of Printer and Yathematical
Physical Science Teacher. This would suggest that the therapist's
interests play an important role in his competence in the therapeutic
setting. It is further suggested that the differences in interest
between the two groups of therapists significantly affected the in-
terchange between the therapist and his patients.
Knupfer, Jackson, and Lrieger (1957) examined the relation-
ship between personality characteristics of forty psychiatric residents
and ratings of competence. A rating of therapeutic competence was made
on each resident by three supervisors. Subjects were divided into high
and low on this rating and each subject gave a self-description by Q-
sort. Some of the characteristics of those rated high in therapeutic
1. Expresses hostility directly.
2. Sarcastic and cynical.
3. Puzhes limits. Sees what he can get away with.
4. Has a rapid personal tempo. Thinks and acts rapidly.
5. High Level of aspiration.
6. Seeks out opposite sex.
7. Able to sense other's feelings.
8. Able to convey personal feelings and thoughts.
Those who were rated low in therapeutic competence had the following
1. Behaves in an indulgent and forgiving way.
2. Is suggestible. Overly responsive to other people's
evaluation rather than his own.
3. Has a brittle ego defense. Small margin of '-tergration.
Parloff (1956) used two "expert" therapists who treated the
same patients at different times. Based on colleagues' ratings of
quality of social relationships established by these experts ihith
their peers, and judges' ratings of relationship established between
therapist and patients it was found that the therapist who was able to
establish the better therapeutic relationships had also established
the better social relationships. Moreover, the therapist who perceived
a patient as approximating his Ideal Patient concept more closely
created the better relationship with that patient.
It was found by Bandura (1956) that anxious therapists were
rated as leas competent than therapists who were low in anxiety. The
results were interpreted as indicating that the presence of anxiety in
the therapist, whether recognized or not, affects his ability to do
successful psychotherapy and that insight into his enxieties alone is
Hiler (1958) studied whether therapists differ with respect
to the Lype of patients bending to remain in treatment or break off
prematurely. He also investigated what characteristics of the
therapists may be responsible for the differences in their patients'
reaction to them. He found that:
1. Therapists in general differ in regard to th- type of
patients who continue or discontinue treatment with them.
2. Whether the th..rapist is a psychiatrist, clinical
psychologist, or psychiatric social -wrker seea unrelated to the
type of patients who continue or discontinue treatment.
3. whether the therapist was male or female did iske a
difference. ::.c female therapists tended to lose slig'iLly more of
the productive patients thrn did the male therapists.
i. Therapists rated as warm and friendly were able to
keep in treatment a larger percentage of unproductive patients than
therapists rated as least warm and friendly.
5. Therapists rated as most competent at analytically
oriented therapy tended to lose fewer productive patients than
therapists rated as least competent.
On the basis of a series of studies Snyder (1953) listed
a number of therapist variables which seem to affect the patient-
doctor relationship. Among these personality characteristics are
self-insight, insight into others, adaptability, and emotional
Golndan-Eisler (1952) investigated individual differences
between interviewers (three senior psychiatrists) and the effects of
these differences cn the interaction patterns of patient-interviewer.
Each of her interviewers used his own pattern of interviewing. She
found that each of the three psychiatrists had his own individual
interaction pattern regardless of the type of patient he was inter-
viewing (depressed versus active patients). It was further found that
the three therapists influenced the interaction patterns of the same
ten patients in different ways such that depressed patients talked more
with one therapist than with another, while these same therapists
had opposite effects on talkative patients.
Luborsky, Holt, and Morrow (1950) found that of 247
psychiatrists in training, those who were rated as the better
therapists established better relationships with their supervisors
and their fellow residents.
In a follow-up study Luborsky (1952) noted that, of these
same 247 psychiatrists, the therapists who were subsequently rated
as most competent were the ones who had also established the better
relationships with ward personnel and with research project staff.
It was noted that the ratings of personal liking for the therapist
by the research staff were better predictors of the doctor's com-
petence as a therapist than were the other measures used.
In summary, a considerable body of literature exists to
show that interviewer personality factors affect the process and out-
come of psychotherapy. These studies do not reveal, however, any of
the interpersonal communication processes whereby the subject is
sti~alated by the therapist's personality and responds accordingly.
It is the purpose of the present study to partially study this
Training and experience
The training and experience of the psychotherapist as it
related to therapeutic efficiency has also attracted considerable
research attention. Fiedler (1949, 1951) found that experts estab-
lished better relationships than their nonexpert colleagues. More-
over the description of the relationships made by experts of one
school more closely approximated those made by experts of other
schools than they resembled relationship as described by nonexperts
of the same scl.ool. In each of the f;.ctor analyeal performed on the
descriptions in the later study, one factor or pair of correlated
factors was fou.J which clearly differentiated experts from nonexperts
regardless of school. These factors were related to the therapist's
ability to communicate with and understand the patient, and to his
security and his isotional distance to the patient. No factors were
found which clearly separate therapists of one school from those of
another. Fiedler concluded from his data that the hypothesis of his
series of Livestigations, that the nature of the therapeutic relation-
ship is a function of expertness rather than school, was supported.
In another study of the same series Fiedler (1950) found
that experienced therapists of markedly divergent orientations (psy-
choanalytic, non-directive, and Adlerian) were closer to one another
in treatment operations than were the novices and the experts of a
homogeneous theoretical persuasion.
Strupp (1955a), addressing himself to the question of the
e_.adLence of systematic differences in technique attributable to pro-
fessional affiliation and level of experience found:
1. The response profiles (using Bales' interaction process
analysis, 1950) of the three professional groups of psychiatry, psy-
chology, and psychiatric social work display a considerable degree of
2. Experienced psychiatrists and psychologists give:
(a) a larger number of interpretive responses,
(b) inexperienced psychiatrist show a preference for exploratory
(c) experienced psychiatrists use more passive rejections.
In a study by Strupp (1955b) the effect of the therapist's
personal analysis with reference to a series of patient communications
and to: (a) suicidal threats, (b) transference reactions, and (c)
schizoid procfctiuons was examined. Therapeutic responses were secured
from twenty-five psychiatrists, seven psychologists, and nine psy-
ciatric social workers of varying degrees of professional experience,
by presenting a series of twenty-seven patient statements extracted
from actual therapeutic interviews. Within the limitations of the
study it was found that personal analysis had a demonstrable effect
on the therapist's verbal behavior. It was also shown that this
effect was independent of the therapist's level of experience.
In a later study Strupp (1958) found that the clinical
evaluations of psychiatrists were similar to those of psychologists
of comparable experience despite marked intragroup differences. In
some respects psychologists as a group appeared to be more passively
expectant in their therapeutic approach than psychiatrists. Ex-
perienced therapists in both groups, however, tended to be warmer
in their communications to the patient. A series of related ex-
periments by Strupp has been reported by him in his chapter in
E,.i-ach (1962). In general, the results showed some differences
between groups of differing professional affilibtion, but these
differences were not as great as the differences between experienced
and nonexperienced therapists of the same profession, nor did the
differences outweigh the similarities between practiced therapists
of different orientations.
In an experiment by Lakin and Lebovitz (1958) seventeen
psychotherapists of three different orientations free ~a~iociated to
the question, "As a psychotherapist, how would you think of this
person?" when given only minimal identifying information of a patient.
Virtually all of the therapists, regardless of school affiliation,
evaluated the patient similarly in terms of potential for recovery.
Beyond this general agreement background factors unique to each
orientation appeared to determine emphasis a well as mode of
Ashby, Ford, and Guerney (1957) studied the effects on
clients of a reflective and a leading type of psychotherapy. They
ummarized their results thusly:
1. The view that a leading and a reflective type of
therapy produce different effects on clients was slightly supported.
2. The view that pretherapy characteristics of clients
relate differentially to client reactions to therapy in reflective
and leading types of therapy was partially supported.
3. The view that individual therapists create different
effects on their clients independent of the type of therapy given
mas partially supported.
4. The view that selected therapist characteristics are
related to the kinds of relationships established, the amount of
defensive or guarded verbal behavior elicited from clients, and the
amount of change in adjustment produced in clients was not supported.
5. The view that the interaction of the therapist as an
individual and the type of therapy he is deploying effects clients
was partially supported.
Fey (1958) examined the extent to which differences in
doctrine and in experience tend to be reflected in the behavior of
psychotherapist. The primary data consisted of therapists' reports
of their handling of issues often arising in treatment. The respondents
were Rogerians, analysts, young eclectics, and older eclectic. Corre-
lations among therapists indicated greatest homogenisty among the
Rogerians, least among the analysts. Comparisons of the responses
typical for each group suggested that the analysts and young eclectic
resembled each other most while older eclectic and Rogerians were
In a questionnaire survey of British psychoanalysts,
Glover (1955) deonstrated that the practice of a relatively homo-
geneous group of therapists could by no means be considered equiv-
In general, the studies on the effect of the interviewer's
professional affiliation and training on patient's behavior have
shown that the experienced interviewer of one discipline bears
greater sidilarity in is behavior to experienced interviewers of
another discipline than he does to less experienced interviewers of
his own discipline. This would suggest that experience itself yields
to the interviewer some as yet undetermined characteristic that
facilitates the interpersonal relationship. We are thus led to be-
lieve that the physical presence and attention of the interviewer
is not the basic variable of psychotherapy as Krasner (1955) suggested,
but rather an ancillary one.
Little research has been done in the area of the effect of
the therapist's physical appearance upon the therapeutic relationship.
One notable exception is the study by Robinson and Rohde (1946). In
this study four interviewer groups were used: (a) Jewish appearing,
(b) non-Jewish appearing, (c) Jewish appearing who introduced then-
selves with Jewish names, and (d) non-Jewish appearing who introduced
themselves with non-Jewish names. The subjects were asked the follow-
ing two questions: (a) "Do you think there are too many Jews holding
government offices and jobs?" and (b) "Do you think the Jews have too
much power?" The frequency of anti-Semitic responses on both questions
was greatest where the interviewer did not appear to be Jewish. As
the Jewish identification increased there occurred a decrease in
the frequency of enti-S3aitic responses.
Relation of previous work to present study
We have thus far seen that the therapist's personality,
experience, professional affiliation, and appearance can and may
influence the behavior of the patient in therapy. These are aspects
of the therapist which are at least partially out of his own control.
There are two prime waya that have been used historically in rsycho-
therapy to minimize the effect of these otherwise uncontrolled
variables. In the Freudian psychoanalytic situation the therapist
does not sit in view of the patient and thus provides lose infor-
nation to the patient about his personality and appearance. The
same effect is achieved by larpuan in his use of written material.
arpman asks the patient, after a few preliminary interviews, to
answer a series of questions taken from his own autobiography. These
answers are subaitted in writing and when they have been received the
therapist prepares a typed memorandum for the patient. This exchange
constitutes the major portion of the therapeutic situation. Incorpo-
rated in this therapy procedure is the use of bibliotherapy in which
the patient's attention is further focused on written material rather
than an interpersonal relationship.
In the licht of Krasner's (1955) statement on the funda-
mental importance of a physically present, attentive, listening
therapist it would appear that both Freud and Karpman have negated
the basic foundatiO'n of psychotherapy, the interpersonal relationship.
There is research, however, that would tend to support the technique
of encouraging physical and social distance between therapist and
patient. Ellis (1948) studied the love relationships of female
college students by personal interview followed by questionnaires
filled out one year later by the same students. In general, the
subjects exhibited less favorable (that is, lees acceptable in our
society) response patterns on the questionnaire than in the inter-
view. Of the sixty-nine subjects, fifty-three gave on the whole
less favorable questionnaire than interview responses, eight about
the same, and only eight more favorable responses.
Further evidence tending to confirm Ellis' general findings
was found in a study by Metaner and Mann (1952). Anonymous question-
naires, group administered, covering the area of satisfaction with
job and supervisor were obtained from workers in a utility company.
Personal interviews with 328 of these respondents were conducted at
a later date, using two questions that were similar to the original
wording in the questionnaire, but not identical. Comparison of the
results revealed a general tendency among workers to report less dis-
satisfaction in the personal interview. The change in procedure had
a differentially greater effect on blue-collar workers than on white-
collar workers. These differential effects supported the notion that
the anonymity of the self-administered questionnaire permits greater
expression of unsanctioned attitudes, since the blue-collar workers
in general were found to be less satisfied with their work.
Lazarsfield and Franzen (1945) summarzied their research on
the differential effects of face-to-face interviewing versus question-
naire interviewing with the statement that, "These findings substantiate
several claims that are usually made for mail answers: (a) bias
that comes from the reeponaent'e desire to impress or conceal from
the interviewer is eliminated, (b) answers to personal questions are
more frequently given an an anonymous mail reply, and (c) a mail
reply is filled out in leisure and thus produces a more thoughtful
While pertinent to the present study in terms of the light
it shed upon the effect of the presence of an interviewer on tl.
quantity and quality of information yielded in an interview, the
extent to which questionnaire research can be generalized to the
psychotherapeutic situation is unknown. More closely approximating
the actual therapy situation is Colby's (1960) research. In his
study normal ale subjects were asked to free associate for a series
of half hour periods. During eome of these periods a silent observer
was present and in the reminder of the periods the subject was alone.
Colby's data suggested that productivity was significantly increased
by the presence of the silent, non-interacting observer.
Martin, Iady, and Lewin (1960) placed their subjects into
three experimental therapy situations. These situations were: (a)
talking to a tape recorder, (b) talking to a therapist who would
respond on a non-verbal level only, and (c) a regular therapeutic
situation, nondirectively oriented. The degree of manifest (as
measured by GSR) and reported anxiety was obtained as well as ratings
concerning the patient's tendency to approach meaningful areas of
discussion and mount of associated affect. Over a period of five
half-hour sessions the "regular' group showed a tendency to approach
more effectively laden content and to experience more anxiety during
the initial approach, but showed overall anxiety reduction. Thus,
these experimenters concluded that meaningfulness of the material
dealt with has a direct relationship to therapist involvement.
The present research proposes to investigate the inter-
viewer variable, in terms of degree of physical presence with the
interviewer, and shall utilize the patient's verbal behavior as the
dependent variable. It has been suggested, from previous literature,
t at the mere presence of an interviewer would tend to increase pro-
ductivity (Colby, 1960). This would be in accord with Krasner's
(1955) notion, as cited earlier, that therapist physical presence
and attention are the basic elements of psychotherapy. It is
questionable, however, whether increase in verbal productivity re-
flects true self-revealing behavior as si ested by Martin, Lundy,
and Lewin (1960), or rather just an increase in verbal productivity
devoid cf meaning. It has been noted above, for example, that
questionnaire techniques in which the questioner is not present
resulted in answers which at least superficially, would appear more
The hypotheses to be tested in this study may be summarized
1. The degree of presence of an interviewer will have a
significant effect upon the amount of verbalization elicited from
the subject, with greater presence yielding more verbal responsive-
ness on the part of the subject.
2. The degree of physical presence of an interviewer will
have a significant effect on the amount of self-revealing verbal
behavior the subject will yield. It is predicted that the greater
the presence the more self-evaluating behavior will be forthcoming
from the subject.
Measurement of verbal behavior
The most striking feature of recent research in psychotherapy.
is the variety of techniques now available for conducting such
studies. Matarazzo has pioneered the methods by which the inde-
pendent variable, therapist or interviewer behavior, can be con-
trolled experimentally. In his research Matararzo carefully
structured the experimenter-therapist's reaction times, interruption
or non-interruption behavior, duration of verbalizations, and fre-
quency and duration of silences so as to study the effect that
various patternings of therapist behavior, on a time dimension, have
upon patient behavior in that same dimension. His use of Chapple's
Interaction Chronograph provides a useful and objective tool for
measuring behavior change in the therapy situation. For a more de-
tailed report of Mataranzo's work the reader is referred to litu-az:c's
chapter in Bachrach (1962).
Goldman-Eioler (1952) used much the sane dependent variable
as does Matarazzo. She found that interviewers with different action-
silence patterns effected differentially the action-silence patterns
of their subjects. Others who have used action-silence as the de-
pendent variable in psychotherapy research have been iaclay and Osgood
(1959) and Timans, Rickard, and Taylor (1960).
Lorenz and Cobb (1952, 1953, 1954), as well as other re-
searchers, have used grammatical distribution and word frequency of
patient and normal groups, as a dependent variable. Comparison of
patient with control language patterns indicated that there were
statistically significant differences, uhich could be justified on the
basis of pathology, between the normal controls and the patients.
Jaffe (1958) analyzed interview material by using what
he called a Type-Token Ratio (TTR). The ratio was obtained by
comparing the number of different words with the total number of
words in blocks of fifty words. The value of such a method lies
in its emphasis upon redundancy on the part of the interviewee
which was considered a negative measure of the amount of new and
different information that is being imparted in the fifty word in-
terval. Thus, if a person were to say only one word over and over
he would obtain a TTR of 1/50, whereas a person who never used the
same word twice in a fifty word block would have a TTR of 50/50.
This procedure avoided the difficulty that Kanfer (1959) faced
while using as his data words per minute. In this case, if a sub-
ject rapidly repeated a single word his score, which was intended
to denote rate of information flow, might be higher than somebody's
who spoke slowly, to the point, and was quite self-revealing.
Larger elements than the single word have as well been
used in content analyses of interview behavior. Lennard and Bernstein
(1960) used propositions as their basic element. They used this unit
making the assumption that a long statement (one containing many
propositions) contained more information than a short statement (con-
taining one or only a few propositions). Colby (196 9 preferred the
sentence, which he called "a single, minimum, free utterance" as the
basic unit in his research.
Bales (1950) introduced a coding method which essentially
broke down the dimension of cooperation-resistance into twelve
separate categories corresponding to a like number of degrees of
cooperativeness. Another method for categorization of verbal exchange
was that proposed by Coffey, Freedman, Leary, and Ossorio (1950)
and elaborated by Freedman, Leary, Ossorio, and Coffey (1951). In
this system there are sixteen scoring categories in which behavior
may be represented. These categories were designed so as to give a
graded representation of all possible aspects of interpersonal ex-
Heyne (1948) presented another scheme for the study of
verbal content in interpersonal interaction. Primarily concerned
with group problem solving behavior he defined twelve separate cate-
gories such as "goal setting", problemm proposals", and "information
seeking" to indicate all the possible activities the members of a
problem solving group might engage in during their interaction with
one another. An approach emphasizing the motivational component of
interpersonal behavior was presented by Steinsor (1949). His final
system consisted of two category sets. The major one is comprised of
eighteen categories and one residual category. Categories such as
'activate and originate", 'structure and delimit", "diagnose by
labeling', were focused on the intent of the actor, ignoring the
motivational consequences of the act in the other members of the group.
Murray (1956) devised a method which encompassed nineteen
categories for interviewee's content and ten for the interviewer.
Porter (1943, 1943a) concerned himself with classification of the
interviewer's responses. Concerned with client-centered therapy, his
classification system emphasized the degree of responsibility which
was assumed by the therapist. Snyder (1945) used a modification of
Porter's categories to designate the techniques used by the interviewer
in a non-directive therapy setting. Some of his categories were
'restating content', clarifying feeling', 'interpreting", 'structur-
ing', and leadingg. Snyder was concerned, as well, with the client's
responses and used a separate classification for these responses
which included such behaviors as simple responses, planning, and in-
sight. A classification system developed by Curran (1945) used as a
measure of insight the relating of two previously unrelated problems
by the client. Curran's classification system is of note in t;.at he
attempted to indicate not only that the interviewee was discussing a
problem, but an attempt was also made to indicate the general nature
of that problem.
Dollard and Mowrer (1947) created a codification system
based on the theoretical tenets of learning theory. Their discomfort-
relief quotient (D.R.Q.) was obtained by taking the ratio of the num-
ber of discomfort and relief words. These authors applied the D.R.Q.
to larger units than words as well, such as sentences or "thought
unite". Rainy (1948) presented a similar syctea to the D.R.Q. with
his positive-negative-ambivalent quotient (PNAvQ). It differs from
Dollard and Eower's system only in that it concerns itself solely
with the client's self-referring comments and not to everything he
says as does the D.R.Q.
Fisher (1956) created a system similar to that of Murray
(1956) in which his main concerns were with depth of interpretation
and plausibility of interpretation. Weeks (1957) reported a study in
which each client's responses were rated on a five point scale of ex-
pressed affect. Strupp (1957) presented a method of analysis of
interview material with special emphasis on the activity of the
therapist. Eac, of the therapist's comments are viewed from five
different vantange points: (a) the kind of technique employeJ by
the therapist, (b) the amount of inference employed by the therb-
pist in his comments to the patient, (c) the focus of the therL-
poutic interventions, (d) the degree of initiative assued by the
therapist, and (e) the warmth versus coldness in the attitude of
the therapist. More comprehensive listing and description of coding
schemes are su.-marized in an article by Strupp in Bachrach (1962),
the Handbook of Social Psychology (Lindzey, 1954) and Auld end
In general, the attempt& to quantify verbal material have
all attempted to cope with the two problems of reliability and
validity in different ays. Cn the one hand we find the studies that
disregard content and measure such aspects of verbal behavior as
auraticn or frequency. This was usually done in the interest of
reliability and in many instanees lost the "what' of the verbal ce-
hbvior in oraer to quantify the '"how" of such behavior. lWile yielding
interesting and valuable information, this approach yields little
that is specific to the problems of psychotherapy. The contrasting
approach evidenced in the cited research emphasizes the content of
the verbal material, but often does this at the expense of reliability
of coding. This is especially true when the researchers have a theo-
retical bias so as to sensitize them to particular aspects of the ver-
bal behavior at the expense of other aspects. Thus, Ahile a content
approach as used in the cited studies is meaninrl ful for psychotherapy,
there remains room for a more quantifiable measure of content in
interviewer research. It is the purpose of this study to attempt
to formulate such a measuring instrument.
The selection of the dependent variable becomes crucial
whea one is attempting to quantify verbal behavior. As noted above,
there have been many attempts in the literature to create reliable
and valid coding schemes for such a quantification. This experimenter
felt, however, that a greater degree of objectivity in the scoring
technique than is available in most of the mentioned systems would be
of considerable value to such research as t.as attempted here. The
first step in creating such a system is the determination of the coding
unit. Piter attempts to use a number of different units it was found
that the simple sentence offered a useful an-d reliable unit. The
following rules were created for defining the unit as it is used in
1. A simple sentence was to be treated as a single udit.
2. A compound sentence wias to be broke down into its main
clauses and each maJi clause was to be treated as a unit.
3. A complex sentence was to be treated like a simple
sentence and thus was to be treated .s one unit.
4. A compound-complex sentence was to be broken down into
its main clauses and each main clause be conaidajed 0on unit.
5. Interrogative and imperative sentences were not to be
considered as units.
In our attempt to determine what would be the most valuable
and yet reliable dimensions for coding the units the question was
asked as to what verbal behavior the psychotherapist most values in
the therapy situation. It was felt that therapists are usually in-
terested in two aspects of the patient's verbalizations. The first of
these may be designated as 'self-descriptive", which has L-'sady re-
ceived some attention in the studies of Lorenz and Cobb (1952, 1953,
1954) and Paimy (1948). It is usually considered to be therapeutically
beneficial for a patient to be talking about himself, rather than of
other people or generalized concepts. Second, when talking of himself,
it is again thought to be most conducive to progress of the therapy
situation if these self-referring statements reveal feelings or internal
states, rather than reporting fact or description.
We thus arrived at the categorization of verbal behavior into
three codable categories: (a) self-descriptive, (b) other-referring,
and (c) self-evaluative behavior. The term evaluative used here is
to be contrasted with descriptive. An evaluative statement is one
which the listener cannot empirically verify. In fact, the usual mean-
ing of truth or falsity cannot be ascribed to such a statenet. It ex-
presses an internal state such as feeling, attitude, or judgement. A
descriptive statement is empirically verifiable. The listener can de-
termine truth or falsity by examination of the world. Thus, if a
person says, "I am ugly" this would be considered an evaluative state-
nent as it is based on a personal standard (of beauty). The statement,
"I em six feet tall" however, is a descriptive statement. It depends
upon other then an individualistic standard and the truth of the state-
ment may be determined by simply measuring the person.
Again a set of coding rules waa established so as to
standardize and objectify the coding procedure. These rules were
1. If the verbal unit yieldE Lnfornmtion about the speaker,
and if it appeared to be the intention of the spel:or to yield such
information about 1imrself, the unit was to be coded as self-descriptive.
2. If the verbal -nit yields information about someone or
something other thai the speaker, or if the unit yields inor-Lation
about the speaker cnl"y by coder inference, that unit was to be coded
3. If the verbal unit met the criteria for the self-de-
ccriptivc ctogcry and in addition was evaluative in nature rather
than descriptive, it was to be treated as self-evaluation and not as
4. If the verbal unit referred to someone else's evaluation,
e.g., 'He felt happy.", it was to be considered as an other-referring
In essence, there wtre two rain aimensions, self-raferring
and cther-referring, with the self-referring dimension further broken
down into the two categories of evaluative and descriptive. A saae-
twhat r-ore detailed description of the coding process :ay be seen in
Forty-five nale University of Florida students in an intro-
ductory psychclogy course were used as subjects in this experiment.
They were all volunteers under a system where the students in the
course were required to Farticipate in at least. two experiments during
the term. Thus, while volunteers, they could not be considered to be
positively motivated towards participation in the experiment.
A tape recorder was used to record all of the subject's
comments. This was .laced so as not to be in the subject's direct
line of vision while sitting in his chair, but there was no effort
made to conceal the machine and the microphone was in plain view on
top of the desk.
Three interview methods were used and each subject w s
randomly assigned to one of the three method groups. All subjects
were seen individually by the same interviewer. The relationship
of this interviewer to the Psychology Department of the university,
as well as other personal data, was not disclosed to th, subject
other than to the extent described below in the statnennt read to
the subject on the nature and purpose of the experiment. The three
interview conditions reflect three degrees of physical presence of
a'md.mum physical presence
The subjects in this group, after being seated in the
interview room and the tape recorder turned on, were given the
We are asking the subjects in different parts of the study
to do different things. We are interested in studying
various factors which affect the process of psychotherapy,
and in one group, the group you are in, we want to study
the effects of having you talk to a therapist who remains
silent. So that, as the therapist in this situation, I
will be listening carefully to what you have to say, but
will respond by non-verbal means only. In other words, I
will indicate that I understand liat you are saying by
nodding my head, but I will not be .ole to say anything
We woula like for 'cIu to ixtaine that you are a client
in psychotherapy. This may seem a little stranr e, but
try to talk about the things that you wculd expect to
in a reou-ir therapy session. People ordinarily talk
about pns. and present events in their lives, the iv-
portant people in their lives, and what sorts of feel-
ings the3e events cr people arouisd within themselves,
that is, their personal emotional reaction to the
people and occurrences. But in general, feel free to
talk about whatever comes to cind. There may be tines
when nothing comes to mind, and that's OK too. You can
Just iait until something does, and then tal: about th.lt.
In general, though, try to keep talking as much as you
Do you have any questions?
ihe subject was seated in the cowmon interview position in relation
to the interviewer, so that by turning approximately forty-five
degrees he would be directly facing the interviewer. D. rin- the
twelve minute period that was devoted to the subject's responding
tre experimenter was permitted no verbalizatlono. He did nod his
head occasionally, but care was taken that no pattern was set up by
the noddinps Ja to reinforce certain response types and discourage
others. Rather, tho purpose of the nodding uas solely to indicate
to the subject that the interviewer was listening and attentive.
Interieriiato nhysicnl presence
The interviewer, with this group, sat behind the subject
and off to one side so that by turring aFproxinately forty-five de-
grees the interviewer would be looking directly at the subject's
bac!. his position was assumed orly after the instnrctions icre
given. T.e instructions to this group of subjects were the srme as
those given to the rA.'=irun physical presence -roup with the sut-
stitution of the following first yarcrgaph:
We are asking the subjects in different parts of the
study to do different things. We are interested in
studying various factors which affect the process of
psychotherapy, and in one group, the group you are in,
we want to study the effects of having you talI to a
therapist who cannot be readily seen. So that, as the
therapist in this situation, I will bo seated behind
you and will be listening carefully to what you have
to say, but I will not respond verbally.
Again, the interviewer was permitted no verbalizations during the
period while the subject was responding.
inimnum physical presence
As explained in this group's instructions, the interviewer
left the room immediately after the orientation of the subject and
did not return until after twelve minutes had elapsed. The members
of this group received the standard instructions with the substitu-
tion of the following first paragraph:
We are asking the subjects in different parts of the
study to do different things. We are interested in
studying various factors which affect the process of
psychotherapy, .nd in one group, the group you are in, we
want to study the effects of not having a therapist in
the room with you. So that as the therapist in this situ-
ation, I will leave the roon and return after a period of
At the end of the twelve minute period the experimenter returned to
the room. There was no attempt on the part of the experimenter to
overhear the subject while the subject was alone in the room.
In order to evaluate intercodcr reliability independent
codings of ten of the experimental interviews were obtained from
two covers. Thus, for each coder there resul6-d ten sets of scores,
these sets composed of the number of self-evaluative, number of
self-referring, and number of other-referring items for the ten
subject coded. These frequency scores were then converted into
percentage scores so that the resulting correlation consisted
essentially of percentage of subject A's responses being self-
referring when coded by one coder as compared to the sane subject'a
percentage of self-referring responses as coded by the other coder.
hien a Pearson Product-tonent correlation was ;e-+formed on this
converted data, using all of each coder's percentages for the ten
subjects (thus yielding an II of thirty pairs of codings) a .96
correlation was obtained. When these coder pairing were analyzed
b;r response (celf-reforring percentcaes coded by one coder as com-
pared with self-referrin percentages coded by the other, etc.)
with a resulting N of ten pairings for each type of response, correla-
tions ranging front .93 to .96 were obtained. Correlation was ueed
instead of percentage agreement because the later would havo been
less useful for comparison with other studies. Correlation figure
are available for mot other coding schaees while percentage agree-
ment between coders has rarely teen reported. A correlation on cn
i.-n by item basis for four subjects ias also lone with independent
codings by these saac tuo coders. In tids instance, the coderu scored
iten by item and the resulting correlations iwre measures of how well
the two coders agreed on how each item should be scored. The resulting
U's for those correlations were thus the total number of codable units
for each of the four subjects. These correlations ranged from .87 to .94.
For tne final coding of the taped interviews, the coder
listened to the tapes and recorded the responses for each of the
three response categories for each subject. A profile was thus
attained for each subject in terms of the frequency in each re-
sponse class for that individual. As an operational note, it was
found that for normal speech rates the coder was able to code with
few pauses to catch up. Thus, it was possible to code a twelve
minute interview in fifteen to twenty minutes.
The distribution of the codable units attained from the
forty-five subjects as distributed amng their groups may be seen
in Fig. 1. In order to perform an analysis of variance the raw
frequency data was adjusted for individual differences in verbal
DESCRIPTIVE OTHER-REFERRING EVALUATIVE
VERBAL BEHAVIOR CATEGORY
-- Maidan Presence Group
---------- Intermediate Preeence Group
-- - Minima Preeence Group
Fig. 1. Frequency of verbal response category by group.
output by converting into proportions. This was done in the sanm
manner as expected frequency is computed in a contingency table.
The percentage of the total group responses of the particular sub-
ject was obtained and this figure was then multiplied by the total
number of responses for the group in a parLicular response category
to obtain the adjusted cell entry for that individual in that re-
sponse category. As a concrete example, if a subject in the maxi-
mum physical presence group gave a total of eighty codable responses,
the subjects of his group gave a total of one thousand codable re-
sponses and five hundred of these were self-descriptive, the indi-
vidual adjusted cell entry for this subject for self-descriptive
responses would be c( 500 = 62.50. The raw frequency data
and adjusted scores may be seen in Appendix B.
The resulting analysis of variance on these converted data
may be seen in Table 1. The source of response refers to the three
-ossible response categories: (a) self-descriptive, (b) self-
evaluative, and (c) other-referring. The group variable refers to
the experimental groups assigned to the subjects, maximum physical
presence, intermediate physical presence, and minimum physical
presence. The performed analysis of variance was of the McNemar,
Case XVII type (MNhemar, 1955, p. 332).
The data were subjected to a series of F-tests to examine
the variance of subjects within groups and also to study subject by
group interaction within groups for each of the three groups. The
ratios proved non-significant at the .05 level of confidence; hence
the assumption of homogeniety of the variances was confirmed and the
legitimacy of the statistical operation involved in an analysis of
variance of this type was demonstrated.
SUW1ARY OF ANAIAbS & VARIANCE FMR IKIK WIMRIim OF
ELXPRIBUN R FTSICAL FIPRSNCE AND 'TREE CGU9ITICES OP
sourom d.c a m F
Respss 2 2,868.7369 1,434.3685 141.8271*
Group 2 3,552.5776 1,776.288S 3.9489w
by roaup 4 33.0603 208.2651 20.592Ba 0
Subjects 42 18,692.4089 449.8211
RmiAder 84 849.5323 10.1135
Total 1 26.996-.396
There mw significant difference, at the .001 level of
oeeoideee, between the fr ency of the various responee types.
In a series of nuing t-tests (see AppeMdix C), it m fund that
the subjects had a siificantly greater aber of self-decriptive
respnas thau etbhr-nrefrring reesp se (p<.001) and a signifi-
eanty greater amber of evaluative respese tha other-rferring
respoe (p< .05) but the different between the self-deeriptive
and self-evaluative reposs not mis ifieent.
As an be seem in Table 1, a sigpificat different in the
verbal output of the thre erpermatal group, using the data ad-
justed for verbal rate, m fewd. This aigifsaut differene
re tmed the .01 level of osefidenee. Wi t-teets ere used to
locate the differeese between grupe that eentributed to the over-
all significant different, it MU ferad that he Maaim Presee
group had a significantly greater number of codable responses, that
is spoke more, than the intermediate presence group (pC.02) but
not than the minimum presence group (p>.10). There was no significant
difference (p>.10) between the intermediate and minimum groups on
number of codable responses.
The interaction between group and response, found to be
significant at the .001 level, indicated that group membership did
influence the type of response made. Again a series of t-tests was
performed (see Appendix C) to find the locus of the difference.
No significant differences were found between the maximum and inter-
mediate presence groups in the frequency of sel'-descriptive responses
(p>.10), while significant differences were found in the frequency of
self-evaluative responses (p<.02, intermediate(maximum) and in the
frequency of other-referring responses (p<.001, intermediate(maximum).
Thus, it was found that the maxinm physical presence group was more
responsive, that is talked more, in two of the three response categories
than did the intermediate physical presence group. When comparing the
maximum to minimum presence groups it was found that there was a sig-
nificant difference between these two groups in the frequency of
other-referring responses (p<.01, minimum(muaxium). There were not
significant differences between the maximum and minimum groups in
either the frequency of se.f-descriptive or self-evaluative responses.
.a the comparison of the intermediate with the minimum presence groups
only the fr quency of self-evaluative responses was found to be
significantly different. This was significant at the .02 level of
confidence with the minimum presence group yielding more self-
evaluative responses than had the intermediate presence group.
There were two hypotheses to be tested in this study.
The first of those was the prediction that the greater the degree
of physical presence of an interviewer, greater the responsiveness
of the subjects. Thus, it was predicted that the maidmu~. presence
group, where the interviewer and subject were in a face to face
relationship, would provide the greatest number of ccdable responses,
and that the intemrediate and minim= presence groups would have pro-
gressively fewer codable respo3es respectively. That there was a
significant difference in responsivity between the three groups was
demonstrated in the analysis of variance. There -as found that the
difference between the groups was significant at the .01 level of
confidence. The ordering of these differences, however, was not
as had been predicted. The maximum presence group did, as predicted,
yield a significantly greater amount of responses than the intermediate
participation group. The differences in verbal output between the
intermediate end mirjmum presence groups, and the maximum and minimum
presence groups were non-significant. This would suggest that face
to face, r.xdimum presence facilitated verbal responsiveness, while
the absence of the stimulation accruing to the face to face situation
inhibits or does not elicit responsiveness. There wes some subjective,
as well as the above quantitative, data available to this experimenter
which tended to support this hypothesis. A number of subjects in the
intermediate and riniraum presence groups, while in the experimental pro-
cess, commented that the situation was like talking to oneself. They
also reported feelinE very uncomfortable in the situation, moreso than
was reported by members of the imaimum presence group. Several subjects
reported that the discomfort they experienced had disrupted their
thought processes. There were other subjects, however, who did not
appear to experience the disruptive effects in the intermediate and
minimum conditions. This would be in accord iith Taylor and Spence's
(1952) findings in the effect of anxiety on performance. Some people
function more efficiently under anxiety arousing conditions while
others experience performance decrement under similar conditions.
There was no attempt in this present study to correlate personality
variables with responsiveness under the experimental conditions,
however, so examination of this factor must depend upon future re-
In ordering the total responses for each of the three
groups it was found that the resulting order was maxinmu, imium,
and intermediate. While the difference between the intermediate and
minimum groups did not reach statistical significance as noted above,
the difference wsa of a magnitude to suggest that with a larger
sample, the minimum presence group might prove significantly more
responsive than the intermediate participation group. The results
of the present study were, thus, in disagreement with Colby's (1960)
findings. Qualifying the discrepancy of results by the fact that
the present study was in no way a replication of Colby's research,
still there were important differences. Colby found the non-
observable, present, silent observer to facilitate the flow of infor-
mation from the subject, while the present study found no significant
difference between the intermediate and minimum groups on the amount
of verbal behavior yielded. There were major motivational as well as
experimental differences between these two studies which might account
for the different findings. Colby's subjects were paid five dollars
per session for serving as subjects. He used only eleven subjects
uhtch could also have had a biasing effect, and all of these eleven
subjects were in the medical profession. It is interesting to note
that Colby's subjects subjectively reported the experience more in
line with the present results than with Colby's. In guessing what
the experiment was attempting to study they concluded that it was
studying the inhibitory effects of the silent observer. This would
be in keeping with the present findings of lowered productivity under
the intermediatu presence condition.
The second hypothesis tested in this ELudy was that the
differing interview conditions would yield a difference in quality,
as well as quantity of verbal material. It was predicted that the
greater the dL-ree of physical presence of the interviewer, the
greater the proportion of self-descriptive behavior, t e greater the
proportion of self-evaluative behavior, and the lessBr the proportion
of other-referring behavior would occur. There did occur a significant
difference between the use of the different forms of response across
subjects as well as a significant interaction effect. MN&bership in
a particular experimental group did affect mode of response. I-Mmber-
ship in one of the three groups thus affects both quantity and quality
of response. This finding would be contrary to those of MHtzner and
. nn (1952). One would e, ect, from their findings, that the minimum
presence gr up would yield the greatest number of self-desc.iptive
and self-evaluative responses, but this was not the case. The max-
immu presence group was equally productive as the iinimui group
in its yielding of self-descriptive responses while no significant
difference was found between the maximum and minimum presence groups
in frequency of self-evaluative responses.
Martin, Lundy, and Lewin (1960) did show a discrepancy be-
tween their nonverbal (our maximum presence) group and their tape
(our minimum presence) group on meaning unit rate. Their meaning
unit rating was designed to assess the revelation of affect laden
material, a unit somewhat similar to the present study's self-evalu
active category. One would thus expect to find a similar relationship
in this study. This is not the case, however, as the t-test showed
that there was not a significant difference between these two groups
on this type of verbal behavior. Martin et al. (1960) however, did
not report on the statistical significance of the discrepancies they
found. If it was the case that their discrepencies did not reach a
statistical level of significance, then the results of that study and
of the present one are in accord. The Martin et al. study did not
report findings on amount of verbal material obtained under the
differing conditions as they measured in time units. This precluded
comparison between the present study and that one on this dimension.
In sumarizing the results, it was found that the maxium
presence group was more responsive and that they talked more about
themselves and about others than either of the other two groups. The
minimum and intermediate groups were alike except in their talking
about self-evaluations. The minimum group spoke significantly more
of self-evaluations than did the intermediate group. Thus, it was
found that in regard to self-evaluative behavior, those subjects who
were face to face with the interviewer behaved like those for whom
the interviewer was not physically present, while under the condition
where the interviewer was physically present but not observable there
were significantly fewer evaluative statements made by the subjects.
This uuld suggest that a psychoanalytic orientation in w ic, the
patient cannot see the uial-yst is the least likely of the three con-
ditions usea here to produce self-evaluative behavior.
The question of generalizability of ths data obtained in this
present study is an important one. As in most studies o. tris nature,
the imposed conditions nake it difficult to use a true patient popula-
tion. In addition, the instructions to the subjects wiic. constituted
an atterLpt to produce a role playing situation where the subject would
act as if he were a patient in psychotherapy has obvious shortcomings.
'he motivation for disclosing personal, perhaps sensitive, emotional
content woa lackidn. There w.s no co.ditment on the art of .ie
subject to a therapy relationship nor, in most cacs, did the subject
see any need for one. Furtheriore, being an experiment, the subject
could easily, in spite of assurance to the contrary,doubt the con-
fidentiality of his remarks. We wouldd also necessarily have to
question how closely the college student population, or even more
specifically, the male college student population at the University
of Florida, would resemble a patient population. Another factor which
compoundcd the motivation problem was that the subjects were, in a
sense, compulsory volunteers. There thus results the possibility of
strong negative motivation towards full cooperation, as a passive
resistance to the course requirement. As one auLject clearly stated,
'I wonder what you'd do if I didn't sa anyt.hinC?"
Another hinderance to generalizability would be the nature
of the situation-itself. The testing situation most resembles the
intake interview and as such would say little about the course of
psychotherapy as a whole. This would be offset by the fact that the
intake often sets the tenor of the later therapy sessions.
In short, care must be taken in generalizin-g the results
beyond the situation as presented in the experiment itself. This
was not a unique fault in design, however. It is extremely difficult
to use a patient population seeking therapy in an experiment which
affords less than optimum opportunity to the patient for relief of
his distress, so that we are forced by circumstance to approximate
such a patient population as closely as possible.
Te intention of the present study was not so much to shed
light onto a great truth of psychotherapy as it was to serve as a
prototype for future research. To this extent it achieved its purpose
as a host of future studies were indicated. It would be of value to
study, as did Martin, Lundy, and Lewin (1960), the longitudinal ef-
fects of therapist interaction on self-evaluative behavior. By this
means it would be possible to ascertain how the initial experience
influences later experience. Thus, in the experiment envisioned here,
the experimenter variable would not remain constant for the subject.
The initial interview mi-ht occur under one condition while all sub-
sequent interviews be under still another condition. Therapist per-
sonality factors would also be of interest in such a paradigm. As has
been noted in the introductory section of this research, a relationship
between therapist personality and therapeutic success has been found
in a number of studies. The unanswered question remains, however, as
to how the differences in therapist personality relate t3 diffor-
ences in patient behavior in the therapy hour. In particular, how
do various therapist personality factors affect the rate of self-
evaluative behavior elicited from the patient? A corollary question
to be answered by further reOsarch would L- the correlation between
self-evaluative patient responses and judced therapeutic success.
It is indeed possible that taldkng about one's feolinrs or thoughts
has no relation at all to remission of symptoms and return to social
productivity. These are just a few of the rzny research ideas that
were engendered by the present research. The shibboleth furtherr
research is indicate" would certainly be warranted here here so
little has been done and so much needs be done.
It is yet premature to make a final judge.ent on the value
of the dependent variable used here. It was found to discriminate
between groups and this would suggest that it was sensitive enough
c pick up behavior differences produced by experimental ccnditiona.
Coder reliability was high and the coding technique was not difficult
to learn. It would be ef value to code inLerviets with this technique
and with a host of others to get some validity estimates. This would,
however, cnly be suggestive of the validity of the measure as it is
difficult to assess, except through the use of factor analysis, what
diicn ions are being uncovered by the instrument. This ezperimenter
feels that such future research would be well warranted in light of
the great need of a reliable, easily codable measure of approach to
affective material in psychotherapy, and by the suggestion that such
an -pproach was to some extent achieved in this study.
SUMJABY AND CONCLUSIONS
Three degrees of interviewer's physical presence were
studied as to their effects upon the subjects' verbal behavior.
In the maximum physical presence situation the interviewer sat
face to face with the subject. The intermediate physical presence
condition differed from the maximd~ situation in that the inter-
viewer was seated behind the subject and then could not be readily
seen. The third condition, that of minimum physical presence, was
one in which the interviewer left the room and thb subject spoke to
a tape recorder. In all of these interviewer conditions the inter-
viewer was permitted no verbal comments beyond the initial reading
The subjects' verbal behavior was evaluated in three
categories. The first of these was a self-referring category which
consisted of a frequency count of the number of times the subject
spoke of himself in a descriptive manner, that is, told facts about
himself that were theoretically subject to empirical verification.
The second category of the subjects' verbal behavior scored was that
of other-referring behavior. This consisted of a frequency count of
the number of statements about someone other than the subject. The
third category was that of self-evaluative statreants. A self-
evaluative status ent was any statement made by the subject about
himself that was based on an internalized reference, that is, was
not subject to empirical verification. In this category were in-
cluded the subjects' feelings, thoughts, a.'itudea, opinions, and
Forty-five rale undergraduate students enrolled in an
introductory psychology course served as subjects. They were
randomly divided into three groups of fifteen subjects each, and
assigned to one of the three treatment groups. The instructions
to the subjects were such as to encourage then .o play the role of
a client .n psychotherapy. All of the interviews were recorded on
tape for later coding.
After the data were adjusted to account for differences
between subjects in gross Caount of talking an analysis of variance
w performed. It was found that there was a siinif'c-nt difference
between ho rl .-c m and intermediate physical presence groups in
amount of verbal behavior elicited, with the mrximum group yielding
more verbal boha.._cr than the iLntermdiate group. There were no
significant differences between the nTnxim~ and mini-an groups on
(aount spoken. A si-iificant interaction between L:-. grcup to
which the subject was as3i nad and the response category profile
ed us to acamiLne the effe:t of group mer.bership on the use of the
thr eo differ ont rosponse categories. It was found, again t ing
the data adjusted for individual differences in -mount spoken,
tha. the maax.-n presence group spoke significantly more than the
intermediate presence group in the self-evaluative and other-
referring categories, but not in the self-descriptive category. In
the comparison between the maximum and minimum presence groups the
only category that was found to yield significant differences was
in the use of other-referring responses. .- the comparison of the
intermediate with the minimum presence groups only the frequency
of self-evaluative responses was found to be significantly differ-
ent. This would suggest that the psychoanalytic positioning, where
the therapist is seated behind the patient, initially yields the
least amount of subjective material of the three situations examined
The results of this study suggested a number of research
ideas which would further test the utility of the particular be-
havioral measures used here. The ability to generalize the findings
of the present study is also limited until future verification can
Ashby, J.D., Ford, D.H., Guerney, B.C., Jr. and Guerney, Louise F.
Effects on clients of a reflective and a leading type of psycho-
therapy. Fsychol. Vono5r., 1957, 1 (24 I.hole Ho. 453).
Auld, F., Jr. and lurray, E.J. Content analysis studies of psychotherapy.
Pnvchol. Full., 1955, .2, 377-395.
Bachrach, A.J. xperirmental Foundations of Clinical Psychology.
New York. Basic Dooks, Inc., 1962.
Bales, R.F. Interaction Process Cnalysis. Ca.-bridge, Mass.:
Bandura, .. Pj:chotherapists anl ety level, self-insight, and psycho-
therapeutic competence. J. Abnorm. Soc. Psychol., 1956, 52, 333-337.
Coffey, H.S., Freedman, M.B., Leary, T.F., and Ossorio, A.G. Cor=unity
service and social research Group psychotherapy in a church
program. J. Soc. Issues, 1950, o, 1-55.
Cclby, K..'. Experiment on the effects of an observer's presence on the
imago system during psychoanalyticc free-association. Behav. Sci.,
Curran, C.A. Personality Factors in Counselin.. New York: Greene and
Dollard, J. and [tower, O.H. A method of measuring tension in written
doc ients. J. Abnorr. Soc. Frychol., 1947, L2, 3-32.
Ellis, A. Questionnaire versus interview method in the study of h-=an
love relationships. ,mer. Soc. Rev., 1948, 12, 61-j5.
Fey, .a'. Doctrine and experience. Their influence upon the psycho-
therapist. J. Consult. Psychol., 1958, 22, 403-409.
Fiedler, F.E. A comparative investigation of early therapeutic re-
lationships created by experts and non-experts of the pujchcanalytic,
non-directive, and Adlerian schools. Unpublished doctoral disserta-
tion, Univer. of Chicago, 1-49.
Fieldler, F.E. A comparison of therapeutic relatior.nhipo in psy-
choanalytic, non-directive, and Adlerian therapy. J. Consult.
Psychol., 1950, 14 436-495.
Fielder, F. E. Factor analyses of psychoanalytic, non-directive,
and Adlerian therapeutic relationship. J. Consult. Psychol., 1951,
Fisher, S. Plausibility and depth of interpretation. J. Consult.
Psychol., 1956, p 249-256.
Freedoan, M.B., Leary, T.F., Ossorio, A.G., Coffey, H.S. The
interpersonal dia"onsion of personality. J. Pers., 1951, 2, 143-161.
Glover, E. Coimaon technical practices: a questionnaire research.
In The Technicue of Psycho- analysis. New York: Internat. Univ.
Press, 1955, Jl-350.
Goldman-Eisler, Frieda. Individual differences between interviewers
and their effect on interviewees' conversational behavior. J.
Ment. Sci.. 1952, a. 66C-671.
Heyns, R.W. Functional Analysis of Group Problem Solving Behavior.
Ann Arbor, Michigan: Conference Research, Department of Psychology,
University of lachigan, 1948.
Hiler, E.W. An analysis of patient-therapist compatibility. J.
Consult., 1958, ?2, 341-347.
Jaffe, J. Language of the dyed. Psychiatry 1958, 21, 249-258.
Kanfer, F. H. Verbal rate, content and adjustment ratings in ex-
perimentally structured interviews. J. Abnorm. Soc. Psychol.,
1959, 58. 305-311.
Knupfer, Genevieve, Jackson, D.D,, and Krieger, G. Personality
differences between more and less competent psychotherapists as
a function of criteria of competence. J. Nerv. Vent. Dis., 1957
(Oct.), 129! 375-384.
Krasner, L. The use of generalized reinforcers in psychotherapy
research. Psycho. J. 1955, 1, 19-25.
Lakdn, H. and Lebovitz, B. Bias in psychotherapists of different
orientations. Amer. J. Psychother., 1958, 12, 79-86.
Lazarsfield, P.F. and Franzen, R.H. Prediction of political behavior
in America. Amer. Sociol. Rev., 1945, J10 261-273.
Lennard, H.L., and Bernstoin A. The AnatoLy of Psychotherapy:
S'sLens of Cotmaniceticn and Excrectation. New York: Col. Univ.
Lindzey, G. (Ed.). Handbook of Soci.l Psycholop CtabridGe, liass.:
Lorenz, M. and Cobb, S. Language behavior in manic patients. A.M.A.
Prch. Ueurol. and P3ychiat., 1952, 67, 763-770.
Lorenz, 1. and Cobb, 3. La-nuage behavior in psychoneurotic patients.
A.M.A. Arch. !eurol and Psychiat., 1953, 69 684,694.
Lorenz, M. and Cobb, S. Language patterns in psychotic and psycho-
neurotic subjects. LA..A. Arch. Neurol. and Pnychint.. 195L, 72,
tuborsIy, L. The prrscnality of the psychotherapist. Fanrnin-er
Qu-rt. 1952, o, 1-j.
Luborsly, L., Holt, ....i., and .orrow, W.R. Ir.nerin report of the
research project on the selection of medical men for psychiatric
Lrainin&. Bull. Menninrer Clin., 1950, L-, 92-101.
1acla3, H., eand Osgood, C.E. llesitation -ic~icena in sp :.'-rt:eous
English speech. Word, 19;9, 15, 19-U.
Martin, B., Lundy, R.1., and Lewin, M.R. Verbal and GSR response in
e-:.er'lental interviews as a function of three degrees of 'tJhcrd~ist"
commuincation. J. Abnorm. Soc. Psychol., 1960, 60, 234-240.
cllemar, Q. Psvchol-ictal Statistics, New York: John 1,ley and Sons,
i!etzner, Helen and 'rann, F. A limited comparison of two methods of
data collection: The fixed alternative questionnaire and the
open-ended inter.-iew. frner. Soc. 'ov., 1952, j. 4E6-.91.
I'urray, L.J. The content-analysis method of studying psychotherapy.
Psychol. fansirr., 1956, 70, No. 13, Whole No. 420.
Parloff, I.B. Same factors affecting the quality of therapeutic
relationships. J. Abnorn. Soc. Psychol.. 195t, 52. 5-10.
Forter, E.H., Jr. iho development and evaluation of a c.oasure of
counseling and interview procedures. I. The development. Fduc.
.nd Psychol. 'eaa-t., 1943, 2, 105-125.
Porter, E.H., Jr. The development and evaluation of a measure of
counseling and interview procedures. II. The evaluation. Educ.
rnd Psychol. ic.-nt., 1943a, ,3 215-233.
rainy,, V.C. Self reference in counselinJ interviews. J. Consult.
Psychol., 1942, 12, 153-163.
Robinson, D. and Rohde, S. Two experiments with an anti-Semitism
poll. J. Abnorm. Soc. Psychol. 1946, ~, 136-144.
Shafer, G.W. and Lazarus, R.Z. Funda-_ntal Conceapo in Clinic -
Psycholorv. New York: McGraw-iill, 1952.
Snyder, W.U. An investigation of the nature of non-directive
psychotherapy. J. Genet. Psychol. 1945, 33, 193-223.
Snyder, W.U. (Ed.),. Group Report of a Program of Research in
Psychotherapy. State Coll; 1953.
Steinzor, B. The development and evaluation of a measure of social
interaction. :-.!. .ilt., 1949, 2 103-122.
Strupp, I.H. Psychotherapeutic technique, professional affiliation,
and experience level. J. Consult. Psychol., 1955a, 12, 97-102.
Strupp, H.H. The effect of the psychotherapist's personal analysis
upon his techniques. J. Consult. Psychol., 1955b, 2, 197-204.
Scrupp, H.H. A multidimensional comparison of therapist activityy
in analytic and client-centered therapy. J. Consult. Psychol.,
1957, a1, 301-308.
Strupp, T.H. The performance of psychiatrists and psychologists i.
a therapeutic interview. J. Clin. Psychol.. 1958, 4, 219-226.
Taylor, Janet A. and Spence, K.W. The relationship of anxiety level
to performance in serial learning. J. '"T. .a'chcl., 195l .,
Timonus, E.O., Rickard, H.C., and Taylor, R.E. Reliability of
content-free, group verbal behavior. Psychol. Record. 1960, 10,
Weeks, J.S. Level of affect in the counseling responses of high
school senior boys. J. Counsel. Psychol., 1957, 4, 297-303.
lfhitehorn, J.C. and Betz, Barbara J. A study of psychotherapeutic
relationships between physicians and schizophrenic patients.
Amer. J. Psychiat., 1954, 11i, 321-331.
CODING OF SUBJECT'S HERBAL BEHAVIOR
The following steps were used for the coding scheme.
They were derived, prior to the coding of the experimental data,
from a pilot study of five subjects.
1. A simple sentence was coded as a single unit. As an
example; "John went to torn." would be considered as one codable unit.
2. A compound sentence was broken donm into its major
clauses and each major clause counted as one unit. Thus the sentence,
"John went to town and bought a book." was coded as, "John went to
town." and "John bought a book."
3. A complex sentence was coded as a single unit. The
sentence, "John went to town whenever he needed a book." was treated
as one unit.
4. A compound-complex sentence was broken down into its
major clauses and each clause coded as a single unit. The sentence,
"Whenever John needed a book he went downtown and bought one." was
coded as, "Whenever John needed a book he went downtown. and "He
5. Interrogative sentences were not coded.
6. Imperative sentences were not coded.
7. A codable unit that referred to the speaker but was
descriptive in nature was coded as self-descriptive. The sentences,
"I am six feet tall." and "I went to the store yesterday." would
both be examples of sentences included in the category.
8. A codable unit that referred to the speaker, but was
as well evaluative in nature was coded as self-evaluative. 'I am
happy." and "I think I'm mart." and "I believe in God." are all
examples of this category.
9. A codable unit that referred to someone other than
the speaker was ceded as other-referring. "Ity either is fifty-five
years old." "He left the house." and "John was sick." are sertences
that would be coded in this category.
10. A codable unit that referred to someone other than
the speaker while still using evaluative terms, was coded as
other-referring. An example of this would be, "They believed it."
11. If a codable unit referred to both the speaker and
someone else it was coded twice, one time for the speaker (either
self-descriptive or self-evaluative) and one coding of other-
referring. An example of this situation would be, "We like baseball."
This sentence would be coded as one self-evaluative response and
one other-referring response.
12. If a codable unit's subject is a generalized other,
including supposedly the speaker, it was coded as other-referring.
An exanmle of this would be, "Everyone likes ice cream."
13. rI a codable unit referred to sonone other than the
= :.-:, L. constituted the speaker's evaluation of l.at other
person, it was coded as self-evaluative. 'He is horrid." is an
example of this situation.
FREQUENCY DATA FOR SUBJECTS IN THE
MAXIMUM PHYSICAL PRESDCE GROUP
13 11.86 39
43 27.07 89
13 18.56 61
72 34.67 114
1 2.43 8
40 41.06 135
14 24.33 80
41 47.14 155
63 45.92 151
20 27.07 89
43 38.93 128
19 35.28 116
.28 41.98 138
28 29.81 98
-17 28.89 95
455 455.00 1,496
FREQUBICY DATA FOR SUBJECTS IN THE
DI1TE IDIATE PHYSICAL PRFESL CE GROUP
29 28. 48
5 8.6 '
213 213.00 934
FREQUENCY DATA FOR SUBJECTS IN THE
MDINIMU PHYSICAL PRESENCE GROUP
27 1 271.02
t-TETJ3 -O ADJUSTED DATA
Comparison of raxinmu and interr.diate rocur..
1. Frequency of self-descriptive responceE.
-39, =3.00 A.nt.enB c:2 *60
"a 31.7207 d.f.=28 Inr .10
2. Frequency of self-evaluative responaem.
I =30.40 A =19.47
C=2.4792 d.f.=28 p< .02
3. Frequency of other-referring responses.
S.-30.33 *rnterm .20
t-4.JL314 d.f.=28 p<.001
4. FrequvnLy of all categories of responses combine.
Mx.=199.73 df nt2 =62.2
t-2.6197 d.f.-2S p(< .02
i. if Kdimun and .
1. Froqusncy of
2. Frequency of
tx 3 0.00O
d.f.=23 p >.10
MLn. 530 .l7
3. Frequency of other-referring responses.
X M =30.33 X =18.07
t=3.2657 d.f.=28 p<.01
4. Frequency of all categories of responses combined.
,lax. 99-73 x6 n.77.40
t=1.6678 d.f.=26 p>.10
Comparison of intermediate Pnd minimum groups.
1. Frequency of self-descriptive responses.
I =28.60 x =28.87
t<1.000 d.f.=28 p>.10
2. Frequency of self-evaluative responses.
t=2.5239 d.f.=28 p <.02
3. Frequency of other-referring responses.
I =14.20 X =18.07
t=1.3452 d.f.=28 p>.10
4. Frequency of all categories of responses combined.
Inter =62.27 i.=77.47
t=1. 2112 d.f.-28 p>.10
Irwin Parman was born on June 2, 1936, in New York,
New York. From 1953 to 1954 he attended the Polytechnic Institute
of Brooklyn. In the Fall of 1954 he attended the College of the
City of New York. Entering New York University in the Spring
of 1955 he received his B.A. from that institution in 1957. He
entered graduate school at the University of Maryland in the
Fall of 1957, in the Department of Psychology, where he remained
until 1959. During this time he was also employed as a research
psychologist by the Child Research Branch of the National Institute
of Mental Health.
In the Fall of 1959 he continued his graduate training
in psychology at the University of Florida. He was an Office of
Vocational Rehabilitation trainee in psychology froe 1960 to 1962.
He received his M.A. from the University of Florida in August, 1961.
He completed his internship at the Teaching Hospital, University of
Florida in 1962. Since that time he has been employed as a staff
psychologist at the Alaehua County Health Department. Irnn
Farbaan is an associate member of the Aerican Psychological
Association and a member of the Florida Psychological Association.
This dissertation was prepared under the direction of
the chairman of the candidate's supervisory committee and hcs
been approved by all members of that coanittee. It was submitted
to the Dean of the College of Arts and Sciences and to the Graduate
Council, and was approved as partial fulfillment of the requirements
for the degree of Doctor of Philosophy.
Dean, College of Arts Sciences
Dean, Graduate School
Super.. ory Committee: