Kiesler 1966

Psychological Bulletin
1966, Vol. 65, No. 2, 110-136
SOME MYTHS OF PSYCHOTHERAPY RESEARCH

AND THE SEARCH FOR A PARADIGM
DONALD J. KIESLER
University of Iowa
3 myths prevalent in psychotherapy research are considered and refuted. These

include the uniformity assumptions, spontaneous remission of psychoneurosis,
and the belief that present theoretical formulations provide adequate paradigms.
Several other confusions are listed and clarified including the process-outcome
distinction, the classification problem, and the expectation of the definitive
study in therapy research. Finally, an attempt is made to delineate the minimal
requirements of any psychotherapy research paradigm by incorporating present
empirical evidence as well as by specifying common sources of confounding
vis-a-vis the therapy interaction.
One of the unfortunate effects of the prolific off prematurely an area of inquiry and give rise to
and disorganized psychotherapy research the illusion that a problem has been solved, whereas
the exploration has barely begun. [Strupp, 1962,
literature is that a clear-cut, methodologically p. 470].
sophisticated, and sufficiently general para-
digm which could guide investigations in the This first section is devoted to enumerating
area has not emerged. Perhaps this is an un- and refuting several misconceptions about
avoidable state of affairs in a new area of psychotherapy research which have tena-
research. Yet a perusal of this literature indi- ciously persisted. These myths have served the
cates that most of the basic considerations unfortunate purposes of confusing the con-
necessary for a general paradigm have ap- ceptualization of psychotherapy, hence im-
peared, albeit in many cases parenthetically, peding research progress; and of spreading
at some place or another. But to date no one pessimism regarding the utility of further
has attempted to integrate empirical findings research. Let us deal with these myths in turn.
and methodological concerns in a way that
might lead to a useful research paradigm. This The Uniformity Assumption Myths
lack of integration of the paradigm ingredi-
ents has minimized their impact on investi- This misconception was first labeled by
gators in the area. Moreover, concomitant Colby (1964) although it has been alluded
with, and perhaps because of, the absence of to by several other authors (Gendlin, 1966;
a paradigm several myths have been perpetu- Gilbert, 1952; Kiesler, 1966; Rotter, 1960;
ated ; and because of these myths, inadequate Strupp, 1962; Winder, 1957). Colby only
designs continue to appear. parenthetically mentioned the myth, and al-
This paper will first attempt to spell out luded to only one of its aspects, that regard-
in some detail several myths current in the ing patients in psychotherapy research. This
area of psychotherapy research which con- paper will extend its meaning to cover the
tinue to weaken research designs and confuse psychotherapy treatment itself. The former
the interpretation of research findings. Sec- is referred to as the Patient Uniformity As-
ondly, it will attempt to present a minimal sumption, the latter the Therapist Uniformity
Assumption.
but general paradigm which takes into ac-
count current theoretical inadequacies and Patient Uniformity Assumption. For Colby,
empirical learnings, and focuses on methodo- the Uniformity Assumption refers to the
logical considerations which can no longer belief that "patients at the start of treatment
continue to be ignored. are more alike than they are different." This
implicit assumption has led to a remarkably
SOME MYTHS OF PSYCHOTHERAPY RESEARCH naive manner of choosing patients for psycho-
We should be wary of pseudo-quantifications and therapy research: patients are not selected,
methodological gimmicks which often tend to close rather they pick themselves by a process of
110
SOME MYTHS OF PSYCHOTHERAPY RESEARCH 111
natural selection. In searching for patients for mals than they were like other schizophrenics
investigating psychotherapy, researchers have (Herron, 1962). In other words, extreme
traditionally chosen available samples, such variability was the usual case when one
as any patient coming to a clinic over a lumped all patients diagnosed schizophrenic.
certain period of time. In some cases patients It was only when this variability could be
are further divided into experimentals and reliably reduced that useful research could
controls. But, in most studies, all patients begin to be done. The important empirical
receive therapy, measures being taken pre finding was that schizophrenic patients dif-
and post in order to reflect its efficacy. In fered markedly with respect to the abruptness
any case, the assumption has been made that of onset of their disorder. Some could be
by this procedure one obtains a relatively reliably classified in terms of their case his-
homogeneous group of patients differing little tories as "process" schizophrenics, those with
in terms of meaningful variables, and homo- a relatively long-term and gradual onset;
geneous simply because they all sought out while others had a quite short-term and
psychotherapy (be it in a counseling service, abrupt onset, "reactive" schizophrenics. It
outpatient clinic, mental health clinic, private was further discovered that the reactive
practice, or what have you). schizophrenics had much better prognosis.
Far from being relatively homogeneous, pa- But most importantly, the radical reduction
tients coming to psychotherapy are almost in variability permitted by this operational
surely quite heterogeneous—are actually distinction made possible for the first time
much more different than they are alike. The research that could lead to replicable differ-
assumption of homogeneity is unwarranted ences among process schizophrenics and other
since on just about any measure one could diagnostic groups.
devise (demographic, ability, personality, Now, few would argue that for some pa-
etc.) these patients would show a remarkable tients psychotherapy is not effective. Clinical
range of differences. This is apparent from experience as well as research data point
clinical experience and from much of the evi- clearly to this conclusion. Many studies have
dence on initial patient characteristics that shown patient differences evident at the be-
is available today. Because of these initial ginning of psychotherapy which are crucially
patient differences, no matter what the effect related to subsequent dropout or failure in
of psychotherapy in a particular study (be therapy.1 If psychotherapy is differentially
it phenomenally successful or a dismal effective depending on initial patient differ-
failure), one can conclude very little if any- ences, as the evidence strongly suggests, then
thing. At best, one can say something such it seems clear that research should take these
as: for a sample of patients coming to this 1
particular clinic over this particular period, In the following discussion extensive bibliograph-
ical citings will be omitted since the resulting list
psychotherapy performed by the clinic staff would be prohibitive. Fortunately, an exhaustive
during that period on the average was either bibliography of the psychotherapy research literature
successful or unsuccessful. No meaningful has now appeared and is available (Strupp, 1964).
conclusions regarding the types of patients From the topical index provided one can find an
exhaustive bibliography of studies for each of the
for whom therapy was effective or ineffective therapy variables discussed below. Also, several spe-
are possible. This is inevitably the case since cific reviews provide some integration of the many
no patient variables crucially relevant for studies and can be usefully consulted (Auld & Mur-
subsequent reactivity to psychotherapy have ray, 1955; Breger & McGaugh, 1965; Cartwright,
I9S7; Eysenck, 1961; Gardner, 1964; Goldstein,
been isolated and controlled. 1962; Grossberg, 1964; Herzog, 1959; Marsden,
This Patient Uniformity Assumption ham- 196S; Stieper & Wiener, 1965; Zax & Klein, 1960).
pered research in the area of schizophrenia Finally, the chapters on "Psychotherapeutic Proc-
for years. The assumption was that patients esses" and "Counseling" in the Annual Review of
diagnosed as schizophrenic are more alike Psychology (1950-1965) as well as the American
Psychological Association's two volumes on Research
than different. Subsequent data showed very in Psychotherapy (Rubinstein & Parloff, 1959;
clearly that some schizophrenics were quite Strupp & Luborsky, 1962) are both indispensible
different from others, in fact more like nor- sources for critical reviews of research.
112 DONALD J. KIESLER
differences into account. This would imply secret, myths of consensus within paradigms
the use of a design with at least two experi- were easily perpetuated. For example, among
mental groups, dichotomizing patients by one psychoanalysts there grew the myth of a
or more patient variables shown to relate to single agreed-upon perfect technique [p.
subsequent outcome; or matching experi- 348]." The advent of general tape-recording
mental^ and controls on these relevant initial has made records of the therapist's behavior
patient measures. Meaningful results will not progressively more available to other cli-
occur if one continues to aggregate patients nicians and researchers, and it has become
ignoring the meaningful variance of relevant apparent that differences in technique and
patient characteristics. personality exist, even within schools, and
Therapist Uniformity Assumption. Perhaps that disagreement prevails.
an even more devastating practice in psycho- Despite this token admission of therapist
therapy research has been the selecting of differences, the Uniformity Assumption still
various therapists for a research design on abounds in much psychotherapy research.
the assumptions that these therapists are Patients are still assigned to "psychotherapy"
more alike than different and that whatever as if it were a uniform, homogeneous treat-
they do with their patients may be called ment, and to psychotherapy with different
"psychotherapy." Theoretical formulations therapists as if therapist differences were
seem to have perpetuated this assumption irrelevant. As Rotter (1960) observes:
since they have traditionally focused on de-
Although the trend is generally away from the notion
scribing "The Therapy" which is ideally ap- of psychotherapy as an entity, there is still too much
propriate for all kinds of patients. The myth concern with the process of therapy. For many there
has been perpetuated that psychotherapy in is an assumption that there is some special process
a research design represents a homogeneous which takes place in patients, accounting for their
treatment condition; that it is only necessary improvement or cure. . . . In a similar way the role
of the therapist is conceived of as one ideal set of
for a patient to receive psychotherapy, and behaviors which will maximally facilitate the mys-
that mixing psychotherapists (whether of the terious process. The alternative conception that psy-
same or different orientations) makes no chotherapy is basically a social interaction which
difference. follows the same laws and principles as other social
interactions, and in which many different effects can
This kind of loose thinking seems to have be obtained by a variety of different conditions, is
grown out of what Raimy (1952) has called frequently neglected [p. 408].
the "egg-shell era" of psychotherapy wherein
In addition, the Myth ignores the growing
it was rare for a therapist to tape record his
body of evidence that psychotherapists are
sessions or in other ways to make public his
quite heterogeneous along many dimensions
therapeutic behavior.
(e.g., experience, attitudes, personality vari-
Almost without exception . . . psychotherapists ables) and that these differences seem to in-
adopted the attitude that patients and clients were fluence patient outcome. Some therapy may
frail, puny beings who would flee the field if anyone be more effective than others; and it seems
except their own private psychotherapist touched
them. In view of the lack of evidence to the con-
not too unlikely that some therapy may be
trary, such an attitude was at one time entirely more deleterious than no therapy at all
justified since, in simple ethics, the welfare of the (Rogers, Gendlin, Kiesler, & Truox, 1966).
patient does come first. The fact that the attitude As Meehl (1955) states: "In our present
also had other supports, particularly the fact that ignorance it is practically certain that clients
it provided defensive measures for the therapist's
unsteady ego-structure, made no difference [Raimy, are treated by methods of varying appropri-
19S2, p. 324]. ateness, largely as a function of which thera-
pist they happen to get to. Also, it is prac-
In this aura of mystery it became easy for tically certain that many hours of skilled
one to think of "the One" psychotherapy therapists are being spent with unmodifiable
which would maximally benefit all patients. cases [374]."
As Colby (1964) relates: "As long as what If psychotherapy research is to advance
went on in therapeutic sessions remained it must first begin to identify and measure
these therapist variables so relevant to what circumstances, responding to what

eventual outcome (personality characteristics, psychotherapeutic stimuli." And Gilbert
technique factors, relationship variables, role (1952) urges:
expectancies, and the like). One can then One of the extremely important problems in psycho-
build therapist differences into his design by logical counseling and therapy is that concerning
having several experimental groups, with the need for being able to describe and quantify
therapists at respectively different levels of different types of psychotherapeutic relationships.
relevant dimensions. Or, one can match thera- The possibility of accomplishing this is basic to
further studies concerning the relative effectiveness
pists on relevant factors. To continue the of different types of counseling procedures. Studies
practice of assigning some patients to psycho- of this nature will be necessary before it will ever
therapy and others to a "control" group be possible to make any scientific statements regard-
seems futile. ing such basic questions as the relative effectiveness
of different counseling procedures with different types
Unfortunately, this kind of naive conceptu- of counselee problems, or the relative effectiveness of
alization has dominated studies evaluating the a single over-all approach to all types of counseling
effects of psychotherapy where therapy and problems as compared with diverse approaches based
control groups are compared. Typically, the upon various diagnostic categories [p. 360].
results have been discouraging in that no Finally, Kiesler (1966) states in summarizing
significantly greater improvement has been the outcome results of a 5-year study of indi-
demonstrated for the "therapy" patients vidual psychotherapy with schizophrenics
(Eysenck, 1961). As Gendlin (1966) argues: (Rogers et al., 1966):
Research in psychotherapy has suffered from the fact The picture emerging is that if we compare the
that psychotherapy was not definable. It has meant therapy patients as a group to the control patients
that, if an experimental therapy group was com- no differences emerge. Yet—and this seems to me to
pared to a non-therapy control group, some of the be the crucial question for outcome studies—if we
supposed therapy subjects were not really receiving divide the therapy cases by means of a theoretically
something therapeutic at all. . . . The effect of aver- relevant therapist variable (in this case level of
aging the changes in the "experimental" group as "conditions" offered) the results are quite consistent
compared with the "control" group often showed no with theoretical expectation. Hence, my final point
significant differences. To bring this home, imagine would be that before we can validly assess the out-
trying to investigate the effects of a drug with an come or therapy evaluation problem, it is vitally nec-
experimental group taking the drug and a control essary that we attempt to isolate therapist dimen-
group receiving a placebo. Imagine that some (per- sions that will accurately reflect the heterogeneity of
haps half) of your "experimental" group are actually therapist performance. If we continue to evaluate
taking a preparation without the effective ingredients therapy as a homogeneous phenomenon we will
of the drug . . . and you don't know which ones continue to obtain invalid results.
these are. Then, too, perhaps one or two "controls"
are actually getting the drug on the side. Your In summary, it would seem quite essential
"experimental treatment" group is not always getting
the treatment. and useful to bury these Uniformity Myths
once and for all. Until our designs can incor-
An implication of this consideration is that porate relevant patient variables and crucial
if one could analyze the variability (rather therapist dimensions—so that one can assess
than or in addition to the mean differences) which therapist behaviors are more effective
of the experimental and control groups in with which type of patients—we will continue
these many studies, the therapy groups would to perpetuate confusion. Psychotherapy re-
be expected to show a much greater range of search must come to grips with the need for
improvement behavior than the controls. This factorial designs—as recommended a decade
would follow if some of the "experimental" ago by Edwards and Cronbach (1952)—
patients indeed were exposed to some quite wherein different types of patients are as-
"bad or indifferent psychotherapy." signed to different types of therapists and/or
Sanford (1953) states this still differently: therapy, so that one can begin to discover
"From the point of view of science, the ques- the parameters needed to fill in a meaningful
tion 'Does psychotherapy do any good?' has paradigm for psychotherapy.
little interest because it is virtually meaning- As a postscript, it is often not remembered
less. . . . The question is which people, in that different theoretical formulations and
techniques have at least originally been de- patients improve without having anything
rived from patients of different types. As Stein done to or for them.
(1961) observes: This percentage represents a rather severe
standard, as the evidence, such as it is, has
One source of difference between schools of psycho-
therapy that is often overlooked and which needs to reflected. Without this base rate for compari-
be made explicit is the difference in types of patients son most therapists and laymen might be
on which the founders of the different schools based satisfied with a two out of three success rate.
their initial observations. Maskin summarizes this But with this base rate of spontaneous remis-
point rather well when he says: "Freud used hysteria sion psychotherapy needs to be almost totally
as the model for his therapeutic method, depression as
the basis for his later theoretical conjectures. Adler's successful. Apparently the assumption has
clinical demonstrations are rivalrous, immature char- taken a rather tight hold on both practitioners
acter types. Jung's examples were constructed to a and researchers of psychotherapy. Yet, the
weary, worldly, successful, middle-aged group. Rank surprising fact is that the entire evidence for
focussed upon the conflicted, frustrated, rebellious this assumption comes from the findings of
artist aspirant. Fromm's model is the man in a white
collar searching for his individuality. And Sullivan's two studies, which are, at best, ambiguous.
example of choice is the young catatonic schizo- Further, the assumption contradicts clinical
phrenic." To this one might add that Rogers' original experience as well as some of the experimental
formulations were based on college students [pp. findings regarding human and animal learning,
6-7].
and has been refuted in the literature on sev-
Assuming these theoreticians were all perspi- eral occasions. Unfortunately, these refuta-
cacious and accurate in their observations, the tions focused on different aspects of the argu-
historical fact that their different formulations ment, and were obscured by their connection
were derived from experience with different with the effectiveness-of-therapy polemic.
classes of patients would seem to reinforce Hence, it is necessary to separate the spon-
strongly the necessity and appropriateness for taneous remission argument from the latter
abandoning the Uniformity Assumption in polemic, and to integrate the various argu-
psychotherapy research. ments against spontaneous remission. This sec-
tion will, therefore, seriously reconsider this
The Spontaneous Remission Myth assumption with the hope that this refutation
will bury it permanently.
This second myth has been perpetuated In the first place, clinical lore indicates that
primarily by Eysenck (19S2, 1954, 19SSa, the phenomenon of spontaneous remission has
195Sb, 1961, 1964). Despite many refutations, been observed for only three diagnostic cate-
it has continued to muddle research regarding gories. The first category is acutely reactive
the effectiveness of psychotherapy, and has schizophrenics, who typically experience an
fostered much of the pessimism that has more abrupt onset of psychosis under usually speci-
recently colored this research. Although this fiable traumatic conditions, and whose pre-
conception was restricted by Eysenck to psy- morbid history is relatively free of gross pa-
choneurosis alone, its implications seem to thology. Lasting recovery is generally rapid
have generalized to most of psychotherapy. for these schizophrenics regardless of treat-
Its more specific statement takes the follow- ment. The other two diagnostic groups in-
ing form (Eysenck, 1961): "We may conclude the reactive and psychotic depressions.
clude with some confidence that about two- After temporary remission of their depressive
thirds of severe psychoneurotics show recovery symptoms these patients characteristically ex-
or considerable improvement without the bene- hibit a regular course of recovery, ordinarily
fit of systematic psychotherapy, after a lapse for a period of about 2 years, after which the
of two years from the time that their disorder depression recurs. It would obviously be es-
is notified, or they are hospitalized [p. 711]." sential in any studies evaluating therapy with
The clear implication of this proposition is any of these three groups that these remis-
that for psychotherapy to be proven worth- sion characteristics be considered. But, as far
while, it has to demonstrate it can beat this as can be ascertained by this author, spon-
two-thirds percentage, since two-thirds of the taneous remission as a typical phenomenon
has not been clinically observed for other variable other than the defined treatment
types of patients. In regard to psychoneurosis, variable is responsible for the effects observed
moreover, clinical tradition indicates quite for the experimental subjects. To remove this
clearly that, rather than spontaneous recovery, possibility of confounding, one traditionally
increased rigidity of symptoms tends to be the uses a group of control subjects. In the present
rule when the patient remains untreated. case, if therapy patients change significantly
Freud was so impressed by the rigidity of the more than controls, one can legitimately con-
resistance encountered in the treatment of clude that some aspect of the treatment,
psychoneurosis that he coined the term repeti- ceteris paribus, is responsible for the differ-
tion compulsion to describe the process. Sec- ences.
ondly, no attempt has been made to explain But, as mentioned, Eysenck found it im-
the phenomenon in other than quite gross possible to find any such study in his 1952
terms. If spontaneous remission of neurosis survey. Hence, as a substitute for the missing
occurs, it must occur via some psychological experimental control groups he looked for
and/or physiological process. What is the na- evaluative studies of untreated psychoneurotics
ture of this process? What is the stimulus (receiving no psychotherapy) where the pa-
which initiates the process of recovery? Are tients had been followed up over time to de-
the stimulus and the process the same for all termine what, if any, improvement occurred
psychoneurotics, or different for various types? "spontaneously" as the result of the "natural
How does it come about that attitudes and healing process." Eysenck found two pub-
habit systems on which one has acted for lished studies which seemed to satisfy these
much of his life are modified so easily without criteria. He then abstracted and used the per-
rather energetic intervention of some sort? centage of cases who improved over time
What makes an habitual maladaptive pattern from these two untreated samples as a base
of behavior suddenly begin to disappear? line with which to compare the changes ob-
These are crucial questions that need to be served in the reported studies of psycho-
considered regarding spontaneous remission. therapy in the literature at that time.
Thirdly, how can one reconcile spontaneous The first of these base-line studies was that
remission with the evidence in the area of of Landis (1937) who reported the ameliora-
learning regarding habit strength and par- tion rate in state mental hospitals (in New
ticularly the extreme difficulty of extinguish- York State as well as in the United States
ing avoidance responses? generally) for patients diagnosed under the
Since this phenomenon seems counter to heading of psychoneurosis. Because of the
clinical experience, is only grossly explained, overcrowding of state hospitals and their
and contradicts evidence from learning re- chronic understaffing problem, it would seem
search, it would seem that the empirical evi- extremely unlikely that these hospitalized
dence for its existence needs to be quite im- neurotics received much, if any, therapy.
pressive indeed before its generality can be Hence, any recovery observed for them could
accepted. Instead, the entire argument for legitimately be considered as spontaneous.
spontaneous remission of neurotic patients Landis reported that the percentage of pa-
comes from two survey studies cited by tients "discharged annually as recovered or
Eysenck (Landis, 1937; Denker, 1947), whose improved" was 10% for New York State
results are interpreted to meet the needs of (during the years 1925-1934) and 68% for
his ineffectiveness-of-therapy polemic. Let us the United States as a whole (1926-1933).
reexamine these two studies critically to see if Eysenck (1961) concludes from these data:
Eysenck's conclusion is justified. "By and large, we may thus say that of
In approaching the problem of evaluating severe neurotics receiving in the main cus-
psychotherapy, in 19S2 Eysenck searched in todial care, and very little if any psycho-
vain for a psychotherapy research study which therapy, over two-thirds recovered or im-
had included a control group in its design. proved to a considerable extent." Quoting
This was a legitimate search, since there is Landis, he continues: "Although this is not,
always the possibility in research that some strictly speaking, a basic figure for 'spontane-
116 DONALD J. KIESLBR
ous' recovery, still any therapeutic method may sometimes last less, we may conclude with some
must show an appreciably greater size than confidence that about two-thirds of severe neurotics
show recovery or considerable improvement, with-
this to be seriously considered." In other out the benefit of systematic psychotherapy, after a
words, Eysenck seems to be saying that al- lapse of two years from the time that their disorder
though this is not a basic figure for "spon- is notified, or they are hospitalized [p. 711].
taneous" remission, we can still treat it as
such. Is this conclusion justified from Landis' and
The second base-line estimate which Ey- Denker's findings? Is Eysenck correct when he
senck offers comes from a study by Denker states that two-thirds of untreated psycho-
(1947). Denker's report concerns 500 dis- neurotics will, over a 2-year period, experience
ability claims taken from the files of the spontaneous remission of their neurotic ill-
Equitable Life Assurance Society of the nesses? Many individuals have questioned this
United States. These claims were made by conclusion, notably Rosenzweig (1954), as
persons who reportedly had been ill of a well as others (Cartwright, 1955; de Charms,
neurosis for at least 3 months before their Levy, & Wertheimer, 1954; Diihrssen & Jors-
claims were submitted. The claimants came wieck, 1962; Luborsky, 1954; Stevenson,
from all parts of the country, had many dif- 1959; Strupp, 1964a, 1964b). Let us examine
ferent occupations, and included all types of in detail these counter arguments.
psychoneuroses. During their disability (de- Rosenzweig provides the most comprehen-
nned as inability to carry on with any "oc- sive and critical attack on the conclusion
cupation for remuneration or profit") these Eysenck draws from the Landis and Denker
patients were regularly treated only by their studies. His basic argument is that before
local general practitioners "with sedatives, these two studies can be considered as repre-
tonics, suggestion, and reassurance, but in no senting a base line for recovery for untreated
case was any attempt made at anything but psychoneurotics (thereby functioning as ex-
this most superficial type of 'psychotherapy' trapolated control groups for studies evaluat-
which has always been the stock-in-trade of ing the effects of psychotherapy) the data of
the general practitioner." The disability bene- these studies must show three experimental
fits the patients received ranged from $10 to characteristics: (a) the patients used in the
$250 monthly. Denker followed up these cases Landis and Denker studies must be com-
for at least a 5-year period after their illness, parable to those treated by psychotherapy—
and often for as long as 10 years after the that is, the definition of psychoneurosis for
period of disability had begun. The criteria the patients in these studies must be the same
he used for "apparently cured" were, (a) as that for patients in psychotherapy, and the
complaint of no further, or very slight, diffi- severity of the neurotic illness must be
culties, and (b) successful social and economic equivalent for the contrasted groups; (b) the
adjustment by the patient. Landis and Denker base-line groups must in
Eysenck (1961) reports: fact have received no psychotherapy; other-
wise the essential meaning of control group
Using these criteria, which are very similar to those here is violated; and (c) the criteria for suc-
usually used by psychiatrists, Denker found that cessful outcome or improvement need to be
45% of the patients recovered after one year, an-
other 21% after two years, making 72% in all. equivalent, so that recovery or improvement
Another 10%, 5%, and 4% recovered during the means the same thing for the Landis and
third, fourth, and fifth years respectively, making a Denker patients as for typical psychotherapy
total of 90% recoveries after five years [pp. 710- patients.
711].
Rosenzweig then proceeds to argue that the
These are certainly very striking figures. Landis and Denker studies violate all three of
Eysenck finally concludes: these necessary conditions; therefore, Ey-
senck's conclusion of two-thirds spontaneous
If we take a period of about two years for each base- recovery is unwarranted. If Rosenzweig is
line estimate, which appears to be a reasonable figure
in view of the fact that psychotherapy does not correct, then the purported phenomenon of
usually last very much longer than two years and spontaneous recovery for psychoneurotic pa-
tients is indeed a myth, since the Landis and And Luborsky (1954) speculates still further
Denker studies are the only evidence offered about the lack of comparability of the Denker
for its existence. Let us look at Rosenzweig's patients to psychotherapy patients:
and others' arguments in detail as to why the
Many of the "insurance" group would probably
two studies do not meet the three essential never have visited the doctor if it were not required.
conditions for a psychotherapy control group. As a whole the group is probably of higher social
1. Are the Patient Groups Comparable? In and economic level than other groups (apparently
the first place, the Patient Uniformity Myth since they were able to carry disability insurance in
the first place). Very likely the choice of a general
is operative in this comparison. It is quite practitioner rather than a psychiatrist to treat their
easy, but incorrect, to assume that patients psychoneurosis reflects a not-to-be-ignored difference
labeled psychoneurotic are more alike than in an attitude to their illness [p. 129],
different, despite the fact that they are natu-
rally selected in both the Landis and Denker or as Cartwright has just argued, reflects the
studies. From an a priori basis alone the scarcity and relative expense of psychiatrists
probability seems quite small that equivalent in depression years.
groups resulted from these several natural Regarding the lack of comparability of the
selection processes. Rosenzweig further argues patients in the Landis study, Rosenzweig
that makes the following comments:
The insurance disability cases were, as a whole, in Here one could reasonably expect that the neuroses
all likelihood less severely ill than any of the others. must have been extraordinarily severe in order for
Denker himself points out that in these cases where these patients to have become eligible for admis-
disability income was a factor the illness may have sion to these crowded institutions. In these in-
been prolonged by this tangible secondary gain stances the outcome of treatment would be ex-
[money]. By the same token the illness may very pected to be far less favorable than for either the
well have been initiated, or at least partly instigated, Denker control group or the experimental groups
by conscious or unconscious prospects of such gains. [p. 300],
To compare psychoneuroses of long standing, dating
in many instances from early childhood (the typical This of course argues for less spontaneous
case treated by psychoanalysis), with such disability recovery for Landis' patients, which is incon-
neuroses is highly dubious, and the fact that the sistent with the percentages reported, at least
latter would have cleared up quickly after brief for the questionable criterion of recovery that
treatment by a general practitioner is thus not sur-
prising [p. 300]. Landis used.
Regarding both the Denker and Landis pa-
Cartwright (19SS) further argues against the tient groups, Rosenzweig summarizes:
psychoneurotic status of Denker's insurance
It may be concluded that, in general, the Denker
patients: base-line group was probably less seriously ill, the
Denker's study was published in 1946, and all cases Landis control group more seriously ill" [than the
were followed-up for at least five years after re- patients who typically are seen in psychotherapy].
covery. If it is assumed that Denker's research took To the degree that this conclusion is sound it may
one year to carry out, then, since some cases were be further inferred that the control and experimental
disabled for five years and others for only one, all groups fail to meet an essential criterion of com-
these cases of neurosis had their onset between 1934 parability—illness severity [p. 300],
and 1940. In 1933 the economic depression was at
its worst in the United States. From that time on, It seems quite clear from the above rebut-
the country's economy tended to improve except for tals how one can get into inextricable in-
a partial relapse around 1937-38. . . . It is evident terpretive difficulties, by operating on a mis-
that this period (which overlapped the period of conception as unfounded as the Patient Uni-
disability of Denker's subjects) was one of general
growth from a condition of severe unemployment to
formity Assumption, (for those cases where
a condition of plentiful employment throughout the patients are naturally selected for various
United States. These data (i.e., employment rates studies and where one attempts to compare
from 1933 to 1944) suggest that it is reasonable to results). It seems quite obvious that the above
ask what proportion of the variance of Denker's itemizations represent serious patient con-
results may be accounted for in terms of national
recovery from economic depression rather than a foundings—possible secondary gain, a con-
personal recovery from neurosis [p. 292], comitantly improving economic milieu, and
social-class contamination of the "psychoneu- group cannot properly be considered a con-

rotic" patients in Denker's study and the more trol group for spontaneous recovery, since
severely ill Landis patients—and indicate at the patients admittedly received some of the
the very least feasible alternatives to Ey- elements of psychotherapy.
senck's claim of comparability of the "control" Similar doubt is cast upon Landis' group.
patients to those usually seen in psycho- Again to quote Rosenzweig:
therapy. Indeed, in view of these confounding
factors, the probability that the groups are To maintain that neurotic patients admitted to state
hospitals receive no psychotherapy is seriously open
comparable seems quite low; and hence the to doubt. These institutions, despite their notorious
use of the Denker and Landis patients as shortage of staff, usually make a special effort to
control groups for base-line comparisons with treat their neurotic admissions, because these cases
psychotherapy seems invalid. One can ap- have a better prognosis, and because they are far
proximate comparability of groups only by more accessible to treatment [p. 301].
random selection and random assignment of De Charms, Levy, and Wertheimer (1954)
patients to treatments; or by careful match- add: "Some of Landis' group did receive psy-
ing of experimental and control patients on chotherapy;" and suggest a further contami-
relevant variables; or by obtaining post facto nation in that "hospital confinement and
measures of relevant patient characteristics, treatment may themselves be therapeutic."
statistically controlling for their influence. Luborsky (1954) elaborates further on the
These procedures represent the reasonable same point:
alternatives to the naive selection dictated by
the Patient Uniformity Myth, as well as the Also as Landis points out (in objecting himself to
recommended designs for any future studies the use of the consolidated amelioration rate as a
base-line for "spontaneous" recovery) neurotics in
attempting to arrive at a base line of "spon- state hospitals are given a variety of treatments, in-
taneous remission" for psychoneurotic pa- cluding some psychotherapy. And, as they are rela-
tients. The incomparability of the control tively unusual occupants of state hospitals, they
groups vitiates the case for spontaneous re- probably get unusual treatment [p. 131].
mission of psychoneurotic disorders based on This point can be underscored further by
the Denker and Landis studies. adding that, since psychiatrists in state hos-
2. Did the Two Groups of Patients Ac- pitals are very likely human, it would not be
tually Receive No Psychotherapy? Let us too unbelievable that they might seek out,
look first at the Denker group, and again and perhaps enjoy a little, some contact with
quote Rosenzweig (1954): a patient who was not divorced from reality,
In Eysenck's words these patients were "regularly who could converse reasonably well, who
seen and treated by their own physicians with seda- presented some hope of recovery, and whose
tives, tonics, suggestion, and reassurance." . . . These treatment-of-choice could appropriately be
various presumably nonpsychotherapeutic techniques
mentioned include suggestion and reassurance—well- traditional psychotherapy.
known methods of psychotherapy; and psychiatrists One again is compelled by these arguments
regularly use sedatives and tonics as adjuncts to their to agree with Rosenzweig (1954):
practice. . . . The only difference between the work
of the general practitioner and of the eclectic psy- It must then be concluded that the control sub-
chiatrist that could be assumed, in the absence of groups cited by Eysenck do not sharply differ from
detailed and specific knowedge, would be a difference the experimental groups in respect to the important
in thoroughness or expertness, not a difference in variable of having received psychotherapy. As before
kind [p. 300]. with regard to illness severity, the necessary con-
trast between the base line and the experimental
In other words, from the Denker data, le- groups becomes markedly attenuated [p. 301].
gitimate comparison could be made between
psychotherapy of different levels of expertness, Further, since the spontaneous recovery phe-
with the prediction being that the more ex- nomenon by definition requires that control
pert therapy would produce greater improve- patients not be treated, the violation of this
ment than that of the general practitioner. essential condition by the Landis and Denker
But the crucial point is that the Denker studies by itself negates their value as
SOME MYTHS OP PSYCHOTHERAPY RESEARCH 119
evidence for the Spontaneous Remission basic indices: (a) complaint of no further,
Assumption. or very slight, difficulties, and (b) successful
3. Are the Criteria for Improvement or social and economic adjustments by the pa-
Recovery Used in the Denker and Landis tient. Further, he followed up these patients
Studies Comparable to Those Used to Evalu- for a S to 10 year period—a procedure that
ate Traditional Psychotherapy? Can the de- would have certainly strengthened Landis'
gree of improvement or recovery reported in outcome data. This seems to represent a
these two studies be regarded as equivalent to careful and sophisticated attempt to evalu-
that reported for traditional psychotherapy? ate the recovery of his insurance patients.
In the first place, it is important to note that However, Cartwright (19SS) asks:
terms like improvement or recovery are at It is of some interest to speculate about what evi-
best ambiguous. As Luborsky (19S4) states: dences were available in the flies of the insurance
company concerning successful social adjustments
The terms say nothing about what the patient was made by persons whose disability benefits had been
like at the beginning and end of treatment; they terminated. Such termination must certainly be taken
can be and are applied to patients at the entire as evidence for the making of successful economic
range of mental health. A schizophrenic patient can adjustments. But "complaint of no further, or very
be called "recovered"; so can a patient with a slight, difficulties" may represent little more than
slight personality problem. Obviously the word no further supportable claims against the company
"recovered" is used differently in each case [p. [p. 291].
1303.
In other words, what motive would make an
The criterion for improvement or recovery insurance company collect careful and de-
for Landis' state hospital patients was "favor- tailed records of social adjustment of patients
able discharge" from the hospital. Rosenzweig after they had withdrawn their claims. If
(1954) reasons that the probability is quite subjective report of the patients was given
low that the criteria used to come to a heaviest weight in these indices, as seems
favorable discharge decision for hospital likely, then this report seems especially sus-
patients are the same as those used for ceptible to the "hello-goodbye" effect (Hatha-
termination of therapy outpatients. way, 1948), particularly if one recalls the
In other words, while patients residentially treated above argument regarding secondary gain
are generally considered in terms of hospital dis- (money) for these patients.
charge and return to the community, the criterion These considerations compel one to agree
of social recovery being highly relevant, patients with Rosenzweig, that "the standards of im-
nonresidentially treated, as by psychoanalysis, live
continuously in the community and are worked with
provement and recovery in Eysenck's various
in terms of radical therapy which, if successful, patient groups, control and experimental, bear
permits them to live not only with others but with so little resemblance to each other that, once
themselves. This difference in therapeutic goal is so again, the basis of his comparisons has little
great that percentage figures for residential and non- demonstrable validity." Since the criteria of
residential treatment are dubiously commensurable
[p. 301]. recovery for the Landis and Denker groups
seem quite divergent from those used for
It could be added, along similar lines, that the evaluation of psychotherapy, the viola-
it is not too unreasonable to assume that tion of this essential condition in the Landis
in many hospitals, especially for voluntary, study, and likely the violation in the Denker
noncommitted psychoneurotic patients, fac- study as well, further destroys their utility
tors other than personality condition—such as evidence for the Spontaneous Remission
as daily patient quotas which determine the Assumption.
hospital budget, pressure from relatives, pres- In summary, the discussion reported seems
sure from the patient himself, etc.—often to lead unequivocally to the conclusion that
come to bear on the decision to discharge a there is no evidence for spontaneous remis-
particular patient. sion of psychoneurosis. Hence, the belief
The criteria of recovery utilized by Denker seems to be nothing more than a myth propa-
are admittedly far superior to Landis' dis- gated by a popularized and naive interpreta-
charge rate. Recall that Denker used two tion of two research studies. The patients
used in the Landis and Denker studies and ous remission rates are not at all consistent. Although
the percentages of recovery reported by these Eysenck used a two year base, we see no reason
why a five year base may not be taken in com-
authors in no way can be considered evidence paring two studies, especially since we found no
of spontaneous remission for untreated psy- other studies utilizing a two year follow-up with
choneurotic patients. Consequently, Eysenck's which to check Eysenck's claim of stability [pp.
use of these percentages as a base line of 234-235],
spontaneous recovery against which to com- It can be added that two more recent follow-
pare the efforts of psychotherapy is invalid. up studies report rates which are also quite
The discussion above has shown that the con- different from Eysenck's two-thirds percentage
trol patients were very likely not comparable, (Hastings, 1958; Saslow & Peters, 1956).
in fact did receive some treatment (psycho- Finally, it is important to emphasize that
therapy) and hence are not controls, and their it would be quite a useful contribution if
recovery was very likely evaluated on sig- valid developmental data could be obtained
nificantly different criteria. As Cartwright for emotionally disturbed individuals. But
(19SS) concludes: the approach must be more sophisticated than
It is a regrettable accident that the question con- those of Landis and Denker. One cannot
cerning the effectiveness of psychotherapy has been operate on the Patient Uniformity Myth and
tied up with the question about spontaneous remis- report spontaneous remission rates for "psy-
sion. It has been assumed that the question about
therapy is dependent for its answer upon the answer choneurosis." Rather an attempt first must be
to the question about spontaneous remission. The made to develop reliable operations by which
regrettable part of this is that the worse assumption one can distinguish different types of psycho-
has been made that the answer to the spontaneous neurotics. Several more recent survey studies
remission question is already known. Of course, it of remission have attempted this kind of
is said, people do recover spontaneously from neu-
rosis and other psychopathological states. Do they? differentiation (Hastings, 1958; Saslow &
How many? How quickly? Certainly there is no Peters, 1956), but unfortunately used tra-
reliable evidence in the studies of Landis and Denker. ditional psychiatric nosologies (hysterics,
Indeed, the general absence of such evidence leaves obsessive-compulsives, etc.) which have been
it possible to conclude that the statement asserting
the existence of spontaneous remission phenomena in shown to be unreliable classifications (Arn-
regard to neursosis is made on a priori grounds, hoff, 1954; Ash, 1949; Dayton, 1940; Doe-
rooted perhaps in loose analogy with the natural ring & Raymond, 1934; Mehlman, 1952;
histories of coughs and colds. It seems to be an Schmidt & Fonda, 1956; Wilson & Deming,
open question of fact as to whether or not there
are spontaneous remission phenomena at all, and if 1927). If reliable measures can be developed
so, what statistical characteristics they possess [pp. which meaningfully differentiate psychoneu-
294-295]. rotic patients, then ideally one could obtain
developmental data covering the entire life-
It should be pointed out that the spontane- span for these respective groups. That is, it
ous recovery rates reported for "psychoneu- would be useful not only to have data chart-
rosis" are far from being reliable. Various ing the course of an untreated disorder after
survey studies do not agree with the two- it has become a debilitating problem, but also
thirds rate that Eysenck presents. As de to obtain data reflecting the prior develop-
Charms, Levy, and Wertheimer (1954) ment of the disorder. With data of this kind
observe: one could not only more validly assess the
Eysenck (19S2) also states that these results are effects of specific therapeutic interventions,
typical and that they are "remarkably stable from but could also be able to predict which indi-
one investigation to another." This statement is viduals will subsequently experience which
questionable in view of the reports of five year kinds of disorders.
follow-ups such as (a) that of Friess & Nelson
(1942) where one may interpret the results. . . .
to mean that 20% is the spontaneous remission rate, The Myth That Present Theories Provide
and (&) that of Denker (1946) where 90% is Adequate Research Paradigms
reported as the spontaneous remission rate for a
five year follow-up. If these two studies differ so Most of the basic deficiencies in psycho-
widely, it appears that existing figures for spontane- therapy research have derived from the at-
tempt to apply relatively unsophisticated (analytical or Rogerian) in explaining the

theoretical formulations. This section will at- known facts about psychotherapy. In either
tempt to demonstrate how deficient our pres- case, these formulations are not comprehen-
ent formulations are, as well as the necessity sive, and hence basically inadequate.
for basic revision before further research Another way of underlining their theoreti-
progress can occur. cal inadequacy is to state that the present
Three theoretical formulations have domi- theories do not offer an adequate paradigm
nated the research approaches to psycho- or model for psychotherapy research. A re-
therapy: Freud's theory of psychopathology search paradigm optimally should perform
and psychotherapy, Rogers' earlier and later several useful functions. Foremost, it should
notions regarding the essential conditions of focus on the apparent independent variables
the therapeutic relationship, and "learning in a domain of research, as well as on the
theory" via the behavior therapy approaches. scaling and methodological problems encoun-
Let us examine the adequacy of these formu- tered in attempts at operationalizing and ma-
lations as guides for research in psycho- nipulating these variables. Secondly, it should
therapy. It is the thesis of this section that provide a similar emphasis regarding the
a critical look at these systems reveals serious dependent variables. Finally, it should deal
inadequacies: basically that the theories are with the problem of implicit and explicit
not comprehensive (i.e., do not exhaust the confounding variables which potentially inter-
domain of variables operative in the therapy act and covary with the dependent variables.
interaction and do not incorporate existing The final product would be an explicit listing
empirical data); and secondly, do not meet and elucidation of the total matrix of vari-
the requirements for an adequate paradigm ables potentially and actually operative in the
or model for psychotherapy research. Let us behaviors being studied in a particular
look at these closely related difficulties. research area.
In order for a theoretical formulation to be Each of the three theoretical positions has
adequate and useful it must first be compre- failed to specify exactly what the independent
hensive, covering in its explanations the or dependent variables are. None has method-
known facts and variables in its empirical ically dealt with the problems of the quality
domain. It seems rather apparent that none or quantity of outcome expected from the
of the three traditional formulations meet this respective therapeutic interventions, or of
basic prerequisite. The basic difficulty seems differential outcome for different kinds of pa-
to be that none of them has explicitly built tients. None has dealt with the sampling and
into its system propositions covering indi- other methodological considerations which,
vidual differences regarding either the patient remaining unspecified, make it impossible to
or the therapist. This oversight seems to be design a test of present constructs. Let us
the direct result of the Uniformity Myths. examine each of these "theories" of therapy
In other words, until quite recently most in order to demonstrate more clearly this
theoreticians and researchers have implicitly quite ambiguous state of affairs.
agreed that there is one ideal form of therapy Freudian Therapy. What is the heart of
for all patients, and the function of research analytical therapy? What is the attitude,
is to tell us which of the above three best technique, or personality characteristic essen-
approximate this ideal. Yet, as argued above, tial before a therapist can be said to be doing
clinical experience and recent empirical data analytical therapy? The answer seems quite
point to the centrality of both patient and unclear and at best multifaceted and inex-
therapist individual differences in the out- plicit. It seems to be an attitude: the thera-
comes of psychotherapy. One searches present pist must present himself as a neutral and
formulations in vain for propositions incor- ambiguous stimulus to the patient, in order
porating these patient and therapist variables, not to distort the patient's task of free-
It seems, therefore, that one is compelled to association and dream production or hamper
conclude that the present theories are either the appearance of transference phenomena.
too specific (behavior therapy) or too general Yet, at subsequent stages of the interaction
the analyst apparently becomes quite non- regularity of time, frequency and duration, three
ambiguous, offering interpretations of child- to five hours per week, use of the recumbent posi-
tion, the confinement of the analyst's activities to
hood experiences and the transference rela- interpretation, the maintenance by the analyst of an
tionship. Where does the one attitude (or attitude of emotional passivity and neutrality in
complex of attitudes) end, and the other accordance with which he offers no gratification of
begin? What are the behavioral cues which the patient's transference wishes, abstention by the
analyst from giving advice or participating in the
determine the shift of set? What is the daily life of the patient, absence of immediate
interrelationship of the various attitudes emphasis upon the curing of symptoms [pp. 217-
(inaction, ambiguity and/or neutrality) pre- 218].
vailing in the earlier stages of therapy? How
do these attitudes relate to the interviewing It is Stone's belief that changes in any one
techniques of questioning and clarification of these features of psychoanalysis might well
which seem to prevail concurrently? At the affect the dynamics of the transference, and
later stages of therapy: Does interpretation hence the whole course of a treatment.
of childhood experiences need to precede in- When we search for the Freudian dependent
terpretations of the transference relationship? variable we encounter similar ambiguities. Is
Does positive transference necessarily precede insight (or making the unconscious conscious)
negative transference in each therapy inter- the crucial in-therapy product of the complex
action? If not, what are the variables deter- pattern of therapist activity in the therapy
mining the sequence? Does the therapist only hour? If so, what are the cues by which
interpret at moderate depth, that is, just insight is evaluated? Is insight regarding one's
beyond the preconscious? What are the guides childhood behavior, or, regarding the transfer-
for the frequency at which he interprets? ence relationship, sufficient? Or is "working
What are the cues for determining the through" essential? How does the therapist
optimal timing of a given interpretation? Do evaluate whether working through has been
interpretations need to be correct in order to accomplished? At what level or degree of
be facilitative of the therapeutic task? What working through does the therapist begin to
are the cues by which the therapist deter- talk about termination: When has there
mines whether a given interpretation has ac- been enough working through? Can working
complished its purpose? Does one interpret through be accomplished without the prior
differently (at different depths, frequencies, accomplishment of insight? Or what kind of
or with different timing) for different kinds insight and what level is required (what cues
of patients? Further, the therapist's person- are used to evaluate these factors?) before
ality seems to be crucial at all stages, in that working through can be effective? Regarding
the analyst himself should undergo therapy. extra-therapy criteria of successful outcome:
But, how does one evaluate (what are the What specific patient characteristics or be-
criteria?) whether an analyst has been "suc- haviors are implied by a concept of a totally
cessfully" analyzed? What are the personality reintegrated or rebuilt personality? How does
characteristics of the ideal analyst? To what the successful patient behave towards other
extent can analytic therapy be accomplished people? What criteria does he employ in
by an unanalyzed or partially analyzed thera- evaluating his own goal-seeking behavior?
pist? How do the personality characteristics What attitudes does he hold toward himself,
relate to the therapist attitudes and tech- toward his family, toward people in general?
niques? Or are all these aspects (person- Can he be unsuccessful at a particular job?
ality, situational, technique, etc.) essential Can he be single? Will he be involved in
ingredients of Freudian technique? social organizations or civic affairs? Do dif-
Stone (1951) argues that all these ingredi- ferent patterns of these specific extra-therapy
ents are essential: criteria of success emerge for different pa-
tients? If so, which ones with which kinds
It is not enough to say that psychoanalysis recognizes of patients? And what are the cues by which
resistance and transference; psychoanalysis has other
technical precepts which, besides the "basic rule" one distinguishes the different kinds of
include the exclusive reliance upon free association, patients?
SOME MYTHS OP PSYCHOTHERAPY RESEARCH 123
To the author's knowledge no one has dealt in resolving these problems. For example,
with these theoretical questions methodically how does one decide where in the therapy
and exhaustively—neither Freud nor his fol- sequence to study his particular variable (the
lowers.2 What is the independent variable— problem of unit or segment location)? Is the
the crucial therapist behavior (attitude, tech- variable operative equally in the early, middle,
nique, personality characteric, or what have and later stages of therapy? Is the variable
you) which brings about patient change limited to a specific content of the patient, or
in the therapy relationship? What is the a particular patient-therapist interaction in
crucial in-therapy patient change that occurs therapy? Does one need to study the entire
as the result of the therapist's behavior? interactive or content sequence, or can one
How and in what manner does this in-therapy sample the sequence? If one can sample, what
change mediate change in behavior outside size sample is necessary for validly reflecting
the therapy hour? Since there are no theo- the dimension in which one is interested?
retical answers to these questions each re- (What is a valid unit or segment size?) Does
searcher attacks the variables most interest- location in the interview hour by itself bias
ing to himself (e.g., interpretation, therapist the kind of sample one obtains? Can a par-
ambiguity, therapist general activity, counter- ticular dimension be rated validly from the
transference, insight, resistance, anxiety, and therapist's (or patient's) verbalizations alone,
the like). Although Freudian theory seems or does one need the preceding and succeeding
to generate many constructs, the explicit comments of the other participant? (Is con-
integration or description of the essential text necessary?) What level of clinical so-
ingredients nowhere occurs. phistication is needed to rate the dimension
Likewise, many methodological questions validly? What are the independent, extra-
arise when one begins to investigate specific therapy criteria for the in-therapy measure?
variables. Since the theory is not explicitly Does the patient's report of what the therapist
elaborated, one has little theoretical guidance is doing need to be congruent with the inde-
2
Fenichel made significant beginnings in the direc- pendent measure obtained?
tion of systematizing the analytic theory of therapy Rogerian Therapy. Roger's theoretical state-
(Fenichel, 1941, 19S3, 1954) but his death interrupted ment appears in several places (Rogers, 1957,
the endeavor. 1959a; Rogers et al., 1966) but perhaps most
Recently, some of these factors have also been succinctly in his "Necessary and Sufficient
considered by analytic investigators. Levy (1961),
reporting a large-scale research project under the Conditions" paper (19S7). In this paper he
directorship of Franz Alexander, recently described specifies clearly the three therapist attitudes
some of the conceptual difficulty their research has which must be communicated to the patient
encountered: "Another important element of the before constructive personality change can
therapeutic process we are investigating is the role occur for that patient. The three attitudes or
of insight in causing changes in the patient's atti-
tudes. . . . It is difficult, however, to be certain conditions are unconditional positive regard,
about both the qualitative and quantitative aspects empathic understanding, and congruence. Ac-
of this feature of the therapeutic experience. Several cording to Rogers, if the therapist communi-
questions need to be investigated. To what extent cates these attitudes to the patient, construc-
do cognitive—intellectual processes and/or emo-
tional experiencing with or without awareness oper- tive personality change will occur. Moreover,
ate? What are their relative importance? How do he specifies the process dimension along which
they vary from case to case? To what extent is the this patient change occurs, describing the
therapist-patient relationship important primarily to
provide support and relieve anxiety so that the seven "strands" of this process and the de-
patient can acquire insight? What are the relative scription of the subsequent levels of each
roles of insight on the one hand and the living strand (Rogers, 19S9b; Rogers, Walker, &
interpersonal experience with a new (substitute)
parent on the other? To what extent is the learning Rablen, 1960). At first glance the theoretical
process unconscious, and how much does conscious statement seems quite simple and easily verifi-
cognition enter into it? How important is insight able. Therapist technique and personality
into genetic factors, i.e., the re-experiencing of and
understanding of the original childhood experiences characteristics are not crucial. Rather, the
in which the neurotic patterns developed? [p. 129]." crucial therapist behavior is the communica-
tion of the three conditions to the patient. is not monotonic, what explains the points of
Hence the independent variable seems clear- acceleration or deceleration of the patient's
cut: it is multidimensional, consisting of three process? Where over the sequence of psycho-
therapist attitudes. Likewise, the in-therapy therapy does an investigator sample in order
dependent variable is clearly delineated, as to validly reflect the process occurring? What
presented by the above-mentioned patient size of sample is necessary, and from what
"process" construct. However, a closer ex- location in the therapy hour, in order to
amination of Rogers' theory from the same validly represent the process occurring in an
points of reference as for the Freudian frame- individual therapy hour? (Rogers' therapist
work reveals that the specificity of Rogerian and patient dimensions are theoretically de-
theory leaves much to be desired. scribed as content-free; hence that issue is at
What level of these therapist attitudes is least resolved.) Can one measure conditions
necessary before constructive personality or process from the individual participant's
change begins to occur? Are they instead all- verbalizations alone, or is it necessary that one
or-none phenomena? (The various operational have context, that is, the other's preceding
definitions of therapist conditions to date all and subsequent verbalizations? (E.g., does
take the form of a continuum). What is the one need the patient's discourse subsequent
interrelationship among the three conditions? to the therapist's statements in order to meas-
Is it necessary for all three conditions to be at ure empathic understanding?) How and in
an equally high level? If so, what are the what manner is in-therapy patient process re-
respective levels? Will a low level of one lated to extra-therapy criteria of successful
condition cancel the effectiveness of other outcome? How does a deeper level of Experi-
high condition levels? Is one of the condi- encing relate to a patient's attitudes toward
tions a precondition? (Rogers seems to sug- his family and others, his performance on his
gest that congruence may operate in this job, his remission of symptoms, etc.?
fashion.) Need the level of conditions be high Kiesler, Klein, & Mathieu (1966) conclude
at every stage of the interaction, or is it suf- from the findings of a S-year study of Ro-
ficient that they average at a high level for gerian therapy with a hospitalized schizo-
the entire interaction? How does one weight phrenic population:
the conditions when combining them for sta-
tistical analyses? Are the conditions related to In the future studies addressed to issues emanating
from Rogerian theory will require more detailed
therapist personality characteristics? Is the definition and elaboration of both conditions and
patient's view of therapist conditions the process factors, as well as their conceptual integra-
crucial and only measure required? How will tion with other aspects of the therapy setting (in-
therapists' and independent observers' views cluding patient and therapist characteristics, inter-
actional factors, and empirical phenomena). Such
agree or disagree with the patient's viewpoint, factors will then require more rigorous experimental
and is agreement or disagreement necessary? control. Before this is possible, however, more ex-
How is the operation of the separate conditions tensive methodological research is necessary in order
balanced by the others? (Is the appropriate- to resolve the many issues presented by the complex
process phenomena. When, as in this study, theo-
ness of congruence evaluated by empathic retically central variables proved to be imbedded in
understanding?) Is the optimal patterning of a more complex framework, exploratory studies are
conditions different for different patient groups necessary to evaluate which of the many theoreti-
or types? cally extraneous factors in the setting require par-
ticular consideration and control. Only with such
Regarding Rogers' patient-process formula- pilot information, and with validly anchored instru-
tion, the dependent variable in his model: ments for the assessment of therapy variables, can
How are the seven strands interrelated? Does more definitive experimental studies be undertaken.
one need to utilize all seven strands in order
to validly reflect constructive personality Behavior Therapy. One expects the para-
change? What function does process change digm here to be quite sophisticated inasmuch
take over the therapy interaction? Is it as it is based on behavior theory. It is rather
monotonic? negatively accelerated? U-shaped? unexpected to find rather that the application
or does it represent some other pattern? If it of learning theory to the psychotherapy inter-
action has not led to a single behavior ther- ality characteristics influencing the effective-
apy, but rather to several kinds of behavior ness of the behavior-therapy techniques? If
therapy: aversion therapy, negative practice, so, how do these factors interact with the be-
operant conditioning, reinforcement with- havior technique for which patients? Given
drawal, and desensitization (Grossberg, 1964). that changes in patient attitudes and feelings
This multiplicity of products of itself makes about himself and others are theoretically
one question the relevance of the original irrelevant (symptom removal is the only
learning theory to the subsequent applica- crucial consideration), do the various behav-
tions. As Colby (1964) ponders: ior techniques nevertheless affect these atti-
tudes and feelings, and in what manner? Do
It is often difficult to see how specific therapeutic
techniques are deducible from the theory. Learning the constructs used to define the technique
theory and behavioristic approaches pride themselves actually lead to standardized and replicable
on their grounding in scientific principles. Yet differ- operational procedures for different therapists
ent learning theorists derive different techniques from applying the same technique? 8
the same theory, and utilize similar techniques from
different theories. The contradictory disagreement In an excellent critical review of the be-
within the paradigm is obvious to all, but each havior therapy literature, Breger and Mc-
fights for the exclusive truth status of his position Gaugh (1965) conclude:
[p. 362].
It is our opinion that the current arguments sup-
Still, at first sight, the various derivations do porting a learning-theory approach to psychotherapy
seem to represent greater clarity and simplic- and neurosis are deficient on a number of grounds.
First, we question whether the broad claims they
ity than either the Freudian or Rogerian make rest on a foundation of accurate and complete
models. However, closer inspection reveals description of the basic data of neurosis and psycho-
several inadequacies in these derivations. therapy. The process of selecting among the data
Are the various techniques equally effective for those examples fitting the theory and techniques
while ignoring a large amount of relevant data
in the therapeutic situations where behavior seriously undermines the strength and generality of
therapies are traditionally applied? If they their position. Second, claims for the efficacy of
are, which are more economical? If they are methods should be based on adequately controlled
not, what is the ordering of the techniques and adequately described evidence. And, finally, when
overall claims for the superiority of behavioral
vis-a-vis effectiveness? How are the respective therapies are based on alleged similarity to labora-
symptom removals or behavior reinforce- tory experiments and alleged derivation from "well
ments related to independent, extra-therapy established laws of learning" the relevance of the
indices of improvement or success (such as laboratory experimental findings for psychotherapy
ability to get along with others, performance data should be justified and the laws of learning
should be shown to be both relevant and valid [p.
on the job, interactions with one's family and 339].
friends)? Is the particular behavior therapy
technique (the independent variable) a uni- In summary, the basic deficiencies in pre-
dimensional manipulation? Or are there other vailing theoretical formulations are that they
implicit facets of the technique besides the perpetuate and do not attack the Uniformity
theoretically described modification? With Myths described in the previous section; do
desensitization, for example, how much of not explicitly deal with the problem of con-
symptom removal might be the result of sug- founding variables; and do not specify the
gestion? Is it possible that the removal of network of independent, dependent, and con-
symptoms occurs not from "desensitization" founding variables in sufficient enough detail
but rather because the patient has learned to to permit researchers to solve sampling and
discriminate his anxieties more clearly as the other methodological problems. In view of
result of being asked to construct an anxiety these considerations, it seems evident that
hierarchy? Is it possible that by simply teach- our formulations about psychotherapy con-
ing the patient to relax, this relaxation abil- 3
ity generalizes to all other situations, includ- "The 'imagination of a scene' is hardly an ob-
jectively defined stimulus, nor is something as gen-
ing the one in which the phobia is apparent? eral as 'relaxation' a specifiable or clearly observable
Are the therapist's attitudes or other person- response [Breger & McGaugh, 1965]."
tain serious inadequacies. Until our present outcome investigations have focused on pa-
theories are brought up to date by being made tient changes in test or other behavior from
more comprehensive and by spelling out in the beginning to termination of therapy.
much detail the variables of the theoretical Two unfortunate effects seem to have fol-
paradigm—or until new formulations are in- lowed from this somewhat ambiguous distinc-
troduced which meet the same requirements tion: "outcome" researchers have tended to
—it seems that psychotherapy investigators focus exclusively on pre-post patient differen-
must continue to make arbitrary decisions re- tiations; and patient process changes have not
garding these parameters, or attempt to fill in been considered legitimate outcome.
the paradigm themselves with much exhaus- In the first place, the exclusive reliance
tive but necessary prior methodological re- upon pre-post measurement in outcome de-
search. As Meehl (19SS) comments: signs may lead to findings that are invalid or
Considering the state of our knowledge, we still do terminate research prematurely. For example,
not seem sufficiently daring and experimental about if patient improvement (as tapped by a par-
therapeutic tactics. Even when practical exigencies ticular criterion measure) is not a monotonic
force a certain amount of trial-and-error, doctri- function but rather curvilinear in some lawful
naire views about therapeutic theory are likely to be
left unquestioned. The lessons would seem to be fashion, a focus on only two points of time
that we know so little about the process of helping may obscure or distort meaningful patient
that the only proper attitude is one of maximum improvement. To the extent that the function
experimentalism. The state of theory and its rela- of outcome change is unknown, it seems that
tion to technique is obviously chaotic whatever our
pretensions [pp. 374-375]. repeated-measures designs have just as legiti-
mate an application to outcome studies as to
Some Miscellaneous Confusions interview-by-interview process changes. Fur-
ther, pre-post designs demand that one have
The distinction traditionally made between highly reliable change measures before he
process and outcome research seems to have can expect to tap sensitively any improvement
clouded the thinking regarding design, par- that occurs. As Chassan (1962) observes:
ticularly in the latter area. The misconception
seems to take the form: process is not out- It becomes apparent that mere end-point observa-
tions for the purpose of estimating change in the
come research, and outcome research is not patient-state after, say, the intervention of some
process research. The position taken here is form of treatment places generally severe limita-
that these propositions are incorrect: to some tions on the precision of the estimation of the
extent process research is outcome research, change. For random fluctuation in the patient state
and outcome research is equivalent to process can then easily be mistaken for systematic change.
To overcome this difficulty, frequent repeated ob-
investigation. servations must be made of each patient in the
The traditional process-outcome distinction study [p. 615].
is made as in the following:
Thus, it first seems apparent that the tradi-
The studies to be summarized here can be roughly tional process-outcome distinction has per-
dichotomized into those with principal concern as to
how changes took place, therefore focusing on the petuated the relatively exclusive use of pre-
interchange between patient and therapist (i.e., the post designs in outcome studies, with the un-
process), and those that focus on the end point, to fortunate effect that information about the
answer the question of what change took place (i.e., form of the function which represents the
the outcome) [Luborsky, 1959, pp. 320-321]. improvement between the two end points, as
Typically, process studies have dealt with the well as for follow-up periods, has not been
therapist-patient interview interaction, while clarified; whereas repeated-measures designs
outcome studies have focused on changes in would offer this essential type of information.
the patient as the result of therapy. Process Secondly, the use of only two measurement
has been studied by various content-analysis points has increased the likelihood that any
procedures as well as by scales or question- differences observed may be only chance fluc-
naires developed to measure therapist, pa- tuations due to unreliability of the measures.
tient, or interactional dimensions; whereas The second unfortunate result of the proc-
ess-outcome dichotomy has been that patient mental illness. The problem stems not only
process change within the interview has not from the fact that the classifications systems
been considered explicitly as legitimate out- are unreliable (Arnhoff, 1954; Ash, 1949;
come. It seems clear, however, that patient Dayton, 1940; Doering & Raymond, 1934;
improvement manifested in his interview be- Mehlman, 19S2; Schmidt & Fonda, 19S6;
havior is just as legitimately outcome as any Wilson Si Deming, 1927) but equally im-
other form of extra-therapy change. Certainly portantly from the fact that differential diag-
not all process investigation is equivalent to nosis makes no difference—that is, leads to no
outcome—for example, if the investigator is prescribed differential psychotherapeutic treat-
focusing exclusively on the therapist, or on ment. A suggested answer to both of these
one point only of the therapy sequence. But nosological difficulties is that we may be look-
to the extent that one is investigating in- ing in the wrong places for a reliable and
therapy patient changes he is concerned di- valid diagnostic scheme. Perhaps the answer
rectly with outcome; and to the extent that to the classification problem lies in differen-
one is interested in outcome, he needs to be tial patient behavior found in the therapy
cognizant of in-therapy patient changes (a hour itself. If therapists in fact deal differ-
point rarely mentioned). To say this differently in therapy with different patients, then
ently, there seem to be two important areas perhaps the patient cues to which the thera-
of patient change: that change manifest in pist differentially responds can be isolated
the therapy hours themselves, and concomi- and reliably measured from that interaction.
tant changes observed outside the therapy If the manner in which the patient talks
interaction (in situ). Process research begins about himself in therapy indeed provides a
with the in-the-interview behavior of the reliable differentiation of patients, then the
patient; outcome investigation begins with likelihood seems good that the process dimen-
his outside-the-interview improvement. The sions isolated would be directly relevant to
crucial implication is that for either to be differential therapeutic techniques. It seems
maximally useful, it must consider the other that this possibility has been overlooked to
focus or perspective. It is necessary for both date.
investigators to formulate some clear para- Finally, an unrealistic hope prevailing in
digm of the dependent variables of psycho- psychotherapy research has been the belief
therapy, both in- and extra-therapy, and their that sooner or later "The Definitive Study"
theoretical interrelationships. will be published which once and for all
It seems, then, that the process-outcome proves the effectiveness of psychotherapy and
confusion has resulted primarily from ignor- defines the process by which it works. This
ing the fact that some interview data reflects belief seems to have motivated the prolific
outcome (patient change); or, said differently, subsidization of the large outcome research
that some of the outcome of therapy may be projects prevalent in the last decade. One of
evident in the interviews. Perhaps it would be the functions of this paper is to demonstrate
helpful to discard these terms, instead refer- the infinitesimal probability that any one-shot
ring to in-therapy (interview) studies (via research attempt will ever significantly ad-
direct observation, movies, tape recordings, or vance our knowledge in this area. The busi-
transcripts) and extra-therapy investigations ness at hand for therapy (just as for any
(dealing with "in situ" observations). It must other) research seems clear: painstaking in-
also be added that since the statistical func- volvement with delineated problems until
tion of these in- and extra-therapy changes repeated replication of individual findings has
is unknown, one should seriously consider the been demonstrated, and subsequent attack of
appropriateness of repeated-measures designs closely related or ancillary questions. As See-
in attempting to evauate the effects of psycho- man (1961) has observed, investigators need
therapy. to "dispel the notion that some single research
A further difficulty in psychotherapy re- package is likely to be devised to answer a
search has been connected with the scientific great many questions all at once." Rather
status of current diagnostic categories for the pattern of research required is "one of
plugging away at small bits of knowledge general terms, the independent and dependent
which, only after an appreciable period of variables as well as the confounding variables
time, might attain a higher order of signifi- that have been shown to be built into and
cance." complicating this system. Concomitantly, an
attempt will be made to derive some sugges-
THE SEARCH FOR A PARADIGM tions regarding experimental design in light of
In the domain of psychotherapy there is no single these considerations—although this is not the
shared paradigm commanding consensus. With con- primary focus since the statistical and design
siderable overlap the leading current paradigms are issues have been addressed quite competently
the psychoanalytic, learning theory and existential. by others (Campbell, 1957, 1963; Edwards &
Signs of crisis are to be found in each in the in- Cronbach, 1952; Patterson, 1956; Under-
creasing recognition and public acknowledgement of
limitations and impasses [Colby, 1964, p. 347]. wood, 1957). The "variable-model" approach
used follows very closely the research model
The previous section described in some de- excellently presented by Underwood with,
tail the theoretical inadequacies present in however, an exclusive focus here on the
the existent models for psychotherapy. Fur- therapy research situation.
ther inadequacy lies in the fact that (with
the possible exception of Rogers) these Independent Variable (s)
theoreticians or their disciples have not modi- What is the independent variable in psycho-
fied their formulations in light of recent em- therapy research? Clearly its choice depends
pirical evidence. For despite the fact that upon one's particular theoretical orientation
conceptualizations have been vague and the or observational hunches. Just as clearly, it
designs of studies leave much to be desired, is evident from the above that present formu-
there has emerged a consistent body of data lations have not specified in sufficient detail
at least regarding patient prognostic variables what these independent variables are: For
(e.g., social class, intelligence, and the like) analytical therapy it lies somewhere among a
which has not been incorporated into any matrix of therapist attitudinal, technique, and
theoretical system. Additional evidence is sug- personality factors (e.g., ambiguity, interpre-
gestive that therapist personality and expectation, personal maturity). For Rogerian
tations, as well as therapist-patient relation- therapy it lies somewhere in an interactional
ship factors are also critically related to thera- matrix of three therapist Conditions or atti-
peutic outcome. Until these various factors tudes (positive regard, empathic understand-
are incorporated into theoretical formulations ing, and congruence). For behavior therapy
(existent or new) these models cannot be it falls somewhere in the communication by
utilized meaningfully in therapy research— the therapist of specific unlearning procedures.
and therapy research will remain in a state of Obviously, more critical thinking needs to be
crisis. It is the purpose of this section to given to the exact delineation of the therapist
spell out the minimum requirements for a variable or variables instrumental in effecting
useful psychotherapy research paradigm by patient change.
attempting to delineate the relevant factors Generally, it seems clear that the independ-
suggested by current empirical data, by elab- ent variable in psychotherapy has to lie some-
orating on the methodological issues that have where in the therapist and his behavior. It
to be dealt with, and by doing this in a lan- seems necessary that some aspect of the thera-
guage sufficiently general to be applicable to pist (attitude, technique, personality charac-
researchers of differing orientations. The goal teristic, and/or the like) be communicated to
is not to construct an adequate paradigm, the patient to some degree before one can
but to try to outline the minimal criteria any expect the patient to change in some manner.
paradigm must satisfy before its adequacy Ideally, there would be but one therapist di-
can even begin to be considered. mension communicated to the patient, thereby
The plan of the following, therefore, is to effecting beneficial changes in that patient.
examine critically the individual psycho- Practically, however, few if any theoreticians
therapy situation in order to delineate, in have talked of a single therapist dimension or
behavior as crucial, but list several or many, Dependent Variable(s)

emphasizing as they go the extreme complex-
Just as much imprecision has been manifest
ity of the therapy relationship. Very likely,
regarding the dependent variable in psycho-
then, the independent variable for psycho-
therapy research. Here the Patient Uniform-
therapy research needs to be multidimen-
sional, in that more than one aspect of the ity Myth has dictated the search for the one
patient "process" dimension along which
therapist are crucial antecedents of beneficial
patient change. An implication of this point, beneficial patient change occurs. For analytic
and an equally crucial one, is that the exact therapy, this dimension has resided some-
where among such variables as insight (mak-
interrelationship or patterning of the variables
ing the unconscious conscious), working
in the multidimensional model needs to be
clearly specified in any theoretical formula- through, reduction in anxiety and resistance,
tion. etc. For Rogers, it is found somewhere among
his seven strands of process or Experiencing.
Now, if psychotherapy were a one-time
event this kind of model of independent For behavior therapy it seems to reside some-
variables might be adequate. However, since where in the process of anxiety reduction and
symptom removal.
therapy is a sequential treatment procedure,
one's model needs to be concerned further Many of the same concerns expressed re-
with the dimension of time. Are the same garding the independent variable are germane
aspects of the therapist operative in the same here. Does patient-beneficial change occur
pattern over the entire therapy interaction? along one dimension or, more feasibly, along
If so, this needs to be explicitly defined theo- several dimensions? Are the same patient-
retically. If, on the other hand, (as seems change dimensions operative at all phases of
more likely) one or more therapist variables psychotherapy? If so, what is their pattern-
are crucial at one phase of the interaction, ing? If not, which dimensions are changing in
and others are indicated at other periods, then what manner at the different phases? Do we
it is necessary to specify these time interac- need different dependent variables of change
tions. Otherwise a researcher may be sampling for different kinds of patients? If so, what
at the inappropriate therapy period in his are the diagnostic dimensions involved, and
attempt to investigate specific theoretical di- what are the respective differential patient-
mensions. Further, the model would be less change processes? Are these dependent varia-
difficult if psychotherapy were an agreed-upon bles manifest in the in-therapy verbal com-
perfect technique effecting changes regardless munications of the patient? If not, what are
of type of patient. However, since it seems the extra-therapy manifestations of this
more likely that "psychotherapy" represents change? If so, how are the in-therapy com-
in practice heterogeneous therapist perform- munication variables related to the extra-
ance depending upon the kind of patient with therapy manifestations? That is, how does the
whom he is dealing, then one's model must in-therapy process mediate changes in extra-
delineate differential levels or classes of inde- therapy patient behaviors? It seems clear that
pendent variables which are correlated respec- the dependent variables of therapy are to be
tively with these patient individual differ- found somewhere in the in- or extra-therapy
ences. verbalizations and/or behavior of the patient,
Hence, if psychotherapy research is to pro- and in changes along these dimensions in a
gress, it seems essential that theoreticians "positive" direction over the therapy sequence.
and/or investigators first define therapist be- In summary, then, the basic skeleton of a
havior in very precise terms: by specifying paradigm for psychotherapy seems to be
the dimensions along which they vary, by spe- something like the following: The patient
cifying the exact interrelationships among communicates something; the therapist com-
these dimensions at separate time-points in municates something in response; the patient
the therapy interaction, and by specifying communicates and/or experiences something
their differentiations for various kinds of lev- different; and the therapist, patient, and
els of patient disorder. others like the change (although they may
like it to different degrees, or for divergent can be evaluated unambiguously [Under-

reasons). What the therapist communicates wood, 1957]."
(the independent variables) is very likely This final section will examine the varied
multidimensional (and the patterning of this sources of confounding that existing para-
multidimensionality needs to be specified), digms have ignored. It will utilize Under-
and may be different at different phases of wood's (1957) three classes of confounding
the interaction for different kinds of patients. variables (subject, task, and environmental)
Similarly, what the patient communicates and/ by applying them directly to the psychother-
or experiences differently (the dependent apy situation. Hopefully, the result will be a
variables) is likely multidimensional (and listing of variables or variable domains which
the patterning of that multidimensionality need to be integrated into the above para-
needs to be clarified) and may be different at digm and/or whose contaminations need to be
distinct phases of the interaction. The enor- eliminated by experimental or statistical con-
mous task of psychotherapy theory and re- trol procedures in individual research.
search is that of filling in the variables of this Subject Confounding Variables. According
paradigm. to Underwood, subject confounding occurs
It should be added at this point that even when stable or temporary factors or dimen-
if formulations were developed to the point of sions in the experimental subjects are per se
filling in this paradigm, there would still be relevant to, or inducive of, differences in the
many basic methodological problems before dependent variable measures. If these subject
one could appropriately begin to process the factors are permitted to go uncontrolled, one
raw data of therapy. These problems have cannot unambiguously interpret dependent
been encountered in many process studies and variable changes as a function of manipula-
concern such issues as unit size, segment loca- tions of the independent variable. Psycho-
tion, context, interdependency of patient and therapy research of the past decade has
therapist measures, data form and presenta- shown that virtually the entire domain of
tion, and parsimonious interpretation. The patient characteristics (demographic, intel-
empirical solution of these methodological lectual, motivational, semantic, perceptual ex-
problems is necessary before one can validly pectancies) are relevant in greater and less
test theoretical propositions. Since these is- degrees to reactivity to psychotherapy. As
sues have been dealt with in detail elsewhere Luborsky (1959) notes:
(Bordin et al., 1954; Kiesler, 1966) they will A number of studies . . . are concerned with
not be elaborated again here. Suffice it to say identifying qualities of patients that are associated
that many process investigators have com- with staying in treatment, improving in treatment,
pletely ignored these basic methodological or returning once having left treatment. There are
issues. many such studies now in the literature. They have
not yet come to anything definitive, but they seem
The Problem of Confounding on the verge of it, for the results of all of these
studies point in the same directions. To put these
Since the minimal requirements for a ther- trends simply: Those who stay in treatment im-
apy paradigm have not been met by theoreti- prove; those who improve are better off to begin
cal formulations, confounding variables have with than those who do not; and one can predict
response to treatment by how well they are to
contaminated much psychotherapy research. begin with [p. 324].
In fact, the major value of investigations to
date has been the demonstration of the multi- And Strupp (1962) observes:
ple and varied sources from which confound- "It is becoming increasingly clear that therapists
ing can occur. Until this contamination is have fairly specific (and valid) notions about the
dealt with, research findings will continue to kinds of attributes a "good" patient should possess
be ambiguous and consequently subject to as well as about those attributes which make a
patient unsuitable for the more usual forms of in-
alternative interpretations. The essential goal vestigative, insight-producing psychotherapy. Patients
of research is to "design the experiment so considered good prognostic risks are described as
that the effects of the independent variables young, attractive, well-educated, members of the
upper middle class, possessing a high degree of ego- prior matching of patients to form equivalent
strength, some anxiety which impels them to seek groups. The former procedure seems to be a
help, no seriously disabling neurotic symptoms,
relative absence of deep characterological distortions good first solution to this difficulty. By ran-
and strong secondary gains, a willingness to talk domly assigning patients, one is statistically
about their difficulties, an ability to communicate assured that the various treatment groups
well, some skill in the social-vocational area, and will be comparable regarding subject char-
a value system relatively congruent with that of acteristics, will not differ on relevant subject
the therapists. Such patients also tend to remain in
therapy, profit from it, and evoke the therapist's variables other than in chance directions.
best efforts [pp. 470-471]. However, one needs relatively large samples
and may not be able to define clearly the
Hence it should be now apparent that pa- population from which his sample came. One
tients manifest individual differences along also has an error term which is larger than
many outcome-relevant and likely interrelated it otherwise need be, in that much of the error
dimensions. A tentative listing of subject con- variance is identifiable—can be traced di-
founding variables would include the follow- dectly to the kind of relevant patient variables
ing: verbal intelligence, age, motivation for listed above. Finally, one is not permitted to
therapy, level of general anxiety, level of ego- ascertain interaction effects between psy-
strength, type of disorder, severity of disorder, chotherapy treatment (or ideally different
type of onset of disorder, socioeconomic back- types and/or amounts of psychotherapy) and
ground, patient expectancies, verbal expres- these known-to-be relevant subject character-
sive ability, level of occupational success, istics.
type of value system, and likely others. In A more efficient procedure for controlling
view of this multiplicity of confounding fac- subject confoundings seems to be by experi-
tors, it should be abundantly clear why per- mental and/or statistical (partial correlation
mitting patients to select themselves for re- or covariance) matching procedures. By these
search studies has made results incomparable procedures, patients in different treatment
and nonreplicable; why the Patient Uni- groups are matched on one or more relevant
formity Assumption is indeed a myth; why dimensions, thereby establishing group equiv-
an adequate paradigm of therapy needs to alency for these measures. The optimal pro-
incorporate these variables into its structure cedure would seem to be to introduce these
by tying together the patterning of these patient variables into the research design in
factors, and by stipulating alternative treat- the form of independent variables, so that
ment procedures for patients at different levels the meaningful interactions between treat-
of these dimensions.4 ments and these organismic variables can be
It also becomes clear why, in investigating ascertained. For example, different depths of
the effects of one's independent variable or interpretation may be more effective for pa-
variables, it is necessary to control for these tients of low in contrast to high levels of gen-
confoundings. The control problem is basically eral anxiety. Or, different levels of therapist
that of equivalent groups (Underwood, 19S7). activity or directiveness may be more benefi-
The experimental alternatives are random as- cial for patients of low in contrast to high
signment of subjects to treatment groups, or socioeconomic background. It seems that if
the psychotherapy research paradigm is to
*As Rapaport (1960) urges: "Therapies or thera- begin to be filled in, this latter kind of inter-
pists . . . end up by establishing their own McCar- actional research (factorial designs) is es-
ran Act: sooner or later they announce that this or
that kind of patient is not the right kind for their sential. This seems to be the conclusion of
kind of therapy. Not rarely they go further and educational research, an area having quite
announce that this or that kind of patient is 'not similar methodological and design problems.
treatable.' In the long run, psychological theories of As Edwards and Cronbach (1952) observe:
therapy must come to a point where they will make
it possible to select the therapy which is good for a Educators spent a generation on studies of the
patient and not the patient who is good for a oversimplified "Is A better than B?" type. They
therapy [p. US]." sought to settle by experimentation whether large
classes were better than small, lectures better than tentative listing would include: therapist ex-
laboratories, frequent tests better than few. Their perience level, prestige, occupational interest
studies led to endless contradiction because, as you
will notice, the question did not specify the organ- pattern, enthusiasm or confidence, verbal re-
ismic variables. . . . [One investigator] showed that inforcement, therapist expectancies, therapist
his Method A was better than B for bright, mediocre "Conditions," depth of interpretation, liking
achievers, but that B was better for those of for the patient, degree of ambiguity, decon-
mediocre intelligence but good past achievement. ditioning, therapist orientation, therapist per-
The inclusion of both organismic variables in the
design was essential if he was not to reach an over- sonality, and likely others. With this extensive
simple, hence untrue, conclusion. . . . The writers list of therapist or task confounding variables,
agree that effort to isolate effects due to organismic it becomes essential that theoretical para-
variables can have only a beneficial effect, and that digms attempt to incorporate these dimen-
cases should be selected to represent as much varia-
tion as can be. It is far more valuable to study ten sions, as well as explain their interrelation-
cases, two each of five identifiable subtypes, than ships.
to study a pool of fifty undescribed and undiffer- Also, in light of the above, it should be
entiated people. . . . The most promising (i.e., the abundantly clear that mere arbitrary defini-
most likely to be relevant) variable should be built tion of one aspect of the therapist as one's
into the design so that gains can be assessed separately
for each variable [pp. 53-54]. independent variable does not excuse an in-
vestigator from considering other therapist
Task Confounding Variables. Task vari- factors that may be concomitantly operative
ables refer to dimensions or aspects of the and of themselves producing the results ob-
experimental apparatus or stimulus, other than tained. One needs to tackle the arduous task
the experimenter-defined independent vari- of attempting to measure and experimentally
able, which per se are relevant to, and in- or statistically control these other factors in
ducive of, changes in the dependent variable order to eliminate task confounding for his
measures. Task confounding comes from particular study. Or, one can incorporate these
aspects of the experimental task (apparatus factors as additional dimensions (additional
or stimulus) on which the experimenter is not independent variables) in his design so that
focusing, aspects other than his arbitrarily possible interactions between these various
defined independent variable. therapist dimensions and patient change can
In psychotherapy research, task confound- be determined. Quite likely the crucial thera-
ing is possible whenever one arbitrarily defines pist communications are multidimensional,
the independent variable. If one's empirical and researchers need to be acutely attuned
hunch or theoretical framework implies that, to the possible covariance and/or interaction
for example, depth of interpretation is the of other therapist dimensions with the par-
crucial therapist dimension (leading to differ- ticular one in which they may be interested.
ences in the dependent variable, e.g., level Environmental Confounding Variables. En-
of patient insight), then one would like to vironmental variables refer to all nontask
conclude that in fact manipulation of therapist (nonapparatus) variables which change con-
depth of interpretation effected the different currently with manipulation of the independ-
levels of patient insight. But task confounding ent variable, and which per se are relevant to,
occurs and confuses the situation, if, for and inducive of, changes in the dependent
example, the therapist's empathic understand- variable measures. Examples of this kind of
ing (rather than depth of interpretation) confounding in psychological research would
could also be responsible for the differences be the influence the examiner has, independ-
obtained. If empathy is related to insight, ent of the ink blots, in Rorschach research;
and if it is not controlled in the above situa- or the effect humidity has on GSR responses
tion, interpretation of the insight differences independent of other manipulations; or time
obtained will be ambiguous. of day, in contrast to type of instruction, in
Recent psychotherapy research has in- educational research.
dicated that a number of therapist variables The basic reason for having a control group
may be related to patient improvement. A in traditional evaluative studies of therapy is
to hold constant this possible contamination Conclusion

from environmental sources. This especially This final section has attempted to spell
needs to be done for psychotherapy research out in detail the classes of variables that need
since time of itself is an important variable to be incorporated into theoretical formula-
vis-a-vis patient changes—more exactly that tions and/or controlled for in psychotherapy
extra-therapy events that may be crucially research before replicable findings can result.
relevant to patient changes can occur over The first crucial implication of these paradigm
time. The death of a close friend or relative considerations is that theoretical formulations
may occur, causing a relapse in the patient.
can no longer ignore the various domains of
He may meet a new friend who may effect an variables shown to be relevant to psycho-
important difference in the patient's percep- therapy outcome.5 The second, and equally
tion of himself and his behavior. His job situ- vital, implication is that psychotherapy re-
ation may change: he could be promoted, find search can no longer ignore the necessity for
the opening he has been waiting for; or on the factorial designs. Although this latter point
other hand, he may be suddenly fired or laid has been emphasized in the past, few re-
off. Obviously, any number of possible inter-
searchers seem to incorporate the recom-
personal events may occur which can be mendation. As recently as 1959 Berdie ob-
directly relevant to the patient's psychopatho-
served:
logical condition, with the result that changes
apparent in a patient may have little to do When this author reviewed current research in
with the psychotherapy he is receiving con- counseling nine years ago, he concluded, an area of
research importance to every counselor, but in
comitantly. Hence, a control group is needed which this reviewer could find no research publica-
which receives no therapy, as a base line of tions, concerns the relationship between diagnosis
improvement from environmental events to and therapy. What diagnostic categories and tech-
which the condition of therapy can be com- niques are most useful in selecting appropriate thera-
pists? Near the conclusion of this nine-year period
pared. one group of authors reported the first substantial
Further environmental contamination can study of therapy that took into account the effects
come from the fact that some of the patients of precounseling attributes, differential counseling
may be receiving concomitant medication methods, and outcomes (Ashby et al., 1957). Al-
which can be responsible for patient behavior though the results of this study were rather meager,
certainly this study itself should be considered a
changes. Other environmental variables not pioneer one [p. 345].
often mentioned would be differences in the
office situation in which therapy occurs (e.g., In short, the research message seems clear:
degree of soundproofing; comfort of chairs; Current paradigms are inadequate. The time
color of room; pictures on the wall or not, is long overdue not only to acknowledge, but
and which kinds; seating arrangement of B
For a detailed discussion of a variable model
therapist and patient; warmth of reception- similar to the one presented here, see Levinson
ist; privacy of reception room and office; (1962), who presents a list of seven domains of
etc.). These variables are likely not exten- therapy variables: (a) relatively stable personal-
sively operative in the therapy interaction, social characteristics of the therapist, (b) relatively
stable personal-social characteristics of the patient,
but to the extent that they are they must be (c) characteristics of the patient-therapist pair, (d)
controlled or varied systematically. stages in the treatment career, (e) overall treatment
In short, since one can seldom have con- outcome, (/) the institutional setting of treatment,
trol over patients' environments (although and (g) the social context of the patient's life. Levin-
son concludes: "If we are to forego theoretical unity,
this ability represents the real potential of we must at least have a common framework on
institutional studies) a control group for en- which to hang our differences. . . . To the extent
vironmental events is essential before one can that [this framework] in fact takes into account
be confident in concluding that either a par- the events and variables dealt with in the diverse
ticular therapy treatment or several treat- studies, it may serve to sharpen our view of the
ways in which these studies overlap, converge, com-
ments are more effective than no treatment plement each other, and stand in direct opposition
at all. [p. 14]."
also to meet head-on in theoretical formula- DAYTON, N. A. New facts on mental disorders.
tions and research investigations the minimal Springfield, 111.: Charles C Thomas, 1940.
requirements of a psychotherapy research DECHARMS, R., LEVY, J., & WERTHEIMER, M. A note
on attempted evolutions of psychotherapy. Journal
paradigm. of Clinical Psychology, 1954, 10, 233-235.
Now that the mystery and aura are being removed, DENKER, P. G. Results of treatment of psychoneu-
the psychotherapy relationship is being seen by roses by the general practitioner: A follow-up
varieties of researchers as a phenomenon which is as study of 500 cases. Archives of Neurology and
fruitful for investigation as are the parent-child, Psychiatry, 1947, 57, 504-505.
peer-peer, teacher-student, experimenter-subject, and DOERING, C. R., & RAYMOND, ALICE F. Reliability of
other important human groups. Also, importantly, it observations of psychiatric and related character-
is being seen in its true perspective: i.e., as no more istics. American Journal of Orthopsychiatry, 1934,
important or mysterious than any of these other re- 4, 249-257.
lationships. [Matarazzo, 1965, p. 219] DUHRSSEN, A., & JORSWIECK, E. Zur Korrektur von
Eysenck's Berichterstattung iiber psychoanalytische
REFERENCES Behandlungsergebnisse. Acta Psychotherapeutica et
Psychosomatica, 1962, 10, 329-342.
ARNHOIT, F. N. Some factors influencing the unre- EDWARDS, A. L., & CRONBACH, L. J. Experimental de-
liability of clinical judgments. Journal of Clinical sign for research in psychotherapy. Journal of
Psychology, 1954, 10, 272-275. Clinical Psychology, 1952, 8, 51-59.
ASH, P. The reliability of psychiatric diagnoses.
EYSENCK, H. J. The effects of psychotherapy: An
Journal of Abnormal and Social Psychology, 1949,
evaluation. Journal of Consulting Psychology,
44, 27-277. 1952, 16, 319-324.
ASHBY, J. D., FORD, D. H., GUERNEY, B. G., JR., &
GUERNEY, L. F. Effects on clients of a reflective and EYSENCK, H. J. A reply to Luborsky's note. British
a leading type of psychotherapy. Psychological Journal of Psychology, 1954, 45, 132-133.
Monographs, 1957, 7 (24, Whole No. 453). EYSENCK, H. J. The effects of psychotherapy: A
AULD, F., JR., & MURRAY, E. J. Content-analysis reply. Journal of Abnormal and Social Psychology,
studies of psychotherapy. Psychological Bulletin, 1955, 50, 147-148. (a)
1955, 52, 377-395. EYSENCK, H. J. Review of C. R. Rogers & R. F.
BERDIE, R. F. Counseling. Annual Review of Psy- Dymond: Psychotherapy and personality change.
chology, 1959, 10, 345-370. British Journal of Psychology, 1955, 46, 237-238.
BORDIN, E. S., CUTLER, R. L., DITTMANN, A. T., (b)
HARWAY, N. I., RAUSH, H. L., & RIGLER, D. Meas- EYSENCK, H. J. The effects of psychotherapy. In H.
urement problems in process research on psycho- J. Eysenck (Ed.), Handbook of abnormal psy-
therapy. Journal of Consulting Psychology, 1954, chology: An experimental approach. New York:
18, 79-82. Basic Books, 1961. Pp. 697-725.
BREGER, L., & McGAUGH, J. L. Critique and re- EYSENCK, H. J. The outcome problem in psycho-
formulation of "learning-theory" approaches to therapy: A reply. Psychotherapy: Theory, Re-
psychotherapy and neurosis. Psychological Bulletin, search and Practice, 1964, 1, 97-100.
1965, 63, 338-358. FENICHEL, O. Problems of psychoanalytic technique.
CAMPBELL, D. T. Factors relevant to the validity of New York: Psychoanalytic Quarterly, 1941.
experiments in social situations. Psychological Bul- FENICHEL, O. Collected papers. New York: Norton,
letin, 1957, 54,297-312. First Series (1922-36), 1953. Second Series (1936-
CAMPBELL, D. T. From description to experimenta- 46), 1954.
tion: Interpreting trends as quasi experiments. In FRIESS, C., & NELSON, M. J. Psychoneurotics five
C. W. Harris (Ed.), Problems in measuring change. years later. American Journal of Mental Science,
Madison: Univer. Wisconsin Press, 1963. Pp. 212- 1942, 203, 539-558.
242. GARDNER, G. G. The psychotherapeutic relationship.
CARTWRIGHT, D. S. Effectiveness of psychotherapy:
A critique of the spontaneous remission argument. Psychological Bulletin, 1964, 61, 426-437.
Journal of Counseling Psychology, 1955, 2, 290- GENDLIN, E. T. The social significance of the re-
296. search. In C. R. Rogers, E. T. Gendlin, D. J.
CARTWRIGHT, D. S. Annotated bibliography of re- Kiesler, & C. B. Truox, (Eds.), The therapeutic re-
search and theory construction in client-centered lationship and its impact: A study of psycho-
therapy. Journal of Counseling Psychology, 1957, therapy with schizophrenics. Madison: Univer.
4, 82-100. Wisconsin Press, 1966, in press. Ch. 21.
CHASSAN, J. B. Probability processes in psycho- GILBERT, W. Counseling: Therapy and diagnosis. An-
analytic psychiatry. In J. Scher (Ed.), Theories of nual Review of Psychology, 1952, 3, 351-380.
the mind. New York: Free Press of Glencoe, 1962. GOLDSTEIN, A. P. Therapist-patient expectancies in
Pp. 598-618. psychotherapy. New York: Macmillan, 1962.
COLBY, K. M. Psychotherapeutic processes. Annual GROSSBERG, J. M. Behavior therapy: A review. Psy-
Review of Psychology, 1964, 15, 347-370. chological Bulletin, 1964, 62, 73-88.
HASTINGS, D. N. Follow-up results in psychiatric ill- RAPAPORT, D. The structure of psychoanalytic theory:
ness, American Journal of Psychiatry, 1958, 114, A systematizing attempt. Psychological Issues,
1057-1066. 1960, 2, 1-158.
HATHAWAY, S. R. Some considerations relative to non- ROGERS, C. R. The necessary and sufficient condi-
directive counseling as therapy. Journal of Clinical tions of therapeutic personality change. Journal
Psychology, 1948, 4, 226-231. of Consulting Psychology, 1957, 21, 95-103.
HERRON, W. G. The process-reactive classification of ROGERS, C. R. A tentative scale for the measurement
schizophrenia. Psychological Bulletin, 1962, 59, of process in psychotherapy. In E. A. Rubinstein
329-343. & M. B. Parloff (Eds.), Research in psychotherapy.
HERZOG, ELIZABETH. Some guidelines for evaluative Washington, D. C.: American Psychological As-
research. (United States Department of Health, sociation, 1959. Pp. 96-107. (a)
Education and Welfare, Social Security Administra- ROGERS, C. R. A theory of therapy, personality, and
tion Children's Bureau, pamphlet) Washington, interpersonal relationships as developed in the
D. C.: Government Printing Office, 1959. client-centered framework. In S. Koch (Ed.), Psy-
KIESLER, D. J. Some basic methodological issues in chology: A study of science. Vol. 3. Formulations
psychotherapy process research. American Journal of the person and the social context. New York:
of Psychotherapy, 1966, in press. McGraw-Hill, 1959. Pp. 184-256. (b)
KIESLER, D. J., KLEIN, MARJORIE H., & MATHIEU, ROGERS, C. R., & DYMOND, R. F. (Eds.), Psycho-
PHILIPP L. A summary of the issues and con- therapy and personality change. Chicago: Univer.
clusions. In C. R. Rogers, E. T. Gendlin, D. J. Chicago Press, 1954.
Kiesler, & C. B. Truox (Eds.), The therapeutic re- ROGERS, C. R., GENDLIN, E. T., KIESLER, D. J., &
lationship and its impact: A study of psycho- TRUOX, C. B. The therapeutic relationship and its
therapy with schizophrenics, Madison: Univer. Wis- impact: A study of psychotherapy with schizo-
consin Press, 1966, in press. Ch. 12. phrenks. Madison: Univer. Wisconsin Press, 1966,
KRASNER, L. Group discussion: Therapist's contribu- in press.
tion. In H. H. Strupp & L. Luborsky (Eds.), Re- ROGERS, C. R., WALKER, A., & RABLEN, R. Develop-
search in psychotherapy. Vol. 2. Washington, D. C.: ment of a scale to measure process change in psy-
American Psychological Association, 1962. Pp. 103- chotherapy. Journal of Clinical Psychology, 1960,
104. 16, 79-85.
LANDIS, C. A. Statistical evaluation of psychothera- ROSENZWEIG, S. A transvaluation of psychotherapy:
peutic methods. In L. E. Hinsie (Ed.), Concepts A reply to Hans Eysenck. Journal of Abnormal
and problems of psychotherapy. New York: Co- and Social Psychology, 1954, 49, 298-304.
lumbia Univer. Press, 1937. Pp. 155-169. ROTTER, J. B. Psychotherapy. Annual Review of
LEVINSON, D. J. The psythotherapist's contribution Psychology. 1960, 11, 381-414.
to the patient's treatment career. In H. H. Strupp RUBINSTEIN, E. A., & PARLOPF, M. B. (Eds.), Re-
& Luborsky (Eds.), Research in psychotherapy. search in psychotherapy. Washington, D. C.:
Vol. 2. Washington, D. C.: American Psychological American Psychological Association, 1959.
Association, 1962. Pp. 13-24. SANFORD, N. Clinical methods: Psychotherapy. An-
LEVY, N. A. An investigation into the nature of nual Review of Psychology, 1953, 4, 317-342.
psychotherapeutic process: A preliminary report. SASLOW, G., & PETERS, ANN D. A follow-up study of
In J. H. Masserman (Ed.), Psychoanalysis and "untreated" patients with various behavior dis-
social process. New York: Grune & Stratton, 1961. orders. Psychiatric Quarterly, 1956, 30, 283-302.
Pp. 125-149. SCHMIDT, H. O., & FONDA, C. P. The reliability of
LUBORSKY, L. A note on Eysenck's article "The ef- psychiatric diagnosis: A new look. Journal of
fects of psychotherapy: An evaluation." British Abnormal and Social Psychology, 1956, 52, 262-
Journal of Psychology, 1954, 45, 129-131. 267.
LUBORSKY, L. Psychotherapy. Annual Review of SEEMAN, J. Psychotherapy. Annual Review of Psy-
Psychology. 1959, 10, 317-344. chology. 1961, 12, 157-194.
MARSDEN, G. Content-analysis studies of therapeutic STEIN, M. I. (Ed.) Contemporary psychotherapies.
interviews: 1954 to 1964. Psychological Bulletin, New York: Free Press of Glencoe, 1961.
1965,63,298-321. STEVENSON, I. The challenge of results in psycho-
MATARAZZO, J. D. Psychotherapeutic processes. An- therapy. American Journal of Psychiatry, 1959,
nual Review of Psychology, 1965, 16, 181-224. 116, 120-123.
MEEHL, P. E. Psychotherapy. Annual Review of Psy- STIEPER, D. R., & WIENER, D. N. Dimensions of psy-
chology. 1955, 6, 357-378. chotherapy: An experimental and clinical approach.
MEHLMAN, B. The reliability of psychiatric diag- Chicago: Aldine, 1965.
noses. Journal of Abnormal and Social Psychology, STONE, L. Psychoanalysis and brief psychotherapy.
1952, 47, 577-578. Psychoanalytic Quarterly, 1951, 20, 215-236.
PATTERSON, C. H. Matching vs. randomization in STRUPP, H. H. Psychotherapy. Annual Review of
studies of counseling. Journal of Counseling Psy- Psychology, 1962, 13, 445-478.
chology, 1956, 3, 262-272. STRUPP, H. H. The outcome problem in psycho-
RAIMY, V. C. Clinical methods: Psychotherapy. An- therapy revisited. Psychotherapy: Theory, Re-
nual Review of Psychology. 1952, 3, 321-350. search and Practice, 1964, 1, 1-13. (a)
STRUPP, H. H. The outcome problem in psycho- WILSON, E. B., & DEMINO, JULIA. Statistical com-
therapy: A rejoinder. Psychotherapy: Theory, Re- parison of psychiatric diagnoses in Massachusetts
search and Practice, 1964, 1, 101. (b) State Hospitals during 1925 and 1926. Bulletin of
STRUPP, H. H. A bibliography of research in psycho- Massachusetts' Department of Mental Disorders,
therapy. Chapel Hill, N. C.: University of North 1927, 11,6-19.
Carolina, Department of Psychiatry, 1964. WINDER, C. L. Psychotherapy. Annual Review of
STRUPP, H. H., & LUBORSKY, L. (Eds.). Research in Psychology, 1957, 8, 309-330.
psychotherapy. Vol. 2. Washington, D. C.: Ameri- ZAX, M. & KLEIN, A. Measurement of personality
can Psychological Association, 1962. and behavior changes following psychotherapy.
Psychological Bulletin, 1960, 57, 435-448.
UNDERWOOD, B. J. Psychological research. New York:
Appleton-Century-Crofts, 1957. (Received August 9, 1965)

Kiesler 1966

Uploaded by

Copyright:

Available Formats

Kiesler 1966

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Kiesler 1966

Uploaded by

Copyright:

Available Formats

Psychological Bulletin

1966, Vol. 65, No. 2, 110-136

SOME MYTHS OF PSYCHOTHERAPY RESEARCH

3 myths prevalent in psychotherapy research are considered and refuted. These

these therapist variables so relevant to what circumstances, responding to what

social-class contamination of the "psychoneu- group cannot properly be considered a con-

tempt to apply relatively unsophisticated (analytical or Rogerian) in explaining the

behavior as crucial, but list several or many, Dependent Variable(s)

like it to different degrees, or for divergent can be evaluated unambiguously [Under-

to hold constant this possible contamination Conclusion

You might also like