Kazdin, 2008

Evidence-Based Treatment and Practice
New Opportunities to Bridge Clinical Research and Practice,

Enhance the Knowledge Base, and Improve Patient Care
Alan E. Kazdin
Yale University
The long-standing divide between research and practice in (e.g., Lambert & Ogles, 2004). There is evidence in support
clinical psychology has received increased attention in of many treatments, and this fact alone draws attention to
view of the development of evidence-based interventions whether some treatments ought to be used more than others
and practice and public interest, oversight, and manage- in clinical practice and under what conditions. The discus-
ment of psychological services. The gap has been reflected sion of treatment and the delivery of services has moved
in concerns from those in practice about the applicability into the public domain as part of the larger health-care
of findings from psychotherapy research as a guide to landscape. There is an effort to provide resources that
clinical work and concerns from those in research about inform and make available current evidence-based inter-
how clinical work is conducted. Research and practice are ventions. Indeed, one prominent Internet link alone encom-
united in their commitment to providing the best of psy- passes over 30 federal, state, professional, and university
chological knowledge and methods to improve the quality Web sites that enumerate these interventions (http://ucoll
of patient care. This article highlights issues in the re- .fdu.edu/apa/lnksinter.html). Perhaps the most well-known
search–practice debate as a backdrop for rapprochement. effort is the Web-based resource provided by the Substance
Suggestions are made for changes and shifts in emphases Abuse and Mental Health Services Administration (http://
in psychotherapy research and clinical practice. The www.nationalregistry.samhsa.gov). There are also efforts
changes are designed to ensure that both research and among third-party payers and states that may prescribe
practice contribute to our knowledge base and provide what treatments are to be allowed and reimbursed. Re-
information that can be used more readily to improve searchers and clinicians alike see dangers in prescriptive
patient care and, in the process, reduce the perceived and and inflexible treatments. Although it is true that several
real hiatus between research and practice. treatments have empirical research to back them up, is the
Keywords: evidence-based treatment, evidence-based prac- evidence adequate, sufficient, and generalizable to practice
tice, psychotherapy, bridging research and practice situations? Also, who (professionals, managed-care agen-
cies) ought to make the decisions about when the evidence
T here is a well-recognized split within clinical psy-

chology between research and practice in profes-
sional work, career paths, and training. The split has
come into sharper focus with the development and evalu-
ation of empirically supported, or evidence-based, inter-
applies to a given case?
Researchers, practitioners, and health-care policy ad-
vocates continue to debate the merits of the evidence in
behalf of various interventions, what counts as evidence,
ventions.1 A central issue is the extent to which findings

from research can be applied to clinical practice. The issue Correspondence concerning this article should be addressed to Alan E.
Kazdin, Department of Psychology, Yale University, P.O. Box 208205,
is not new, but multiple influences within and outside of New Haven, CT 06520-8205. E-mail: [email protected]
psychology have heightened awareness of it and the stakes
involved. 1
The focus on evidence-based interventions spans multiple disci-
Within the profession, there have been enormous ad- plines (e.g., medicine, education, social work, nursing, dentistry, and
vances in psychotherapy research. Approximately 50 years others). In relation to psychological interventions, several countries and
ago, seminal literature reviews noted that psychotherapy many organizations, committees, and task forces within this country have
delineated evidence-based treatment, interventions, and practice. Now that
did not seem to produce benefits that exceeded changes many agencies at the state and federal levels are involved, multiple and
occurring over time without treatment (Eysenck, 1952, often incompatible definitions are in use for the same terms. The varied
1966; Levitt, 1957, 1963). Their conclusions, even though definitions and criteria for delineating treatments as evidence supported
challenged at the time, lingered and were not easily refuted. are not central to the theses of this article. This article focuses on
Since these reviews, thousands of well-controlled outcome psychosocial interventions in the context of clinical work, and I use the
terms psychotherapy and treatment to refer to these interventions. Psy-
studies (randomized controlled trials, or RCTs) have been chological interventions play a critical role in other areas of psychology
completed, reviewed, and meta-analyzed. Indeed, reviews (e.g., school, counseling, and correctional psychology) that are beyond the
of the reviews are needed just to keep track of the advances scope of the discussion.
146 April 2008 ● American Psychologist

Copyright 2008 by the American Psychological Association 0003-066X/08/$12.00
Vol. 63, No. 3, 146 –159 DOI: 10.1037/0003-066X.63.3.146
latter aspects reflect differences in what research and prac-
tice are trying to accomplish and how they realize their
goals, as elaborated below. No one is really well served by
this debate and the split of research and practice. Perhaps
the greatest casualty is the public at large.
The purpose of this article is to address the debate
about EBTs and EBPs and the research–practice split. First,
both research and practice share a keen interest in provid-
ing the highest quality care. This interest includes identi-
fying and using the most effective and cost-effective treat-
ments. Second, there are opportunities for a rapprochement
betweeen research and practice that will not only foster
improved clinical care but will also develop and strengthen
the knowledge base. Recommendations are made in this
article that will help to make research and practice much
more useful to each other, to our profession, and to the
public at large. I begin by highlighting concerns about
evidence-based interventions and clinical practice. These
issues set the stage for conveying how shifts in both re-
Alan E. search and practice can have a significant impact in en-
Kazdin
Photo by Joel Benjamin hancing patient care, improving the knowledge base, and
Photography bridging research and practice.
Evidence-Based Treatments and
Clinical Practice: Illustrative Concerns
and how the evidence is to be used and integrated (e.g.,
Concerns About Evidence-Based Treatments
Burns & Hoagwood, 2005; Goodheart, Kazdin, & Stern-
berg, 2006; Hunsley, 2007; Tanenbaum, 2005; Wampold, The concerns about EBTs have been the subject of many
2001; Westen, Novotny, & Thompson-Brenne, 2004). A excellent articles, chapters, and books (e.g., Hunsley, 2007;
heavy stream of articles continues to emerge that critically Norcross, Beutler, & Levant, 2005; Westen et al., 2004). A
evaluate the evidence or clinical work, each one underscor- frequently voiced and enduring concern is that key condi-
ing the split between research and practice. Key terms that tions and characteristics of treatment research (e.g., thera-
are used to discuss treatments and the use of evidence pists, patients, treatment, and contexts) depart markedly
reflect the differences between and the different priorities from those in clinical practice and bring into question how
of research and practice. For example, empirically sup- and whether to generalize the results to practice (e.g.,
ported or evidence-based treatment (EBT) refers to the Hoagwood, Hibbs, Brent, & Jensen, 1995). For example,
interventions or techniques (e.g., cognitive therapy for de- patients in controlled trials have been characterized as
pression, exposure therapy for anxiety) that have produced having less severe disorders and fewer comorbid disorders
therapeutic change in controlled trials. Evidence-based than patients who routinely come to treatment. In addition,
practice (EBP) is a broader term and refers to clinical recruiting, selecting, and enrolling cases for research (e.g.,
practice that is informed by evidence about interventions, soliciting and obtaining informed consent, conveying that
clinical expertise, and patient needs, values, and prefer- the treatment provided will be determined randomly) differ
ences and their integration in decision making about indi- considerably from the processes leading individuals to
vidual care (e.g., APA Presidential Task Force on Evi- come to clinical services for their treatment (e.g., Westen &
dence-Based Practice, 2006; Institute of Medicine, 2001; Morrison, 2001). Apart from the participants, controlled
Sackett, Rosenberg, Gray, Haynes, & Richardson, 1996). studies often introduce several features of treatment (e.g.,
EBP is not what researchers have studied. From the stand- standardization and manualization of treatment, fixed num-
point of research, one might say that there is evidence for ber and content of the sessions) that differ from the cir-
specific interventions in the highly controlled contexts in cumstances and conditions of treatment in clinical work.
which they are studied but not yet much evidence for EBP Another concern about research on psychotherapy
in the clinical contexts where judgments and decisions are pertains to the focus on symptoms and disorders as the
made by individual clinicians informed by evidence, ex- primary ways of identifying participants and evaluating
pertise, and patient considerations. The issues embedded in treatment outcomes. In clinical practice, much of psycho-
this difference are part of the impetus for this article—these therapy is not about reaching a destination (eliminating
differences can be bridged. symptoms) as it is about the ride (the process of coping
The debate has focused on polarities or differences in with life). Psychotherapy research rarely addresses the
priorities of psychotherapy research and clinical practice. broader focus of coping with multiple stressors and nego-
Some aspects of the debate reflect testable hypotheses, but tiating the difficult shoals of life, both of which are aided
others cannot be resolved through empirical tests. These by speaking with a trained professional. In clinical practice,
April 2008 ● American Psychologist 147

sometimes symptoms are the focus; even when they are the and useful measures may not necessarily tell us how a
focus, over half of patients seen in therapy add new target patient is doing in the world.
complaints or change their complaints over the course of To combat the inherent limitations of statistical sig-
treatment (see, e.g., Sorenson, Gorsuch, & Mintz, 1985). nificance and interpretation of measures, therapy research-
Outcomes that seem loose and fuzzy (e.g., angst, quality of ers have devised indices of clinical significance. These
life, coping) or that are moving targets are rarely addressed indices are intended to reflect whether a change is a large
in controlled therapy trials. Understandably, there is con- and important difference. The three most common indices
cern about applying many of the findings from research to are showing that (a) high symptom scores at pretreatment
practice. fall within the normative range of nonclinic samples by the
There are less frequently voiced concerns that might end of treatment; (b) large changes (e.g., two standard
be considered more fundamental than the usual concerns deviations) are made within individuals over the course of
about generalizability. These concerns are more fundamen- treatment; or (c) individuals no longer meet criteria for a
tal in the sense that those voicing them do not begin by psychiatric diagnosis (e.g., major depression) that served as
accepting the basic findings and then asking whether the a key selection criterion (see Kendall, 1999). Analyses of
results are generalizable. Rather, they look at the methods each of these indices, which are beyond the scope of the
of analysis or the results among several studies and ques- present article, have also indicated that they do not neces-
tion whether these are satisfactory bases for concluding that sarily reflect palpable changes in everyday life. By gedan-
treatment is effective or efficacious. The following three ken experiment alone, one can show that a client can make
issues convey the point. a change that would meet criteria for clinical significance
First, there are many different criteria for delineating (on one of the commonly used indices) but still not change
whether a treatment is evidence based or empirically sup- in everyday life in a way that makes a palpable difference,
ported. A common criterion is that the treatment produces and vice versa (Kazdin, 2001).
an outcome that differs from the outcome of a no-treatment The conclusion about the metrics of evaluation (sta-
control or treatment-as-usual condition. This criterion re- tistical significance, effect size, clinical significance) is that
quires showing statistically significant differences when in most cases it is difficult to tell the extent to which
groups are compared at the end of treatment. Yet statistical patients have been helped in their daily lives. There are
significance (or even the more informative measure of many exceptions. When a problem (e.g., panic attacks) is
effect size) does not necessarily mean that patients have measured directly in everyday life and treatment effects are
improved in ways that are reflected in their everyday func- strong (e.g., elimination of the attacks), one can say treat-
tioning. Statistical significance is a function of sample size ment really helped. Also, many outcome measures (e.g.,
and variability within and between subjects. The difference glucose level, blood pressure, number of cigarettes
required for significance in the outcome (e.g., on measures smoked) can be interpreted more easily in terms of impact
of anxiety, marital discord) may not reflect a detectable or because they map directly onto other metrics that are not
real difference in the everyday life of any individual client arbitrary.
or even of the group. In short, conclusions about treatment A third fundamental issue has come from looking at
that are based on studies showing statistical differences are results within a given study or between multiple studies
difficult to translate into effects on the lives of participants that have been used as the basis for concluding a treatment
in the study, let alone to generalize to patients seen in is evidence based (De Los Reyes & Kazdin, 2006). Typi-
practice. cally, in a single study, multiple measures are used to
Second, apart from statistical issues, the outcome evaluate outcome, and only some of these show that the
measures in most psychotherapy studies raise fundamental treatment and control conditions are statistically different.
concerns. Changes on rating scales, even well-established At the end of a study, one could say the treatment was
ones such as the Beck Depression Inventory or the Minne- efficacious, was not efficacious, or was mixed depending
sota Multiphasic Personality Inventory and its derivatives, on which measures were examined or whether all measures
are difficult to translate into changes in everyday life. A were examined. EBTs are established because some mea-
change of one standard deviation on a measure from pre- to sures in the study showed the expected effect. Replication
posttreatment does not clearly portray (or map onto) how is salvation in science and perhaps redresses the concern.
the client is functioning in everyday life. For example, if a Yet if one looks at two or more studies of the same
patient with obsessive-compulsive symptoms who is treatment, one reconfirms that only some of the outcome
checking electrical and gas outlets for 4 or 5 hours per day measures reflect change within each of the studies, and a
shows a large change on a rating scale of compulsions, one new issue emerges. The measures that show change and no
cannot really know whether the data translate into actual change within a study are not necessarily the same mea-
reduced checking of outlets (e.g., down to 2–3 hours per sures that show these effects between or among the studies
day?) or whether the patient is functioning better in the of the same treatment. In short, an EBT may have support
world. Many valid and reliable measures of psychotherapy for its effects, but within individual studies and among
are “arbitrary metrics” (Blanton & Jaccard, 2006). That multiple studies, the results often are mixed (i.e., show
means we do not know how changes on standardized different effects or no effects).
measures translate to functioning in everyday life. Thus, a Some of the concerns about EBTs, their interpretation,
statistically significant change on standard, popular, valid, and their applicability to clinical practice are subject to

empirical evaluation. For example, one concern is that for later discussion. First is the concern about clinical
individuals seen in controlled trials are not as severely decision making, judgment, and expertise as a guide to
disturbed as those seen in clinical practice, show fewer individual treatment. In contemporary terms, EBP consists
comorbid disorders, and come from situations and contexts of integrating evidence, clinical expertise, and patient con-
that are not as dysfunctional or disturbed. This is not siderations and then making a judgment of what to do.
invariably true (Stirman, DeRubeis, Brody, & Crits-Chris- Clinical judgment as a way of integrating information has
toph, 2003). Also, direct empirical tests have shown that not fared well over decades of evaluation. There are models
severity, complexity, and comorbidity do not impede ther- for case formulation and decision making but little empir-
apeutic change among EBTs (e.g., Doss & Weisz, 2006; ical evidence in two key areas: reliability in decision mak-
Kazdin & Whitley, 2006). Yet there are multiple venues ing (consistency within therapists over time and among
(patients, domains of functioning, contexts, and settings) different therapists at a given point in time) and validity
where the concern would need to be tested to be put to rest. (that the decision makes a difference in the outcome when
Even then, it is always possible to say, “But my patients are compared with a less flexible algorithm or an alternative
worse, different, older, more complex” than those in such case formulation model). The absence of research cannot
studies, and no set of studies can resolve the concern be attributed to clinicians, but it conveys a more subtle
definitively. facet of the research–practice split, showing that many
There are inherent limitations in the ways EBTs are critical clinical issues and concerns are not heavily re-
discussed. Large segments of the literature usually are searched.
grouped together. In many ways, the concerns about EBTs The challenge of clinical decision making can be
reflect the uniformity myths that we were warned of de- conveyed by the effort to “tailor treatment to meet the
cades ago (e.g., Kiesler, 1971)—namely, that conclusions needs of individual patients.” This statement is one we
ought not to be applied equally (uniformly) across the areas make and accept routinely in our clinical work, but re-
of interest (e.g., patients, problems, therapies). Objections searchers have yet to help us do that. A clinician might use
and concerns are likely to be differentially relevant or an eclectic therapy and draw on multiple resources (includ-
applicable as a function of critical distinctions. For exam- ing EBTs) to develop a treatment package suited to the
ple, researchers in the area of my work (the treatment of individual. There are no formal or clearly replicable pro-
conduct disorder—serious aggressive and antisocial behav- cedures for how to do this, in terms of selecting one or
ior among children and adolescents) have identified several more treatments or components of treatment among all
EBTs (e.g., multisystemic therapy, the multidimensional available therapies and deciding in what proportion and
treatment foster care model, parent management training, sequence they ought to be delivered to patients. As clini-
functional family therapy). Many controlled studies have cians, we have an idea of how to do this, but it is not yet
included outpatient, inpatient, and adjudicated and incar- well established that different clinicians would select the
cerated samples with severely aggressive, violent, and an- same or a similar individualized treatment plan (i.e., reli-
tisocial behaviors, comorbid disorders, and impairment in ability) when presented with the same case. And even if the
multiple domains (Kazdin, in press). It would be difficult to treatment selected were the same, we do not know that it
challenge the work with the question “But are they real would make a difference or achieve palpable change in this
patients or really very disturbed?” Some of the objections patient’s life (i.e., validity). Another treatment, one recom-
might not apply to this particular area of research. mended by the best available evidence, might do just as
A central concern about EBTs involves the generali- well as, or better than, an individually tailored treatment,
zation of the results from controlled research to clinical perhaps especially so if treatment technique is not the most
practice. This concern is a cogent one. Although the con- critical influence (Wampold, 2001). We know from every-
cern is not new, it is newly relevant because of the progress day life that when we are told “one size fits all,” the
of psychotherapy research and the challenges to clinical garment in question tends not to fit anybody very well. One
practice and service delivery (e.g., managed care, reim- assumes this is true of therapy too, but how to individualize
bursement issues, the likelihood of clinicians being told therapy for each person and how to show that doing so
what they can and cannot do). Debates about whether EBTs makes a difference are topics researchers have still not
ought to be extended to practice are not likely to be helped elaborate and that are difficult to defend.
resolved, partly because of the way the discussion is Second, there is the issue of generalization of results
framed and partly because of the methods by which re- in clinical practice. Those of us involved in clinical work
search and practice are conducted, as I elaborate later. are apt to say that the results from a controlled trial may not
generalize to our patients because of differences in recruit-
Concerns About Clinical Practice
ment and patient characteristics. This point is cogent. How-
There are parallel concerns about clinical practice; they are ever, the argument is a two-edged sword. Given the unique-
parallel in the sense that they reflect objections and reser- ness of the client in front of me who is about to begin
vations with a history spanning decades and involve sub- treatment, it is not clear on what basis I can generalize from
stantive and methodological issues, that is, what is done in a prior client or several prior clients.
therapy and how treatment is evaluated (e.g., Dawes, 1994; Suppose two patients come in with similar anxiety
Garb, 2005; Hayes, Follette, Dawes, & Grady, 1995; complaints and personality styles. It is likely that the indi-
Meehl, 1954). I highlight a few issues here as background viduals will differ from each other in other ways (e.g., age,

stage of life, marital history, gender, ethnic and cultural psychiatric disorders, smudge art therapy to provide an
identity, socioeconomic status, family history of anxiety, outlet for fear). One should question when a new technique
and available support systems). The possibilities for differ- is warranted, when it ought to be provided, and when it is
ences are dramatic in the clinical service I direct. Given the a treatment of choice.
criteria for diagnosing conduct disorder, there are over
32,000 combinations of symptoms that individuals can General Comments
have and still meet the diagnosis (Perepletchikova & Clinical research and practice both raise concerns about
Kazdin, 2005). On the basis of the diagnosis alone, and treatment and what can be inferred, generalized, and ap-
leaving aside scores on individual-difference, family, and plied to patients. Concerns about the research–practice split
contextual variables, how can one decide to generalize often can be reduced to empirical questions about treat-
from one case to the next? If we really believe that “every ment. One challenge to researchers occurs when clinicians
patient is different” in ways that influence treatment deci- question the generalizability of EBT findings—namely,
sions, then there is a problem in knowing how to make a whether an EBT works with “really disturbed patients.”
decision that is well based and defensible. Of course, the Some studies answer affirmatively, but as I mentioned,
decision is our own best guess, and that brings in the old there is an indefinite number of venues in which to test this.
but still pertinent topic of statistical versus clinical predic- In fact, such studies could not test all plausible moderators
tion as a way of integrating information in order to make a in light of limited time and resources. Knowing that, re-
decision. searchers might ask, “Do we need to consider context and
Deciding what to do with a given patient in clinical the individual or special features of individuals?” Why not
practice is analogous to a multiple regression equation with just bolster, make more effective, and better codify (manu-
many variables to consider, each with a beta weight (i.e., a alize) the treatments? However, we do need to consider
value showing how much it contributes to the decision or contextual aspects of the interventions. For example, psy-
outcome). Restated in English, one must determine what chotherapy is more effective for some ethnic groups (if
variables ought to be considered, how they should be they show low acculturation to European American cul-
weighed in one’s decision making, and how one reached ture) when treatment is provided in their native languages
the conclusion about those variables and their relative and is specifically designed for minority groups (Griner &
importance. In research and in practice, a variable that may Smith, 2006). Culturally insensitive treatments can cause
make a difference is always probabilistic. That is, individ- therapists unwittingly to select goals or embrace values that
ualized therapy based on my judgment of what is needed reflect the culture of the therapist rather than that of the
will not necessarily help this patient. Moderators, whether patient (Comas-Dı́az, 2006).
identified from research or clinical experience, are a set of A bridge between research and practice is not likely to
variables related to an outcome but not invariably related to come from a finding that refutes a specific concern about an
outcome in any individual case. empirically supported treatment (“See, we told you that
Third, the way cases are evaluated in clinical work comorbid patients can be treated effectively.”) or a specific
raises an important assessment issue. Patient progress is concern about clinical practice (“See, we told you that you
often evaluated on the basis of clinician impressions, as have to know more about patients than their symptoms and
opposed to systematic observations using validated mea- what technique can ameliorate them.”). However, there are
sures. Without systematic measures, the reliability, valid- avenues where critical issues of concern to both researchers
ity, and replicability of results in clinical work are easily and practitioners come together and where advances in
challenged. In my own clinical work, therapeutic changes I research and practice can build bridges between science
often see seem stark and qualitative (e.g., children stop and practice.
hitting parents and teachers, no longer destroy property, or
begin to interact prosocially with peers). Yet cognitive Rapprochement: Refocusing Research
heuristics and normal memory processes (of recoding and Practice on Patient Care
rather than recording events) no doubt influence my per-
ception and rendition of the changes (e.g., Gilovich, Grif- There are some natural tensions between research and
fin, & Kahneman, 2002; Pohl, 2004; Roediger & McDer- practice, and perhaps it would be wise not to attempt to
mott, 2000). The absence of systematic assessment raises resolve or even lament them. For example, we need the
many obstacles to making claims about what happens in ascetic rigor of controlled experimental research to under-
therapy and the accumulation of knowledge, as discussed stand (e.g., test theory, identify predictors) and to evaluate
later in this article. the outcome effects of our interventions. This rigor in-
There are additional concerns about clinical practice. cludes RCTs, operational definitions, meticulous control of
One of them is the proliferation of new treatments. In child the intervention, quantification, precision in evaluating out-
and adolescent therapy, for example, over 550 treatments comes, and statistical significance, among other features.
are in use (Kazdin, 2000). Moreover, the number continues All those features that make us worry about the generali-
to grow. The vast majority of treatments have never been zation of the findings from research to practice are strate-
studied in controlled or uncontrolled trials. Not all are gically pivotal to experimental methodology. Psychology
derived from an explicit theory of etiology or theory of also needs the experience and expertise of those engaged in
change (e.g., horticulture therapy for a broad range of clinical work. Findings must be applied before all of the

critical answers are available. Judgment, expertise, and peutic change, (b) the study of moderators of change in
extrapolation of the findings are needed if we are to be able ways that can be better translated to clinical practice, and
to act in the most informed, even if incompletely informed, (c) qualitative research.
way. As psychologists, we cannot wait until we have com- Study of mechanisms of change. Psycho-
plete answers before addressing the immediate needs of our therapy research focuses on many questions including the
friends, relatives, and selves in navigating the shoals of life. following:
Interestingly, in their personal lives, researchers and
1. What is the impact of treatment relative to no
clinicians convey more openness to the concerns of the
treatment?
“other side” and slightly less preoccupation with the ten-
2. What components contribute to change?
sions of research versus practice. For example, probably
3. What treatments can be added (combined treat-
few researchers would refuse therapy for their children
ments) to optimize change?
solely on the basis of the absence of efficacy trials. We
4. What parameters can be varied to improve out-
move to treatment—almost any treatment—and alternative
come?
therapies as needed with the hope of help when the evi-
5. How effective is this treatment relative to other
dence cannot guide us or when it has guided us but to no
treatments for this problem?
avail. The researcher is not likely to say, “There is no solid
6. What patient, therapist, treatment, and contextual
evidence for any treatment, so I am going to withhold best
factors moderate or are correlated with outcome?
guesses by experienced professionals.” Similarly, practic-
7. What processes within or during treatment are re-
ing clinicians, in need of help for their relatives, are likely
sponsible for (not just correlated with) outcome
to search the Web, read extensively, and make phone calls
(mechanisms of therapeutic change)?
to medical centers and experts to identify what the evidence
8. To what extent are treatment effects generalizable
is for the various surgical, pharmacological, and other
across populations, problem areas, settings, and
alternatives for their parents or children with significant
other contexts?
medical problems. The clinician is not likely to say, “My
relative is different and unique and the evidence really has Debates about EBTs and their utility have emphasized
not been tested with people like her, so I am going to forgo Questions 1 and 8 in efficacy studies (Does treatment
that treatment.” Indeed, the participants in the trials sup- work?) and in effectiveness studies (Do the findings extend
porting the treatment may not be scrutinized at all or to practice settings?). Although a broad portfolio with the
enough. full range of questions is a wise investment, psychotherapy
Psychologists need both research and practice, but we research could contribute enormously to clinical work by
also need not assume that key differences and natural focusing more on the mechanisms of change (Question 7).
sources of tension are the only dimensions or facets. The The study of mechanisms of change has received the least
unifying goals of clinical research and practice are to attention even though understanding mechanisms may well
increase our understanding of therapy and to improve pa- be the best long-term investment for improving clinical
tient care. Also, differences in what research and practice practice and patient care.
do in relation to these goals actually can be mitigated. For By mechanisms, I refer to the processes that explain
example, a common way of conveying the differences why therapy works or how it produces change. An RCT
between or different emphases of research and practice is to comparing treatment versus no treatment can establish a
note the following: Research contributes to the knowledge causal relation between an intervention and therapeutic
base, and clinical practice applies that base to help people. change. Yet demonstrating a causal relation does not nec-
Although this is true, it is not complete or very inspired. essarily provide the construct to explain why the relation
More important, this statement and the broader split it was obtained. The treatment may have caused the change,
reflects do not achieve the following three critical goals: but we do not know whether the change can be attributed
to specific or conceptually hypothesized components of
● Optimally develop the knowledge base, treatment (e.g., cognitive restructuring, habituation, stress
● Provide the best information to improve patient reduction, mobilization of hope) and how the change came
care, and about.
● Materially reduce the divide between research and The distinction between cause and mechanism is
practice. readily conveyed with the familiar example of cigarette
smoking. Cross-sectional and longitudinal studies with hu-
I believe that expansion and slight shifts of emphases in
mans and experiments with nonhuman animals have estab-
both research and practice might better address these three
lished a causal relation between cigarette smoking and lung
goals.
cancer. Establishing a causal relation does not explain the
Psychotherapy Research mechanisms, that is, the process(es) through which lung
cancer develops. The mechanism was shown by describing
I suggest three shifts in emphasis in research to advance the what happens in a sequence from smoking to mutation of
knowledge base, improve patient care, and reduce the gulf cells into cancer (Denissenko, Pao, Tang, & Pfeifer, 1996).
between research and practice. These include giving A chemical (benzo[a]pyrene) found in cigarette smoke
greater priority to (a) the study of mechanisms of thera- induces genetic mutation at specific regions of the DNA

that is identical to the damage evident in lung cancer cells. receptor in the amygdala (N-methyl-D-aspartate; see Davis,
This finding conveys how cigarette smoking leads to can- Myers, Chhatwal, & Ressler, 2006). In nonhuman animal
cer. The example is one from biology, but biology is not research, chemically blocking the receptor shortly before
critical to the larger point. Mechanisms of action come extinction training blocks (i.e., interferes with) extinction.
from psychological influences as well. For example, the Blocking the receptor after extinction training also blocks
role of corporal punishment in the development and ame- extinction, which suggests that the consolidation process
lioration of child aggression has been demonstrated in can be interrupted. A compound (D-cycloserine) binds to
cross-sectional and longitudinal observational studies as the receptor and makes the receptor work better by enhanc-
well as intervention studies (e.g., Patterson, Reid, & Dish- ing extinction when given before or soon after extinction
ion, 1992; Reid, Patterson, & Snyder, 2002). training.
In the context of psychotherapy, process research is Exposure-based psychotherapies are empirically sup-
often thought to identify mechanisms of change. Such ported therapies for anxiety and are based on an extinction
processes as the therapeutic alliance, transference, and model. Laboratory research has moved to therapy trials to
changes in cognition have been studied, in some cases enhance the mechanism responsible for extinction. RCTs
extensively (e.g., Luborsky & Luborsky, 2006; Norcross, comparing enhanced versus regular exposure therapy have
2002). For example, the therapeutic relationship is often shown that activating the critical receptor (with oral doses
considered to be a key mechanism that explains therapeutic of D-cycloserine) improves the therapeutic effects of ex-
change. An often cited feature in support of this mechanism posure therapy (i.e., augments extinction) among patients
is the fact that the therapeutic relationship accounts for a referred for acrophobia (Ressler et al., 2004) and social
large proportion (approximately 30%) of the outcome vari- anxiety (Hofmann, Meuret, et al., 2006). The effects are
ance in psychotherapy. However, the percentage of vari- evident immediately after treatment and at follow-up,
ance accounted for cannot answer the question of the months later. In short, understanding mechanisms of
mechanism of therapeutic change.2 change can enhance the effects of treatment in clinical
In most cases, some intervening processes (relation- application. Evidence-based mechanisms of change could
ship, cognition) are shown to emerge or change during prove to be even more interesting or important than EBTs.
treatment and to predict outcome effects at the end of We might be able to use multiple interventions to activate
treatment. Two constraints have limited the identification similar mechanisms once we know the mechanisms of
of mechanisms in psychotherapy research. First, studies change and learn how to optimize their use.
rarely establish the timeline, that is, that the proposed cause Study of moderators and translation to
(e.g., alliance, cognitive change) comes before the changes clinical care. Moderators refer to those characteristics
in symptoms. Symptom changes and processes have to be that influence the intervention– outcome relation. For ex-
evaluated repeatedly and concurrently in a given study to ample, if therapy is more effective for patients with a
establish this. Merely assessing one variable (process) early particular characteristic (e.g., sex, ethnicity, gender iden-
and the other variable (outcome) later does not establish the tity, socioeconomic status, comorbidity), that characteristic
timeline. Second, studies do not explain how the process is said to be a moderator. Moderators, including character-
unfolds to alter patient functioning, that is, how the process istics of the patient, therapist, and contexts, have been
moves along a pathway that directly affects a particular studied extensively for several decades. However, it can be
outcome or set of outcomes (Kazdin, 2007).
There are many reasons to study mechanisms, but one
in particular will help clinical work and patient care. 2
If two variables are correlated (r), then one can identify the pro-
Knowing critical factors of treatment and the processes portion of common or shared influence (r2). For example, therapeutic
through which they operate can optimize therapeutic processes (e.g., alliance) “predict” therapeutic change. Researchers often
note that alliance accounts for a significant proportion of variance and
change. What ought to be fostered and optimized in therapy sometimes even more variance than other influences (e.g., treatment
to effect change? An obvious priority in health care is to technique). Further interpretation is often added to suggest that this must
transplant what is learned from research to clinical practice. mean that the alliance is why treatment leads to change or is the most
All agree on this, but psychologists ought to clarify pre- significant or important influence in therapy. These interpretations may be
cisely what should be extended and why. Knowing how true, but they do not follow from the metric. There is nothing in the
measure of percentage of variance that speaks to mediators or mecha-
therapy works will allow us to optimize the processes nisms. First, the shared variance between alliance and outcome could be
critical to change. huge, but that could be due to symptom change occurring before the
Is there any instance in which we do understand alliance. Second, the therapeutic alliance can “account for” treatment
mechanisms and in which that knowledge makes a differ- outcome variance but itself be explained by one or more other variables
such as symptom change that occurred before or at the same time the
ence in helping patients? Yes—the elegant work on fear alliance began to form, by common method variance in the alliance
conditioning and psychotherapy (Davis, 2006). There have outcome measures, or by characteristics of the patients before they began
been decades of research on Pavlovian conditioning of fear treatment. Each of these latter influences partially accounts for the per-
in humans and nonhuman animals. Conditioning as an centage of variance connection (see Kazdin, 2007, for an elaboration). I
explanation of fear acquisition and extinction as an expla- am not asserting that the relationship is unimportant, but rather am
commenting on the overinterpretation and misinterpretation of percentage
nation of fear reduction or elimination are useful paradigms of variance. It is useful to know multiple influences and the relative
for the processes involved in treatments of fear and anxiety. strengths of their effect sizes, but this information is different from
Conditioning and extinction of fear depend on a particular evidence supporting a mechanism of action.

difficult to translate the findings to clinical work and deci- predicts (moderates) responsiveness to a particular treat-
sion making to help individual clinicians and their patients. ment or to multiple treatments. We cannot be confident
There are two problems that make translation of re- about the answer because the pertinent research is rarely
search to practice difficult: (a) the ways in which modera- done. We would need to study a given moderator among
tors are studied and reported and (b) not knowing how the different techniques, patient samples, and in different con-
moderator works across multiple conditions or treatments. texts. Without this information, we do not know whether
My work, perhaps typical of research on moderators of the moderator will impede a particular therapy, several
psychotherapy, nicely illustrates the problem. Our research therapies, or all therapies that we might offer.
group has identified over 10 moderators of therapeutic Third, it would be helpful to understand what facet of
change among children diagnosed with conduct disorder the moderator is relevant or how the moderator works.
and referred clinically for outpatient services. Some of the Sometimes a moderator can be a proxy variable, that is, one
variables are child age, reading achievement, severity of that stands for or represents some other influence. For
dysfunction, parent psychopathology, parent quality of life, example, patient age, sex, ethnic identity, or cultural iden-
and parent child-rearing practices. As an example, in one of tity might be moderators of dropping out of treatment or of
our recent therapy outcome studies, barriers to participation therapeutic change among those who remain in treatment.
in therapy were shown to predict therapeutic change in Each of these variables may be accounted for or explained
separate samples of children diagnosed with oppositional by influences with which they are highly correlated (e.g.,
defiant disorder and conduct disorder (and with many other severity of clinical dysfunction, stress, mismatch between
comorbid disorders as well; Kazdin & Whitley, 2006). therapist and patient ethnicity). Moderators are only corre-
Barriers to participation in treatment consist of obstacles lates of outcome, but knowing more precisely the basis of
families experience once they enter treatment (e.g., seeing the moderators may provide the option to intervene to see
treatment as too demanding, questioning the relevance of if those moderators that can be altered (e.g., parent stress)
treatment) and are readily distinguishable from other mod- bear a causal relation to treatment outcome.
erators of treatment outcome (e.g., severity of child dys- Clinical decision making is criticized for not relying
function, parent and family dysfunction, and stress and life heavily on research. Yet how variables are studied and the
events in the family). All families received an EBT. Fam- ways the data are evaluated and reported often make the
ilies high in perceived barriers responded less well than did translation of findings difficult. The usual t test or regres-
families low in such barriers when other potentially con- sion equation that shows that a variable makes a difference
founding moderators, as noted above, were controlled. does not tell us what to do in decision making. Presenting
Whereas the research part is clear-cut—a significant mod- the data in ways that facilitate decision making more di-
erator was identified, the clinical part is not so clear-cut— rectly, presenting precisely what proportions of cases are
what should be done with this finding when making deci- likely to respond in one way rather than another, and
sions in clinical work? showing the strength of the effects for those who responded
Closer scrutiny showed that the higher barriers group well or poorly are some of the options that would help
responded significantly (statistically) less well to treatment clinical decision making. All decision making in clinical
than did the lower barriers group but still showed a large work will still be probabilistic; no matter what the research
effect size (d ⬎ 1 across multiple outcome measures).3 In shows, the results may not apply to a particular patient even
other words, barriers to treatment was a moderator, but if he or she is very similar to those in the original study.
what did we learn to help us apply this knowledge clini- Still, our best guess is to draw from the research— but we
cally? These patients still changed a lot on multiple out- need more helpful ways of making that translation from
come measures. Moreover, the study told us nothing about research to practice.
whether the moderator was specific to the treatment we Qualitative research. There is a third strategy
offered or is a characteristic these patients carry with them for research that I believe would bridge the gap between
to all forms of treatment. A clinician learning of the finding research and practice. I mention this only briefly in part
might silently ponder at the initial interview with such a because quantitative research and null hypothesis testing
patient, “I can see that this family coming to treatment is a dominate contemporary psychotherapy research and psy-
‘very high-barriers’ family.” However, the finding may not chology in general. Qualitative research is not routinely
help in decision making. taught in graduate schools (e.g., in the United States and
Three changes would improve the research on mod- Canada), and hence this option, while possibly useful in
erators, improve patient care, and help bridge research and principle, cannot be as easily adopted as the other recom-
practice. First, it would be useful to report findings in a way mendations. Even so, in relation to bridging research and
that makes them applicable to clinical work. A variable practice, qualitative research could play a special role.
may moderate treatment, but that does not necessarily Qualitative methods meet the desiderata of science; the
mean that some individuals (those on the “unfortunate” methods are systematic, replicable, and cumulative (see
side of the moderator, such as those with high barriers) will
Berg, 2001; Denzin & Lincoln, 2005). (For those who
respond poorly to treatment. We need to know more than
just the fact that a variable can significantly influence
outcome. 3
Convention has placed small, medium, and large effect sizes (d) at
Second, it would be helpful to know if a variable .2, .5, and .8, respectively (Cohen, 1988).

might be converted, I hasten to add that qualitative research effective. Generalizing from research, experience, and their
even has software for quantitative data analysis, so all of us combination is always probabilistic and does not guarantee
trained in the quantitative tradition can use numbers and an outcome. EBTs of all sorts (e.g., aspirin, bypass surgery,
statistics as a raft as we drift in the uncharted waters of plastic surgery, chemotherapy, antidepressant medication)
experience.) The methods look at phenomena in ways that cannot be depended on to produce the desired outcome
reveal many facets of human experience that the quantita- without exception. We consider systematic evaluation as
tive tradition has been partially designed to circumvent— pivotal in research. Evaluation is as important in patient
in-depth evaluation, subjective views, and how individuals care because the individual is so important and because we
represent (perceive, feel) and react to their situations and do not have a guarantee of the result, no matter what the
contexts. For example, qualitative research can look at the research or experiential base of the treatment(s) we use.
experience of those who go through treatment and the Second, it is important to monitor treatment effects in
thematic ways in which their lives and the lives of their an ongoing way to make decisions about continuing, alter-
partners are influenced. This information would be tremen- ing, or terminating treatment on the basis of how well the
dously useful for connecting measures to new metrics that patient is doing. It is now well documented that some
are not arbitrary and that reflect genuine changes in func- patients make rapid changes quite early in treatment (so-
tioning, perception, experience, and meaning. called sudden therapeutic gains), as has been shown, for
Qualitative research methods and their many varia- example, in patients with depression or anxiety (e.g., Hof-
tions are well suited to providing an understanding of the mann, Schultz, Meuret, Moscovitch, & Suvak, 2006; Tang
individual experience of patients, to codifying treatment & DeRubeis, 1999); others may not make expected
changes, and to doing so in replicable ways. Also, quali- changes and are unresponsive even to extended treatment
tative research can both test and generate conceptual mod- (so-called signal-alarm cases; Lambert et al., 2003). And,
els and specific hypotheses. The methods could benefit of course, there are the gradations in between and the cases
clinical research by providing a rigorous way to codify the in which change occurs in some areas of functioning but
experiences of individuals (patients, therapists) and to do not in others or at different rates among the various areas.
so in replicable ways. I am not advocating replacing RCTs Systematic assessment would permit finer delineations of
with qualitative research. However, investing narrowly, therapeutic change than would more global clinical judg-
whether in only one stock for a retirement plan or in a ments and unsystematic assessment.
single methodological tradition such as quantitative psy- Third, systematic evaluation is intended to comple-
chology, invariably bears a cost. Different methods can ment clinical judgment. Systematic measures are no sub-
reveal different facets of a phenomenon. In the context of stitute for clinical judgment, which may catch critical is-
psychotherapy research, scientific study of the individual sues that a given measure was not designed to identify.
would bring research much closer to the context of clinical However, the need for systematic evaluation stems in part
practice. Among the benefits, it is likely that intensive and from the limitations of judgment, perception, and memory
systematic study of experience would generate hypotheses and their implications for gathering information, as I men-
directly applicable to clinical work. These hypotheses tioned earlier. We want systematic evaluation because of
could be tested in quantitative and qualitative studies. the complementary information it provides, quite apart
Qualitative research would seem to be a natural way to from another advantage it confers in relation to bridging
bridge the divide between research and practice. research and practice (adding to the knowledge base),
which I note later in the article.
Clinical Practice
Recommendations and guidelines for using systematic
I have mentioned research emphases or shifts in priorities evaluation of the individual case in the context of treatment
that could improve patient care, advance the knowledge are not new (e.g., Bloom & Fischer, 1982; Fishman, 2001;
base, and reduce the gulf between research and practice. Hayes, Barlow, & Nelson, 1999; Kazdin, 1996; Meier,
Here I propose two parallel changes in clinical work to 2003). What are relatively new are the availability of
accomplish the same goals. measures that are validated for clinical use and the impend-
Use of systematic measures to evaluate ing pressures on clinical practice to monitor patient
patient progress. Patient progress in therapy is con- progress. As one example, the Outcome Questionnaire 45
stantly evaluated. By systematic evaluation, I mean the use (OQ-45) is a self-report measure designed to evaluate client
of psychological or other measures (e.g., if there are appli- progress (e.g., weekly) over the course of treatment and at
cations to physical health) that have or in principle could termination (see Lambert, Hansen, & Finch, 2001; Lambert
have reliability and validity and provide replicable infor- et al., 2003). The measure requires approximately five
mation about the status of patient functioning. There are minutes to complete and provides information on four
three related reasons to lobby for systematic assessment of domains of functioning: symptoms of psychological distur-
individual therapy in clinical practice. bance (primarily depression and anxiety), interpersonal
First and foremost, the key argument for systematic problems, social role functioning, (e.g., problems at work),
evaluation pertains to the primary goal of clinical practice: and quality of life (e.g., facets of life satisfaction). Total
to provide high-quality care. Whether one uses an EBT or scores across the 45 items present a global assessment of
one’s own brand of individualized treatment, one cannot be functioning. The measure has been evaluated extensively
sure, in principle or practice, that the treatment will be and applied to over 10,000 patients. There are other avail-

able measures, including ones that can be individualized to hypotheses, and to provide outstanding and astounding
patients and one’s own practice, that can help with decision findings that will add to the knowledge base (see Sechrest,
making and improve the quality of patient care. But they Stewart, Stickle, & Sidani, 1996). There are ways of ar-
can do much more, as I note next. ranging the clinical situation that even constitute quasi-
Clinical practice can contribute uniquely to experiments, a term legitimized by group research. Our
our knowledge base. We have taken as a given current training feeds a clinical–research split by not con-
that research contributes to the knowledge base and that veying that evaluation can help individual patients and
clinical practice is the application of that base. This is an contribute directly to the accumulation of knowledge in
exceedingly unfortunate way of conceptualizing the con- ways that span a continuum of scientific rigor. In general,
tributions of each domain because it fosters and maintains data obtained in clinical practice could directly contribute
the research–practice gap. Clinical work can contribute to knowledge, generate hypotheses to be studied in re-
directly to the scientific knowledge base. search, and, in the process, make treatment research more
A lamentable feature of our field is the knowledge lost aligned with and relevant to clinical practice. Although I
in clinical practice. The accumulated experience of a given am emphasizing the contribution to the knowledge base,
therapist retires when the therapist does. Typically, the the contribution of the database to clinical care is also
knowledge cannot be used in ways that will help future significant. A clinician can draw on prior cases and make
patients. There are, of course, exceptions. For example, inferences about what treatments to consider and combine
clinical supervision allows a clinician to pass on accumu- and what the outcomes were in seemingly similar patients.
lated experiences and beliefs. Yet this information is not Accumulated data can be analyzed in partnerships
accumulated knowledge and does not add to the knowledge with researchers— even if the data were saved until the end
base in the archival sense that I mean here. As another of one’s career, the database could still contribute to
exception, surveys ask practitioners about what they know, knowledge. Objections are easy to raise (e.g., “What about
believe, and have learned. Surveys too have their place in HIPAA [Health Insurance Portability and Accountability
codifying views and experiences, but they do not accumu- Act]?” “What about informed consent?”). The same issues
late substantive findings. emerge in research, and there are solutions (e.g., data can
Our field would profit enormously from codifying the be coded, records in research or practice must be protected)
experiences of clinicians in practice so that the information that can make these manageable and feasible within a
is accumulated and can be drawn on to generate and test practice setting. Let us not begin with what cannot be done.
hypotheses. There is no need for clinicians to become The task is to identify constructive changes that might be
researchers and to do complex data analyses. Yet clinicians made in therapy research and practice to do what we do
already are researchers in the sense of hypothesizing that a even better. Clinical experience, wisdom, novel hypothe-
particular treatment combination will have particular ef- ses, and knowledge are often lost because they are not in a
fects and testing this hypothesis with the individual case. form that we codify and accumulate. We are letting knowl-
This work is not evaluated, codified, and accumulated in an edge from practice drip through the holes of a colander. We
archival way and therefore is lost. can plug up those holes to retain critical information, and
Apart from improving patient care, systematic evalu- we can feed this information into research designed to test
ation in clinical practice can make novel, important, and hypotheses and add further support for what seems to be
scientifically sound contributions to the knowledge base. true from the data gathered in practice.
The accumulation of cases over time, each of which is
systematically evaluated, can yield new insights about Direct Collaborations
treatment processes and outcome. The fact that the condi- I have highlighted directions in which both research and
tions are not controlled, in the sense of experimentation, practice could move to reach across the research–practice
does not preclude their role in adding significantly to the gap. However, through the way in which I chose to orga-
knowledge base. In most sciences (e.g., epidemiology, ge- nize these recommendations (separate sections on research
ology, meteorology, anthropology, economics, and infrared and practice), I may have unwittingly underscored the gap.
astronomy), major conclusions are drawn from careful an- A genuine collaborative bridge could address that gap
alytic work and tests of hypotheses in uncontrolled situa- directly.
tions rather than from experiments, although many fields We need collaborations between colleagues who iden-
have variables that can be controlled and tested experimen- tify themselves as primarily from research and those who
tally. identify themselves as primarily from practice to work
I mentioned previously the availability of measures. directly on this bridge and evaluate clinical practice. Those
Perhaps there is more than a measurement issue here. A coming from a strict research perspective often lament the
critical deficit in our clinical training is in the evaluation of term EBP because there is evidence for many treatments
clinical cases in the context of “real” therapy and clinical (EBTs) but not much in the way of evidence that draws on
practice. Indeed, we have learned some things about study- and modifies the application of these treatments in light of
ing the individual that are not quite true. It is possible, for clinical judgment, expertise, and contextual considerations
example, to draw causal inferences from studying the in- in practice. Therapy will invariably involve judgment and
dividual, to bring to bear information from the case, to help experience. We may always want evidence seasoned by
make rival hypotheses implausible, to test and generate experience and clinical judgment; when the evidence is

unusually weak or barely existent, we want experience There are three interrelated ways to improve the out-
seasoned by evidence. Although medicine and medical comes of treatment. First, identifying effective and the
analogies for some reason have become inherently offen- most effective interventions can improve outcome. The
sive, they convey the point even more clearly than therapy. development of EBTs is an illustration of this way. Second,
A medical EBT for a disease or dysfunction may be firmly understanding how and why an effective treatment works
established (e.g., surgery, organ transplant, medication can improve outcome. Such research identifies what to
with strong side effects), but moderating variables that focus on, manipulate, and optimize in treatment, as pro-
have yet to be studied, including individual characteristics vided in an example earlier in this article. The third way to
(e.g., age of the patient, comorbid disorders, likelihood of improve outcome is to identify moderators of treatment.
recovery) and contextual variables (e.g., family supports, Therapy works well for some individuals and in some
living conditions), may influence or indeed dictate the contexts and works less well or not at all in others. The
decision as to whether to apply the EBT. Given the nature variables that delineate these different outcomes, that is,
of clinical care, the role of judgment, expertise, and context moderators, permit better triage. Outcomes for the popula-
ought to be studied directly. tion receiving treatment are better because one can direct
There are many questions about EBP, but asking even patients to treatments from which they are more likely to
the most basic would be a constructive beginning. Al- profit. Clinical–research collaboration on EBP and on how
though not necessarily the most informative, the “horse- moderators are utilized in practice can help improve out-
race” question might be the place to mobilize interest. come in this latter way.
Researchers often agree that comparisons of competing The movement toward individualized medicine illus-
treatments are not the best place to begin. Yet these com- trates the utility of identifying moderators in clinical ap-
parisons often galvanize attention and bring people into the plications of treatment. Individualized medicine is not yet
area with more nuanced questions. The horse-race question really individualized in the sense that each patient is
would compare the application of an unadulterated EBT uniquely evaluated. Rather, current work focuses on iden-
(e.g., exposure-based therapy for anxiety) with the ap- tifying a moderator (e.g., genetic characteristic, family
plication of the same treatment (or principles or modules history) that influences the impact of some intervention. (A
of that treatment) but with clinical judgment, expertise, more accurate term might be moderated medicine.) For
example, levels of high density lipoprotein (HDL, or the
and context considered as a supplement. The decision-
“good cholesterol”) can be increased or decreased (good
making process could be codified to identify themes and
and bad news, respectively) with consumption of polyun-
how they are applied. From experience with research,
saturated fats in one’s diet. The beneficial or deleterious
one can surmise the likely result. For many patients, the
effects are moderated by a subtle genetic variation (poly-
standard treatment (unadulterated) would be fine, but for
morphism of apolipoprotein [Apo A-1]). Knowing about
other patients, integrating the intervention with clinical
this genetic moderator is important for treatment recom-
judgment, expertise, and context would make a palpable mendations because the same recommendation (increase
difference in treatment outcome. A researcher– clinician polyunsaturated fats in one’s diet) could have opposite
collaboration that helped identify the circumstances effects depending on the gene allele (Corella & Ordovas,
(moderating variables) in which judgment, expertise, 2005). Other examples have emerged in which practices
and context are important would be enormously helpful (e.g., exercise, reducing cholesterol) appear to reduce risk
in patient care and of course would bridge research and for untoward outcomes (e.g., Alzheimer’s disease) but only
practice. for individuals without a genetic susceptibility to the dis-
A second line of direct bridging work would under- order (see Gatz, 2007). These examples illustrate how
score the distinction between the technique and its method knowledge of moderators can guide patients to and away
of delivery. I am considering these as separate to make a from practices that may and may not make a difference in
point, although I readily acknowledge their overlapping outcome; the knowledge can also lead researchers to pursue
(Venn diagram) nature. It might well be that a given EBT subgroups in which new information is lacking in order to
would be effective for many different populations and in improve clinical outcomes.
many different contexts if it could be applied. However,
contextual influences related to delivery (e.g., ethnic and Conclusions
cultural compatibility of the treatment with patient values, The hiatus between clinical research and practice has been
delivery in schools rather than clinics, recasting the inter- heightened by advances in research, pressures on practice,
vention outside the context of “psychotherapy” or mental and their combination. By combination, I refer to the public
health services) may be critical. The effective intervention scrutiny of research. State legislatures and third-party pay-
might not be deliverable unless clinical judgment, exper- ers, for example, are drawing on research to decide what is
tise, and context modify how the treatment is presented and appropriate to do in practice, what is reimbursed, and what
described. I mentioned that delivery and technique over- the rates of reimbursement will be. Our internal discussions
lap— key ingredients of the therapy might need to be about the merits of this or that treatment or set of studies
modified as well, because without modification they might and the generalizability of findings now have a larger
drive patients away from treatment or make adherence to audience. I mention these influences to convey that the
treatment unlikely. issues addressed in this article are not occurring in a vac-

uum, free from priorities related to the funding of research benefit as a field from the accumulated practice of clini-
and mental health services. cians, with the rare exception of those whose groundbreak-
Two guiding questions served as the impetus for this ing treatments may spawn empirical research. For the rest
article: (a) Are there better ways or special opportunities to of us, there is a potentially rich data set lost when our
bridge the divide between clinical research and practice practices end. We do not need the clinician to become a
than those currently in use? and (b) How can we improve researcher any more than we want the researcher to become
the quality of patient care? I have suggested some changes a practitioner. Both the clinician and the researcher can
in both research and practice that relate to these questions. mine the data for practical and scientific questions.
In relation to research, more work is needed on the mech- There are many ideas and models of integrating re-
anisms of change—not correlates of change alone but ex- search and practice for clinical care, training in clinical
planations of precisely how therapy works. The best prac- psychology, and evaluating clinical work (e.g., Borkovec,
tice will continue to be based on the best science. Let us 2004; Chorpita, Daleiden, & Weisz, 2005; Kendall & Bei-
attempt to understand more about the many change pro- das, 2007; Nelson & Steele, 2006; Schaughency & Ervin,
cesses and how they can be triggered, activated, exploited, 2006; Trierweiler, 2006; Weisz, Jensen, & McLeod, 2005).
and trained. This is different from disseminating treatment The efforts and need to integrate research and practice are
manuals and prescribing specific interventions as our pri- timely and more important than ever before because of the
mary focus. stakes involved in academic and clinical training, research,
Research can do more to identify moderators of treat- practice, and health care in general. This article suggests
ment and how they make a difference (i.e., across one emphases in different areas to improve patient care as well
treatment or all treatments) and to report these results so as to better unite research and practice. We can continue to
that clinicians can make better decisions. I mentioned qual- argue why research may not reflect the conditions of clin-
itative research as a third priority because it provides a ical practice and how clinical decision making and judg-
methodology in which rigor and clinical relevance unite. I ment are difficult to defend. No one is served by such a
recognize that accredited programs in psychology may not dialogue, and the patients and the public at large are the
even mention qualitative research yet may offer courses on most poorly served. The task for psychologists is to con-
the subject. I have taught research methodology at the sider how we can bring together more constructively the
graduate level at three universities. Within the quantitative/ critical facets of our field to improve our understanding of
null hypothesis tradition there is so much to teach. Ongoing patient care and delivery of services to make a difference.
advances must constantly be added to the canon to prepare Although efforts to bridge research and practice are not
students competently. There is little time to train in other new, there are now special opportunities as improved treat-
traditions (e.g., qualitative research, single-case experimen- ments and measures have become available for clinical use
tal designs) given the scope of courses required, perhaps and as the dialogue among those involved primarily in
especially so for clinical and counseling psychology stu- research or practice has increased (e.g., APA Presidential
dents, who have additional courses and experiences re- Task Force on Evidence-Based Practice, 2006; Goodheart
quired for state licensing. Still, it is worth mentioning here et al., 2006).
because qualitative research allows intensive study of in- In the process of bridging research and practice in the
dividuals in a scientifically rigorous way and for these ways I am suggesting, we would serve psychology in
reasons bridges the research–practice divide. general as well. Our training in theories, methods, hypoth-
Shifts in emphases in clinical practice were also sug- esis testing, assessment, and data evaluation is unique
gested. Monitoring treatment with systematic assessment among the many disciplines that deliver services. Not only
was the first suggestion. Using an EBT, whether integrated do our research-generated treatments make our work spe-
with experience, judgment, and contextual considerations cial, but our methods of evaluation can improve patient
or not, of course, does not guarantee a positive treatment care in ways that render psychology and clinical practice by
outcome. This is a major reason why the patient’s progress psychologists unique.
should be monitored in a systematic way if at all possible.
Our many unique contributions as psychologists include REFERENCES
remarkable literatures on cognitive heuristics, memory, and
APA Presidential Task Force on Evidence-Based Practice. (2006). Evi-
perception that teach us why we need such tools. Research dence-based practice in psychology. American Psychologist, 61, 271–
on measurement has provided reliable and valid tools that 285.
can be used in clinical and other applied settings to benefit Berg, B. L. (2001). Qualitative research methods for the social sciences
directly the people we serve. Research on psychological (4th ed.). Needham Heights, MA: Allyn & Bacon.
Blanton, H., & Jaccard, J. (2006). Arbitrary metrics in psychology. Amer-
treatments conveys what we can do to increase the likeli- ican Psychologist, 61, 27– 41.
hood of producing therapeutic change. There is no other Bloom, M., & Fischer, J. (1982). Evaluating practice: Guidelines for the
discipline that can claim any of this or that is in such a accountable professional. Englewood Cliffs, NJ: Prentice Hall.
position to provide empirically supported treatment and Borkovec, T. D. (2004). Research in training clinics and practice research
assessment. networks: A route to the integration of science and practice. Clinical
Psychology: Science and Practice, 11, 211–215.
Finally, assessment in the context of patient care will Burns, B., & Hoagwood, K. E. (Eds.). (2005). Evidence-based practice,
overcome what I consider to be a very regrettable loss of Part II: Effecting change [Special issue]. Child and Adolescent Psychi-
accumulated knowledge from clinical practice. We do not atric Clinics of North America, 14(2).

Chorpita, B. F., Daleiden, E. L., & Weisz, J. R. (2005). Modularity in the Kazdin, A. E. (Ed.). (1996). Evaluation in clinical practice [Special
design and application of therapeutic interventions. Applied and Pre- section]. Clinical Psychology: Science and Practice, 3, 144 –181.
ventive Psychology, 11, 141–156. Kazdin, A. E. (2000). Psychotherapy for children and adolescents: Di-
Cohen, J. (1988). Statistical power analysis for the behavioral sciences rections for research and practice. New York: Oxford University Press.
(2nd ed.). Hillsdale, NJ: Erlbaum. Kazdin, A. E. (2001). Almost clinically significant (p ⬍ .10): Current
Comas-Dı́az, L. (2006). Cultural variation in the therapeutic relationship. measures may only approach clinical significance. Clinical Psychology:
In C. D. Goodheart, A. E. Kazdin, & R. J. Sternberg (Eds.), Evidence- Science and Practice, 8, 455– 462.
based psychotherapy: Where practice and research meet (pp. 81–105). Kazdin, A. E. (2007). Mediators and mechanisms of change in psycho-
Washington, DC: American Psychological Association. therapy research. Annual Review of Clinical Psychology, 3, 1–27.
Corella, D., & Ordovas, J. M. (2005). Single nucleotide polymorphisms Kazdin, A. E. (in press). Psychosocial treatments for conduct disorder in
that influence lipid metabolism: Interaction with dietary factors. Annual children and adolescents. In P. E. Nathan & J. M. Gorman (Eds.), A
Review of Nutrition, 25, 341–390. guide to treatments that work (3rd ed.). New York: Oxford University
Davis, M. (2006). Neural systems involved in fear and anxiety measured Press.
with fear-potentiated startle. American Psychologist, 61, 741–756. Kazdin, A. E., & Whitley, M. K. (2006). Comorbidity, case complexity,
Davis, M., Myers, K. M., Chhatwal, J., & Ressler, K. J. (2006). Pharma- and effects of evidence-based treatment for children referred for dis-
cological treatments that facilitate extinction of fear: Relevance to ruptive behavior. Journal of Consulting and Clinical Psychology, 74,
psychotherapy. NeuroRx, 3, 82–96. 455– 467.
Dawes, R. M. (1994). House of cards: Psychology and psychotherapy Kendall, P. C. (Ed.). (1999). Clinical significance [Special section]. Jour-
built on myth. New York: Free Press. nal of Consulting and Clinical Psychology, 67, 283–339.
De Los Reyes, A., & Kazdin, A. E. (2006). Conceptualizing changes in Kendall, P. C., & Beidas, R. S. (2007). Smoothing the trail of dissemi-
behavior in intervention research: The Range of Possible Changes nation of evidence-based practices for youth: Flexibility within fidelity.
Model. Psychological Review, 113, 554 –583. Professional Psychology: Research and Practice, 38, 13–20.
Denissenko, M. F., Pao, A., Tang, M., & Pfeifer, G. P. (1996). Preferential Kiesler, D. J. (1971). Experimental designs in psychotherapy research. In
formation of benzo[a]pyrene adducts at lung cancer mutational hotspots A. E. Bergin & S. L. Garfield (Eds.), Handbook of psychotherapy and
in P53. Science, 274, 430 – 432. behavior change: An empirical analysis (pp. 36 –74). New York:
Denzin, N. K., & Lincoln, Y. S. (Eds.). (2005). The SAGE handbook of Wiley.
qualitative research (3rd ed.). Thousand Oaks, CA: Sage. Lambert, M. J., Hansen, N. B., & Finch, A. E. (2001). Client-focused
Doss, A. J., & Weisz, J. R. (2006). Syndrome co-occurrence and treatment research: Using client outcome data to enhance treatment effects. Jour-
outcomes in youth mental health clinics. Journal of Consulting and nal of Consulting and Clinical Psychology, 69, 159 –172.
Clinical Psychology, 74, 416 – 425. Lambert, M. J., & Ogles, B. M. (2004). The efficacy and effectiveness of
Eysenck, H. J. (1952). The effects of psychotherapy: An evaluation. psychotherapy. In M. J. Lambert (Ed.), Bergin and Garfield’s handbook
Journal of Consulting Psychology, 16, 319 –324. of psychotherapy and behavior change (5th ed., pp. 139 –193). New
York: Wiley.
Eysenck, H. J. (1966). The effects of psychotherapy [with commentaries].
Lambert, M. J., Whipple, J. L., Hawkins, E. J., Vermeersch, D. A.,
New York: International Science Press.
Nielsen, S. L., & Smart, D. W. (2003). Is it time for clinicians to
Fishman, D. B. (2001). From single case to database: A new method for
routinely track patient outcome? A meta-analysis. Clinical Psychology:
enhancing psychotherapy, forensic, and other psychological practice.
Science and Practice, 10, 288 –301.
Applied & Preventive Psychology, 10, 275–304.
Levitt, E. E. (1957). The results of psychotherapy with children: An
Garb, H. N. (2005). Clinical judgment and decision making. Annual evaluation. Journal of Consulting Psychology, 21, 189 –196.
Review of Clinical Psychology, 1, 67– 89. Levitt, E. E. (1963). Psychotherapy with children: A further evaluation.
Gatz, M. (2007). Genetics, dementia, and the elderly. Current Directions Behaviour Research and Therapy, 60, 326 –329.
in Psychological Science, 16, 123–127. Luborsky, L., & Luborsky, E. (2006). Research and psychotherapy: The
Gilovich, T., Griffin, D., & Kahneman, D. (2002). Heuristics and biases: vital link. Lanham, MD: Jason Aronson.
The psychology of intuitive judgment. Cambridge, England: Cambridge Meehl, P. E. (1954). Clinical vs. statistical prediction. Minneapolis:
University Press. University of Minnesota Press.
Goodheart, C. D., Kazdin, A. E., & Sternberg, R. J. (Eds.). (2006). Meier, S. T. (2003). Bridging case conceptualization, assessment, and
Evidence-based psychotherapy: Where practice and research meet. intervention. Thousand Oaks, CA: Sage.
Washington, DC: American Psychological Association. Nelson, T. D., & Steele, R. G. (2006). Beyond efficacy and effectiveness:
Griner, D., & Smith, T. B. (2006). Culturally adapted mental health A multifaceted approach to treatment evaluation. Professional Psychol-
intervention: A meta-analytic review. Psychotherapy: Theory, Re- ogy: Research and Practice, 37, 389 –397.
search, Practice, Training, 43, 531–548. Norcross, J. C. (Ed.). (2002). Psychotherapy relationships that work:
Hayes, S. C., Barlow, D. H., & Nelson, R. O. (1999). The scientist– Therapist contributions and responsiveness to patients. New York:
practitioner: Research and accountability in clinical and educational Oxford University Press.
settings (2nd ed.). New York: Pergamon. Norcross, J. C., Beutler, L. E., & Levant, R. F. (Eds.). (2005). Evidence-
Hayes, S. C., Follette, V. M., Dawes, R. M., & Grady, K. E. (1995). based practices in mental health: Debate and dialogue on the funda-
Scientific standards of psychological practice: Issues and recommen- mental questions. Washington, DC: American Psychological Associa-
dations. Reno, NV: Context Press. tion.
Hoagwood, K., Hibbs, E., Brent, D., & Jensen, P. J. (1995). Efficacy and Patterson, G. R., Reid, J. B., & Dishion, T. J. (1992). Antisocial boys.
effectiveness in studies of child and adolescent psychotherapy. Journal Eugene, OR: Castalia.
of Consulting and Clinical Psychology, 63, 683– 687. Perepletchikova, F., & Kazdin, A. E. (2005). Oppositional defiant disorder
Hofmann, S. G., Meuret, A. E., Smits, J. A., Simon, N. M., Pollack, M. H., and conduct disorder. In K. Cheng & K. M. Myers (Eds.), Child and
Eisenmenger, K., Shiekh, M., & Otto, M. W. (2006). Augmentation of adolescent psychiatry: The essentials (pp. 73– 88). Philadelphia: Lip-
exposure therapy with D-cycloserine for social anxiety disorder. Ar- pincott Williams & Wilkins
chives of General Psychiatry, 63, 298 –304. Pohl, R. F. (Ed.). (2004). Cognitive illusions: A handbook on fallacies and
Hofmann, S. G., Schultz, S. M., Meuret, A. E., Moscovitch, D. A., & biases in thinking, judgment, and memory. New York: Psychology
Suvak, M. (2006). Sudden gains during therapy of social phobia. Press.
Journal of Consulting and Clinical Psychology, 74, 687– 697. Reid, J. B., Patterson, G. R., & Snyder, J. (Eds.). (2002). Antisocial
Hunsley, J. (2007). Addressing key challenges in evidence-based practice behavior in children and adolescents: A developmental analysis and
in psychology. Professional Psychology: Research and Practice, 38, model for intervention. Washington, DC: American Psychological As-
113–121. sociation.
Institute of Medicine (2001). Crossing the quality chasm: A new health Ressler, K. J., Rothbaum, B. O., Tannenbaum, L., Anderson, P., Graap,
system for the 21st century. Washington, DC: National Academy Press. K., Zimand, E., Hodges, L., & Davis, M. (2004). Cognitive enhancers

as adjuncts to psychotherapy: Use of D-cycloserine in phobic individ- Tang, T. Z., & DeRubeis, R. J. (1999). Sudden gains and critical sessions
uals to facilitate extinction of fear. Archives of General Psychiatry, 61, in cognitive– behavioral therapy for depression. Journal of Consulting
1136 –1144. and Clinical Psychology, 67, 894 –904.
Roediger, H. L., III, & McDermott, K. B. (2000). Distortions of memory. Trierweiler, S. J. (2006). Training the next generation of psychologist
In E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of memory clinicians: Good judgment and methodological realism at the interface
(pp. 149 –162). New York: Oxford University Press. between science and practice. In C. D. Goodheart, A. E. Kazdin, & R. J.
Sackett, D. L., Rosenberg, W. M. C., Gray, J. A. M., Haynes, R. B., & Sternberg (Eds.), Evidence-based psychotherapy: Where practice and
Richardson, W. S. (1996). Evidence-based medicine: What it is and research meet (pp. 211–238). Washington, DC: American Psycholog-
what it isn’t. British Medical Journal, 312, 71–72. ical Association.
Schaughency, E., & Ervin, R. (2006). Building capacity to implement and Wampold, B. E. (2001). The great psychotherapy debate: Models, meth-
sustain effective practices to better serve children. School Psychology ods, and findings. Mahwah, NJ: Erlbaum.
Review, 35, 155–166. Weisz, J. R., Jensen, A. L., & McLeod, B. D. (2005). Development and
Sechrest, L., Stewart, M., Stickle, T. R., & Sidani, S. (1996). Effective and dissemination of child and adolescent psychotherapies: Milestones,
persuasive case studies. Cambridge, MA: Human Services Research methods, and a new deployment-focused model. In E. D. Hibbs & P. S.
Institute. Jensen (Eds.), Psychosocial treatments for child and adolescent disor-
Sorenson, R. L., Gorsuch, R. L., & Mintz, J. (1985). Moving targets: ders: Empirically based strategies for clinical practice (2nd ed., pp.
Patients’ changing complaints during psychotherapy. Journal of Con- 9 –39). Washington, DC: American Psychological Association.
sulting and Clinical Psychology, 53, 49 –54. Westen, D., & Morrison, K. (2001). A multi-dimensional meta-analysis of
Stirman, S. W., DeRubeis, R. J., Brody, P. E., & Crits-Christoph, P. treatments for depression, panic, and generalized anxiety disorder: An
(2003). Are samples in randomized controlled trials of psychotherapy empirical examination of the status of empirically supported treatments.
representative of community outpatients? A new methodology and Journal of Consulting and Clinical Psychology, 69, 875– 889.
initial findings. Journal of Consulting and Clinical Psychology, 71, Westen, D., Novotny, C. M., & Thompson-Brenne, H. (2004). The em-
963–972. pirical status of empirically supported psychotherapies: Assumptions,
Tanenbaum, S. J. (2005). Evidence-based practice as mental health policy: findings, and reporting in controlled clinical trials. Psychological Bul-
Three controversies and a caveat. Health Affairs, 24, 163–173. letin, 130, 631– 663.

Kazdin, 2008

Uploaded by

Document Information

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

Kazdin, 2008

Uploaded by

Copyright:

Available Formats

Evidence-Based Treatment and Practice

New Opportunities to Bridge Clinical Research and Practice,

T here is a well-recognized split within clinical psy-

ventions.1 A central issue is the extent to which findings

146 April 2008 ● American Psychologist

April 2008 ● American Psychologist 147

148 April 2008 ● American Psychologist

April 2008 ● American Psychologist 149

150 April 2008 ● American Psychologist

April 2008 ● American Psychologist 151

152 April 2008 ● American Psychologist

April 2008 ● American Psychologist 153

154 April 2008 ● American Psychologist

April 2008 ● American Psychologist 155

156 April 2008 ● American Psychologist

April 2008 ● American Psychologist 157

158 April 2008 ● American Psychologist

April 2008 ● American Psychologist 159

You might also like