ReeseUsherBowmanetetal

Training and Education in Professional Psychology © 2009 American Psychological Association
2009, Vol. 3, No. 3, 157–168 1931-3918/09/$12.00 DOI: 10.1037/a0015673

Using Client Feedback in Psychotherapy Training:
An Analysis of its Influence on Supervision and Counselor Self-Efficacy

Robert J. Reese and Ellen L. Usher Derek C. Bowman, Larry A. Norsworthy,
University of Kentucky Jackie L. Halstead, Steven R. Rowlands,
and Rachael R. Chisholm
Abilene Christian University

Tracking client outcome and the therapeutic relationship across treatment (i.e., client feedback) has become
a recommended practice for clinicians. This study investigated whether the utility of this practice would extend
to trainees if the data gained from clients was provided to their supervisor for use within supervision. Trainees
(N 28) were assigned to a continuous feedback condition or no-feedback condition for 1 academic year.
Results indicated that trainees in both conditions demonstrated better client outcomes at the end of their
practicum training than at the beginning, but those in the feedback condition improved more. However, those
in the feedback condition did not rate the supervisory alliance or satisfaction with the supervision process
differently. The relationship between counselor self-efficacy and outcome was stronger for trainees in the
feedback condition than for those in the no-feedback condition, perhaps indicating that feedback may facilitate
a more accurate assessment of one’s skills. Implications of how counseling self-efficacy, the supervisory
alliance, and satisfaction with supervision are related to effective supervision are addressed.

Keywords: psychotherapy supervision, counselor trainee, counseling effectiveness, client progress and
outcome, counselor self-efficacy

Supervised practice is typically perceived as the sine qua non The lack of supervision research in this area is certainly not
of excellent graduate psychotherapy training, particularly when the result of failing to recognize its importance. For example,
it comes to helping students develop their therapy skills. Al- Ellis and Ladany (1997) stated that client outcome is the “acid
though the developmental needs of trainees (we use supervisee test” of effective supervision (p. 485). Other reviews of the
and trainee interchangeably) influence the specific tasks and supervision literature have also consistently noted the lack of
goals of training (Stoltenberg, 2005), the general purpose of supervision-related studies that address client outcome (Good-
psychotherapy supervision is to ensure ethical and competent year & Bernard, 1998). Freitas (2002) conducted a review of the
treatment (i.e., meet the needs of the client) and to promote skill literature on the influence of supervision on client outcome and
and professional development (i.e., meet the needs of the found only 10 studies, all of which were conducted before 1998.
trainee). Much of the supervision research to date has focused The studies he reviewed focused on client outcome but were
on the latter, but there is a paucity of research that has focused otherwise disparate in nature. For example, 1 study found that
on the benefits of psychotherapy supervision for clients (Lam- client-focused supervision led to a better outcome than an
bert & Hawkins, 2001). administrative approach (Iberg, 1991); another found that su-

ROBERT J. REESE is an assistant professor in the Department of Educa- Marriage and Family Therapy. Research interests include emotional intel-
tional, School, and Counseling Psychology at the University of Kentucky. ligence as predictor of therapeutic training outcome and intrinsic and
He earned his PhD in counseling psychology from Texas A&M University. extrinsic motivation factors in faculty at faith-based universities.
His research interests include psychotherapy outcome, supervision, and the STEVEN R. ROWLANDS is director of the University Counseling Center,
use of telehealth media for providing services to underserved populations. Abilene Christian University. He is a clinical member of the American
ELLEN L. USHER is an assistant professor in the Department of Educa- Association for Marriage and Family Therapy. Steve is a licensed marriage
tional, School, and Counseling Psychology at the University of Kentucky. and family therapist and a licensed professional counselor. Additionally,
She earned her PhD in educational studies from Emory University. Her Steve is an adjunct faculty member for the Department of Marriage and
research interests include social– cognitive theory, motivation, and sources Family Therapy.
of self-efficacy beliefs. RACHAEL R. CHISHOLM received her master’s degree in marriage and
DEREK C. BOWMAN received his master’s degree in clinical psychology family therapy from Abilene Christian University. She is currently in
from Abilene Christian University. His research interests include career independent practice in Terrell, Texas.
development and the psychotherapeutic process. THE AUTHORS wish to acknowledge the Institute for the Study of Ther-
LARRY A. NORSWORTHY is a professor of psychology in the Department apeutic Change (http://www.talkingcure.com) for providing access to the
of Psychology at Abilene Christian University. He earned his PhD in ORS and SRS used in this study.
clinical psychology at Walden University. Research interests include psy- CORRESPONDENCE CONCERNING THIS ARTICLE should be addressed to
chotherapy outcome, supervision, and chronic pain. Robert J. Reese, Department of Educational and Counseling Psychology,
JACKIE L. HALSTEAD is a department chair and an associate professor at University of Kentucky, 235 Dickey Hall, Lexington, KY 40506-0017.
Abilene Christian University. Her PhD is from Iowa State University in E-mail: jeff.reese@uky.edu

157

158 REESE ET AL.

pervisee attraction to the supervisor improved client outcome cum training settings, supervisees are placed in environments that
(Dodenhoff, 1981). The lack of a larger, cohesive focus in this do not allow audio or video recording. In such cases, supervisors
area is surprising given the centrality of supervision to psycho- are left to rely on a trainee’s perspective. Moreover, research has
therapy training. consistently shown that therapists, regardless of experience, have
difficulty judging whether their clients are deteriorating in treat-
Client Feedback in Psychotherapy ment (Grove, Zald, Lebow, Snitz, & Nelson, 2000).
From a supervisory process perspective, client feedback may
One advance in the psychotherapy outcome literature has been also help the supervisor provide feedback to supervisees. Effective
the development of continuous outcome assessment to track client supervision is generally assumed to require that the supervisor
progress in psychotherapy. Continuous outcome assessment, or provide a balance of both positive and challenging feedback
client feedback, refers to using a psychotherapy outcome measure (Falender & Shafranske, 2004). However, most supervisors with-
every session rather than in a typical pre–post outcome format hold feedback, particularly feedback regarding performance (Hoff-
(Lambert, 2001). The premise of using such an approach is that it man, Hill, Holmes, & Freitas, 2005), and if feedback is provided,
helps identify clients who are not progressing as expected in it is often general and positive (Friedlander, Siegel, & Brenock,
treatment and affords the opportunity to proactively address their 1989).
lack of improvement. A number of research studies have indicated
that continuous outcome assessment leads to fewer premature
Purpose of Study
terminations and improved outcomes for clients who are at risk for
terminating treatment early (e.g., Brown & Jones, 2005; Whipple Although the use of client feedback data has been proffered to
et al., 2003). These findings prompted the American Psychological be of value in the supervisory process, there is no research to
Association’s Division 29 (Psychotherapy) Task Force on Empir- support this contention. The purpose of this study was to investi-
ically Supported Relationships to advise practitioners “to routinely gate the use of continuous client outcome data in psychotherapy
monitor patients’ responses to the therapy relationship and ongo- supervision. The study was also designed to help redress the lack
ing treatment. Such monitoring leads to increased opportunities to of research on the influence of psychotherapy supervision on client
repair alliance ruptures, to improve the relationship, and to avoid outcome, as well as to assess how using such data in supervision
premature termination” (Ackerman et al., 2001, p. 496). might influence known correlates of effective supervision (i.e., the
supervisory alliance and satisfaction with supervision) and coun-
Using Client Feedback Data in Supervision: selor self-efficacy. In the following sections, we review the liter-
Possible Advantages ature in these areas and provide hypotheses for the study.

Lambert and Hawkins (2001) have discussed the potential ad- Supervisory Alliance and Satisfaction With Supervision
vantages of using client outcome data for counselor training and
supervision. They have suggested that supervisees could use client The supervisory relationship is an important variable to consider
outcome data to inform treatment and to discuss progress with in relation to the use of client feedback data in supervision.
their clients. The client outcome data could also be used by the Previous research has found that specific and clear supervisory
supervisor and supervisee to help inform supervision. Lambert and feedback can contribute to an improved supervisory alliance and
Hawkins noted that tracking outcome data can provide feedback to increased satisfaction with supervision (Lehrman-Waterman &
supervisees regarding their performance, provide both supervisor Ladany, 2001). In fact, Ellis (1991) found that trainees rated the
and supervisee with more information that may facilitate training, supervisory relationship as the most important component of a
provide beginning therapists with information about where to positive supervisory experience. Although the supervisory alliance
begin with clients, and also help ensure that clients are not harmed is perceived to be integral to the supervisory process, those who
by therapy. Beginning therapists want to know whether they are have measured its relationship to counselor affect and develop-
being helpful to clients; using outcome data provides a specific ment have reported mixed results. For example, Ladany, Ellis, and
means for broaching this question. It is conceivable that a trainee Friedlander (1999) found that the supervisory alliance was unre-
could complete his or her training without really having an answer lated to counselor self-efficacy. Yet others have found that a strong
to this question beyond some general notion. supervisory working alliance was positively related to counseling
Another possible advantage of using client outcome data in self-efficacy (Efstation, Patton, & Kardash, 1990). Despite this
supervision is that it may help make more efficient use of super- ambiguity, a strong supervisory alliance has been found to be
vision time. Client outcome data provide a quick “dashboard related to a positive therapeutic alliance (Patton & Kivlighan,
indicator” for a supervisee’s client load and may help quickly 1997). Clearly, more evidence is needed to assess the potential
identify clients who require more attention during supervision. benefits of the supervisory alliance.
Accurate assessment of a supervisee and his or her caseload can be Counseling researchers have also sought to examine the degree
challenging, and time consuming, when dealing with multiple of satisfaction counselors in training feel with the supervision they
supervisees with large caseloads. Not only can using outcome data receive. Evaluating trainees’ level of satisfaction is a straightfor-
provide key information for assisting supervisees who are strug- ward way to measure their perception of their supervisor and the
gling with a client, it can also highlight and reinforce the growth supervisory process. Satisfaction with supervision has been an
of supervisees with clients who are faring well in therapy. Out- integral outcome variable in many studies evaluating the supervi-
come data can also assist in granting supervisors more direct sory process (Ellis & Ladany, 1997). However, some have sug-
access to their trainees’ performance. For example, in many practi- gested that supervisory satisfaction should be considered not as an

CLIENT FEEDBACK AND SUPERVISION 159

outcome variable but as a process variable that may influence Client Feedback Systems
therapy outcomes (Fernando & Hulse-Killacky, 2005). Either way,
continuing to evaluate the relationship between perceptions of The Outcome Questionnaire 45 (OQ45; Lambert et al., 1996) is
supervision (i.e., both the supervisory alliance and satisfaction the most researched instrument used for client feedback (e.g.,
with supervision) and client outcome is certainly warranted. We Lambert et al., 2001; Whipple et al., 2003). The OQ45 is com-
expected that trainees who received continuous assessment feed- pleted by the client right before every session and provides feed-
back would rate the supervisory alliance and satisfaction with back to the therapist on the client’s progress. The thrust of the
supervision more favorably than would trainees in a no-feedback instrument’s use is to identify clients who are not progressing and
condition. This is because we believe continuous assessment mea- who are projected, on the basis of normative data, to terminate
sures would likely facilitate the supervisor’s ability to provide prematurely or not to benefit from treatment. The feedback from
more specific and clear feedback to trainees regarding their per- the client allows the therapist to intervene by changing the course
formance with clients; previous research has linked specific and of treatment, assessing the reason(s) for the lack of progress such
clear feedback to favorable ratings of supervision (Lehrman- as a poor therapeutic alliance (see Lambert, 2001, for a full
Waterman & Ladany, 2001). description), or both. In their meta-analysis, Lambert et al. (2003)
concluded that clients who were identified as at risk for terminat-
ing prematurely or having a poor treatment outcomes attended
Feedback and Counseling Self-Efficacy more sessions and had better treatment outcomes when using
continuous assessment; client feedback conditions demonstrated a
From a social– cognitive theoretical perspective, feedback does mean medium effect size of .39 when compared with no-feedback
seem to matter. For example, Bandura (1997) theorized that eval- conditions.
uative judgments from others can have a profound effect on an Concerned with the feasibility of asking that clients complete 45
individual’s self-efficacy beliefs. Indeed, supervisory feedback has items before every session, Miller and Duncan (2000) developed a
been found to influence counseling self-efficacy for trainees briefer outcome measure, which they called the Outcome Rating
(Daniels & Larson, 2001). Bandura also argued that compared Scale (ORS). Consistent with the recommendations from the Di-
with the task-specific judgments of others, general feedback rarely vision 29 task force to monitor the therapeutic alliance, the four-
has any lasting effect on self-efficacy and that the potency of social item Session Rating Scale (SRS; Miller, Duncan, & Johnson,
persuasions depends on the interpretation a trainee makes of a host 2000) was also developed for use after every session to monitor the
of factors. “Persuasory efficacy appraisals have to be weighted in therapeutic relationship. The use of these two measures in con-
terms of who the persuaders are, their credibility, and how knowl- junction is referred to as the Partners for Change Outcome Man-
edgeable they are about the nature of the activities” (Bandura, agement System (PCOMS; Miller & Duncan, 2004). Although
1997, p. 104). many factors influence clients’ compliance in completing outcome
Using client feedback as the focal mechanism in supervision has measures, initial research has indicated that using shorter measures
been suggested as a means by which the supervisory process can may be especially influential. For example, Miller, Duncan,
remain more objective (Worthen & Lambert, 2007). Bandura Brown, Sparks, and Claud (2003) reported that compliance rates in
(1997) would concur that “deferral to supposedly objective indi- a training clinic for the ORS and SRS measures were 86%, as
cants of capability can boost persuaders’ influence on people’s compared with 25% for the OQ45.
beliefs in their efficacy” (p. 104). Instead of imposing their own Although based on Lambert et al.’s (1996) system of continuous
subjective standards, supervisors and trainees can refer to client assessment, PCOMS has two other major differences than just
feedback as a starting point for a discussion of the training coun- brevity. First, PCOMS assesses both client outcome and the ther-
selor’s efficacy. For this reason, using client feedback in the apeutic relationship; the OQ45 only assesses client outcome. Sec-
supervisory process could prove to be a potent source of counselor ond, PCOMS is prescribed to benefit all clients, not just clients
self-efficacy, particularly for trainees who lack an experiential who are not progressing as expected.
basis for judging their capabilities. Research findings have shown PCOMS to be related to a host of
Bandura (1997) has long contended that self-efficacy beliefs are positive therapeutic outcomes. Similar to the OQ45 research, stud-
strong predictors of subsequent behavior. We were only able to ies using the ORS and SRS have reported fewer premature termi-
identify one study that investigated the relationship between nations and increased effectiveness. For example, in a study of
trainee counseling self-efficacy and outcome with actual clients, 6,424 clients who received services through an employee assis-
however, the results of which were far from definitive (Heppner, tance program, researchers found that effect sizes increased from
Multon, Gysbers, Ellis, & Zook, 1998). In general, counseling .37 to .79 when PCOMS was implemented (Miller, Duncan,
self-efficacy did not correlate with most of the outcome variables. Brown, Sorrell, & Chalk, 2006). Investigating how such a measure
However, confidence to build a relationship with a client was might affect the supervision process could potentially provide an
related to clients being more motivated to address their career excellent resource to those interested in the effective training of
goals. Heppner et al. (1998) noted that the relationship between psychotherapists.
self-efficacy and client outcome was not linear, that simply pos- With these findings in mind, we undertook the study with three
sessing more counseling self-efficacy did not yield better out- major hypotheses. First, we expected that supervisees in a feed-
comes. Given the few findings in this area, we followed Bandura’s back condition (using a continuous assessment feedback measure
theorizing and expected that counselors’ self-efficacy beliefs with clients and in psychotherapy supervision) would demonstrate
would be predictive of their clients’ judgments of whether their greater client outcomes (as measured by the ORS) than would
therapy was indeed effective. supervisees in the no-feedback condition over the course of an

160 REESE ET AL.

academic year. Second, we expected that supervisees in a feedback A clinical cut score of 25 was determined (77th percentile) on
condition would rate the supervisory alliance and their satisfaction the basis of a sample of 34,790 participants, indicating that those
with supervision more favorably than would their counterparts in scoring higher than 25 are more consistent with a nonclinical
a no-feedback condition. We also conjectured that the supervisory population and less likely to benefit from psychotherapy (Miller &
alliance and satisfaction with supervision would be related to client Duncan, 2004). Miller et al. (2003) also found that the ORS
outcome. Third, we hypothesized that supervisees in the feedback discriminates well among clients and nonclients. Their research
condition would report higher counseling self-efficacy than would indicated that the ORS generates reliable scores among individuals
those in the no-feedback condition, but in keeping with the tenets who receive therapy in a community mental health center (
of social– cognitive theory, we expected the relationship between .93). Test–retest reliability from the first to second therapy session
counselor self-efficacy and client outcome to be similar for super- was .60. The correlation coefficient of .59 between the ORS and
visees in both conditions. OQ45 also provided evidence of construct validity. Later evidence
also demonstrated that clients made gains across therapy (Miller et
al., 2006). Lambert et al. (1996) stated that evidence for con-
Method
struct validity on outcome measures can be established by
Participants showing that scores later in treatment differ from those obtained
at the beginning of treatment. In this study, we obtained coeffi-
Twenty-eight trainees (10 men and 18 women; 79% White, cient alphas of .82 and .90 for the first two sessions of the current
10.5% African American, 10.5% Asian American; mean age study, respectively.
25.14, SD 3.49) participated in the study during their 2nd year SRS (Miller et al., 2000). The SRS consists of four items that
of training in a master’s-level marriage and family (n 19) or are also measured via a visual analog scale. Based on Bordin’s
clinical– counseling psychology (n 9) program at a medium- (1979) pantheoretical definition of the therapeutic alliance and an
sized, private southwestern university. Participants’ theoretical inclusion of the client’s theory of change, the scale assesses the
orientations included cognitive– behavioral, solution-focused, and therapeutic relationship (“I felt heard, understood, and respected”),
variations on a general systems perspective. goals and topics covered in therapy (“We worked on or talked
Supervisors (N 9) who participated in the study were either about what I wanted to work on or talk about”), the approach used
full-time faculty or adjunct faculty within each program (marriage in therapy (“The therapist’s approach is a good fit for me”), and the
and family 6, clinical– counseling psychology 2, both 1; 5 overall rating of the session (“Overall, today’s session was right
men and 4 women; 100% White; doctorate in counseling– clinical for me”). Clients make a hash mark on each of the four analog scales
psychology 3, doctorate in marriage and family therapy 3; as described earlier. Marks on the left of the scale indicate less
master’s in marriage and family therapy 3). Client participants satisfaction for each item, and marks on the right indicate higher
(N 110) attended individual therapy at either the graduate satisfaction. Once again, a ruler is used to measure the distance
program’s marriage and family training clinic that serves the from the left end of the scale to the hash mark. The individual
community (n 95; 25 men and 68 women, and 2 who did not items are then recorded and totaled, ranging from 0 to 40. Prob-
indicate gender; 79.5% White, 15.7% Hispanic–Latino, 3.6% Af- lem(s) with the therapeutic alliance are indicated by either a
rican American, 1.2% Asian American; age range 15– 69 years, clinical cut-score of 36 or a rating below 9 on any one item.
M 34.05, SD 11.70) or the university’s counseling center (n Miller and Duncan (2004) found that increases on the SRS
15; 5 men and 10 women; 86.6% White, 6.7% Asian American, during the course of treatment were statistically significantly as-
6.7% international; age range 18 –24, M 19.54, SD 2.15). sociated with better outcome. When compared with clients who
Clients presented with problems similar to those that would be did not use the SRS, clients who used the SRS were three more
seen at a typical community mental health center or university times likely to attend a next session and experienced more change
counseling center, including mood and anxiety disorders, relation- during treatment. The cut-score of 36 was derived from a sample
ship or marital problems, grief, and a variety of other Axis I and of 15,000 clients, of whom only 24% scored below 36 and were “at
Axis II disorders. a statistically greater risk for dropping out of or experiencing a
negative or null outcome from treatment” (p. 14).
Initial research has indicated that the SRS generates reliable and
Measures
valid scores. In their sample of 337 community mental agency
ORS (Miller & Duncan, 2000). The ORS is a four-item mea- clients, Duncan et al. (2003) found that the SRS had a coefficient
sure designed to assess client progress in treatment. Three items alpha of .88 and possessed a correlation coefficient of .48 with the
are taken from the three subscales of the OQ45 and address social Helping Alliance Questionnaire–II (Luborsky et al., 1996). Test–
(work, school, and friendships), interpersonal (family and close retest reliabilities averaged .74 across the first six sessions with the
relationships), and individual (personal well-being) domains. The SRS as compared with .69 for the Helping Alliance Questionnaire–
fourth item asks the client to evaluate his or her general sense of II. For this study, coefficient alphas for the first two sessions were
well-being. Each item is rated using a visual analog scale. Clients .97 and .92, respectively.
make a hash mark on each of the four analog scales that are 10 cm Supervision Outcomes Survey (SOS; Worthen & Isakson, 2003).
in length, with scores on the left side of the scale indicating lower The SOS is a 20-item measure designed to assess the supervisee’s
functioning and scores on the right indicating higher functioning. satisfaction with the supervisory process. Items are measured using
A ruler is then used to measure the distance from the left end of the a 7-point Likert-type scale ranging from 1 (not at all) to 7 (greatest
scale to the client’s hash mark. The measures for the four items are degree possible). Sample items include “My supervisor helps me
then summed to provide a total score, ranging from 0 to 40. develop by providing both challenge and support,” “Supervision


helps me see my mistakes as learning experiences,” and “Overall was not possible for the 9 trainees in the clinical– counseling
I feel satisfied with my supervision.” Scores for the items are psychology program; however, the 2 supervisors were randomly
totaled for a composite score ranging from 20 to 140. Worthen and assigned. Two trainees from this program were eligible to be in the
Isakson (2003) found evidence for construct validity in that SOS feedback condition because they were placed at a site that permit-
scores increased across the supervisory process in a manner similar ted the use of outcome measures. The other 7 trainees were at
to the supervisee and trainee versions of the Supervisory Working practicum sites where the use of outcome measures was not
Alliance Inventory (trainee version: SWAI–T; Efstation, Patton, & appropriate or permitted (e.g., acute care facility for adults or
Kardash, 1990). For the current sample, internal consistency was adolescents). Therefore, they were placed in the no-feedback
evaluated for the first administration (middle of the first semester) group.
and last administration (end of the second semester); coefficient Trainees in the marriage and family program conducted all of
alphas were .92 and .98, respectively. their practicum training in the marriage and family clinic, whereas
SWAI–T. The SWAI–T consists of 19 items designed to mea- trainees in the clinical– counseling program completed practicum
sure supervisees’ perceptions of the supervisory relationship. The training in a variety of clinical settings (e.g., university counseling
measure consists of two subscales, Rapport (13 items; e.g., “I feel center, career and academic development center, acute care facility
comfortable with my supervisor”) and Client Focus (6 items; e.g., for adults and adolescents, and neuropsychology clinic). Because
“My supervisor helps me work within a specific treatment plan of the inability to randomly assign the clinical– counseling gradu-
with my clients”), that are rated on a 7-point Likert-type scale ate students, their data were not included in the first research
ranging from 1 (almost never) to 7 (almost always). Efstation et al. question that addresses comparing outcomes in the feedback and
(1990) reported evidence for validity via a confirmatory factor no-feedback conditions. However, their outcome data were in-
analysis that revealed a two-factor structure. The Rapport subscale cluded for the second and third research questions because those
accounted for 30% of the variance, whereas the Client Focus scale questions are focused on how the use of the PCOMS data influ-
accounted for 8%. Both factors were found to be statistically enced the supervisory process and counseling self-efficacy.
significantly correlated with the Supervisory Styles Inventory and In this study, the trainees provided all of the counseling sessions
had low correlations with a supervisor version of the same mea- individually. Clients were randomly assigned to trainees in each
sure. The authors also reported internal consistency estimates for condition. Trainees in the feedback condition used PCOMS with
the items in each subscale of .90 and .77, respectively. Despite their clients and the resulting data in supervision. That is, each
these findings, other researchers have opted to combine the items trainee in the feedback condition administered, scored, graphed,
from the two subscales because of high interfactor correlations and then discussed the ORS scores at the beginning of each session
(Patton & Kivlighan, 1997). In keeping with this approach, we with the client and then repeated the process with the SRS at the
elected to use a composite scale made up of items from the two end of each session. The feedback was discussed directly with the
subscales. Coefficient alphas for the current sample were .87 client. Consistent with the PCOMS protocol, trainees in the feed-
(middle of first semester) and .97 (end of second semester).
back condition used the following four categories to help track
Counseling Self-Estimate Inventory (COSE; Larson et al.,
client progress and frame how to proceed with clients (see Miller
1992). The COSE is a 37-item, 6-point Likert-type scale ranging
& Duncan, 2004, for a complete description):
from 1 (strongly disagree) to 6 (strongly agree). The COSE is
designed to measure a counselor’s self-efficacy across five coun- No change: For a client who has not shown reliable change (a
seling areas: microskills, counseling process, dealing with difficult gain of 5 points) after three sessions, therapists are directed to
client behaviors, cultural competence, and (awareness of one’s address the therapeutic alliance and the course of treatment. If
own) values. In their initial validation study, Larson et al. (1992) the client has not demonstrated reliable improvement after six
reported that five factors accounted for 36% of the variance with sessions, the manual suggests consultation, supervision, or
eigenvalues all above 1.0. Additional support for construct validity staffing.
was found via convergent validity (moderate correlation with a
measure of self-concept) and discriminant validity (weak correla- Deteriorating: Clients in this category are considered to be at
tions with measures of personality and social desirability). Internal risk for terminating prematurely or having a poor outcome.
consistency estimates ranged from .88 (microskills) to .62 (values) Therapists are directed to discuss possible reasons for the
and demonstrated an overall alpha of .93 (Larson et al., 1992). For drop in score, review the SRS items with the client to assess
the current study, scores at the initial administration generated a the therapeutic alliance, or consider changing the treatment
coefficient alpha of .91 for the total score, with the subscale alphas approach, frequency, mode, or even therapist if no improve-
ranging from .86 (microskills) to .62 (values). ment is noted after three sessions.

Procedure Reliable change: Treatment is progressing as expected. Ther-
apists are advised to reinforce changes and to continue treat-
The study took place over the course of 1 academic year ment until progress begins to plateau, whereupon a therapist
consisting of two 16-week semesters. Assignment to the feedback should consider reducing the frequency of sessions.
or no-feedback condition for supervisors and trainees was different
for each program. For the marriage and family program, supervi- Clinically significant change: The client is likely no longer
sors and trainees were randomly assigned to the client feedback struggling with issues that led to therapy. Therapists are
condition (n 4 supervisors and 9 trainees) or to the no-feedback advised to consolidate changes, anticipate potential setbacks,
condition (n 3 supervisors and 10 trainees). Random assignment and consider reducing the frequency of sessions.

162 REESE ET AL.

Trainees in the no-feedback condition at the marriage and Results
family clinic administered the ORS at the beginning of each
session (to track outcome) but did not observe, score, or discuss Did Outcomes Improve Over the Course of an
the results with clients. The measures were placed in a central- Academic Year?
ized collection area and were scored by a coinvestigator. To Pre–post means for the feedback and no-feedback conditions for
reduce the potential for demand characteristics (i.e., concern trainees in the marriage and family program help provide a context
that people would not answer honestly for fear of hurting the for the use of PCOMS with trainees. Trainees in the feedback
trainee’s feelings or grade), trainees in both groups were in- group demonstrated better outcomes with clients than did trainees
structed to encourage clients to complete the measures honestly in the no-feedback group. Clients in the feedback group had a
and to tell clients that their scores would not be tied to the mean gain of 8.90 points from pretest to posttest (range 19.25 to
trainee’s grade or evaluation. Trainees in the no-feedback con- 28.15), whereas clients in the control group had a mean gain of
dition in the clinical– counseling program did not have clients 4.01 points (range 18.56 to 22.56). These findings are consistent
complete the measures and provided services in a business-as- with research in which professional staff have used PCOMS (see
usual format. Miller, Duncan, Sorrell, & Brown, 2005). Pretest score differences
All supervisees received weekly individual and group super- for both groups were not statistically significant. A repeated mea-
vision (trainees were not separated by treatment condition for sures factorial analysis of variance, using supervisor as a covariate,
group supervision). Individual supervision typically consisted found that trainees in both groups demonstrated statistically sig-
of providing updates on a supervisee’s caseload and viewing nificant improvement overall with clients, F(1, 92) 47.76, p
selected video and audio recordings of a session for discussion. .000, 2 .34. Clients in the feedback condition showed better
More focus was placed on the supervisee’s goals and strengths outcomes than did those in the no-feedback condition (Time
or weaknesses in the individual supervision format. Group Condition), F(1, 92) 6.88, p .05, 2 .07. The eta-square of
supervision typically relied on a case presentation format with .07 is considered to be a small to medium effect size (Cohen,
1988). The supervisor covariate was not statistically significant,
supervisees designated to present on clients with a selected
F(1, 92) 0.11, p .05, 2 .00.
video or audio recording segment to facilitate the presentation.
To look at individual trainee effectiveness, we computed effect
Time was also provided to discuss pressing trainee concerns
sizes for each client who completed treatment. Cohen’s d was com-
regarding clients and other trainee needs.
puted by using the highest original standard deviation for each group
The client feedback group used PCOMS in both individual
mean rather than using a pooled standard deviation to avoid inflating
and group supervision to facilitate the aforementioned supervi- the effect size common with measures (Dunlap, Cortina, Vaslow, &
sion formats. For the feedback group in individual supervision, Burke, 1996). A mean treatment outcome effect size for each trainee
supervisors and supervisees reviewed each client’s scores to- was computed by aggregating the outcome effect size for each client
gether using the PCOMS categories to denote whether a client in the trainee’s caseload. Trainees in the feedback condition had a
was progressing as expected and, if not, identified strategies for mean effect size of 0.92 (range 0.43 to 1.72) versus a mean of 0.23
how to proceed. In general, the data were used to help direct (range 0.22 to 1.02) for trainees in the no-feedback condition.
which clients needed more attention in supervision. Supervisors However, we found no statistically significant differences in the mean
used the client feedback data to organize the supervisory ses- number of sessions attended by clients in each condition (feedback
sions in a triage fashion, prioritizing those clients demonstrat- 5.40, no-feedback 4.19), t(113) 1.42, p .05.
ing less progress (i.e., deteriorating or no change) or having To assess whether supervisees progressed during their training,
lower scores. Clients identified as progressing as expected (i.e., we compared their effectiveness with clients across academic
reliable change or clinically significant change) or having semesters. Specifically, we examined whether trainees in the feed-
higher scores were discussed later within the supervisory ses- back condition demonstrated more improvement (as evidenced by
sion to identify or reinforce the trainee’s areas of strength. In client outcomes) during the academic year as compared with
group supervision, the PCOMS data were used similarly, but trainees in the no-feedback condition. Trainees in both the feed-
trainees in the feedback condition (rather than both supervisor back and the no-feedback conditions evidenced improvement from
and trainee) selected which client(s) needed the most attention. the first to the second semester as shown by aggregated effect sizes
Both ORS and SRS scores were used to provide more informa- (see Figure 1). The feedback group demonstrated an effect size
tion about the client to be discussed. Trainees in the no-
feedback condition received individual and group supervision 0.97
1.00 0.92
as usually provided. The time spent in individual and group
supervision was equal for both conditions. 0.80 0.70
Effect Size

Trainees in both conditions completed the counselor self- 0.60
Feedback
efficacy measure (i.e., COSE) at the beginning and end of the 0.40 0.30
0.37
0.23 No-Feedback
fall semester and again at the end of the spring semester. The
0.20
SOS and the SWAI–T were completed by trainees in both
conditions at the middle and end of the fall semester and again 0.00
Fall Spring Overall
at the end of the spring semester to evaluate their perceptions of
their supervisory experience. Supervisors were not permitted to Figure 1. Aggregate effect sizes on the Outcome Rating Scale (ORS) by
review the results of these measures. academic semester and overall for feedback and no-feedback conditions.


gain in effectiveness from 0.70 (n 35 clients) in the first Table 2
semester to 0.97 (n 26 clients) in the second semester, whereas Correlations of Trainee Supervisory Alliance, Therapeutic
the no-feedback group demonstrated smaller effect size gains from Alliance, Supervision Satisfaction, and Counselor Self-Efficacy
0.30 (n 31 clients) in the first semester to 0.37 (n 23 clients) With Trainee Outcome
in the second. Not only did the trainees in the feedback group have
better outcomes from the beginning, they also evidenced more Scale 1 2 3 4 5 6 7 8
improvement in the second semester. 1. Trainee outcome
(overall) —
Did PCOMS Influence the Supervisory Relationship or 2. SRS (Session 2) .23 —
3. SWAI–T (end of fall) .06 .37 —
Satisfaction With Supervision? 4. SWAI–T (end of
spring) .15 .89 .04 —
Scores on the SWAI–T and SOS indicated that trainees’ per- 5. SOS (end of fall) .05 .01 .83 .12 —
ceptions of the supervisory relationship and satisfaction with su- 6. SOS (end of spring) .24 .72 .15 .92 .01 —
pervision were similar for the no-feedback group across the aca- 7. COSE (beginning of
demic year but more variable for the feedback group (see Table 1). year) .02 .04 .01 .02 .14 .22 —
However, a repeated measures multivariate analysis of variance 8. COSE (end of year) .19 .01 .02 .24 .35 .16 .04 —
revealed no differences between the groups, Wilks’s .03, F(2, Note. SRS Session Rating Scale; SWAI–T Supervisory Working
12) 0.15, p .05, 2 .03. The interaction (Time Condition) Alliance Inventory—Trainee Version; SOS Supervision Outcomes Sur-
also indicated no statistically significant differences. It appears that vey; COSE Counseling Self-Estimate Inventory.
contrary to our hypothesis, using the continuous assessment sys- p .05. p .01.
tem in supervision did not influence the supervisory alliance or
overall satisfaction with supervision.
To investigate whether the supervisory alliance or satisfaction Did PCOMS Influence Counselor Self-Efficacy?
with supervision was related to trainee effectiveness as measured As can be seen in Table 3, the COSE scores for trainees in both
by client outcome, we examined correlations between the conditions increased across the academic year. A repeated mea-
SWAI–T, the SOS, and the mean treatment outcome effect size for sures analysis of variance found the differences across time to be
each trainee. Correlations between the SWAI–T and the SOS were statistically significant within each group, F(2, 12) 4.50, p
.82 (middle of fall), .83 (end of fall), and .92 (end of spring). These .05, 2 .56. However, the interaction effect (Time Condition)
strong correlations are consistent with previous research investi- showed that the degree of change between the groups was not
gating the supervisory alliance and satisfaction with supervision different, F(2, 12) 0.13, p .05, 2 .04. Contrary to our
(Ladany et al., 1999). Correlations between these variables and expectations, trainees who used PCOMS data in supervision did
trainee effect sizes were much lower, ranging from .05 to .24 (see not report larger increases in counseling self-efficacy when com-
Table 2). A correlation of .23 was observed between outcome and pared with trainees who did not use PCOMS.
the therapeutic alliance as measured by the SRS at Session 3. It is A secondary hypothesis was to investigate the relationship be-
also worth noting that the correlation between the SWAI–T and tween counselor self-efficacy and client outcome. Although the
SRS (r .89) at the end of the academic year was statistically correlations between the COSE and outcome increased over time,
significant. These results indicate that as expected, the quality of all of the correlations were low (see Table 2), with coefficients
the supervisory alliance was related to the quality of the therapeu- ranging from .02 (beginning of the year) to .19 (end of the year).
tic alliance. With few exceptions, COSE scores did not correlate highly with
measures of satisfaction with supervision or the supervisory
alliance.
Table 1 We also conducted the same correlational analyses for each
Supervisory Working Alliance Inventory: Trainee and condition (see Table 4). The correlations between counseling self-
Supervision Outcomes Survey Means and Standard Deviations efficacy and outcome were surprisingly different for each condi-
for the Client Feedback and No-Feedback Conditions Across the tion; at the end of training, the self-efficacy scores reported by
Academic Year trainees in the feedback condition were strongly related to their
aggregate therapy outcome (i.e., effectiveness) effect sizes (r
Middle of fall End of fall End of spring

Condition M SD M SD M SD Table 3
SWAI–T
Counseling Self-Estimate Inventory (COSE) Means and
Standard Deviations for the Client Feedback and No-Feedback
Client feedback 103.67 8.89 95.57 14.54 100.00 27.96
No feedback 99.09 15.69 99.10 19.08 106.45 24.02 Conditions Across the Academic Year

SOS Start of fall End of fall End of spring
Client feedback 111.00 19.77 98.28 20.66 118.40 21.15
Condition M SD M SD M SD
No feedback 112.91 19.42 109.50 23.27 115.36 26.13
Client feedback 140.90 17.32 150.50 14.83 167.78 13.71
Note. SWAI–T Supervisory Working Alliance Inventory—Trainee No feedback 143.23 15.82 151.81 12.65 165.71 16.65
Version; SOS Supevision Outcomes Survey.

164 REESE ET AL.

Table 4 the first semester and demonstrated even more improvement in the
Correlations of Counseling Self-Estimate Inventory (COSE) With second semester than did those in the no-feedback group.
Trainee Outcomes by Condition Comparing the client psychotherapy outcomes of trainees is
fraught with threats to internal validity given that each supervis-
Trainee outcome ee’s caseload consists of clients from diverse backgrounds who
COSE Feedback No feedback
differ in any number of ways, including presenting problem and
level of problem severity. Therefore, conclusions from this study
Beginning of year .04 .06 about how to improve supervision training and therapy outcomes
End of year .51 .38 should be interpreted with caution. Ronnestad and Ladany (2006)
noted that “there are significant obstacles to gathering data in a
controlled fashion on the relationship between supervision process
and outcome and psychotherapy process and outcome” (p. 263).
.51). For trainees in the no-feedback condition, however, self- However, having a better gauge of how trainees perform with their
efficacy scores had a correlation of .38 with aggregate therapy clients is one way to track trainees’ progress and serves as a means
outcome. Interestingly, the self-efficacy of trainees in both condi- of ensuring that trainees are capable of demonstrating a minimum
tions measured at the beginning of the study demonstrated near- level of competence. Direct outcome comparisons could be mis-
zero correlations with therapy outcome at the end of the study. leading but may also be meaningful if made in an ipsative fashion.

Discussion Supervisory Alliance and Satisfaction With Supervision
Few psychotherapy supervision studies have investigated Although trainees who used PCOMS in supervision demon-
whether and how client outcome might inform psychotherapy strated better client outcomes, they did not report a better super-
training and supervision. Using outcome data that are collected visory alliance or more satisfaction with supervision than those
from clients every session has been proffered as a way to help who did not use the feedback system. The results were surprising
redress this limitation. Such continuous assessments offer other given the expectation that the use of outcome data in supervision
potential advantages as well. Monitoring client feedback could would lead to more specific feedback, which is considered a
help provide rich information for use in supervision by enabling marker of effective supervision (Lehrman-Waterman & Ladany,
supervisors to monitor supervisee progress more closely (Lambert 2001). A concern expressed by both supervisors and supervisees
& Hawkins, 2001). The purpose of this study was to investigate the before the study was that the likelihood of negative feedback
promise of using continuous outcome data in supervision and the would increase and hurt the supervisory relationship. Perhaps this
manner in which these data influence the supervisory process. anticipation led to heightened sensitivity on the trainee’s part when
Results indicated that although most trainees experience positive discussing a client who was not progressing. Another possibility is
outcomes with their clients and supervisors, trainees who used a that the supervisor focused more on the data and information from
continuous assessment system (i.e., PCOMS) with clients and the measures and less on the supervisee, which may have affected
shared the feedback data in supervision demonstrated better out- the supervisee’s perception of the supervisory relationship.
comes with their caseloads than did trainees in a control group who A more positive way of looking at the data is that the use of
did not use continuous assessment. The treatment gain scores for continuous feedback in the supervisory process did not negatively
trainees in the continuous feedback condition (8.90) versus those influence trainees’ satisfaction with supervision or their percep-
in the no-feedback condition (4.01) and the number of clients who tions of the supervisory alliance. When asked about the supervi-
demonstrated clinically significant improvement in each group sory process after the study, supervisors explained that using the
(28% vs. 15%) indicates that trainees who received continuous continuous assessment data in supervision made providing critical
feedback from their clients were approximately twice as effective feedback easier and served as a starting point for providing more
as those who did not receive feedback over the course of therapy. specific feedback. Hoffman et al. (2005) found that most supervi-
This is consistent with previous research (Miller et al., 2005) and sors withhold feedback regarding performance. Our results may
indicates that the usefulness of the ORS and SRS extends to help quell concerns that giving such feedback will harm the
master’s-level psychotherapy trainees. supervisory relationship.

Trainee Improvement Importance of the Supervisory Alliance
Few supervision studies have directly examined the relationship The supervisory alliance has been conceptualized to function
between client outcome and the quality of training. Successful similarly to the therapeutic alliance in that a good working rela-
outcome in supervision has typically been operationalized from the tionship is foundational to effective training. Although the litera-
supervisee’s perspective. However, the overarching goal of super- ture is clear that successful supervision is predicated on a good
vision is to improve one’s counseling skills, thereby increasing supervisory alliance (Nelson & Friedlander, 2001), the literature is
one’s ability to help clients. The current study demonstrated that less clear on how successful supervision is defined, particularly as
irrespective of group, trainees were generally helpful to their regards the perspectives of the other stakeholders in the therapy
clients (average Cohen’s d 0.56). Client outcomes improved process. The supervision literature is also decidedly mixed on how
throughout treatment regardless of the trainee’s group assignment. the supervisory relationship influences training outcomes (Ladany
However, trainees in the feedback group had better outcomes in et al., 1999). The results of this study provide evidence that the


supervisory alliance is related to client outcome. The correlation Comparing the correlation between client outcome and counsel-
between the supervisory alliance and client outcome is similar to ing self-efficacy for the feedback group to the low correlation
correlations typically found between the therapeutic alliance and obtained for the no-feedback group reveals that perhaps the effi-
client outcome (Baldwin, Wampold, & Imel, 2007; Martin, Gar- cacy beliefs of counselors in the feedback group was better cali-
ske, & Davis, 2000). However, correlations between the supervi- brated with the clients. Those in the no-feedback condition, how-
sory relationship and counselor self-efficacy were all quite low. ever, had limited information on which to base their efficacy
Past evidence for this relationship is unclear; some researchers judgments, namely their perceptions of the client’s progress in
have found such a relationship (Efstation et al., 1990), and others therapy and the comments they received during supervision. With
have not (Ladany et al., 1999). Although this study provides this in mind, it does not seem surprising that trainees who received
evidence that the supervisory relationship is important, the feed- regular feedback from clients on outcome and the therapeutic
back condition was small and results cannot be generalized at this relationship had levels of self-efficacy that were more closely
time. In general, this study’s data indicated that trainees were related to their actual performance with clients. These results
generally happy with their relationships with supervisors and the nevertheless point to the complexity of counseling self-efficacy
quality of supervision received as indicated by the SWAI–T mean and suggest that increasing self-efficacy during training is not a
scores. More research is clearly needed to understand the role of panacea that automatically results in better client outcomes. This
the supervisory alliance in promoting psychotherapy training. contention is supported by previous findings that outcome with
career counseling clients may not have a linear relationship with
counseling self-efficacy (Heppner et al., 1998).
Counselor Self-Efficacy and Trainee Development

Two questions guided our investigation of trainees’ counseling Limitations of Study
self-efficacy beliefs. First, we sought to examine whether the There are limitations in this study that warrant mention. First,
condition to which trainees were assigned, with or without client the sample of trainees and supervisors was small and consisted of
feedback, would differentially affect trainees’ efficacy beliefs. Our only two master’s programs. Tracking trainees’ outcomes with
results indicate that trainees in both conditions experienced statis- their clients for an entire academic year is a strength of the study,
tically significant gains in counseling self-efficacy across the train- but the generalizability of the results, particularly for doctoral
ing year. Several explanations could account for this. First, con- training programs, is limited. The small sample size also limited
sistent with Bandura’s (1997) theorizing, the practical experience the ability to conduct more powerful analyses. Hierarchical linear
all trainees obtained could have served to boost their self-efficacy. modeling would have been the preferred analysis to assess growth
Indeed, mastery experience has been shown to be the most pow- curves across the academic year for both clients and trainees and
erful source of self-efficacy, and in this study such experiences would have helped better capture individual growth, but the sam-
may have overshadowed the social persuasions that trainees in ple size was not adequate. A second limitation is the issue of
each group received. treatment integrity. This study involved three levels of partici-
A second explanation could be that although trainees in both pants: clients, trainees, and supervisors, which made it difficult for
groups reported similar self-efficacy levels at the conclusion of the us to ensure that participants complied with the study’s protocol.
study, the self-efficacy beliefs reported by trainees in the no- Trainees were most enthusiastic about the project and expressed
feedback group were inflated. This seems plausible given the excitement that they would have an additional tool to use in
lower outcome scores for these trainees’ clients. This is also therapy. Their supervisors were generally less enthusiastic, how-
consistent with past research. Researchers who asked therapists to ever, about an additional responsibility in supervision. Protocol
rate their level of effectiveness found that therapists rated them- adherence was evidenced by the positive anecdotal feedback from
selves equally, regardless of their actual level of performance supervisors and supervisees and that the majority of clients com-
(Hiatt & Hargrave, 1995). Perhaps this sort of “Lake Wobegon pleted measures consistently throughout the academic year. Par-
effect” was at work for those trainees who did not receive explicit ticipant compliance was encouraged via e-mail reminders sent to
feedback from their clients and who, given this missing feedback, the supervisors during the midpoint of each semester.
assumed themselves to be equally capable of treating their clients Another limitation is that trainees from the marriage and family
despite their actual differences. program were all placed in one setting and the clinical– counseling
Our second aim for investigating counselor self-efficacy beliefs trainees were placed in different training sites. Although client
was to test Bandura’s (1997) hypothesis that self-efficacy beliefs outcomes with the clinical– counseling trainees were not used in
are positively related to actual performance. Specifically, we comparing differences in outcome between treatment conditions,
sought to examine whether trainees who reported higher counselor their placement in a variety of settings may have led to different
self-efficacy also experienced better client outcomes throughout levels of success with clients and, consequently, different experi-
the year. Research in a number of disciplines has found that those ences and perceptions of their training and supervision. However,
who believe themselves to be capable tend to put forth more effort, trainees from all the programs rated their supervisory experiences
persevere in the face of difficulty, and set more challenging goals and counseling self-efficacy similarly. A fourth limitation was that
for themselves (Bandura, 1997). We found that self-efficacy was we used a continuous assessment system that has less empirical
related to outcome (r .52), but only for trainees in the client support than the OQ45 (Lambert et al., 1996), although initial
feedback condition. Providing consistent feedback on performance research on PCOMS has been encouraging. The reason for using
(e.g., Locke & Latham, 2002) has been found to lead to improved PCOMS was twofold: The study sites were resistant to using
performance. longer measures, and PCOMS provides additional information

166 REESE ET AL.

about the therapeutic alliance that we believe was valuable feed- Using client feedback data in supervision did not enhance the
back for trainees to receive. Last, it was not possible to know how supervisory relationship, satisfaction with supervision, or coun-
to connect the effects of supervision directly to client outcome. selor self-efficacy reported by trainees when compared with re-
Using a continuous outcome model appears to have been helpful ports from trainees in the no-feedback condition. Interestingly, the
for trainees to use with clients, but we are uncertain of its utility correlation found between the supervisory working alliance and
within the supervisory process. It is plausible that outcomes would client outcome is similar to correlations found for the therapeutic
have been equally as impressive if the client feedback data were alliance and outcome. Additionally, we found moderate correla-
not used in the supervisory sessions. tions between counseling self-efficacy and outcome for trainees
who received client feedback data, whereas we found negative
correlations between counseling self-efficacy and outcome for
Future Study those in the no-feedback condition.
For practicum sites that have limited ability to video- or audio-
We considered this study to be an exploratory investigation. We
tape consistently or for supervisors who have a large number of
could not identify other studies that have assessed the utility of
supervisees that makes keeping track of client caseloads cumber-
continuous client feedback in counselor training and supervision.
some, the use of a continuous assessment model for trainees may
Nevertheless, our findings point to several fruitful research possi-
be especially appealing. In both situations, a supervisor is often left
bilities to be pursued. First, a similar design with the addition of a
to rely on a trainee’s perceptions. We know such perceptions are
third condition in which trainees use a continuous assessment
not always accurate and that therapists tend to overestimate client
model with clients but not in their supervision would help assess
progress (Grove et al., 2000). Trainees’ ambivalence about want-
whether using outcome data in supervision leads to improved
ing to share mistakes in the evaluative context of supervision also
client outcomes. Second, studies in this area may also include
makes sole reliance on trainees’ self-reported performance prob-
using other continuous assessment measures such as the OQ45 and
lematic. Using outcome data consistently in supervision may help
evaluating the utility of the different components of client feed-
therapists have a more accurate perception of their abilities and
back provided. For example, one of the reasons PCOMS was
may also serve as a means of normalizing the process of sharing
selected was to include a measure of the therapeutic relationship.
information about clients, particularly when things are not going
How is the inclusion of this measure useful to trainees? Third, one
well. Trainees need to know that even the best therapists are not
of the proposed advantages of using continuous assessment is that
successful with all of their clients. Conversely, seeing concrete
it lends itself to more specific feedback in supervision. Using a
evidence that clients are benefiting and improving from therapy
measure that directly assesses this process may shed light on how
can be motivating and may bolster one’s sense of self-efficacy.
using continuous assessment data contributes to the supervision
Specific feedback is considered a key ingredient in effective
process. Also, future research should continue to investigate how
supervision, and the use of client outcome in supervision holds the
counselor self-efficacy and other person-level variables may me-
promise of facilitating this process. Despite this optimism, caution
diate both supervisory and client outcomes. Although conducting
should be taken in how client data are used as a supervisory tool.
supervision research that focuses on client outcome is difficult and
Care should be taken to help trainees recognize that monitoring
challenging, such research is important and necessary given the
client outcome is not to reduce their counseling skills to an effect
centrality of supervision to counselor training. The question of
whether supervisory training is beneficial is in part still unan- size but rather to help them gain a clearer understanding of how
swered because of limited research in this area (Stein & Lambert, they work with clients and to help them become more effective as
1995). therapists.

Conclusions References
Ackerman, S. J., Benjamin, L. S., Beutler, L. E., Gelso, C. J., Goldfried,
The main purpose of psychotherapy supervision is to help
M. R., Hill, C., et al. (2001). Empirically supported therapy relation-
trainees become more effective clinicians. It makes sense that
ships: Conclusions and recommendations of the Division 29 Task Force.
tracking client outcome would be helpful in this process. The Psychotherapy: Theory, Research, Practice, Training, 38, 495– 497.
results of this study demonstrated that with a sample of master’s Baldwin, S. A., Wampold, B. E., & Imel, Z. E. (2007). Untangling the
program trainees, using continuous assessment measures, the ORS alliance outcome correlation: Exploring the relative importance of ther-
and SRS, resulted in better client outcomes. Trainees in both apist and patient variability in the alliance. Journal of Consulting and
conditions demonstrated improved client outcomes over the course Clinical Psychology, 75, 842– 852.
of an academic year, although those in the feedback condition Bandura, A. (1997). Self-efficacy: The exercise of control. New York:
showed twice as much improvement. An informal qualitative W. H. Freeman.
analysis found that trainees in the feedback condition were pleased Bordin, E. S. (1979). The generalizability of the psychoanalytic concept of
the working alliance. Psychotherapy: Theory, Research, and Practice,
that they could monitor how they were doing with their clients as
16, 252–260.
novice therapists. They also felt that feedback from clients gave
Brown, G. S., & Jones, D. R. (2005). Implementation of a feedback system
them more structure and a good place to start with new or “diffi- in a managed care environment: What are patients teaching us? Journal
cult” clients. Supervisors generally felt that client feedback was of Clinical Psychology, 61, 187–198.
less useful, but they agreed that client feedback assisted them in Cohen, J. (1988). Statistical power analysis for the behavioral sciences
providing trainees with challenging feedback and helped to iden- (2nd ed.). Hillsdale, NJ: Erlbaum.
tify clients who needed more attention in supervision. Daniels, J. A., & Larson, L. M. (2001). The impact of performance


feedback on counseling self-efficacy and counselor anxiety. Counselor Lambert, M. J., & Hawkins, E. J. (2001). Using information about patient
Education and Supervision, 41, 120 –130. progress in supervision: Are outcomes enhanced? Australian Psycholo-
Dodenhoff, J. T. (1981). Interpersonal attraction and direct supervisor gist, 36, 131–138.
influence as predictors of counselor trainee effectiveness. Journal of Lambert, M. J., Whipple, J. L., Hawkins, E. J., Vermeersch, D. A., Nielsen,
Counseling Psychology, 28, 47–52. S. L., & Smart, D. W. (2003). Is it time for clinicians to routinely track
Duncan, B. L., Miller, S. D., Sparks, J. A., Claud, D. A., Reynolds, L. R., patient progress? A meta-analysis. Clinical Psychology: Science and
Brown, J. B., et al. (2003). The Session Rating Scale: Preliminary Practice, 10, 288 –301.
psychometric properties of a “working alliance” measure. Journal of Lambert, M. J., Whipple, J. L., Smart, D. W., Vermeersch, D. A., Nielsen,
Brief Therapy, 3, 3–12. S. L., & Hawkins, E. J. (2001). The effects of providing therapists with
Dunlap, W. P., Cortina, J. M., Vaslow, J. B., & Burke, M. J. (1996). feedback on client progress during psychotherapy: Are outcomes en-
Meta-analysis of experiments with matched groups or repeated measures hanced? Psychotherapy Research, 11, 49 – 68.
designs. Psychological Methods, 1, 170 –177. Larson, L. M., Suzuki, L. A., Gillespie, K. N., Potenza, M. T., Bechtel,
Efstation, J. F., Patton, M. J., & Kardash, C. M. (1990). Measuring the M. A., & Toulouse, A. L. (1992). Development and validation of the
working alliance in counselor supervision. Journal of Counseling Psy- Counseling Self-Estimate Inventory. Journal of Counseling Psychology,
chology, 37, 322–329. 39, 105–120.
Ellis, M. V. (1991). Critical incidents in clinical supervision and in super- Lehrman-Waterman, D., & Ladany, N. (2001). Development and valida-
visor supervision: Assessing supervisory issues. Journal of Counseling tion of the Evaluation Process Within Supervision Inventory. Journal of
Psychology, 38, 342–349. Counseling Psychology, 48, 168 –177.
Ellis, M. V., & Ladany, N. (1997). Inferences concerning supervisees and Locke, E. A., & Latham, G. P. (2002). Building a practically useful theory
clients in clinical supervision: An integrative review. In C. E. Watkins of goal setting and task motivation: A 35-year odyssey. American
(Ed.), Handbook of psychotherapy supervision (pp. 447–507). New Psychologist, 57, 705–717.
York: Wiley. Luborsky, L., Barber, J. P., Siqueland, S., Johnson, S., Najavits, L. M.,
Falender, C., & Shafranske, E. (2004). Clinical supervision: A competence- Frank, A., & Daley, D. (1996). The revised Helping Alliance Question-
based approach. Washington, DC: American Psychological Associa- naire (HAQ-II): Psychometric properties. Journal of Psychotherapy
tion. Practice and Research, 5, 260 –271.
Fernando, D. M., Hulse-Killacky, D. (2005). The relationship of supervi- Martin, D. J., Garske, J. P., & Davis, K. M. (2000). Relation of the
sory styles to satisfaction with supervision and the perceived self- therapeutic alliance with outcome and other variables: A meta-
efficacy of master’s level counseling students. Counselor Education and analytic review. Journal of Consulting and Clinical Psychology, 68,
Supervision, 44, 293–304. 438 – 450.
Freitas, G. J. (2002). The impact of psychotherapy supervision on client Miller, S. D., & Duncan, B. L. (2000). The Outcome Rating Scale.
outcome: A critical examination of two decades of research. Psycho- Chicago: Author.
therapy: Theory, Research, Practice, Training, 39, 354 –367. Miller, S. D., & Duncan, B. L. (2004). The Outcome and Session Rating
Friedlander, M. L., Siegel, S. M., & Brenack, K. (1989). Parallel process Scales: Administration and scoring manual. Chicago: Author.
in counseling and supervision: A case study. Journal of Counseling Miller, S. D., Duncan, B. L., Brown, J., Sorrell, R., & Chalk, M. (2006).
Psychology, 36, 149 –157. Using formal client feedback to improve outcome and retention: Making
Goodyear, R. K., & Bernard, J. M. (1998). Clinical supervision: Lessons ongoing, real-time assessment feasible. Journal of Brief Therapy, 5,
from the literature. Counselor Education and Supervision, 38, 6 –22. 5–22.
Grove, W. M., Zald, D. H., Lebow, B. S., Snitz, B. E., & Nelson, C. (2000). Miller, S. D., Duncan, B. L., Brown, J., Sparks, J. A., & Claud, D. A.
Clinical versus mechanical prediction: A meta-analysis. Psychological (2003). The Outcome Rating Scale: A preliminary study of the reliabil-
Assessment, 12, 19 –30. ity, validity, and feasibility of a brief visual analog measure. Journal of
Heppner, M. J., Multon, K. D., Gysbers, N. C., Ellis, C. A., & Zook, C. E. Brief Therapy, 2, 91–100.
(1998). The relationship of trainee self-efficacy to the process and Miller, S. D., Duncan, B. L., & Johnson, L. D. (2000). The Session Rating
outcome of career counseling. Journal of Counseling Psychology, 45, Scale 3.0. Chicago: Author.
393– 402. Miller, S. L., Duncan, B. L., Sorrell, R., & Brown, G. S. (2005). The
Hiatt, D., & Hargrave, G. E. (1995). The characteristics of highly effective partners for change outcome management system. Journal of Clinical
therapists in managed behavioral provider networks. Behavioral Health- Psychology, 61, 199 –208.
care Tomorrow, 4, 19 –22. Nelson, M. L., & Friedlander, M. L. (2001). A close look at conflictual
Hoffman, M. A., Hill, C. E., Holmes, S. E., & Freitas, G. F. (2005). supervisory relationships: The trainee’s perspective. Journal of Coun-
Supervisor perspective on the process and outcome of giving easy, seling Psychology, 48, 384 –395.
difficult, or no feedback to supervisees. Journal of Counseling Psychol- Patton, M. J., & Kivlighan, D. M. (1997). Relevance of supervisory
ogy, 52, 3–13. alliance to the counseling alliance and to treatment adherence in
Iberg, J. R. (1991). Applying statistical control theory to bring together counselor training. Journal of Counseling Psychology, 44, 108 –115.
clinical supervision and psychotherapy research. Journal of Consulting Ronnestad, M. H., & Ladany, N. (2006). The impact of psychotherapy
and Clinical Psychology, 59, 575–586. training: Introduction to the special section. Psychotherapy Research,
Ladany, N., Ellis, M. V., & Friedlander, M. L. (1999). The supervisory 16, 261–267.
working alliance, trainee self-efficacy, and satisfaction. Journal of Stein, D. M., & Lambert, M. J. (1995). Graduate training in psychotherapy:
Counseling and Development, 77, 447– 455. Are therapy outcomes enhanced? Journal of Consulting and Clinical
Lambert, M. J. (2001). Psychotherapy outcome and quality improvement: Psychology, 63, 182–196.
Introduction to the special section on patient-focused research. Journal Stoltenberg, C. D. (2005). Enhancing professional competence through
of Consulting and Clinical Psychology, 69, 147–149. developmental approaches to supervision. American Psychologist, 60,
Lambert, M. J., Hansen, N. B., Umphress, V. J., Lunnen, K., Okiishi, J., 857– 864.
Burlingame, G. M., et al. (1996). Administration and scoring manual for Whipple, J. L., Lambert, M. J., Vermeersch, D. A., Smart, D. W., Nielsen,
the OQ 45.2. Stevenson, MD: American Professional Credentialing S. L., & Hawkins, E. J. (2003). Improving the effects of psychotherapy:
Services. The use of early identification of treatment failure and problem-solving

168 REESE ET AL.

strategies in routine practice. Journal of Counseling Psychology, 50, Advantages of adding systematic client tracking to supportive consulta-
59 – 68. tions. Counselling and Psychotherapy Research, 7, 48 –53.
Worthen, V. E., & Isakson, R. L. (2003, August). Enhancing supervisory
relationships (Report No. CG-032–566). Paper presented at the 111th
Annual Conference of the American Psychological Association, To- Received July 31, 2008
ronto, Canada. (ERIC Document Reproduction Service No. ED479919) Revision received January 12, 2009
Worthen, V. E., & Lambert, M. J. (2007). Outcome-oriented supervision: Accepted February 9, 2009

ReeseUsherBowmanetetal

More Related Content

What's hot (20)

Viewers also liked (8)

Similar to ReeseUsherBowmanetetal (20)

Recently uploaded (20)

ReeseUsherBowmanetetal