Giving Youth a Voice: A Preliminary Study of the Reliability and Validity of a Brief Outcome Measure for Children, Adolescents, and Caretakers

Abstract

Policy makers and payers are insisting that to be paid, therapists must "deliver the goods." Concurrently, there is a worldwide movement to involve consumers in their care. Consequently, the measurement of change, from the client's perspective, has become an important topic. Unfortunately, no self-report outcome measure has been available for children under 13. This article describes the development, validation, and psychometric properties of a brief outcome measure, the Child Outcome Rating Scale (CORS), for use with children 6-12 and their caretakers as well as the Outcome Rating Scale (ORS), for youth 13 and above and their caretakers. Results indicate that the CORS/ORS represent a balanced trade-off between the reliability and validity of longer measures, and the feasibility of these brief scales, and that youth do indeed deserve a voice in the delivery of services. We argue that outcome management is a pivotal part of delivering consumer-driven services, the therapeutic relationship, and change itself.

Full text

Journal of Brief Therapy
Volume 5 • Number 2 • 2006
71
Giving Youth a Voice: A Preliminary
Study of the Reliability and
Validity of a Brief Outcome Measure for
Children, Adolescents, and Caretakers*
Barry L. Duncan, Psy.D.
Institute for the Study of Therapeutic Change
Ft. Lauderdale, Florida
Jacqueline A. Sparks, Ph.D.
University of Rhode Island
Kingston, Rhode Island
Scott D. Miller, Ph.D.
Institute for the Study of Therapeutic Change
Chicago, Illinois
Robert T. Bohanske, Ph.D.
Southwest Behavioral Health Services
Phoenix, Arizona
David A. Claud, MS
The Center for Family Services of Palm Beach County
West Palm Beach, Florida
Policy makers and payers are insisting that to be paid, therapists must “deliver the goods.”
Concurrently, there is a worldwide movement to involve consumers in their care. Consequently, the
measurement of change, from the client’s perspective, has become an important topic. Unfortunately,
no self-report outcome measure has been available for children under 13. This article describes
the development, validation, and psychometric properties of a brief outcome measure, the Child
Outcome Rating Scale (CORS), for use with children 6-12 and their caretakers as well as the
Outcome Rating Scale (ORS), for youth 13 and above and their caretakers. Results indicate that the
CORS/ORS represent a balanced trade-off between the reliability and validity of longer measures,
and the feasibility of these brief scales, and that youth do indeed deserve a voice in the delivery
of services. We argue that outcome management is a pivotal part of delivering consumer-driven
services, the therapeutic relationship, and change itself.
* The authors wish to thank Jeb Brown for his invaluable statistical consultation.
Requests for offprints should be sent to Dr. Sparks at University of Rhode Island, Transition
Center, 2 Lower College Rd., Kingston, RI 02881 or jsparks@uri.edu

72 Giving Youth a Voice
More than any previous time in the history of the eld, policy makers and payers
are stridently insisting that to be paid, therapists, and the systems of care in which
they operate, must “deliver the goods.” Accountability is the watchword of the
day and “return on investment” the guiding metric. Like it or not, mental health
and substance abuse services have become a commodity. Those footing the bill
want proof of the effectiveness and value of the product being purchased (Miller,
Duncan, Sorrell, & Brown, 2005).
Concurrently there is a growing worldwide movement, both private and governmental,
to involve consumers in mental health and substance abuse care (Bohanske, in press). Con-
sequently, the measurement and management of change, from the client’s perspective, has
become an important topic in the delivery of mental health services (Brown, Burlingame,
Lambert, Jones, & Vaccaro, 2001; Duncan, Miller, & Sparks, 2004)), and for good reason:
Monitoring client-based outcome, when combined with feedback to the clinician, increases
the effectiveness of clinical services by an amazing 65% in real clinical settings (Lambert et
al., 2003; Miller, Duncan, Brown, Sorrell, & Chalk, 2006).
Unfortunately, these substantial benets of outcome management and the move toward
consumer-driven services have not included children; there are no valid self-report outcome
tools for children under the age of 13. Although numerous child assessment instruments are
available, few measure the outcome or benet of provided services—most were developed
to assess psychosocial dysfunction or psychological disorders (Burlingame, Mosier, et al.,
2001). Moreover, even if designed for outcome assessment, available measures for children
are lengthy and perhaps too cumbersome to be feasible—for both clients and clinicians—for
everyday use. Finally, all available outcome measures for children, like the Youth Outcome
Questionnaire 30 (YOQ) (Burlingame, Jasper, et al., 2001) are parent-report measures.
Adolescents can report on their progress, but children under 13 have no such opportunity.
Indeed, background research conducted for this study revealed an astonishing lack of refer-
ences for not only child self-report outcome measures but also about children’s perceptions
about therapy in general (Day, Carey, & Surgenor, 2006).
In short, children have little voice in the services they receive. More often than not,
children and adolescents receive services as mandated clients—someone else thought they
needed it. This often leaves the youth with little control over the process, and perhaps, little
reason to engage. It is no wonder, then, that 40-60% of youth drop out of treatment (Kasdin,
2004). To ll the void regarding self-report measures for the under 13 age group, address
the feasibility problem in all youth outcome management, and extend the privilege of client-
based outcome feedback to children—to give youth a voice in the services they receive—the
Child Outcome Rating Scale (CORS) (Duncan, Miller, & Sparks, 2003) was developed.
This article reports the psychometric properties of the CORS for children 6-12 and the Out-
come Rating Scale (ORS) (Miller & Duncan, 2000), previously validated for adults (Miller
et al., 2003), for youth 13 and above for use in tracking the effectiveness of mental health
and substance abuse services. The current study also sought to conrm the validity of using
the youth’s voice in the delivery of services. With more and more children participating in
therapy, it becomes even more important that their voices are solicited if their engagement
and ultimate treatment outcome is valued.
Methods
Development of the Child Outcome Rating Scale (CORS)
Over the last several years, the Institute for the Study of Therapeutic Change (ISTC)

Duncan, et al. 73
has developed a system for both monitoring and improving the effectiveness of treatment
(Duncan et al., 2004; Duncan et al, 2003; Duncan & Sparks, 2007; Miller et al, 2003; Miller
et al., 2006). The approach builds on two key ndings regarding the prediction of outcome
as well as the movement toward consumer involvement (Duncan, Miller, & Wampold, in
press). First is the robust predictive relationship between the alliance and outcome (Norcross,
in press). And second is the nding that the client’s subjective experience of improvement
is a reliable predictor of eventual treatment outcome (Brown et al., 2001; Howard, Kopte,
Krause, & Orlinsky, 1986; Howard, Moras, Brill, Martinovich, & Lutz, 1996).
Measures of client progress and experience of the alliance can be used to “determine
the appropriateness of the current treatment…the need for further treatment…[and] prompt
a clinical consultation for patients who [were] not progressing at expected rates” (Howard
et al. 1996, p. 1063). Providing clinicians with client-based outcome feedback dramatically
increases both the effectiveness and efciency of clinical services (Bohanske, in press;
Duncan et al, in press; Lambert, in press; Miller et al., 2006). For example, Whipple et
al. (2003) found that clients at risk for a negative or null outcome were less likely to
deteriorate, more likely to stay longer, and twice as likelyto achieve a clinically signicant
change when their therapists had access to outcome and alliance information. The efciency
of services—length of stay, productivity, cancellations, no shows, and dropouts—are also
signicantly improved by systematic outcome feedback (Bohanske, in press).
The ISTC outcome management system was specically designed for utilization in
partnership with clients, actively incorporating the now over 1100 studies that conrm
the importance of client engagement and participation for treatment success (Duncan et
al., in press). Such a partnership between the provider and consumer ts the ever more
powerful worldwide movement to view mental health and substance abuse services from
the perspective of the client—to insure that consumers have both “choice and voice” in the
services they receive (Bohanske, in press). Outcome management, rather than a provider
evaluation of the client, becomes a pivotal part of delivering consumer-driven services, the
therapeutic relationship, and change itself.
In addition to establishing a system that is valid, reliable, and consumer driven, a major
goal of the ISTC has been making the collection and use of outcome data user-friendly for
both providers and consumers. As is news to no one on the front lines, and especially in
the public sector, the number of forms and oversight procedures has exploded. Few have
the time to devote to the repeated administration, scoring, and interpretation of lengthy
measures. Brown, Dreis, and Nace (1999), for example, found that practitioners did not
consider any measure that took more than ve minutes to complete, score, and interpret
practical. After experimenting with a number of outcome measures, we found that similar
tolerance levels apply to consumers. Clients quickly tire of measures that lack obvious face
validity, require more than a few minutes to complete, or appear to take away from time
spent with the counselor. Low compliance rates are the most frequent result.
Feasibility, therefore, is a critical issue in outcome management. Though it may be
distressing to researchers, the ease with which an instrument can be explained, completed,
interpreted, and then integrated into ongoing care is much more likely to inuence utilization
than either validity or reliability. Seeking to make the measurement process feasible to the
consumer and front-line-clinician so that outcome could be tracked and the benets of
improved effectiveness and efciency could be realized, the ORS (Miller & Duncan, 2000)
was developed as a brief alternative to the Outcome Questionnaire 45.2 (OQ) (Lambert
et al., 1996). The specic items on the ORS were adapted from the three areas of client
functioning assessed by the OQ; specically, individual, relational, and social domains.
Changes in these three areas are widely considered to be valid indicators of successful
treatment outcome (Lambert & Hill, 2004). With regard to the specic items on the ORS,
the three areas of client functioning were simply translated into a visual analog format

74 Giving Youth a Voice
with instructions to place a mark on the corresponding 10 cm line, with low estimates to
the left and high to the right (see Appendix 1). Research has demonstrated the reliability
and validity of brief visual analog scales (e.g., Ger, Ho, Sun, Wang, & Cleeland, 1999;
Zalon, 1999). In addition to their brevity, ease of administration, and scoring, such scales
frequently enjoy face validity with clients typically missing from longer and more technical
measures that seem distant from the client’s experience.
Miller et al. (2003) reported results of an initial investigation of the reliability and validity
of the ORS. Pearson product moment correlation between the ORS and the OQ yielded a
concurrent validity coefcient of .58, a gure considered adequate given the brevity of the
ORS. Reliability of the measure, as assessed by Cronbach’s coefcient alpha, was .93. An
independent study of a normative population found similar results (Bringhurst, Watson,
Miller, & Duncan, 2006). Perhaps more importantly, Miller et al. (2003) also compared the
feasibility of the OQ and the ORS at two community mental health centers and reported
signicant differences. Utilization of the 4-item ORS reached 86% at the end of one year
while the 45-item OQ dropped signicantly by six months and nished the year at 25%.
The problem of feasibility also plagues youth outcome assessment—although reliable and
valid, available outcome measures for youth are long and present an arduous task for both
clinicians and clients. For example, the Ohio Scales (Ogles, Melendez, Davis, & Lunnen,
2001) “short” form is 48 items while the short form of the Youth Outcome Questionnaire
is 30 items.
Seeking to simplify the assessment of youth outcome for everyday use, the CORS was
developed to track the effectiveness of clinical services offered to children, as reported by
children and their parents or caretakers, from age 6 through 12. The CORS (see Appendix
2) is similar in format to the ORS but contains child friendly language as well as smiley
and frowny faces to aid the child’s understanding. The CORS rates at a third grade reading
level, making it not suitable with some children. The ORS rates at an eighth-grade reading
level, making it feasible for most adolescents who seek or are referred for services. Parents
and caretakers use the ORS to rate youth over 12 and the CORS for children 12 and under.
Children, adolescents, parents, and caretakers have little difculty connecting their day-
to-day lived experience to the ORS/CORS and translating it into the specics of their
circumstances. Indeed, a recent qualitative study of children’s experiences of the measures
found a surprising level of understanding of both ORS/CORS and the therapeutic process
(Crystal, 2007). The youth in this study found the measures to be practical, simple and
helpful. They believed the measures opened important discussions, gave credence to their
perspective, and encouraged active participation.
The Comparison Measure
The Youth Outcome Questionnaire 30 (YOQ). To assess concurrent validity of the
CORS/ORS, the YOQ, the 30 item form of the YOQ-2.0, was selected because of its strong
psychometric qualities and widespread use (Burlingame, Mosier, et al., 2001). It contains
thirty questions that are scored on a ve-point Likert scale. The YOQ provides a total score
or global index of behavioral and emotional distress in a child/adolescent’s life. Critical
items alert clinicians to potential high-risk behaviors (e.g. suicide, substance abuse). It has
a broad normative sample and has been used extensively in research. The YOQ is available
in a parent-rating version which is appropriate for ages 4 through 17, and a self-report
version, which is appropriate for ages 12 through 17. Most of the variance on the YOQ,
like most outcome measures, can be accounted for on a single dimension: distress (Mueller,
Lambert, & Burlingame, 1998). Parents or others with extensive interaction with the client
complete the YOQ at intake to establish a severity baseline and then complete it at regular
intervals to track the progress of the child’s treatment (Burlingame, Jasper, et al., 2001).

Duncan, et al. 75
Participants
Participants in this study were recruited from three non-clinical and three clinical sites.
Non-clinical Group
The rst non-clinical site was a prevention project in a public school. Fifty-nine
dyads of caretakers and youth completed the questionnaires on two separate occasions
for a total of 236 administrations of the measures. The second non-clinical sample was
recruited from graduate courses at Ottawa University (116 dyads of caretakers and youth).
Youth and caretakers completed the measures on two separate occasions for a total of 464
administrations. The third non-clinical sample was recruited from employees of Southwest
Behavioral Health (23 dyads), totaling 92 administrations of the measures over two separate
occasions. Of the total 199 youth/caretaker dyads, 154 were children 12 and under and
45 were adolescents 13 and above. The under 13 group consisted of an equal amount of
males and females and the 13 and above group was composed of 23 males and 22 females.
Parents completed the ORS for youth 13 and above and the CORS for children 12 and
under as well as the YOQ.
Clinical Group
Adolescent. The rst clinical sample consisted of 1,495 adolescents treated for at least
two sessions at outpatient treatment centers. Fifty-ve percent of this sample was male and
forty-ve percent was female. A total of 9,917 youth completed ORSs were available for
analysis.1 This sample consists of adolescent ratings only.
Caretaker. The second clinical sample consisted of 1961 children and adolescents
receiving publicly funded outpatient treatment services for at least two sessions within a
ve county rural area in Oregon. A total of 11, 737 completed measures were available for
analysis. This sample consists of caretaker ratings only of both children and adolescents.2
Child, Adolescent, and Caretaker. The nal clinical sample consisted of 155 children
and adolescents and a matched pair of 155 caretakers at a South Florida community mental
health center3 that completed at least two sessions. A total of 1860 completed C/ORSs were
originally available for analysis but only the rst and last sessions were entered into the
data base. This sample includes child, adolescent, and caretaker ratings.
Procedure
Nonclinical Group
In each non-clinical sample, participants received two concurrent administrations of
the CORS (if age 6-12) or the ORS (if 13-17) and the YOQ (if over 12) over a period
ranging from 10 days to 3 weeks. Caretakers were administered either the CORS or ORS,
depending on the age of their child, and the YOQ.
Clinical Groups
In all clinical samples, therapists or other staff collected data as part of standard agency
policy. Data from the rst sample occurred over a four year span; the second over a three
span; and the third represented two years of data collection. Cases where either an initial
or nal ORS score was missing were excluded from the sample, as were those cases that
marked the maximum score throughout service.

76 Giving Youth a Voice
Results
Normative Data
Table 1 displays the means and standard deviations for the non-clinical and clinical
samples. As expected, a two-tailed t-test comparing initial ORS and CORS scores for the
non-clinical and clinical samples was highly signicant (p < .0001).
Table 1.
Comparison Of Clinical And Nonclinical Samples
Sample N Instrument Mean SD
Non-clinical
199 dyads or 796
administrations
45
45
154
154
45
45
154
199
199
(rst administration mean)
ORS Adolescent (28.5)
ORS Adol. Caretaker (31.0)
CORS Child (33.0)
CORS Caretaker (33.5)
YOQ Adolescent (24.6)
YOQ Adol, Caretaker (21.7)
YOQ Child Caretaker (16.0)
Total C/ORS Caretaker (32.6)
Total YOQ Caretaker (17.1)
29.5
30.7
33.4
33.7
25.4
20.9
15.6
32.9
16.8
7.9
7.7
7.0
6.9
16.3
18.7
13.0
7.2
14.7
Clinical (over 20,000
administrations)
Adolescents only
Caretakers only
Children, Adolescents,
And Caretakers
1495
1961
38
38
119
119
157
(clinical cutoff)
ORS Adolescent (27.7)
C/ORS Caretaker (27.2)
ORS Adolescent (27.6)
ORS Adol. Caretaker (27.5)
CORS Child (31.8)
CORS Caretaker (29.7)
Total C/ORS Caretaker (29.0)
25.9
21.1
25.3
23.9
30.2
24.0
23.9
8.1
7.8
9.3
8.9
7.8
9.8
9.6
t-test of CORS and ORS scores for clinical and non-clinical samples yielded
signicant difference (p< .001)
For the non-clinical group, the mean YOQ score was comparable for both caretakers
(16.8 v. 17.3) and adolescents (25.4 v. 23.4) to that reported in the test manual for the
large community normative sample of the measure (Burlingame, Jasper, et al., 2001). This
lends support to the premise that the non-clinical sample in this study, though relatively
small, was similar. Gender differences in scores in the non-clinical sample proved to be
nonsignicant. For this reason, gender specic norms are not broken out here. However,
there was a tendency toward less distress reported for both females and participants who
identied themselves as Hispanic (self-report and caretakers). Future research will continue
to explore the possibility of gender and ethnicity-specic norms.
Reliability of the ORS/CORS
Reliability, based on 1495 adolescents and 1961 children (over 20,000 administrations
of the CORS and ORS) was estimated using Cronbach’s coefcient alpha, a measure of

Duncan, et al. 77
the internal consistency of the measure. The ORS and the CORS display strong evidence
of reliability, with coefcient alpha estimates of .93 and .84 respectively. These are very
high coefcients of reliability for such brief measures, comparing very favorably with that
reported for the YOQ, suggesting the CORS and ORS tap the factor that most if not all
outcome measures tap, global distress.
Table 2
Test-Retest Correlations of The CORS, ORS, and YOQ
Measure CORS CTCORS ORS CTORS YOQ CTYOQ
(child)
CTYOQ
(adol.)
Admin.2 .60 .51 .78 .72 .75 .77 .87
An estimate of test-retest reliability was obtained by correlating the test scores at the
rst administration with those at the subsequent administration in the non-clinical sample.
Table 2 presents the test-retest correlations for the C/ORS and YOQ. As depicted in
Table 1, the mean scores changed little from administration to administration. As might
be expected from a brief measure, the rest-retest reliability was generally lower for the
CORS than the YOQ although the ORS compared favorably. However, test-rest reliability
is an inappropriate measure of reliability for a questionnaire designed to be sensitive to
client’s perception of subjective change. All outcomes measures will tend show a pattern
of declining correlations the rst administration and each subsequent administration. This
is referred to as autocorrelation. For example, Table 3 presents the autocorrelation matrices
for the rst ve administrations of two of the clinical samples.
Table 3
Correlations Between First and Subsequent Administrations of Adolescent and
Caretaker Samples
Session 2 Session 3 Session 4 Session 5
Adolescents 0.6 0.5 0.44 0.38
n= (1495) (1420) (1210) (889)
Caretakers 0.63 0.57 0.51 0.48
n= (1961) (1550) (1236) (1009)
Validity of the ORS
Tables 4 and 5 present the concurrent validity correlation matrix for children and
adolescents from the non-clinical sample that received two concurrent administrations of
the ORS/CORS and the YOQ for both youth and caretakers. Note that correlations between
the C/ORS and YOQ will be negative since the low score reect low distress on the YOQ
while on the ORS high scores are low distress.

78 Giving Youth a Voice
Table 4
CORS and YOQ Correlation Matrix (12 and under)
CORS CT-CORS CT-YOQ
CORS 1 0.63* -0.43*
CT-CORS 0.63* 1 -0.61*
CT-YOQ -0.43* -0.61* 1
* represents signicance
Table 5
ORS and YOQ Correlation Matrix (age 13 to 17)
ORS CT-ORS YOQ CT-YOQ
ORS 1.0 0.45* -0.53* -0.31
CT-ORS 0.45* 1.0 -0.46* -0.61*
YOQ -0.53* -.0.46* 1.0 0.69*
CT-YOQ -0.31 -0.61* 0.69* 1.0
* represents signicance
An inspection of Tables 4 and 5 reveal that the CORS and ORS were signicantly related
to the YOQ in all cells of the matrix demonstrating moderate concurrent validity with the
well researched but much longer YOQ. Similar to the results obtained comparing the ORS
for adults to the OQ (.58), a Pearson product moment correlation yielded a coefcient
between the CORS/ORS and YOQ caretaker scores of .61; the ORS and YOQ completed
by adolescents resulted in a .53 correlation. These correlations provide evidence of the
concurrent validity of the CORS/ORS as brief alternatives for assessing global subjective
distress similar to that measured by the full-scale score on the YOQ. Interestingly, the
correlation was also signicant between both children (.63) and adolescents (.45), and
caretaker ratings, suggesting that giving voice to youth via outcome measures is supported
by the evidence. Although adolescents have long had this opportunity, the CORS is the rst
outcome measure that taps into the perspective of children ages 6-12. Similarly, change
scores (the difference between rst and last session scores) of youth and their caretakers in
the clinical sample were also signicantly correlated.
The construct validity of the CORS/ORS rests in part on the nding that the items load
on a common factor shared with other similar outcome measures. Correlations between the
ORS/ORS and Caretaker CORS/ORS scores and the well-validated YOQ provide further
evidence of construct validity. This aspect of construct validity rests on the assumption
of an underlying unobservable trait or state that the questionnaire purports to measure. It
the case of outcome questionnaires, the underlying state appears to be subjective global
distress. More specic constructs such as “depression,” “anxiety,” and “interpersonal
problems” can be shown to share a large percentage of variance with the global distress
factor. From a purely statistical point of view, constructs based on diagnostic nomenclature
such as “anxiety disorder symptoms” or “symptoms of depression” appear to have very
little predictive value, as both anxiety and depression symptoms load heavily on the global
distress factor.
Other indications of construct validity include: the ability of the measures to differentiate

Duncan, et al. 79
between normative and clinical samples, and the demonstration of stability in non-clinical
populations vs. change sensitivity, beyond regression to the mean, in clinical populations.
Ability to Discriminate Between Client and Nonclient Samples. Comparing rst
administration scores for the clinical and non-clinical groups can be used to provide evidence
of construct validity. Were the CORS/ORS able to accurately discriminate between the two
samples, initial scores would be expected to be signicantly lower (more distressed) for the
clinical group. The results presented in Table 1 conrm that this is the case.
Sensitivity to Change. If valid, the ORS/CORS should reect change following
psychotherapy, but remain stable in an untreated population. Therefore, it was expected
that ORS/CORS scores in the clinical sample would increase while those in the non-clinical
sample would vary only minimally from a pre- and post-test. Such a nding would provide
additional evidence of construct validity for the instrument.
Table 6
Change Sensitivity in Clinical Samples
Clinical Sample First Session
Mean (SD)
Last Session
Mean (SD)
Pre-post Change
(Signicance)
ORS Adolescent
N=1495
25.9 (8.1) 33.6 (6.5) 7.9 (p< .001)
C/ORS Caretaker (CT)
N=1961
21.1 (7.8) 24.4 (7.7) 3.3 (p< .001)
ORS Adolescent
CT Adolescent
N=38
25.3 (9.3)
23.9 (8.8)
29.7 (10.2)
29.0 (8.1)
4.4 (p< .05)
5.1 (p< .001)
CORS (12 and under)
CT (12 and under)
N=119
30.2 (7.8)
24.0 (9.8)
35.2 (6.3)
31.2 (7.6)
5.0 (p< .001)
7.2 (p< .001)
An inspection of Table 6 reveals that across all three samples with children, adolescents,
and caretakers, this was indeed the case. A t-test for correlated samples tested the hypothesis
that CORS/ORS scores would increase following therapy intervention. As expected, the
t-test between the mean of the client pre-test scores and their post-test scores revealed
statistically signicant improvement. Both self-rated and caretaker-rated measures showed
signicant improvement between pre- and post-test scores. (p<.01; one-tailed t-test) The
caretaker-rated measures averaged 6.7 points improvement, compared to 4.8 on the self-
rated measures. The mean change on the self rated measures was likely limited due to the
skewed nature of the scores at pre-test.
Conversely, a t-test between mean pre- and post- CORS/ORS test scores from the non-
clinical sample proved nonsignicant. Therefore, the CORS/ORS was sensitive to change
in those clients receiving psychotherapy and relatively stable for those not receiving
intervention.
Sensitive to Change beyond Regression to the Mean. To test if the observed changes
exceeded regression to the mean a time-reversed regression was employed (Cohen & Cohen,
1983). Pure regression to the mean is a measurement artifact that is time symmetrical,
meaning that the equation used to predict the last score from the rst score is the same if
time is reversed and the last score is used to predict the rst score. If the change between the
rst and last score is greater in the forward prediction than in the time-reversed prediction,
then the observed change exceeded regression to the mean.

80 Giving Youth a Voice
Figure 1. Regression to the Mean
Figure 1 presents the results of the analyses. The x-axis displays C/ORS child/
adolescent and caretaker scores. For the purpose of this analysis children and adolescents
were considered together. The y-axis displays the predicted score resulting from the
regression analysis. The purple lines (solid and dotted) represent the forward predictions
for each measure, while the red lines are the time-reversed predictions. The results clearly
demonstrate that change on each measure exceeded regression to the mean.4 Strikingly, the
forward predictions are virtually identical for the CORS/ORS and caretaker ratings. Even
though the distributions of these scores on these measures differed in this sample, with
youth reporting less distress than their parents, the underlying relationship between intake
score and nal score (or change score) is very similar, and another indication that the voice
of youth should be incorporated.
Discussion
All the problems typically associated with brief self-report tests (Boulet & Boss, 1991)
apply to the CORS/ORS; for example, interpretation relies on clients’ accurate assessment
of their levels of distress and there are no controls for response sets like social desirability.
Additionally, the CORS/ORS does not measure nor is intended to identify clinical risk
factors such as suicide or alcohol or drug use. Research with more diverse clinical and non-
clinical samples is underway and should further identify the strengths and weakness of the
measure. Clearly, evaluation of outcome via the ORS and CORS is far from comprehensive
and does not contain multiple perspectives (e.g., therapists, outside judges, community
criteria, etc.).
Although a short measure like the CORS/ORS cannot be expected to achieve the
same precision or depth of information as a longer measure like the YOQ, this study found
that the CORS/ORS has moderate validity and solid reliability. The CORS for children
under 13 and their caretakers and the ORS for adolescents and their caretakers provide a
brief measure of global distress suitable for assessing treatment outcomes. The measure
shows unusually high reliability for a 4-item questionnaire, comparing favorably to well
establish outcome measures containing many more items. The high coefcient alpha
provides evidence of construct validity; the four items all correlate with one another and
appear to measure the broad construct of global distress.

Duncan, et al. 81
The non-clinical sample was used to assess the concurrent validity of the CORS/
ORS. The overall correlation with the YOQ demonstrated that the C/ORS is moderately
related to this gold standard of self-report scales. Lower scores (more distressed) at the rst
administration were anticipated for the clinical sample as compared to the non-clinical.
The difference found between the clinical and non-clinical samples suggests that the
CORS/ORS measures what it purports to: psychological distress. Changes in client scores
between pre-and post-test as compared to the stable scores for the non-clinical sample also
provide evidence of construct validity for the C/ORS
This study also addressed the issue of including the youth’s voice in clinical decision
making—whether a child/adolescent self-report measure is an appropriate source of
information. Developmental considerations, such as verbal skills and reading ability,
motivation for treatment, social desirability, and perception of the problem (i.e. whether the
child sees a problem v the caretaker perspective) have all combined to make caretaker report
the norm for children under twelve and the probability for adolescents. The assumption is
that parents will be the most reliable source for obtaining data about a child’s functioning.
In adolescents, Achenbach and Edelbroch (1991) suggest that the major differences
between parent and self-report are: 1) adolescents under-report problems as compared to
parents and 2) parents are better sources regarding objective behaviors (e.g. oppositional
attitude, school failures, etc) while adolescents are more accurate about their subjective
states (moods, feelings, etc). It is our view that any a priori dismissal of the child (too
young) or adolescent’s (objective behaviors) perspective is seriously misguided. Children,
much like adults diagnosed with a severe mental illness, have had a profound absence of
voice in the delivery of services under the justication that they do not know what is best
for them. This injustice is compounded by the fact that youth are most often mandated for
services, and thus are subjected to the whims, well-intended as they are, of the adults who
decide on their behalf. This study demonstrated that while youth do rate themselves higher
(less distressed) than their caretakers, their views are nonetheless positively correlated
with caretakers and are reliable and valid markers of treatment of success. Youth scores
increased with psychotherapy just as the caretakers’ scores.
The systematic incorporation of the youth’s voice may be criticized as naïve or ill-
informed. However, treating young people as reliable informants and helping them make
sense of their experience in ways that generate hope and engagement is well grounded
in empirical evidence. Recent meta-analyses of the child outcome literature indicate that
no one approach is superior to another for resolving child problems (Miller, Wampold, &
Varhely, in press). We simply do not know what will be helpful for the individual client
unless we obtain feedback about the effectiveness of any intervention from the client’s point
of view. Additionally, research conrms the pivotal role the alliance plays in the outcome
of child intervention (Shirk & Karver, 2003). When children and adolescent experience
that their opinions matter, participation is enhanced, enabling the practitioner to harness the
most powerful therapist provided-means to promote change—the alliance.
The brevity of the measures greatly aids the partnership process—feasibility is critical
to outcome monitoring and management. The length of the YOQ and its lack of face
validity made the completion of this study doubtful at times. The original study included
7 normative sites, but they quickly dropped out as they discovered the difculty in getting
children, adolescents, and their families to complete more than one pass at the YOQ. A few
of the sites offered incentives: rafes, dinners, and payment, but the result was the same—a
majority of those who volunteered dropped out. In one school site, following a donation
to the school, 500 youth/parents dyads volunteered for the study. At the rst assessment,
only 200 completed the measures. Of that 200, only 25 returned for a second assessment.
In total, over 2500 research packets were disseminated which nally resulted in a non-

82 Giving Youth a Voice
clinical sample of 199 dyads. We believe this illustrative of the feasibility of the measures
involved.
On the practitioner side of things, many therapists see outcome measurement as an
“add-on” separate from actual clinical work and relevant only to management and other
overseers. In addition to wanting measures to be brief, easy to integrate, and have face
validity, counselors want measures that are clinically useful. Is the measure intended
to improve the effectiveness of rendered services or merely monitor them? Most if not
all other youth outcome measures were developed primarily as pre-post and/or periodic
outcome measures. Such instruments, like the 48 item Ohio Scales, provide an excellent
way to measure program effectiveness but are not feasible to administer frequently, and
therefore, do not provide real-time feedback for immediate treatment modication before
clients drop out or suffer a negative outcome—in short they are not clinical tools as much
as they are oversight tools.
Longer measurement systems largely intended for oversight can also create signicant
management problems. In reaction to a managed care company’s introduction of the
Outcome Questionnaire 30, just 30 items, it was recently reported in the New England
Psychologist (Hanlon, 2005) that providers complained about its length and frequent
administration, that it cut into sessions and increased workload, and that some items were
intrusive. The response by clinicians was so severe that it led the State Psychological
Association president to say, “I have never seen such negative reaction from providers”
(Hanlon, 2005, p. 11). This is not an infrequent reaction in our experience.
The ORS/CORS was designed as a clinical and outcome tool to provide immediate
feedback to both clients and providers to improve the effectiveness of services, and as a
way to measure outcome at individual, program, and agency levels. Given their feasibility,
the ORS/CORS can provide immediate feedback not only based on client scores but also
in comparison to normative trajectories of change of a large and growing clinical data base
(over 300,000 administrations). In addition, the measures were designed to be used in
collaboration with clients, to encourage a partnership between the client and therapist for
monitoring the effectiveness of services. Accountability becomes a joint endeavor, integral
to alliance building, rather than simply more paperwork.
To be sure, the ORS and CORS are weaker psychometrically than the YOQ and other
longer measures. Neither do these brief visual analogue scales offer the same breadth of
assessment as the longer scales (e.g., the absence of “critical items” related to suicide or
alcohol and drug use). At the same time, a measure that goes unused is worthless regardless
of its strengths. In the real world of clinical practice, measuring outcome means striking
a balance between the competing demands of validity, reliability, and feasibility. The
development of the ORS, and subsequently the CORS reect the ISTC’s attempt to nd
such a balance.
Moreover, the ORS and CORS represent our efforts to bring consumers into the loop
in decisions about their care to allow a true partnership between clients and providers of
service—to improve outcome one client at a time by assigning those we serve key roles.
The current study demonstrated that young people can also be given a voice in mental
health and substance abuse services, and assume their rightful roles in the forefront of their
own change.
References
Achenbach, T. M., & Edelbroch, C. (1991). The child behavior checklist and revised child
behavior prole. Burlington, VT: University Associates in Psychiatry.
Bohanske, R.T. (in press). Transforming public behavioral health care. In B. Duncan, S.

Duncan, et al. 83
Miller, & B. Wampold (Eds.), The heart and soul of change: Delivering what
works, 2nd ed. Washington, DC: APA Press.
Boulet, J., & Boss, M. (1991). Reliability and validity of the Brief Symptom Inventory.
Journal of Consulting and Clinical Psychology, 3(3), 433–437.
Bringhurst, D.L., Watson, C.W., Miller, S.D., & Duncan, B.L. (2006). The reliability
and validity of the Outcome Rating Scale: A replication study of a brief clinical
measure. Journal of Brief Therapy, 5, 23-30.
Brown, J., Dreis, S., & Nace, D. (1999). What really makes a difference in psychotherapy
outcome? Why does managed care want to know? In M. Hubble, B. Duncan, & S.
Miller (Ed.), The heart and soul of change (pp. 389–406). Washington, DC: APA
Press.
Brown, G. S., Burlingame, G. M., Lambert, M. J., Jones, E., & Vaccaro, J. (2001). Pushing
the quality envelope: A new outcomes management system. Psychiatric Services.
52 (7), 925-934.
Burlingame, G. M., Jasper, B. W., Peterson, G., Wells, M. G., Lambert, M. J., Reisinger,
C. W. & Brown, G. S. (2001). Administration and Scoring Manual for the Y-LSQ.
Stevenson, MD: American Professional Credentialing Services LLC and Pacicare
Behavioral Health, Inc.
Burlingame, G. M., Mosier, J. I., Wells, M. G., Atkin, Q. G., Lambert, M. J., Whoolery,
M., & Latkowski, M. (2001). Tracking the inuence of mental health treatment:
The development of the Youth Outcome Questionnaire. Clinical Psychology and
Psychotherapy. 8, 361-379.
Cohen, J., & Cohen, T. (1983). Applied multiple regression/correlation analysis for the
behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates.
Crystal, S.C. (2007). Exploring children’s experiences of the use of outcome measures in
therapy. Unpublished masters dissertation, University of Birmingham, Birmingham,
UK.
Day, C., Carey, M., & Surgenor, T. (2006). Children’s key concerns: Piloting a qualitative
approach to understanding their experience of mental health care. Journal of
Clinical Child Psychology and Psychiatry, 11 (1), 139-155.
Duncan, B. L., Miller, S.D., Sparks, J. A. (2003). The Child Outcome Rating Scale. Authors:
Ft. Lauderdale, FL.
Duncan, B. L., Miller, S.D., Sparks, J.A. (2004). The heroic client: A revolutionary way to
improve effectiveness, rev. ed.. San Francisco: Jossey Bass.
Duncan, B.L., Miller, S.D., & Wampold, B.E. (Eds.) (in press). The heart and soul of
change: Delivering what works, 2nd ed.. Washington, DC: APA Press.
Duncan, B.L., & Sparks, J.A. (2007). Heroic clients, heroic agencies: Partners for change,
rev. ed. www.talkingcure.com: ISTC Press
Ger, L. P., Ho, S. T., Sun, W. Z., Wang, M. S., & Cleeland, C. S. (1999). Validation of
the Brief Pain Inventory in a Taiwanese population. Journal of Pain & Symptom
Management, 18(5), 316–322.
Hanlon, P. (2005). PaciCare screening tool, policies raise concerns. New England
Psychologist, 13, 1,11-12.
Howard, K.I., Kopte, S.M., Krause, M.S., & Orlinsky, D.E. (1986). The dose-effect
relationship in psychotherapy. American Psychologist, 41, 159-164.
Howard, K., Moras, K., Brill, P., Martinovich, Z., & Lutz, W. (1996). Evaluation of
psychotherapy: Efcacy, effectiveness, and patient progress. AmericanPsychologist,
51(10), 1059–1064.
Kazdin, A. E. (2004). Psychotherapy for children and adolescents. In M. J. Lambert (Ed.).
Bergin and Gareld’s handbook of psychotherapy and behavior change, 5th ed. (pp.
543-589). New York:Wiley.

84 Giving Youth a Voice
Lambert, M.J. (in press). Yes, it is time for clinicians to routinely track outcome. In B.
Duncan, S. Miller, & B. Wampold (Eds.), The heart and soul of change: Delivering
what works, 2nd ed. Washington, DC: APA Press.
Lambert, M. J., Burlingame, G. M., Umphress, V., Hansen, N. B., Vermeersch, D. A.,
Clouse, G. C., & Yanchar S. C. (1996). The reliability and validity of the Outcome
Questionnaire. Clinical Psychology and Psychotherapy. 3 (4), 249-258.
Lambert, M. J., Okiishi, J. C., Finch, A. E., & Johnson, L. D. (1998). Outcome assessment
from conceptualization to implementation. Professional Psychology: Research
and Practice. Vol. 29, No. 1, 63-70.
Lambert, M. J., Whipple, J. L., Hawkins, E. J., Vermeersch, D. A., Nielsen, S. L., & Smart,
D. W. (2003). Is it time for clinicians routinely to track patient outcome? A meta-
analysis. Clinical Psychology, 10, 228-301.
Lambert, M. J., & Hills, C. E. (2004). Methodological issues in assessing psychotherapy
processes and outcomes and processes. In M. J. Lamberts (Ed.), Bergin and
Gareld’s Handbook of psychotherapy and behavior and change, 5th ed. (pp. 84–
136). New York: John Wiley.
Lambert, M. J., Whipple, J., Smart, D., Vermeersch, D., Nielsen, S., & Hawkins, E. (2001).
The effects of providing therapists with feedback on patient progress during
psychotherapy: Are outcomes enhanced? Psychotherapy Research, 11(1), 49–68.
Miller, S. D., & Duncan, B. L. (2000). The Outcome Rating Scale. Chicago: Author.
Miller, S. D., Duncan, B.L., Brown, J., Sorrell, R., & Chalk, M. B. (2006). Using outcome
to inform and improve treatment outcomes. Journal of Brief Therapy. 5, 5-22.
Miller, S. D., Duncan, B. L., Brown, J., Sparks, J. A., & Claud, D. A. (2003). The Outcome
Rating Scale: A preliminary study of the reliability, validity, and feasibility of a
brief, visual, analog measure. Journal of Brief Therapy, 2, 91-100.
Miller S., Duncan, B., Sorrell, R., & Brown, J. (2005). The Partners for Change Outcome
Management System. Journal of Clinical Psychology: In Session, 61, 199-208.
Miller, S.D., Wampold, B.E., & Varhely, K. (in press). Direct comparisons of treatment
modalities for youth disorders: A meta-analysis. Psychotherapy Research.
Mueller, R. M., Lambert, M. J., & Burlingame, G. M. (1998). Construct validity of the
Outcome Questionnaire: A conrmatory factor analysis. Journal of Personality
Assessment, 70(2), 248–262.
Norcross, J.C. (in press). The therapeutic relationship. In B. Duncan, S. Miller, & B.
Wampold (Eds.), The heart and soul of change: Delivering what works, 2nd ed.
Washington, DC: APA Press.
Ogles B.M., Melendez G., Davis D.C., & Lunnen K.M (2001). Journal of Child and Family
Studies, 10 (2), 199-212.
Shirk, S. R., & Karver, M. (2003). Prediction of treatment outcome from relationship
variables in child and adolescent therapy: A meta-analytic review. Journal of
Consulting and Clinical Psychology, 71(3), 452-464.
Whipple, J. L., Lambert, M. J., Vermeersch, D. A., Smart, D. W., Nielsen, S. L., &
Hawkins, E. J. (2003). Improving the effects of psychotherapy: The use of early
identication of treatment and problem-solving strategies in routine practice.
Journal of Counseling Psychology, 50(1), 59–68.
Zalon, M. L. (1999). Comparison of pain measures in surgical patients. Journal of Nursing
Measurement, 7(2), 135–152.

Duncan, et al. 85
Appendix 1:
Outcome Rating Scale (ORS)
Name ________________________Age (Yrs):____ Sex: M / F
Session # ____ Date: ________________________
Who is lling out this form? Please check one: Self_______ Other_______
If other, what is your relationship to this person? ____________________________
Looking back over the last week, including today, help us understand how you have been feeling
by rating how well you have been doing in the following areas of your life, where marks to the left
represent low levels and marks to the right indicate high levels. If you are lling out this form for
another person, please ll out according to how you think he or she is doing.
Individually
(Personal well-being)
I----------------------------------------------------------------------I
Interpersonally
(Family, close relationships)
I----------------------------------------------------------------------I
Socially
(Work, school, friendships)
I----------------------------------------------------------------------I
Overall
(General sense of well-being)
I----------------------------------------------------------------------I
Institute for the Study of Therapeutic Change
_______________________________________
www.talkingcure.com
© 2000, Scott D. Miller and Barry L. Duncan
Note: Working copies of the measures are available for free download from www.talkingcure.com.

86 Giving Youth a Voice
Appendix 2:
Child Outcome Rating Scale (CORS)
Name ________________________Age (Yrs):____
Sex: M / F_________
Session # ____ Date: ________________________
Who is lling out this form? Please check one: Child_______ Caretaker_______
If caretaker, what is your relationship to this child? ____________________________
How are you doing? How are things going in your life? Please make a mark on the scale
to let us know. The closer to the smiley face, the better things are. The closer to the
frowny face, things are not so good. If you are a caretaker lling out this form, please ll
out according to how you think the child is doing.
Me
(How am I doing?)
I----------------------------------------------------------------------I
Family
(How are things in my family?)
I----------------------------------------------------------------------I
School
(How am I doing at school?)
I----------------------------------------------------------------------I
Everything
(How is everything going?)
I----------------------------------------------------------------------I
Institute for the Study of Therapeutic Change
_____________________________________
www.talkingcure.com
© 2003, Barry L. Duncan, Scott D. Miller, & Jacqueline A. Sparks

Duncan, et al. 87
(Endnotes)
1. The data from this site was released on the condition of total anonymity.
2. This sample from Accountable Behavioral Health Alliance used a derivative of the ORS/
CORS called the Oregon Change Index (OCI). The ORS/CORS utilizes a 10-centimeter
visual analog scale with anchors at the extremes of the scale. The OCI is scored on a 10-
point Likert scale, also with anchors at the extremes. A Likert scale with discrete points
was used to reduce labor costs associated with scoring and data entry. Both methods of
presenting and scoring the scales appear to create very similar results, though differences
in response patterns to a visual analog versus 10-point Likert scale may result in small
differences in the psychometric properties of the scores produced.
3. The Center for Family Services of Palm Beach County, Inc. is a not-for-prot family
services agency serving Palm Beach County of South Florida. The agency provides an
array of services including individual and family counseling, substance abuse treatment,
sexual abuse and domestic violence treatment, EAP services, homeless assistance/shelter,
a school readiness program, and home-based family therapy to families at risk.
4. Cautionary note: The skewed nature of the C/ORS scores violates the assumption
of normality underlying the use of regression techniques. While these techniques are
generally robust and not overly sensitive to deviations from normality, some caution
should be used when interpreting regression results for the C/ORS data. A sample with
less skewed scores may yield different results.

88 Giving Youth a Voice