ERIC EJ697210: Effects of Immediate Performance Feedback on Implementation of Behavior Support Plans

2005·0.19 MB·English

by  ERIC

JOURNALOFAPPLIEDBEHAVIORANALYSIS 2005, 38, 205–219 NUMBER2 (SUMMER2005)
EFFECTS OF IMMEDIATE PERFORMANCE FEEDBACK ON
IMPLEMENTATION OF BEHAVIOR SUPPORT PLANS
ROBIN S. CODDING
CITYUNIVERSITYOFNEWYORKGRADUATECENTER
AND
ADAM B. FEINBERG, ERIN K. DUNN, AND GARY M. PACE
MAYINSTITUTE
Research has focused on increasing the treatment integrity of school-based interventions by
utilizing performance feedback. The purpose of this study was to extend this literature by
increasing special education teachers’ treatment integrity for implementing antecedent and
consequence procedures in an ongoing behavior support plan. A multiple baseline across
teacher–student dyads (for two classrooms) design was used to evaluate the effects of
performance feedback on the percentage of antecedent and consequence components
implemented correctly during 1-hr observation sessions. Performance feedback was provided
every other week for 8 to 22weeks after a stable or decreasing trend in the percentage of
antecedent or consequence components implemented correctly. Results suggested that
performance feedback increased the treatment integrity of antecedent components for 4 of 5
teachersandconsequencecomponentsforall5teachers.Theseresultsweremaintainedfollowing
feedback for all teachers across antecedent and consequence components. Teachers rated
performance feedback favorably with respect to the purpose, procedures, and outcome, as
indicated byasocialvalidity rating measure.
DESCRIPTORS: treatment integrity, performance feedback, behavior support plans,
specialeducation
_______________________________________________________________________________
Treatment integrity, the implementation of bertson,Ranier,&Freeland,1997;Witt,Noell,
aninterventionasintended,isatopicofinterest LaFleur, & Mortenson, 1997). One successful
amongresearchersandchangeagents(Gresham, intervention strategy for enhancing treatment
Gansle, & Noell, 1993; Watson, Sterling, & integrity is performance feedback (Mortenson
McDade, 1997). This interest in treatment & Witt, 1998; Noell et al., 1997, 2000,
integrity is accompanied by a recent focus on 2002; Witt et al., 1997). In the literature,
accountability by school systems with the performance feedback has encompassed several
enactment of No Child Left Behind (U.S. components including (a) review of data, (b)
Department of Education, 2001). Con- praiseforcorrectimplementation, (c)corrective
sequently, researchers have attempted to iden- feedback, and (d) addressing questions or
tify parsimonious, effective, and time-efficient comments.
strategies to ensure the integrity of school- Performance feedback has been used in
based interventions (Cossiart, Hall, & Hop- regular education settings to improve the
kins, 1973; Gillat & Sulzer-Azaroff, 1994; implementation of academic interventions
Mortenson & Witt, 1998; Noell, Duhon, (Mortenson & Witt, 1998; Noell et al., 2000;
Gatti, & Connell, 2002; Noell, Witt, Gil- Witt et al., 1997), peer tutoring (Noell et al.,
2000), the use of contingent praise (Jones,
Address correspondence to Robin Codding, Educa- Wickstrom, & Friman, 1997; Martens,
tional Psychology, CUNY Graduate Center, 365 Fifth Hiralall, & Bradley, 1997), and implementa-
Avenue,NewYork,NewYork10016(e-mail:rcodding@
tion of behavior-management interventions
gc.cuny.edu).
doi:10.1901/jaba.2005.98-04 (Noell et al., 2002). Noell et al. (2002)
205
206 ROBIN S. CODDING et al.
examined the integrity with which 4 general The purpose of the present investigation was
education teachers employed interventions to to extend the current research on treatment
decreaseout-of-seatbehavior andtalking outin integrityby(a) directlyexaminingtheeffectsof
8 elementary school students. Integrity was performance feedback on special education
measuredbypermanentproductsthatconsisted teachers’ implementation of antecedent and
of data collected either by students’ self-report consequence procedures of ongoing individual-
or by teachers. This study represents an ized behavior support plans, (b) administering
example of the effects of performance feedback performance feedback every other week and
on one aspect of behavior interventions: data during the same day as the observation, and
collection. In other situations, it may to be (c) assessing short-term maintenance effects of
important to investigate integrity of specific performance feedback.
intervention components for complex individ-
ualized behavior support plans.
METHOD
Two other aspects of plan implementation
Participants and Setting
that deserve consideration are (a) what ante-
Thepresentstudywasconductedinaprivate
cedent interventions are implemented to de-
school for students with acquired brain injury
crease the likelihood of problem behavior and
who exhibited significant behavior problems.
(b) the teacher’s response to students’ target
Students were all male and ranged in age
behaviors. Teachers may not accurately imple-
from 10 to 19 years. Three of the students
ment both types of procedures with equal inte-
hadnontraumaticacquiredbraininjuries,and2
grity because implementation of antecedent
students had been diagnosed with a traumatic
and consequence procedures may require
brain injury. The sample of students observed
different skills. Implementation of antecedent
comprised approximately 10% of the total
procedures requires teachers to plan and pre-
population of the school.
vent the occurrence of behavior (Luiselli &
Observational data were collected in two
Cameron, 1998), whereas consequence proce-
special education classrooms on a total of 5
dures require teachers to impose contingencies
teacher–student dyads. Two dyads (i.e., Mr.
after the occurrence of problem behaviors
Canton and Seth; Mr. Martin and Philip) were
(Kern, Choutka, & Skol, 2002).
identified from Classroom 1, and 3 dyads (i.e.,
The frequency and structure of performance
Ms. Lowell and Brian; Ms. Malden and Jason;
feedback have been a focus of several studies. Mr. Mack and Darrin) were identified from
For example, research has demonstrated that Classroom 2. All teachers participating in the
weekly feedback (Mortenson & Witt, 1998) study had earned a bachelor’s degree and were
leads to increases in treatment integrity and enrolled in a master’s level program in special
may be more practical for supervisors, clin- education.Teachers’ experience working in this
icians, and consultants than daily performance environmentrangedbetween6and30 months.
feedback. The schedule and immediacy of
performance feedback may also influence the Materials
maintenance of treatment integrity. For exam- Behavior support plans. Individualized be-
ple, in many studies performance feedback haviorsupportplanswerepreviouslycreatedfor
has consisted of reviewing data from the each student and were ongoing at the time of
previous day (Noell et al., 1997, 2002; Witt the investigation. The plans had been in place
et al., 1997) or week (Mortenson & Witt, for an average of 4 months at the time of the
1998), rather than providing feedback on the study. These plans consisted of individualized
same day. multicomponent interventions that prescribed
TREATMENT INTEGRITY 207
both antecedent and consequence procedures on shoulder, praise, conversation) contingent
for students’ targeted problem behaviors. on appropriate social behavior (e.g., sharing,
Teachers were expected to employ these be- complimenting peers) or planned ignoring
havior support plans under specific conditions following the occurrence of inappropriate
(e.g.,activities,timesof day,andcontingenton speech. Table 1 lists all the antecedent
student behaviors). Six behaviors were targeted and consequence components that were pre-
for Seth (tantrums, perseverative speech, in- scribed in various students’ behavior support
appropriate speech, inappropriate social behav- plans.
ior,teasing,andnoncompliance),Philip(major The integrity data sheet divided each plan
aggression, minor aggression, inappropriate into 10 to 13 individualized components (see
speech, inappropriate touching, public expo- the Appendix for Brian’s integrity data sheet;
sure, and instigation), and Brian (aggression, otherparticipants’datasheetsareavailablefrom
destruction, peer instigation, inappropriate the authors on request). One of three levels of
speech, inappropriate sexual behavior, and implementation integrity was scored under the
noncompliance). Four behaviors were targeted implementation rating section of the integrity
for Jason (tantrums, inappropriate speech, in- data sheet: (a) implemented as written (i.e., the
appropriate social behavior, and invasion of entire component was implemented every time
space), and six behaviors were targeted for the target behavior occurred or when the
Darrin (physical aggression, property destruc- situation required an antecedent component),
tion, wandering, inappropriate verbalizations, (b) not implemented as written (i.e., sometimes
mimicking, and noncompliance). implemented the entire component as written,
Integrity data sheet. The integrity data sheet implementedpartofthecomponent aswritten,
was three pages long and included (a) the type or did not employ the component as written),
of procedure (i.e., antecedent or consequence) and (c) no opportunity to observe (i.e., the target
that was prescribed for that particular student, behavior did not occur, or the antecedent was
(b) an operational definition of each compo- notpresent).Forexample,Ms.Maldenreceived
nent of the intervention (copied directly from a rating of implemented as written if she
each student’s behavior support plan), (c) provided all three transition warnings every
observer ratings of the teacher implementation time Jason was required to change from
of each component (see below), and (d) a space a preferred to a nonpreferred activity during
for the observer to record comments or the observation period. However, she received
examples.Eachcomponentconsistedofspecific a rating of not implemented as written if she
behaviors that teachers were instructed to provided the three warnings for only two of
engage in, either as a result of the activity a three opportunities that Jason had to change
student was required to perform or in response from a preferred to a nonpreferred activity or
to students’ behavior. For example, when asked if she provided only two transition warnings.
to change from a preferred to a nonpreferred Mr. Mack earned a rating of implemented as
activity, Ms. Malden was instructed to provide written if he blocked Darrin from leaving the
Jason with three transition warnings (i.e., at classroom, redirected him to the scheduled
2 min, 1 min, and 30 s). Another example of activity, and provided a warning that Darrin
an antecedent component specified that Mr. would lose a penny for every time that he
Mack explain all expectations to Darrin using attempted to elope from his classroom. If Mr.
first-then statements for every activity that he Mack did not carry out all parts of this
engaged in. Consequences included reinforce- component every time that Darrin attempted
mentsuchasreceivingsocialattention(e.g.,pat toelopefromhisclassroom,hereceivedarating
208 ROBIN S. CODDING et al.
Table 1
BehaviorSupport PlanComponents ClassifiedasAntecedents andConsequences
Classroom Antecedents Consequences
1 Transitionwarnings Self-monitoring
Explicitinstructions DRA
Modificationofpresentationofdemands Time-out
Breaksincorporatedintoworkperiods Plannedignoring
Choices Ignoreandredirect
Scheduledactivities(limitdowntime) Guidedcompliance
Modeling Contingentprotectivehold
Contingentescort
2 Scheduleofpreferredandnonpreferredtasks DRO
Transitionwarnings DRA
Explicitinstructions Plannedignoring
Modificationofpresentationofdemands Ignoreandredirect
Choice Blockandredirect
Preteaching Guidedcompliance
Breaksincorporatedintoworkperiods Responsecost
One-to-oneattention Contingentprotectivehold
Seatingwithfewdistracters Contingentescort
Note.DRO5 differential reinforcement ofother behavior;DRA 5differential reinforcement ofalternative behavior.
of not implemented as written. In the comments procedures of the observation and feedback
andexamplessectionoftheintegritydatasheet, sessions, and three items asked teachers to rate
consultants could provide an example of how the outcomes of the feedback sessions (e.g., was
the teacher implemented the component. it helpful to increase integrity of plan imple-
Percentage of correct implementation was mentation) (Schwartz & Baer, 1991). Respon-
used as the measure of treatment integrity, and dentswererequiredtorateeachitemonaLikert
was calculated by dividing the the number of scale ranging from 1 (strongly disagree) to 5
plan components that were implemented as (strongly agree). Teachers were instructed not to
written divided by the total number of plan place their names on the questionnaires, and
components teachers had the opportunity to aboxwasplacedinacommonareaforteachers
exhibit. The average percentage of the plan to return the forms.
components that teachers had the opportunity
to exhibit was 61% (range, 50% to 90%) for Training
Classroom1and57%(range,40%to77%)for All teachers had received both general and
Classroom 2 across teacher–student dyads. student-specific formal training in implement-
There were no opportunities for teachers to ing behavior support plans prior to this
exhibit all the components in the behavior investigation. After being hired at the school,
support plan in any single observation period. teachers received 4 hr of training in basic
It should be noted that consultants took notes principles of applied behavior analysis.
on specific deviations from the written pro- Teachers also received ongoing in-service train-
cedure. ing on general behavior principles four times
Social validity. A 10-item questionnaire that annually. One in-service session was conducted
examined the acceptability of the performance during the course of this project, and the topic
feedback and corresponding integrity observa- (i.e., task analysis) was unrelated to these
tions was administered to each teacher after students’ behavior support plans. Teachers
performance feedback was terminated. Three received2 weeksofspecifictrainingonindivid-
items addressed the purpose of performance ual students’ plans at the time they were
feedback, four items inquired about the developed. This training included reviewing
TREATMENT INTEGRITY 209
the written plan with a consultant (i.e., one of were not told the purpose of the observation
the authors), modeling by the consultant in the during baseline. This was not unusual, because
classroom with the target student, prompting observationswereaconsistentpartoftheduties
the teacher-in-training to employ the compo- of experimenters.
nents as written, and immediate performance Intervention. Performance feedback was im-
feedbackgivenbytheconsultant.Teacherswere plemented after stable or decreasing perfor-
not trained to a standardized criterion; howev- mance in baseline was demonstrated by either
er, training continued until they could verbally the percentage of antecedent components or
report each component of the plan and consequence components implemented as writ-
reported that he or she could implement the ten. The first feedback session was provided to
plan. In addition, consultants were present in
each teacher after the last baseline session to
the classrooms on a daily basis and provided
address inadequate plan implementation as
informal feedback to teachers.
soon as possible for the benefit of the student.
On the same day as each observation, the
Procedure
experimenter spent an average of 12 min with
Observation sessions. Observations were con-
the target teacher outside the classroom. At this
ducted an average of every 2.1 weeks for each
time, components in the behavior support plan
dyad (range, 1 to 3 weeks). Variations in the
were reviewed, and feedback was provided on
time between observation sessions reflected
all the components that were observed.
teachers’ earned leave time and school breaks.
Feedback included providing praisefor compo-
Each observation was planned to be 60 min
nents followed as written and constructive
long. There were five occasions on which
feedback for those components that were
observations were less than 60 min (range, 45
followed sometimes or not at all. Constructive
to 55 min). The length of time for each
feedback consisted of reviewing the specific
observation session and time between observa-
components observed and explaining how the
tion sessions remained consistent throughout
component should have been implemented.
the baseline, intervention, and maintenance
phases of data collection. As recommended by Performance feedback was terminated after
Gresham et al. (1993), we attempted to reduce improved performance had stabilized. Two
reactivity by conducting observations on a vari- maintenance sessions were conducted for Mr.
able-time schedule and used consultants who Canton and Mr. Martin, three maintenance
routinely worked in the classrooms as primary sessions were conducted for Ms. Lowell and
observers. The consultant assigned to Class- Ms. Malden, and one maintenance session was
room 1 spent 50%of his totaltime conducting conductedforMr. Mack. Maintenancesessions
observations. Nine percent of the consultant’s were identical to baseline sessions. The first
time in Classroom 1 was devoted to this maintenance session occurred 5 weeks after the
project. The consultant assigned to Classroom last feedback session. Each subsequent follow-
2 devoted 33% of her time to conducting up session occurred at 5-week intervals across
observations. Six percent of the consultant’s dyads.
time in Classroom 2 was devoted to this Experimental design. A concurrent multiple
project. baseline across teacher–student dyads design
Baseline. Baseline consisted of observing was selected for each classroom to evaluate the
each student–teacher dyad and completing the efficacy of performance feedback. Maintenance
integrity data sheet without the teachers’ sessions were employed to examine the short-
knowledge of the observation. Feedback was term effects of the intervention when feedback
provided following the observation. Teachers was no longer provided.
210 ROBIN S. CODDING et al.
Interobserver Agreement second observer and the primary observer’s
Interobserver agreement was collected across responses. Agreement was calculated by di-
20% of the sessions. A second independent viding the number of agreements per compo-
observer was present and used the same nent by the number of agreements plus
treatmentintegrityformtoobservetheteachers disagreements per component and multiplying
directly. On a component-by-component basis, by 100%. Mean agreement across dyads was
comparisons were conducted between the 95% (range, 91% to 100%).
Figure 1. Percentage of antecedent and consequence components implemented as written across teacher–student
dyadsfor Classroom 1.
TREATMENT INTEGRITY 211
RESULTS performance feedback, substantial improve-
ments were demonstrated in the percentage of
Teacher Implementation of the Behavior
correctly implemented antecedent (M 5 78%)
Support Plan
and consequence components (M 5 95%).
Classroom 1. Figure 1 presents the percent-
During baseline, Ms. Malden’s correct
age of antecedent and consequence compo-
implementation of antecedent components
nents implemented as written (correctly
was stable (M 5 47%). Correct implementa-
implemented) for 2 student–teacher dyads in
tion of consequence components appeared to
Classroom 1. During baseline, a decreasing
be increasing in baseline (M 5 27%). Every
trend was evident for the percentage of con-
attempt was made to have stable or decreasing
sequence components that Mr. Canton imple-
trends in baseline prior to implementation of
mentedas written (M5 31%).Performanceof
performance feedback. However, a clinical de-
antecedent components implemented correctly
cision was made to employ performance
was stable (M 5 69%) during baseline.
feedback when the implementation of anteced-
Following performance feedback there was an
ent components was stable. Following perfor-
immediate increase in the percentage of
mance feedback, correct implementation of
consequence components correctly implemen-
antecedent (M 5 95%) and consequence
ted (M 5 92%) and a gradual increase in the
(M 5 95%) components improved.
percentage of antecedent components correctly
The sequence of baseline and performance
implemented (M 5 92%).
feedback was varied for Mr. Mack and Darrin.
During baseline, Mr. Martin’s performance
During the initial baseline phase, Mr. Mack
was variable for the correct implementation of
displayed stable performance for the imple-
antecedent and consequence components. Mr.
mentation of antecedent components as writ-
Martin correctly implemented 100% (eight
ten.Mr.Mackdidnotcorrectlyimplementany
sessions), 50% (three sessions), or none (two
consequence components. Following the initial
sessions) of the consequence components dur-
baseline,anincreaseinthefrequencyandinten-
ingbaseline,whichmakesthesedatadifficultto
sity of Darrin’s physical aggression required
interpret. The mean percentage of components
several revisions in his behavior support plan.
correctly implemented during baseline was
Following Revision 1, there was an increase in
85% for antecedents and 73% for conse-
the correct implementation of consequence
quences. Following performance feedback,
Mr. Martin demonstrated 100% correct imple- components, and antecedent performance
mentation across consequence components for remained at baseline levels. Revision 2 encom-
three consecutive sessions (M 5 100%). There passedbothbaselineandperformancefeedback.
was little change in Mr. Martin’s correct Performancefeedbackresulted inan increase in
implementation of antecedent components the percentage of antecedent and consequence
(M 5 89%). components correctly implemented. Improve-
Classroom 2. Figure 2 presents the percent- ments in performance were maintained during
age of antecedent and consequence components Revision 3 across antecedent and consequence
implemented as written for 3 student–teacher components.
dyads in Classroom 2. During baseline, a de- Maintenance observations. Treatment integ-
creasing trend was evident for the percentage of rity was maintained at high rates during the
antecedent components correctly implemented follow-up phase across all teachers (Figures 1
by Ms. Lowell (M 5 40%). Ms. Lowell did and 2). The percentages of antecedent (M 5
not correctly implement any consequence com- 83%) and consequence (M 5 100%) compo-
ponents during baseline (M 5 0%). Following nents correctly implemented were maintained
212 ROBIN S. CODDING et al.
Figure 2. Percentage of antecedent and consequence components implemented as written across teacher–student
dyadsfor Classroom 2.
5 and 10 weeks following the termination of and consequence (M 5 100%) components 5
performancefeedbackforMr.Canton.ForMr. and 10 weeks after feedback. For Ms. Lowell,
Martin, high levels of correct implementation high percentages of correctly implemented
were maintained across antecedent (M 5 83%) antecedent (M 5 89%) and consequence
TREATMENT INTEGRITY 213
(M592%)components were maintainedat 5, Performance feedback resulted in greater
10, and 15 weeks without performance feed- percentages of both antecedent and conse-
back. Ms. Malden also implemented high quence components correctly implemented for
percentages of antecedent (M 5 92%) and 4 of 5 teachers. For the 5th teacher (Mr.
consequence (M 5 100%) components as Martin), performance feedback resulted in
written during three follow-up observations increases in correct implementation of conse-
(i.e., 5, 10, and 15 weeks after intervention). quence components. These data also illustrate
Five weeks after performance feedback was that during baseline, 4 of the 5 teachers
terminated, Mr. Mack correctly implemented correctly implemented more antecedent than
100% of antecedent and consequence com- consequence components, although implemen-
ponents. tation was variable in some cases and at lower
rates than desirable. In the absence of perfor-
Social Validity
mance feedback, teachers may have more
Average ratings across all teachers ranged
successfully implemented antecedent compo-
from 4.5 to 5.0 for each item. For the purpose
nents because these interventions were similar
of feedback sessions, intervention procedures,
acrossstudents.Forexample,inClassroom2,3
and intervention outcomes, mean ratings of
students had daily scheduling with preferred
4.8, 4.7, and 4.9, respectively, were obtained,
following nonpreferred activities and choices
indicating that teachers strongly agreed with
built into all tasks, 2 students needed demands
items related to the purpose and procedures as
to be presented in individualized ways, and 2
well as the benefits of the intervention out-
studentsneededtransitionwarnings.Asaresult,
comesontheirskillsandthesubsequentimpact
these antecedent procedures may have operated
on their students.
as part of the daily routine. In addition, the
complexity (e.g., various schedules and types of
DISCUSSION reinforcement) and number of consequence
procedures may account for lower treatment
The results revealed that accurate implemen-
tation of ongoing behavior support plans integrity of consequence components in base-
improved across 5 student–teacher dyads fol- line. Which consequence components were
lowing performance feedback. This research more likely to be correctly employed (e.g.,
extends the literature by demonstrating the positive vs. negative reinforcement) was not
efficacy of performance feedback on the examined; however, research has previously
treatmentintegrityof individualized multicom- demonstrated that teachers tend to focus their
ponent behavior support plans in a special attention on students’ inappropriate behavior
education setting. This study also showed (Cooper,Thomson,&Baer,1970).Futureres-
that the results of performance feedback were earch might consider assessing teachers’ imple-
maintained for up to 15 weeks. In contrast to mentationofspecificconsequencecomponents.
other studies (Mortenson & Witt, 1998; Differences in implementation of antecedent
Noell et al., 1997, 2000; Witt et al., 1997), and consequence procedures evident in this
treatment integrity was assessed using direct study may have interesting implications for the
observation, and performance feedback was types of interventions used to increase teachers’
provided every other week rather than daily treatment integrity. These findings suggest that
(Noell et al., 1997, 2000; Witt et al., 1997) future studies should continue to examine the
or weekly (Mortenson & Witt, 1998) and integrity of individual treatment components
on the same day that the observation (Gresham et al., 1993). Analysis of whether
occurred. treatment integrity is a function of teacher
214 ROBIN S. CODDING et al.
skill or motivational deficits is also a worthy that these results would be maintained over the
topic for investigation. For example, exploring long term. We suspect that periodic collection
whether treatment integrity increases as a func- of treatment integrity data and subsequent
tion of performance feedback, teachers’ role in performance feedback are necessary for high
intervention planning, contingent reinforce- rates of intervention integrity to persist.
ment, training, or a combination of these may The importance of investigating the social
begin to address whether treatment integrity is validity of performance feedback has evolved
a result of a skill or motivational deficit. fromsuggestions(Mortenson&Witt,1998)that
Despite its documented success, the potency feedback may serve as a negatively reinforc-
of performance feedback has been criticized for ingeventforsometeachers.Alloftheteachersin
several reasons, including the following: (a) the current study rated performance feedback
Results have not been maintained at high rates favorablywithrespecttothepurpose,procedures,
(Mortenson & Witt, 1998; Witt et al., 1997). andoutcome;thisisconsistentwiththefindings
(b) It may be time consuming (Mortenson & ofNoelletal.(2002).Assessingtheacceptability
Witt, 1998). (c) There may be a lack of of performance feedback provides a prelim-
acceptability by change agents (e.g., teachers) inary attempt to identify how teachers per-
(Mortenson & Witt, 1998) as well as (d) ceive treatment integrity interventions. Future
reactivity to observers (Mortenson & Witt, researchers may consider experimentally analyz-
1998; Witt et al., 1997). To increase the ing the function that performance feedback
sophistication of treatment-integrity interven- serves for teachers, rather than relying on verbal
tions and specifically performance feedback, report.
attempts to address these concerns were The current investigation has several limita-
included in the current study. tions. First, it is possible that treatment effects
Improvements in the integrity of behavior were confounded by reactivity to being
support plan implementation were maintained observed. Consistent with recommendations
in the current study. This may be due to by Gresham et al. (1993), we attempted to
the length of time performance feedback reducereactivitybyconductingobservationson
was employed and the latency between ses- a variable-time schedule and used consultants
sions. Previous research implemented perfor- who had been assigned to these classrooms as
mance feedback for approximately 6 weeks primary observers. Therefore, it was common
(Mortenson & Witt, 1998; Noell et al., 2000) for teachers to see these consultants in the
compared to 8 to 22 weeks in our study. classroom conducting observations. Although
Whereas other studies utilized daily (Noell we attempted to make these observations un-
et al., 1997, 2002; Witt et al., 1997) or weekly obtrusive, teachers were aware during the
(Mortenson & Witt, 1998) performance feed- intervention phase that they were being
back, we provided feedback every other week. observed. A second limitation of the study is
Importantto noteisthat performancefeedback that implementation of performance feedback
was provided during the same days that direct occurred when either antecedent or conse-
observations occurred. Consequently, recent quence components were stable or decreasing.
examples of teachers’ behavior could be As a result, experimental control may not be
reviewed during feedback sessions. This sched- as robust for Ms. Malden, whose correct
ule of performance feedback may lead to the implementation of consequence components
maintenance of treatment effects in the short appeared to be improving in baseline. Every
term and is time efficient and practical for attempt was made to have stable or decreasing
clinicianstoimplement.However,itisunlikely trends in baseline prior to feedback, but