Content uploaded by SungWoo Kahng
Author content
All content in this area was uploaded by SungWoo Kahng
Content may be subject to copyright.
181
JOURNAL OF APPLIED BEHAVIOR ANALYSIS
2000, 33, 181–194
NUMBER
2(
SUMMER
2000)
SKILL ACQUISITION IN THE IMPLEMENTATION OF
FUNCTIONAL ANALYSIS METHODOLOGY
B
RIAN
A. I
WATA
,M
ICHELE
D. W
ALLACE
,S
UNG
W
OO
K
AHNG
,
J
ANA
S. L
INDBERG
,E
ILEEN
M. R
OSCOE
,J
ULIET
C
ONNERS
,
G
REGORY
P. H
ANLEY
,R
ACHEL
H. T
HOMPSON
,
AND
A
PRIL
S. W
ORSDELL
THE UNIVERSITY OF FLORIDA
Functional analysis methodology is a powerful assessment tool for identifying contingen-
cies that maintain a wide range of behavior disorders and for developing effective treat-
ment programs. Nevertheless, concerns have been raised about the feasibility of con-
ducting functional analyses in typical service settings. In this study, we examined the issue
of skill acquisition in implementing functional analyses by evaluating an instructional
program designed to establish a basic set of competencies. Eleven undergraduate students
enrolled in a laboratory course in applied behavior analysis served as participants. Their
performance was assessed during scripted simulations in which they played the roles of
‘‘therapists’’ who conducted functional analyses and trained graduate students played the
roles of ‘‘clients’’ who emitted self-injurious and destructive behaviors. To approximate
conditions under which an individual might conduct an assessment with limited prior
training, participants read a brief set of materials prior to conducting baseline sessions.
A multiple baseline design was used to assess the effects of training, which consisted of
reading additional materials, watching a videotaped simulation demonstrating correct
procedural implementation, passing a written quiz, and receiving feedback on perfor-
mance during sessions. Results showed that participants scored a relatively high percentage
of correct therapist responses during baseline, and that all achieved an accuracy level of
95% or higher following training that lasted about 2 hr. These results suggest that basic
skills for conducting functional analyses can be acquired quickly by individuals who have
relatively little clinical experience.
DESCRIPTORS: assessment, functional analysis, staff training
Assessment procedures that identify the
functional characteristics of behavior are
powerful tools for identifying sources of re-
inforcement that maintain behavior disor-
ders and for developing subsequent treat-
ment programs. In addition to having gen-
erated a great deal of interest in the research
literature (e.g., see special issue of the Journal
of Applied Behavior Analysis, Vol. 27, No. 2,
1994), these approaches to assessment have
been mandated in recent court decisions and
legislation on behalf of persons with devel-
opmental disabilities who have severe behav-
This research was supported in part by a grant from
the Florida Department of Children and Families.
SungWoo Kahng is now at the Kennedy Krieger In-
stitute, Johns Hopkins University School of Medicine.
Reprints may be obtained from Brian Iwata, Psy-
chology Department, The University of Florida,
Gainesville, Florida 32611.
ior disorders (e.g., Individuals with Disabil-
ities Education Act, 1975/1997). Thus,
functional approaches to behavioral assess-
ment have come to be regarded as best prac-
tice in both clinical research and application.
Many techniques have been subsumed
under the general label functional assessment,
including indirect methods such as inter-
views and questionnaires, descriptive analy-
ses via direct observation, and systematic
manipulations of environmental conditions
(for reviews, see Iwata, Vollmer, & Zarcone,
1990; Mace, Lalli, & Lalli, 1991). The most
precise methods for identifying behavioral
function are derived from procedures that
are commonly used in applied behavior anal-
ysis and are characterized by objective mea-
sures of ongoing behavior taken under mul-
tiple test and control conditions, in which
182 BRIAN A. IWATA et al.
antecedent and consequent events are clearly
prescribed and are arranged in such a way as
to identify functional relations between en-
vironment and behavior. These latter meth-
ods are commonly referred to as ‘‘experi-
mental analysis’’ or ‘‘functional analysis’’
methodologies, and their utility has been
demonstrated repeatedly over the past 15
years.
Nevertheless, concerns have been raised
about the feasibility of conducting function-
al analyses in typical service settings. The
most commonly cited limitations of func-
tional analyses are the fact that they may be
lengthy (Applegate, Matson, & Cherry,
1999; Durand & Crimmins, 1988; Pyles,
Riordan, & Bailey, 1997; Sturmey, 1994)
and that their implementation may require
a considerable amount of training and clin-
ical expertise (Crawford, Brockel, Schauss,
& Miltenberger, 1992; Durand & Crim-
mins, 1988; Spreat & Connelly, 1996; Stur-
mey, 1994). Although conducting a func-
tional analysis is a more lengthy process than
is administering a questionnaire, it is not
necessarily any more time consuming than
is performing a descriptive analysis (Lerman
& Iwata, 1993). Furthermore, recent re-
search has yielded several ways for increasing
the efficiency of functional analyses by con-
ducting either fewer (Northup et al., 1991)
or briefer (Wallace & Iwata, 1999) sessions.
By contrast, the issue of procedural com-
plexity has not been addressed systematically.
Thus, suggestions that ‘‘many staff members
in positions to develop such programs lack
the professional training to conduct such
complex analyses’’ (Spreat & Connelly,
1996, p. 528) remain largely unanswered.
A logical basis for routinely teaching staff
how to conduct functional analyses can be
found through consideration of the skills
needed to implement assessment and treat-
ment programs. A functional analysis re-
quires the ability to deliver a prescribed se-
quence of antecedent and consequent events
while interacting with a client, and it could
be argued that staff who lack this skill would
also be unable to implement most behavioral
interventions with any degree of consistency.
Logical arguments aside, little is known
about the extent to which (or the speed with
which) staff can acquire a basic set of skills
for conducting functional analyses. Staff
training has rarely been a topic of empirical
investigation in research on the assessment
and treatment of behavior disorders because
a high degree of competence is usually a pre-
requisite for conducting a study. In other
words, implementation errors in clinical re-
search are typically minimized as unwanted
sources of risk to both clients and staff as
well as potential sources of experimental
confounding and, as a result, have not been
examined systematically as dependent vari-
ables (see Shore, Iwata, Vollmer, Lerman, &
Zarcone, 1995, as an exception).
In this study, we evaluated the effects of
a training program designed to establish ba-
sic competence in conducting functional
analyses. Of particular interest were (a) the
extent to which untrained individuals could
implement a functional analysis with mini-
mal instruction, and (b) the amount of
training that was necessary to produce a high
degree of technical accuracy.
METHOD
Participants and Setting
Therapists. Although staffing arrange-
ments vary across settings, the entry-level in-
dividual responsible for conducting assess-
ments, developing service plans, and imple-
menting interventions is usually the teacher
or the ‘‘qualified mental retardation profes-
sional’’ with a BA degree. Therefore, we se-
lected as therapists upper-level undergradu-
ate students who were representative of the
types of staff who might be expected to con-
duct functional analyses. The participants
were 11 students (10 women and 1 man)
183FUNCTIONAL ANALYSIS METHODOLOGY
who were enrolled in an applied behavior
analysis laboratory course. All participants
were junior or senior psychology majors who
had one prior course in behavior analysis but
no practicum experience in behavior analysis
and no experience in the use of functional
analysis methodology. The study was con-
ducted at the beginning of the term, before
participants had the opportunity to observe
ongoing assessment (functional analysis) or
treatment sessions. Although participants
were required to complete this project as
part of their course activities, their perfor-
mance in the study did not count toward
their course grades. All participants provided
written informed consent for the use of their
data as part of a research project.
Clients. In order to collect data on the
performance of therapists who were relative-
ly untrained, it was necessary to create a con-
text for conducting functional analyses that
did not place either therapists or actual cli-
ents who had severe behavior problems at
risk. Therefore, 8 graduate students, all of
whom had extensive experience conducting
functional analyses, played the roles of cli-
ents throughout the study.
Settings. Training sessions for therapists
were conducted either in a university class-
room or in various therapy rooms of a day-
treatment program located on the grounds
of a state residential facility for persons with
developmental disabilities, where most of the
students’ laboratory work was conducted. All
functional analysis sessions were conducted
at the day program.
Functional Analysis Conditions
The experimental task involved imple-
mentation of procedures that typically com-
prise a functional analysis. Three assessment
conditions (attention, demand, and play)
were selected as representative of those com-
monly used in functional analyses involving
the manipulation of antecedent and conse-
quent events associated with behavior dis-
orders (see Iwata, Dorsey, Slifer, Bauman, &
Richman, 1982/1994). A fourth condi-
tion—alone—was deleted because it did not
require the presence of a therapist. Correct
implementation of the assessment conditions
involved the delivery and removal of pre-
scribed antecedent and consequent events by
a therapist, as indicated below. (Note that
condition descriptions depict ideal respons-
es, which were not necessarily observed dur-
ing baseline.)
During the attention condition, the client
was given free access to several leisure items
throughout the session. The therapist ig-
nored the client throughout the session, ex-
cept to deliver attention in the form of state-
ments of concern (e.g., ‘‘Stop, don’t do that;
you’ll hurt yourself’’) and brief physical con-
tact (e.g., a pat on the back) following each
occurrence of the target behavior. During
the demand condition, the therapist pre-
sented learning trials to the client through-
out the session, which were initiated at 30-
s intervals using a three-prompt sequence
(verbal instruction, instruction plus demon-
stration, instruction plus physical prompt).
The therapist delivered praise if the client
complied, continued the prompting se-
quence if the client did not comply, and ter-
minated the trial (by removing the task ma-
terials and turning away from the client until
the next trial) if the client exhibited a target
behavior at any time during the trial. During
the play condition, the client had free access
to several leisure items throughout the ses-
sion. The therapist delivered attention on a
fixed-time (FT) 30-s schedule throughout
the session, responded to any client-initiated
appropriate social interactions, and ignored
all occurrences of inappropriate behavior
(target and nontarget). If the client was en-
gaged in inappropriate behavior when the
FT schedule called for the delivery of atten-
tion, attention was delayed until inappro-
priate behavior had ceased for 5 s.
184 BRIAN A. IWATA et al.
Table 1
Sample Scenario for the Attention Condition
Time
(minutes:
seconds)
Scripted ‘‘client’’ behavior
Response class Topography
0:16
0:18
0:30
0:43
0:56
1:21
1:50
2:01
2:12
2:22
2:33
2:44
2:58
3:35
4:02
4:12
4:23
4:35
4:45
4:56
SIB
SIB
SIB
SIB
SIB
Disruption
Appropriate play
SIB
SIB
SIB
SIB
Social initiation
SIB
Disruption
SIB
SIB
SIB
SIB
Appropriate play
SIB
Hit self
Hit self
Hit self
Hit self
Hit self
Kick wall
Put toy in therapist’s
hand
Hit self
Hit self
Hit self
Hit self
Ask ‘‘Can we play?’’
Hit self
Throw chair
Hit self
Hit self
Hit self
Hit self
Tug therapist’s sleeve
with toy in hand
Hit self
Assessment Simulations
We developed a series of scripts that spec-
ified the occurrence of various client behav-
iors during 5-min simulated assessment ses-
sions representing each of the three condi-
tions described above. The behaviors exhib-
ited by clients included (a) self-injurious
behavior (SIB), which was the target behav-
ior during the functional analysis and con-
sisted of striking an arm, leg, or part of the
torso with the hand; (b) disruption, a non-
target problem behavior, which consisted of
kicking a wall; kicking, throwing, or knock-
ing over furniture; or jumping up and down
while screaming; (c) appropriate play, which
consisted of either manipulation of leisure
materials in a nondestructive manner or at-
tempts to hand the materials to the thera-
pist; (d) compliance with an instruction giv-
en by the therapist; and (e) appropriate ini-
tiation of social interaction, which consisted
of asking the therapist to play or tugging on
the therapist’s sleeve. Fifteen scenarios were
created, five each for the attention, demand,
and play conditions. Each scenario con-
tained the same number and distribution of
client behaviors; however, both the sequence
and temporal occurrence of these behaviors
throughout a session were varied from one
scenario to the next. For example, each sce-
nario called for 15 occurrences of SIB, which
occurred at different times across scenarios.
A sample scenario for an attention session is
provided in Table 1.
Target Therapist Behaviors
The behaviors of primary interest were
the performances of the undergraduate stu-
dents, who played the roles of therapists
while interacting with clients during the
simulated functional analyses. The designat-
ed therapist behaviors consisted of (a) atten-
tion, defined as any social interaction, in-
cluding reprimands, unrelated to a specific
task; and (b) instruction, defined as a verbal
directive to perform a task, delivered either
with or without a supplemental prompt.
These two behaviors generated a larger num-
ber of scoring categories because both could
be initiated or terminated as either anteced-
ent or consequent events with respect to cli-
ent behavior. Therapist behaviors were
scored as either correct or incorrect based on
their occurrence, nonoccurrence, or termi-
nation relative to either a prescribed tem-
poral sequence or the occurrence (or non-
occurrence) of client behavior. For example,
the delivery of an antecedent event (e.g.,
noncontingent attention during the play
condition or instructions during the demand
condition) was scored as correct if it oc-
curred at the appropriate time or as incorrect
if it did not occur. The delivery of conse-
quences (e.g., contingent attention during
the attention condition or termination of a
trial during the demand condition) was
scored as correct if it followed a designated
185FUNCTIONAL ANALYSIS METHODOLOGY
client behavior within 5 s or as incorrect if
it did not occur following a target client be-
havior. Finally, the nondelivery of conse-
quences (e.g., the absence of attention for
appropriate behavior during the attention
condition) was scored as correct if it did not
occur or as incorrect if it did occur.
Response Measurement and Reliability
All sessions lasted for 5 min and were vid-
eotaped and scored later by graduate stu-
dents, who recorded occurrences of therapist
and client behaviors using a 10-s partial-in-
terval recording procedure. Client behaviors
were scored during all sessions, and compar-
isons of these data with sequences specified
in the scripts provided a measure of proce-
dural fidelity; these comparisons always ex-
ceeded 90% accuracy. A second observer
scored 26% of the videotapes to collect in-
dependent data on the occurrence of thera-
pist behaviors. Observers’ records were com-
pared on an interval-by-interval basis, and
percentage agreement scores were calculated
by dividing the number of scoring agree-
ments (for the occurrence or nonoccurrence
of a behavior) by the number of agreements
plus disagreements, and multiplying by
100%. The mean agreement score for ther-
apist behaviors was 93.9% (range, 86.0% to
100%).
Baseline
Although exposure to any training during
baseline would preclude the assessment of
naive performance, we presumed that such
performance would be uninformative from a
practical standpoint because it is unlikely
that anyone would conduct a functional
analysis with no information whatsoever.
Thus, the baseline was constructed to ap-
proximate a situation in which a therapist
who was attempting to conduct a functional
analysis had access to published material but
received no formal instruction, assistance, or
feedback.
Participants were given the method sec-
tion of the Iwata et al. (1982/1994) article
and were asked to read it several days prior
to initiating the collection of baseline data
(the exact number of days varied from 1 to
5 because of scheduling differences). On the
first day of baseline, participants were given
an additional 30 min to study the materials,
after which they were asked to serve as ther-
apists while conducting functional analysis
sessions (attention, demand, and play con-
ditions) with graduate students playing the
roles of clients. Prior to each baseline ses-
sion, participants were told which type of
assessment condition they were to conduct
but were given no further instructions. Par-
ticipants received no feedback on their per-
formance following baseline sessions.
Training
Training was conducted by one or more
graduate students in two phases. The first
phase was conducted in a group format in a
university classroom. Participants were first
given written descriptions and outlines of
the assessment conditions to read. The de-
scriptions were brief summaries based on the
Iwata et al. (1982/1994) article that high-
lighted salient components: the purpose of a
condition, target behaviors, and how to con-
duct a session. Descriptions used for each
condition are contained in Appendix A, and
their corresponding outlines are contained in
Appendix B. A graduate student then re-
viewed the key components of each condi-
tion and showed a videotaped simulation of
each condition, in which one graduate stu-
dent played the role of therapist while an-
other played the role of client. Following this
exercise, participants took a 20-item written
quiz containing simple factual questions
about the assessment process (a sample quiz
can be found in Appendix C). If a partici-
pant scored below 90% correct on the quiz,
a graduate student reviewed the quiz answers
with the participant, showed the videotape
186 BRIAN A. IWATA et al.
again, and administered another quiz. This
phase of training was complete when all par-
ticipants scored 90% correct or higher on a
quiz. All participants achieved a passing
score on either their first or second quiz, and
all training activities in this phase were com-
pleted in approximately 1.5 hr.
Following the classroom training, partici-
pants again conducted functional analysis
sessions as in baseline, with two exceptions.
These two changes from baseline procedures
represented the second phase of training,
which was conducted at the day program.
First, participants were allowed to bring the
outline into a session, which they could use
as a cue. Second, they were given feedback
on their performance by a graduate student
immediately following each set of three ses-
sions (one session of each condition). If a
participant scored less than 95% correct re-
sponses during a session, the graduate stu-
dent replayed the videotape of that session
while pointing out both correct and incor-
rect aspects of the participant’s performance.
This sequence continued until a participant
completed two consecutive sessions of each
of the three conditions (excluding the first
posttraining session of each condition) at or
above 95% accuracy. The total duration of
training in this second phase (including
feedback and repeat viewing of videotapes
but excluding actual session time) varied
across participants from 10 min to 30 min.
Experimental Design
Participants conducted simulated func-
tional analysis sessions, as described above,
in a fixed repeating sequence (attention,
play, demand). The effects of training were
evaluated in a multiple baseline across sub-
jects design. Five participants conducted
three sessions (one set of functional analysis
conditions) under baseline conditions prior
to receiving training; the remaining 6 par-
ticipants conducted six sessions (two sets of
functional analysis conditions) prior to re-
ceiving training.
RESULTS
Figure 1 shows participants’ performance
during the simulated functional analysis ses-
sions, expressed as the percentage of correct
therapist responses. Baseline performances
were generally high (M
5
69.9%; range of
individual means, 50.0% to 89.5%), al-
though the data reflect a great deal of vari-
ability. In addition, the performances of sev-
eral participants, especially Carrie, Cindee,
and Ricki, showed noticeable upward trends
during baseline. All participants improved
following training (M
5
97.5%; range of
individual means, 92.1% to 100%) and
completed the study after conducting the
minimum number of posttraining sessions.
A comparison of individual performances
across conditions showed that none of their
baseline means exceeded 90%, whereas all of
their posttraining means exceeded 95%,
with the exception Karen, whose posttrain-
ing mean was 92.1% because of one low
score (her first demand condition following
training). Across all participants, 33 of the
51 baseline scores were below 80%; by con-
trast, 69 of the 99 posttraining scores were
100%.
DISCUSSION
It has been suggested that the skills re-
quired to conduct functional analyses of be-
havior disorders exceed those of staff who are
typically responsible for performing assess-
ments and, furthermore, that the training
necessary to develop such skills may be ex-
cessive. We examined these issues in the
present study by observing the performance
of undergraduate students playing the roles
of therapists while conducting simulated
functional analyses with graduate students
playing the roles of clients who exhibited
187FUNCTIONAL ANALYSIS METHODOLOGY
Figure 1. Percentage of correct therapist responses during simulated functional analysis sessions across base-
line and posttraining conditions.
problem behaviors. After reading excerpts of
previously published research describing one
method for conducting a functional analysis,
participants averaged about 70% correct re-
sponses during baseline sessions. Subse-
quently, all participants achieved a 95% ac-
curacy criterion after receiving about 2 hr of
training, which consisted of reading addi-
188 BRIAN A. IWATA et al.
tional materials, watching a videotape, pass-
ing a quiz, and receiving feedback on their
performance. These results suggest that staff
with academic preparation equivalent to that
of a BA degree can quickly acquire a basic
set of skills for conducting functional anal-
yses.
The fact that our participants exhibited a
high percentage of correct responses during
baseline deserves further comment. First, be-
cause their baseline performance was not
completely naive, it was probably much
higher than that of individuals who would
attempt to conduct a functional analysis
with absolutely no prior instruction. As not-
ed earlier, however, we felt that truly naive
performances would provide highly unrep-
resentative samples of behavior likely to be
exhibited by staff who actually attempt to
conduct functional analyses. Second, the
amount of information participants received
prior to conducting baseline sessions was
limited to brief exposure to a single source
(Iwata et al., 1982/1994). We selected the
source because it is frequently cited in the
literature and because it serves as the basis
for functional analyses conducted in our
program. In reality, it is likely that staff who
attempt to initiate assessments based on
functional analysis methodology would do
so only after having read more extensively.
In spite of this limited exposure, participants
performed rather well during the baseline
sessions. Third, the performance of several
participants showed increasing trends during
baseline. We predicted (somewhat inaccu-
rately) that participants’ baseline perfor-
mance would be uniformly poor and would
not improve; therefore, we arbitrarily set the
baseline lengths at the beginning of the
study. It would have been preferable to con-
tinue taking baseline data until trends sta-
bilized, but this was not possible from a
practical standpoint. Because the partici-
pants’ laboratory responsibilities included
observing and, in some cases, conducting
functional analyses, we postponed these ac-
tivities as long as possible at the beginning
of the term to insure that sources of infor-
mation (e.g., additional readings, extended
observation of sessions) other than those de-
scribed as training in this study would not
influence the data. Nevertheless, the baseline
trends represent a weakness in the study and
raise the possibility that some participants
eventually may have achieved the posttrain-
ing performance criterion with no additional
instruction. All of these factors suggest that
participants did reasonably well in their ini-
tial attempts to conduct sessions with min-
imal exposure to information (access to read-
ing material) and that they may have im-
proved further with practice.
The training procedures used in this study
were neither novel nor complex. All of the
instructional components, including written
materials, video simulations, quizzes, and
performance feedback, are common features
of many staff training curricula (Reid, Par-
sons, & Green, 1989) and, with the excep-
tion of the quizzes, bear close resemblance
to training programs implemented with
paraprofessional staff such as direct-care ser-
vice providers (e.g., Ducharme & Feldman,
1992) and respite-care workers (e.g., Neef,
Trachtenberg, Loeb, & Sterner, 1991). In
addition, graduate students prepared all of
the training materials used in this study and
conducted all of the training sessions. Thus,
the level and extent of training provided in
this study were probably representative of
those available in many service settings that
employ behavior analysts with advanced de-
grees.
Although the participants in this study ac-
quired a very high degree of proficiency in
conducting functional analysis sessions, the
findings are limited in two significant re-
spects. First, all performance was assessed
under simulated conditions. Although re-
sults from several studies suggest that staff
training conducted under simulated condi-
189FUNCTIONAL ANALYSIS METHODOLOGY
tions generalizes to clinical situations (e.g.,
Ducharme & Feldman, 1992; Jones & Ei-
mers, 1975; Neef et al., 1991), the extent to
which our participants could adequately
conduct functional analyses with actual cli-
ents immediately following training is un-
known. We did not attempt to measure per-
formance under actual clinical conditions for
two reasons. First, unlike other research on
staff training, in which participants (train-
ees) were actual service providers or in which
interventions involved teaching adaptive be-
havior to clients, the participants in this
study had virtually no prior experience
working with any clinical population, and
the actual clients all exhibited behavior
problems (i.e., SIB or aggression) that placed
either them or therapists at risk. Therefore,
the use of performance probes with actual
clients immediately after training was
deemed unacceptable. Second, although a
number of participants eventually did con-
duct functional analysis sessions during the
course of their laboratory work, several in-
tervening educational experiences (e.g., read-
ing a series of related articles, collecting pri-
mary and reliability data from assessment
and treatment sessions, conducting prefer-
ence assessments with clients) had transpired
since the completion of training provided in
this study, making it difficult to identify the
necessary components required to produce
competence under clinical conditions. As an
extension of the work described in this
study, however, each of the authors has par-
ticipated in the training of therapists (in-
cluding paraprofessional staff) in residential
or school programs for persons with devel-
opmental disabilities. The instructional pro-
gram has usually been conducted in a group
context and has included the textual, role-
play, videotape, and performance feedback
components used in the present study, with
an additional component consisting of dem-
onstration and practice with actual clients.
Using these procedures, we have found that
it is possible to establish levels of compe-
tence that are comparable to those reported
here in less than half a day. The consultative
nature of this work precluded an experimen-
tal arrangement, so the collection of con-
trolled data in clinical contexts should be
considered an important area for future
work.
The second limitation in the present
study was that training focused on a circum-
scribed set of skills. It is important to note
that the implementation of functional anal-
ysis methodology under actual clinical con-
ditions may require professional judgments
related to data interpretation, modification
of assessment conditions to identify idiosyn-
cratic maintaining variables, risk manage-
ment, and so forth. Although these addi-
tional skills were not examined in the pres-
ent study, to the extent that they can be de-
scribed objectively, it is likely that they could
be taught in component fashion using meth-
ods similar to those employed here (e.g., see
the recent work of Hagopian et al., 1997,
on interpretation of functional analysis out-
comes). Of course, the full benefit of adopt-
ing functional analysis methodology is real-
ized when it is placed within the context of
a program that integrates empirical ap-
proaches to individual assessment with ad-
vanced therapeutic technology, and an anal-
ysis of the skills required to develop such a
program extends far beyond the scope of any
single study.
Given these limitations, the present results
are nevertheless encouraging in light of re-
cent suggestions that functional analysis
methodology is too complicated for use un-
der typical clinical conditions. Most service
settings to which clients with severe behavior
disorders are referred employ one or more
therapists with graduate training that in-
cludes course work in behavior analysis. The
procedures and data presented here suggest
that these therapists should be sufficiently
skilled to train both themselves and others
190 BRIAN A. IWATA et al.
to implement a number of the experimental
approaches to behavioral assessment that
comprise functional analysis methodology.
REFERENCES
Applegate, H., Matson, J. L., & Cherry, K. E. (1999).
An evaluation of functional variables affecting se-
vere problem behaviors in adults with mental re-
tardation by using the Questions about Behavioral
Function Scale (QBAF). Research in Developmenal
Disabilities, 20, 229–237.
Crawford, J., Brockel, B., Schauss, S., & Miltenberger,
R. G. (1992). A comparison of methods for the
functional assessment of stereotypic behavior.
Journal of the Association for Persons with Severe
Handicaps, 17, 77–86.
Ducharme, J. M., & Feldman, M. A. (1992). Com-
parison of staff training strategies to promote gen-
eralized teaching skills. Journal of Applied Behavior
Analysis, 25, 165–179.
Durand, V. M., & Crimmins, D. B. (1988). Identi-
fying the variables maintaining self-injurious be-
havior. Journal of Autism and Developmental Dis-
orders, 18, 99–117.
Hagopian, L. P., Fisher, W. W., Thompson, R. H.,
Owen-DeSchryver, J., Iwata, B. A., & Wacker, D.
P. (1997). Toward the development of structured
criteria for interpretation of functional analysis
data. Journal of Applied Behavior Analysis, 30,
313–326.
Individuals with Disabilities Education Act, 20 U.S.C.
§1400 et seq. (1975, as amended, 1997).
Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K.
E., & Richman, G. S. (1994). Toward a func-
tional analysis of self-injury. Journal of Applied Be-
havior Analysis, 27, 197–209. (Reprinted from
Analysis and Intervention in Developmental Dis-
abilities, 2, 3–20, 1982)
Iwata, B. A., Vollmer, T. R., & Zarcone, J. R. (1990).
The experimental (functional) analysis of behavior
disorders: Methodology, applications, and limita-
tions. In A. C. Repp & N. N. Singh (Eds.), Per-
spectives on the use of nonaversive and aversive in-
terventions for persons with developmental disabili-
ties (pp. 301–330). Sycamore, IL: Sycamore.
Jones, F. H., & Eimers, R. C. (1975). Role playing
to train elementary teachers to use a classroom
management ‘‘skill package.’’ Journal of Applied
Behavior Analysis, 8, 421–433.
Lerman, D. C., & Iwata, B. A. (1993). Descriptive
and experimental analyses of variables maintaining
self-injurious behavior. Journal of Applied Behavior
Analysis, 26, 293–319.
Mace, F. C., Lalli, J. S., & Lalli, E. P. (1991). Func-
tional analysis and treatment of aberrant behavior.
Research in Developmental Disabilities, 12, 155–
180.
Neef, N. A., Trachtenberg, S., Loeb, J., & Sterner, K.
(1991). Video-based training of respite care pro-
viders: An interactional analysis of presentation
format. Journal of Applied Behavior Analysis, 24,
473–486.
Northup, J., Wacker, D., Sasso, G., Steege, M., Cig-
rand, K., Cook, J., & DeRaad, A. (1991). A brief
functional analysis of aggressive and alternative
behavior in an outclinic setting. Journal of Applied
Behavior Analysis, 24, 509–522.
Pyles, D. A. M., Riordan, M. M., & Bailey, J. S.
(1997). The stereotypy analysis: An instrument
for examining environmental variables associated
with differential rates of stereotypic behavior. Re-
search in Developmental Disabilities, 18, 11–38.
Reid, D. H., Parsons, M. B., & Green, C. W. (1989).
Staff management in human services: Behavioral re-
search and application. Springfield, IL: Charles C.
Thomas.
Shore, B. A., Iwata, B. A., Vollmer, T. R., Lerman, D.
C., & Zarcone, J. R. (1995). Pyramidal staff
training in the extension of treatment for severe
behavior disorders. Journal of Applied Behavior
Analysis, 28, 323–332.
Spreat, S., & Connelly, L. (1996). Reliability analysis
of the Motivation Assessment Scale. American
Journal on Mental Retardation, 100, 528–532.
Sturmey, P. (1994). Assessing the functions of aber-
rant behaviors: A review of psychometric instru-
ments. Journal of Autism and Developmental Dis-
orders, 24, 293–304.
Wallace, M. D., & Iwata, B. A. (1999). Effects of
session duration on functional analysis outcomes.
Journal of Applied Behavior Analysis, 32, 175–183.
Received October 8, 1999
Final acceptance February 23, 2000
Action Editor, David P. Wacker
APPENDIX A
P
ROCEDURAL
D
ESCRIPTIONS OF
A
SSESSMENT
C
ONDITIONS
Attention Condition
Purpose
This condition is designed to determine
whether the target behavior is maintained by
contingent attention delivered by a therapist.
The condition involves remaining in a room
with a client and ignoring all client behavior,
except for the target behavior, which is fol-
lowed by attention.
191FUNCTIONAL ANALYSIS METHODOLOGY
Target Behavior
The target behavior being assessed con-
sists of self-hitting, defined as one body part
(arm, hand, etc.) striking against another
body part.
How to Conduct a Session
1. Begin a session by directing the client
toward the leisure materials that are present
in the room. Tell the client that he or she
should play with the toys while you do some
work.
2. After issuing the initial instruction,
move away from the client, sit in another
chair, read or do some paperwork (or pre-
tend to do so), and completely ignore all
behaviors exhibited by the client except as
noted below.
3. If the target behavior does not occur
during the session, you will ignore the client
for the entire session. Someone will inform
you when the session is over.
4. If any behaviors other than the target
behavior occur, ignore these also. Examples
include appropriate behaviors (e.g., playing
with the toys, smiling at you, or any at-
tempts to talk to you or to interact with you
in an appropriate manner) and inappropriate
behaviors other than self-hitting (e.g.,
screaming, throwing materials, running
around the room, aggression, etc.).
5. The only time you will attend to the
client is when he or she engages in self-hit-
ting. If the client exhibits the target behavior
of self-hitting at any time during the session,
do the following: (a) Go over to the client
and verbally express concern and disapprov-
al. For example, you could say something
like, ‘‘Stop that, you’re going to hurt your-
self,’’ ‘‘[Name], you shouldn’t hit yourself;
play with your toys,’’ ‘‘[Name], I don’t want
you to do that; you’re going to get hurt,’’ or
something similar. (b) While you express
concern, briefly touch the client’s arm, place
your hand on the client’s shoulder, or phys-
ically block the hitting response, but do not
physically restrain the client. The general
idea is to express concern, briefly interrupt
the behavior, and calm the client. Do not
shout at the client and do not handle the
client roughly.
6. After a target behavior occurs and you
have responded as indicated above (Step 5),
resume ignoring the client until another tar-
get behavior occurs or until the session is
over.
Demand (Escape) Condition
Purpose
This condition is designed to determine
whether the target behavior is maintained by
escape from task demands. The condition
involves presenting a series of instructional
trials to a client. Compliance produces
praise, noncompliance produces a series of
prompts, and occurrence of the target be-
havior immediately terminates the trial.
Target Behavior
The target behavior being assessed con-
sists of self-hitting, defined as one body part
(arm, hand, etc.) striking against another
body part.
How to Conduct a Session
1. Begin a session with you and the client
seated at a table. Using the materials that are
available, you will implement a series of tri-
als to teach the client to perform a task. The
task selected for this simulation is putting
blocks in a bucket.
2. Activate a stopwatch at the beginning
of the session. At the beginning of every 30-
s interval (starting at 0), you will initiate an
instructional trial. Thus, there will be ap-
proximately 10 trials during a 5-min session.
Begin each training trial with the bucket and
a block on the table in front of the client.
The sequence to be used during each trial is
as follows: (a) First deliver a clear instruction
to the client, such as ‘‘[Name], put the block
192 BRIAN A. IWATA et al.
in the bucket.’’ If the client performs the
response within 5 s (count to 5 slowly to
determine this), or at least begins to initiate
the response during that time, deliver praise
(e.g., say ‘‘nice job,’’ ‘‘that’s great,’’ ‘‘good,’’
etc.) when the client has finished. (b) If the
client does not perform the response within
5 s, repeat the instruction and simultaneous-
ly demonstrate the response (i.e., you put a
block in the bucket). If, following this dem-
onstration, the client performs the response
in 5 s, deliver praise as noted above. (c) If
the client does not perform the response
within 5 s of your demonstration, repeat the
instruction again and simultaneously provide
physical assistance. That is, use your hands
to help the client pick up the block and put
it in the bucket. Do not deliver praise if you
used physical assistance. (d) If, at any time
during this sequence, the client emits the
target behavior (self-hitting), immediately
terminate the trial. Remove the materials
from the table, turn away from the client,
and ignore the client until it is time to begin
a new trial. (e) If the client emits other in-
appropriate behaviors (screaming, throwing
things, aggression, etc.), continue with the
sequence; do not terminate the trial when
these responses occur.
3. Repeat the above sequence after 30 s
have elapsed since the trial began, and con-
tinue until the session is over.
Play Condition
Purpose
This is designed to be a general control
condition, in which no demands are placed
on the client, continuous access to leisure
materials is available, and attention is deliv-
ered frequently independent of the client’s
behavior.
Target Behavior
The target behavior being assessed con-
sists of self-hitting, defined as one body part
(arm, hand, etc.) striking against another
body part.
How to Conduct a Session
1. Begin a session by activating a stop-
watch and directing the client toward the
leisure materials that are present in the
room. You may say something like, ‘‘Here
are some nice toys; why don’t you play with
them for a while?’’ or ‘‘Would you like to
play with these toys?’’ (as you hand one to
the client), or anything similar.
2. At least once every 30 s, deliver some
form of attention to the client. For example,
you can tell the client that he or she is play-
ing nicely, ask if he or she is having fun, and
so forth. You can also hand the client an-
other toy, pat the client briefly on the shoul-
der, or smile at the client. The general idea
is to provide some type of friendly, nonde-
manding interaction (lasting about 5 s) at
30-s intervals.
3. If the client attempts to interact with
you appropriately (e.g., asks for something,
hands you a toy, etc.), reciprocate.
4. If the client emits any form of inap-
propriate behavior, including the target be-
havior, do not deliver attention.
5. If the target behavior occurs precisely
at the end of a 30-s interval (just as you are
about to deliver attention), do not deliver
attention. Instead, wait until the behavior
has stopped for 5 s, then deliver attention.
APPENDIX B
O
UTLINE OF
A
SSESSMENT
C
ONDITIONS
Attention Condition
1. Instruct client to play with toys; then
ignore.
2. Client emits appropriate behavior: Ig-
nore.
3. Client emits inappropriate behavior
other than the target: Ignore.
4. Client emits target behavior (hits self):
193FUNCTIONAL ANALYSIS METHODOLOGY
Express concern paired with brief physical
contact.
Demand (Escape) Condition
1. Activate stop watch and begin trials.
2. First instruction (prompt): Instruct cli-
ent to put block in bucket.
(a) Client complies: Deliver praise.
(b) Client emits target behavior (hits self):
Withdraw materials and turn away until
next trial.
(c) Client emits any other behavior: Contin-
ue sequence.
(d) Client emits no response: Go to second
prompt.
3. Second prompt: Repeat instruction and
demonstrate.
(a) Client complies: Deliver praise.
(b) Client emits target behavior (hits self):
Withdraw materials and turn away until
next trial.
(c) Client emits any other behavior: Contin-
ue sequence.
(d) Client emits no response: Go to third
prompt.
4. Third prompt: Repeat instruction and
physically guide.
(a) Do not deliver praise.
(b) Client emits target behavior (hits self):
Withdraw materials and turn away until
next trial.
(c) Client emits any other behavior: Contin-
ue sequence.
5. Begin a new trial when the 30-s inter-
val has ended.
Play Condition
1. Activate stopwatch and direct client to-
ward toys.
2. Deliver attention at least once every
30 s.
3. Client initiates appropriate social inter-
action with therapist: Deliver attention.
4. Client emits any inappropriate behav-
ior: Do not deliver attention.
5. Client emits inappropriate behavior
just as you are about to deliver attention:
Wait until the behavior has stopped for 5 s.
APPENDIX C
S
AMPLE
Q
UIZ
1. Which assessment condition (attention,
demand, alone, play) is considered the con-
trol condition for the other three conditions?
2. In which assessment condition or con-
ditions does the client have access to leisure
items?
3. During all conditions, what should you
do if the client engages in a disruptive or
aggressive behavior (e.g., tips over a table or
tries to kick the therapist) that is not a target
behavior during the functional analysis?
4. What should you do if a client becomes
injured during a session?
5. How do you begin an attention session
(what do you say and do)?
6. When do you deliver attention to the
client during the attention condition?
7. Give two examples of what you might
say or do when delivering attention during
the attention condition.
8. What should you do if the client asks
a question or requests help during the atten-
tion condition?
9. How often do you deliver attention to
the client during the play condition?
10. Give two examples of what you might
say or do when delivering attention during
the play condition.
11. What should you do if the client en-
gages in the target behavior (SIB) during the
play condition just as you are about to de-
liver attention?
12. What should you do if the client asks
you a question during the play condition
when you are not scheduled to deliver atten-
tion?
13. What should you do if the client en-
gages in disruptive behavior (e.g., knocks over
furniture, throws objects, etc.) that is not a
target behavior during the play condition?
194 BRIAN A. IWATA et al.
14. How often should you initiate train-
ing trials during the demand condition?
15. If the task during the demand con-
dition is putting a puzzle piece into a puzzle,
what should you say when initiating a trial?
16. If, during the demand condition, the
client does not respond to your first prompt
within 5 s, what should you do?
17. If, during the demand condition, the
client does not respond to your second
prompt within 5 s, what should you do?
18. What should you do if the client en-
gages in the target behavior (SIB) while you
are trying to get the client to work on a
puzzle during the demand condition?
19. Should you praise the client during
the demand condition if you had to physi-
cally guide the client to complete the task?
20. What should you do if the client asks
for help completing the assigned task during
the demand condition?
STUDY QUESTIONS
1. What are the three general methods used for conducting functional assessments? Which
method is considered superior, and what is its main advantage?
2. What are the most commonly cited limitations of functional analysis methodology, and
which of these was addressed in the current study?
3. What were the authors’ rationales for their selection of therapists and clients, and for the
procedure they used to assess therapist performance?
4. The behaviors of primary interest were those of the therapist. What target responses were
scored, and what criteria were used to determine whether these responses were correct or
incorrect?
5. Briefly describe the baseline procedures. Why was performance not assessed under completely
naive conditions?
6. What were the major components of the training program, and what was the total duration
of training?
7. Describe the participants’ performance in baseline and posttraining. What aspect of the data
may have diminished the demonstration of experimental control? In discussing this limita-
tion, what did the authors suggest about the likelihood that participants might have achieved
high performance levels without any training?
8. What additional skills (not addressed in this study) may be important to the successful
implementation of functional analyses?
Questions prepared by Gregory Hanley and Rachel Thompson, The University of Florida
