You are on page 1of 39

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2003, 36, 147185 NUMBER 2 (SUMMER 2003)

FUNCTIONAL ANALYSIS OF PROBLEM


BEHAVIOR: A REVIEW
GREGORY P. HANLEY
UNIVERSITY OF KANSAS

BRIAN A. IWATA
UNIVERSITY OF FLORIDA

AND

BRANDON E. MCCORD
ARLINGTON DEVELOPMENTAL CENTER

Functional analysis methodology focuses on the identification of variables that influence


the occurrence of problem behavior and has become a hallmark of contemporary ap-
proaches to behavioral assessment. In light of the widespread use of pretreatment func-
tional analyses in articles published in this and other journals, we reviewed the literature
in an attempt to identify best practices and directions for future research. Studies included
in the present review were those in which (a) a pretreatment assessment based on (b)
direct observation and measurement of (c) problem behavior was conducted under (d)
at least two conditions involving manipulation of an environmental variable in an attempt
(e) to demonstrate a relation between the environmental event and behavior. Studies that
met the criteria for inclusion were quantified and critically evaluated along a number of
dimensions related to subject and setting characteristics, parametric and qualitative char-
acteristics of the methodology, types of assessment conditions, experimental designs, to-
pographies of problem behaviors, and the manner in which data were displayed and
analyzed.
DESCRIPTORS: functional analysis, assessment, problem behavior

Functional analysis methodology identi- cies of reinforcement or punishment over ex-


fies variables that influence the occurrence of isting but often unknown sources of rein-
problem behavior and has become a hall- forcement for problem behavior (Mace,
mark of behavioral assessment (see the spe- 1994). By contrast, by identifying contin-
cial issue on functional analysis in the Jour- gencies that currently maintain problem be-
nal of Applied Behavior Analysis [JABA], havior, relevant consequences and their as-
1994, Vol. 27). Prior to the advent of func- sociated discriminative stimuli (SDs) and es-
tional analysis approaches to assessment, tablishing operations (EOs) may be altered
problem behavior was typically treated by to reduce problem behavior. In essence,
superimposing powerful arbitrary contingen- functional analysis methodology reempha-
sized the importance of applied research in
This research was supported in part by a grant from contributing to an understanding of the de-
the Florida Department of Children and Families. We terminants of behavior as the basis for iden-
thank Marc Branch and several anonymous reviewers
for their insightful comments and suggestions on ear- tifying effective treatments that produce gen-
lier versions of this article. eralized results.
Reprints may be obtained from Greg Hanley, De- Since the development of comprehensive
partment of Human Development and Family Life,
4001 Dole Center, University of Kansas, Lawrence, models for conducting functional analyses
Kansas 66045 (e-mail: ghanley@ku.edu). (i.e., those that examined multiple sources

147
148 GREGORY P. HANLEY et al.

of influence; E. G. Carr & Durand, 1985; about how the behavior operates on the en-
Iwata, Dorsey, Slifer, Bauman, & Richman, vironment.
1982/1994), hundreds of direct and system- Although early conceptual analyses (Bach-
atic replications, as well as extensions across man, 1972; E. G. Carr, 1977; & Smolev,
populations, settings, and topographies of 1971) suggested that self-injurious behavior
problem behavior, have been reported in the (SIB) might be the product of reinforcement
literature. However, the extent of these var- contingencies that differed across individuals
iations has not yet been systematically eval- who exhibited these behaviors, methods for
uated or critically examined. The purpose of identifying various conditions that were cor-
this review is to provide a quantitative and related with SIB and other problem behav-
qualitative analysis of research on the func- iors prior to intervention were not described
tional analysis of problem behavior and to until years later. Nevertheless, several note-
identify unanswered questions that may be worthy studies included systematic empirical
addressed in future research. investigations of environmental influences
The term functional analysis was used by on problem behavior and laid the ground-
Skinner (1953) to denote empirical dem- work for a comprehensive functional analysis
onstrations of cause-and-effect relations methodology. Lovaas and colleagues (Lovaas,
between environment and behavior; howev- Freitag, Gold, & Kassorla, 1965; Lovaas &
er, the term has been extended by behavior Simmons, 1969) were the first to demon-
analysts and psychologists in general to de- strate the effects of social-positive reinforce-
ment (attention) on the SIB of children who
scribe a wide range of procedures and op-
had been diagnosed with autism and mental
erations that are different in many important
retardation. Similar studies demonstrated
ways (see Haynes & OBrien, 1990, and
the effects of attention on problem behaviors
Iwata, Kahng, Wallace, & Lindberg, 2000,
common to the classroom, such as aggres-
for two different but comprehensive discus-
sion (Pinkston, Reese, LeBlanc, & Baer,
sions). In addition, the term evokes different 1973) and disruption (Thomas, Becker, &
responses through somewhat different uses Armstrong, 1968). Sailor, Guess, Ruther-
in other disciplines such as medicine, math- ford, and Baer (1968) provided an early
ematics, physics, and biology. In the behav- demonstration that problem behavior also
ior analysis literature, the term function has could be maintained by negative reinforce-
been used in two ways. One use conveys the ment (escape from difficult instruction) in a
effect that a behavior has on the environ- young girl with mental retardation; this
ment or, speaking loosely, the purpose the work was extended by E. G. Carr, Newsom,
behavior serves for an individual (e.g., the and Binkoff (1976, 1980), who showed that
function of behavior is to terminate an on- aggression and SIB were correlated with the
going event). The second use describes a re- presentation and removal of demands, and
lation between two variables (typically be- by Weeks and Gaylord-Ross (1981), who
tween some environmental event and a class showed that SIB was positively correlated
of behavior) in which one varies given the with task difficulty. In addition to demon-
presence or absence of the other (e.g., re- strating the effects of specific contingencies
sponding as a function of an event). Both on problem behavior, these studies illustrat-
uses of the term are relevant to a functional ed the general value of identifying the con-
analysis of existing behavior, in that relations ditions under which problem behavior may
between behavior and environmental events actually worsen: If one could specify which
are demonstrated in the context of learning aspects of a procedure led to more problem
FUNCTIONAL ANALYSIS 149

behavior, one should then be able to change seat) of 4 children with developmental dis-
the procedure so as to effect a reduction in abilities in which two antecedent variables,
problem behavior (a similar notion was pre- amount of attention and difficulty of in-
sented by Baer, Wolf, & Risley, 1968). struction, were manipulated. Several patterns
The preceding studies established the ba- of problem behavior were observed, suggest-
sic methodological features of a functional ing that influential variables differed across
analysis of problem behavior: direct obser- participants.
vation and measurement of problem behav- The functional analysis methods de-
ior under test and control conditions in scribed by Iwata et al. (1982/1994) and E.
which some environmental variable is ma- G. Carr and Durand (1985) marked the be-
nipulated. From these strategies, a relation ginning of a comprehensive approach to in-
between an environmental event and behav- tervention in which control techniques de-
ior was demonstrated. However, all of the rived from the experimental analysis of be-
studies described above focused on single re- havior were applied, not only to the treat-
sponsereinforcer relations. ment of problem behavior but to its
The first comprehensive analysis of the assessment as well. In addition, both assess-
determinants of problem behavior was re- ment models represented an improvement
ported by Iwata et al. (1982/1994), who over arbitrary approaches to the treatment of
proposed a general model for concurrently problem behavior and led to the develop-
assessing the sensitivity of SIB to contingen- ment of more precise reinforcement-based
interventions and an apparent decrease in
cies of positive, negative, and automatic re-
the use of punishment (Pelios, Morren,
inforcement. More specifically, direct obser-
Tesch, & Axelrod, 1999). In essence, func-
vation and repeated measurement of behav-
tional analysis has provided a means to de-
ior were conducted across four conditions
termine in advance which treatments should
(three tests and one control) arranged in a and should not work, as well as why. What
multielement, single-subject experimental follows is a review of functional analysis
design (Ulman & Sulzer-Azaroff, 1975). methodology, a presentation of guidelines
Each test condition contained an EO, SD, for best practice, and a discussion of areas
and source of reinforcement for a given con- that warrant further research attention.
tingency, whereas these same operations and
contingencies were absent from the control
condition. This methodology was applied to METHOD
various forms of SIB (e.g., head banging, Functional analysis studies were identified
biting, eye gouging, face slapping, hair pull- through a search of Current Contents,
ing) exhibited by 9 children with develop- PsychInfo, and ERIC using the key words
mental disabilities. Results showed that lev- function, analysis, and behavioral assessment
els of SIB varied widely across participants. through 2000. The reference section of each
article so identified was then examined to
More important, they showed that SIB was
identify additional functional analysis arti-
higher in particular test conditions relative
cles. Finally, all identified studies were re-
to the control for 8 of the 9 participants.
viewed to determine if they met criteria for
E. G. Carr and Durand (1985) described inclusion in the present review.
another model for conducting a functional
analysis of problem behavior. The influences INCLUSION AND
of three assessment conditions were evalu- EXCLUSION CRITERIA
ated on varied problem behaviors (aggres- Studies included in the present review
sion, tantrums, SIB, opposition, and out of were those in which (a) a pretreatment as-
150 GREGORY P. HANLEY et al.

sessment based on (b) direct observation and Hanson, Chamberlain, & Thompson,
measurement of problem behavior was con- 1985), questionnaires (e.g., Matson, Bam-
ducted under (c) at least two conditions in- burg, Cherry, & Paclawskyj, 1999), or clin-
volving manipulation of some environmen- ical interviews (e.g., ONeill, Horner, Albin,
tal variable in an attempt (d) to demonstrate Storey, & Sprague, 1990) were not included
a relation between the environmental event due to their reliance on anecdotal reports
and behavior. The criteria for inclusion (and from caregivers in lieu of direct observation
exclusion) are described more fully below. of problem behavior.
Pretreatment Assessment Manipulation
Pretreatment assessment refers to an at-
tempt by the researcher to identify variables By limiting the review to studies involving
that affected rates of problem behavior (an at least two conditions in which some en-
evaluation of a treatment was not required). vironmental variable was manipulated, all
This criterion ruled out studies in which a studies that relied exclusively on descriptive
functional relation was established only in analysis were not included. Descriptive anal-
the context of treatment (e.g., SIB decreased ysis involves direct observation of behavior
when a particular intervention was used). under naturally occurring (uncontrolled)
That is, studies were excluded if functional conditions in an attempt to identify envi-
relations were not demonstrated indepen- ronmental correlates of problem behavior.
dent of treatment. Examples of this approach include continu-
ous observation methods (e.g., Bijou, Peter-
Direct Observation and Measurement of son, & Ault, 1968), antecedent-behavior-
Problem Behavior consequence (ABC) recording (e.g., Groden,
The focus of the current review is on the 1989), and scatter-plot recording (e.g.,
functional analysis of problem behavior, de- Touchette, MacDonald, & Langer, 1985).
fined as behavioral excess that is socially sig- Studies that included descriptive analysis in
nificant to the extent that someone com- addition to other types of analysis that met
plains of its occurrence. These behaviors are the criteria noted above were included in the
typically of sufficient intensity or frequency present review (e.g., Lerman & Iwata, 1993;
that the safety of the person or others is Mace & Lalli, 1991).
threatened, the ability of the person or oth-
ers to acquire new skills is hindered, or more FUNCTIONAL ANALYSIS METHODOLOGY
restrictive living arrangements are warranted.
The requirement for direct observation Studies that met the criteria for inclusion
and measurement specified that the primary in the present review varied along a number
data used in the analysis were collected by of dimensions related to subject and setting
observers who recorded responses of study characteristics, parametric and qualitative
participants (either live or from videotape). characteristics of the methodology, types of
Thus, reviews, commentaries, and discussion conditions arranged, experimental designs
papers were excluded because they did con- used, types of problem behaviors evaluated,
tain data of this type. In addition, studies and the manner in which data were dis-
that relied exclusively on indirect means to played and analyzed. Studies that met cri-
identify functional variables were excluded. teria for inclusion were quantified and crit-
More specifically, studies in which data were ically evaluated along the following dimen-
based solely on rating scales (e.g., Weiseler, sions.
FUNCTIONAL ANALYSIS 151

Population and Setting Characteristics If several conditions were used to evaluate


Participants. Data were collected on par- multiple functions, we documented which
ticipant age, level of functioning, and diag- ones were assessed (e.g., attention, tangible
nosis. Participants were categorized as either [materials or food], escape, automatic [pos-
child (1 to 18 years) or adult (19 and over). itive or negative]) and whether a relevant
It was noted whether participants fell within control condition was included. Descrip-
a normal range of functioning or if a partic- tions of atypical test conditions (e.g., escape
ular developmental disability (e.g., mental from noise) were also noted.
retardation) was noted. Diagnoses of autism
Assessment Duration
also were documented.
Settings. Settings in which assessment oc- Data were collected on the total number
curred were categorized as home, school, of analysis sessions for each participant and
outpatient clinic, inpatient hospital unit, in- were categorized according to the number of
stitution, or vocational program. observations per condition. An analysis was
considered brief (e.g., Northup et al., 1991)
Response Topographies if two or fewer observations were conducted
Data were collected on the specific topog- in each condition, whereas an analysis was
raphies of problem behavior included in the considered full if three or more observations
functional analyses. Based on the authors were conducted in at least two conditions.
description, behaviors were categorized as
SIB, aggression, property destruction, pica, Session Duration
motor disruptions, vocalizations (either bi- Data were collected on the duration of
zarre or disruptive), elopement, stereotypy, each observation session in the analysis.
tantrums, aberrant (this was scored if several
topographies were combined into one re- Experimental Design
sponse class), or other. The type of single-subject design used to
demonstrate the effects of a variable on
Type of Functional Analysis problem behavior was noted. Each analysis
Data were collected on which of the two was categorized as a reversal, multielement
general types of functional analysis, the AB (i.e., rapid alternation between two or more
(antecedent-behavior) model (E. G. Carr & conditions), or pairwise (sequential evalua-
Durand, 1985) or the ABC model (Iwata et tion of each test condition through rapid al-
al., 1982/1994), characterized the structure ternation between a single test and control
of the functional analyses in each study. condition; Iwata, Duncan, Zarcone, Ler-
Data also were collected on whether addi- man, & Shore, 1994) experimental design,
tional pretreatment assessment data, such as or as sharing features of more than one de-
those derived from indirect or descriptive sign (noted as a combination).
types of functional assessment, were includ-
ed. Finally, it was noted if a comparison of Data Display and Analysis
any of the methodologies (indirect assess- Data were collected on the method used
ment, descriptive analysis, or functional to present data from the functional analyses.
analysis) was conducted. All of the data conformed to one of three
types of displays. Data were presented as (a)
Condition Types condition means exclusively (typically pre-
Data were collected on whether single or sented as a bar graph, table, or numerical
multiple functions of behavior were assessed. data in the text), (b) values for each session
152 GREGORY P. HANLEY et al.

typically displayed on a graph (i.e., a point Behavioral Function by Topography


per session), or (c) within-session values (i.e., For analyses in which the behaviors main-
the data binned in sequential time segments taining reinforcer was identified, the specific
during one or more sessions). It was also behavioral function (as noted by the authors
noted whether data analysis relied exclusively of each study) was catalogued across behav-
on visual inspection or was aided or replaced ioral topography. Maintaining reinforcer cat-
by some descriptive or inferential statistical egories included attention, tangible (edible
procedures. items, toys), escape, automatic, or multiple
(two or more behavioral functions).
Stimulus Parameters
Antecedent variables. Descriptions of an- INTERRATER AGREEMENT
tecedent variables that were manipulated A second reader independently analyzed
were recorded (e.g., instructional types, 12.6% of the articles as a basis for assessing
manner in which attention was diverted). interrater agreement on the categorization of
Consequence variables. If consequences studies. Agreement was then assessed by an
were programmed for the occurrence of item-by-item comparison of score sheets
problem behavior during test conditions, the generated by the two readers in which the
specific type of consequence provided (e.g., number of agreements between the two (i.e.,
verbal reprimand or physical interaction), as same subcategory scored) was divided by the
well as its duration, was noted. Schedules number of agreements plus disagreements
were categorized as either continuous (i.e., and multiplied by 100%. The mean inter-
consequences delivered following each in- rater agreement was 98.0% (range, 92.1%
stance of problem behavior) or intermittent to 100%) across score sheets.
(consequences delivered following a propor-
tion of responses). RESULTS
FUNCTIONAL ANALYSIS OUTCOME SUMMARY Given the search strategy described above,
a total of 790 published works were identi-
To avoid summarizing functional analysis fied. Of these, 215 articles were excluded
data published in more than one study from the analysis because they were book
(Fisher, Piazza, & Hanley, 1998) or small chapters, reviews, discussion pieces, or com-
data sets also published as aggregates (e.g., mentaries on functional analysis methodol-
Derby et al., 1992), only functional analysis ogy that did not contain any data. Although
data that appeared in a line chart format data based, an additional 298 studies were
were included in the outcome summary. By excluded because they lacked critical func-
so doing, any data that appeared in more tional analysis elements described above as
than one study were easily identified and in- inclusion criteria. From this, a total of 277
cluded only once in the summary. empirical studies were identified and includ-
ed in the quantitative review.
General Outcome across Topography
Based on the authors conclusions pre- JOURNALS PUBLISHING FUNCTIONAL
sented in each study, the number of differ- ANALYSIS STUDIES
entiated (i.e., those assessments that yielded Table 1 lists the journals in which func-
a determination of behavioral function) and tional analysis studies have been published.
undifferentiated functional analyses was not- A total of 34 journals have published at least
ed for each topography of problem behavior. one functional analysis study (as defined in
FUNCTIONAL ANALYSIS 153

Table 1 ure 1); however, the overwhelming majority


Journals Publishing Functional Analysis Studies of functional analysis studies have been pub-
Number of Percentage
lished in JABA (64.9%). This is not surpris-
Journal studies of sample ing given that the procedures (operant con-
Journal of Applied Behavior 180 64.9 tingencies) and methodology (single-subject
Analysis designs) on which functional analysis is
Research in Developmental Dis- 21 7.6 based are also the cornerstones of applied
abilitiesa behavior analysis and the most common
Behavior Modification 10 3.6
Journal of Behavior Therapy 9 3.2 characteristics of articles published in JABA.
and Experimental Psychiatry Although functional analysis may still be
Journal of the Association for 6 2.2 limited to the research and practice of a
Persons with Severe Handi-
capsb small number of individuals (Gable, 1996;
Behavioral Interventionsc 5 1.8 Gresham, Quinn, & Restori, 1999), the
Education and Training in 5 1.8 present database shows that over 400 indi-
Mental Retardation and De-
velopmental Disabilities viduals have coauthored data-based func-
Journal of Autism and Develop- 4 1.4 tional analysis studies.
mental Disordersd
Journal of Developmental and 4 1.4 ANALYSIS METHODOLOGY
Physical Disabilitiesa
School Psychology Quarterly 4 1.4 Population and Setting Characteristics
Behavioral Disorders 3 1.1
Behavior Therapy 3 1.1 Although a substantial proportion of
Journal of Intellectual Disability 2 0.7 functional analysis studies (37.2%) included
Research f adults, the majority of studies included chil-
Number of other journals with 21 7.6
one publication dren (70.0%) with some form of develop-
Total number of functional 277 mental disability (91.3%; see Table 2). Giv-
analysis studiesa en the high prevalence of problem behavior
a Applied Research in Mental Retardation merged with Anal-
in persons with developmental disabilities,
ysis and Intervention in Developmental Disabilities to become
Research in Developmental Disabilities. the fact that the majority of functional anal-
b Formerly Journal of the Association for Education of Persons
ysis studies have focused on this population
with Severe and Profound Handicaps.
c Formerly Journal of Autism and Childhood Schizophrenia. is not surprising. A much lower percentage
d Formerly Behavioral Residential Treatment.
e Formerly Journal of the Multihandicapped Person.
of studies included functional analyses of
f Formerly Journal of Mental Deficiency Research. problem behaviors exhibited by persons
g Functional analysis studies as defined in the Method Sec-
without disabilities (9.0%), showing that
tion.
this is a relatively underresearched area.
However, functional analysis methodology
the present review); 13 journals have pub- has been applied in 25 studies to evaluate
lished two or more studies. Combined with problem behaviors that are more common
the fact that numerous discussion and review among typically developing children (some
papers have appeared in still more journals of these studies will be described below).
(especially those in the area of education), Most functional analysis studies have been
the data suggest that a large number of read- conducted in hospital (inpatient) facilities
ers of psychological literature have come in (32.5%), schools (31.4%), or institutions
contact with functional analysis methodolo- (25.3%; see Table 2); much less research
gy. The number of different journals that (17.4%) has been conducted in other set-
published a functional analysis study shows tings (e.g., homes, vocational programs, and
a slight increasing trend across years (see Fig- outpatient clinics). It is unclear whether
154 GREGORY P. HANLEY et al.

Figure 1. Total number of functional analysis publications (dark bars) and total number of publishing
journals (light bars) for 5-year periods between 1961 and 2000.

choice of setting has been due to the greater Response Topographies


degree of control afforded by institutional The majority of functional analysis studies
environments or the fact that persons with have either included some form of SIB in
more severe problem behaviors are more the cluster of behaviors undergoing assess-
likely to be treated in these settings. ment or exclusively evaluated the controlling

Table 2
Participant and Setting Characteristics

Number of Percentage
studies of sample

Participants Child 194 70.0


Adult 103 37.2
Developmental disability 253 91.3
Autism 58 20.9
No disability 25 9.0
Setting Hospital (inpatient) 90 32.5
School 87 31.4
Institution 70 25.3
Home 21 7.6
Clinic (outpatient) 21 7.6
Vocational program 6 2.2
FUNCTIONAL ANALYSIS 155

Table 3 1993), tantrums (Vollmer, Northup, Ring-


Prevalence of Behavior Topographies dahl, LeBlanc, & Chauvin, 1996), drug in-
Number of Percentage
gestion (Chapman, Fisher, Piazza, & Kurtz,
Topography studies of sample 1993), elopement (Piazza, Hanley, et al.,
Self-injury 179 (130) 64.6 (4.6) 1997), and property destruction (Fisher,
Aggression 113 (46) 40.8 (1.6) Lindauer, Alterson, & Thompson, 1998). In
Disruption 53 (19) 19.1 (6.9) addition, topographies of problem behavior
Vocalizations 35 (16) 12.6 (5.8)
Property destruction 29 (2) 10.5 (0.7) more commonly exhibited by either typically
Stereotypy 25 (17) 9.0 (6.1) developing children or children with mild
Noncompliance 12 (1) 4.3 (0.3) disabilities have also been assessed via func-
Tantrums 10 (1) 3.6 (0.3)
Elopement 8 (1) 2.9 (0.3) tional analysis; examples include disruptive
Pica 7 (3) 2.5 (1.1) behavior of children in regular education
Other 10 (0) 3.6 (0) classrooms (Broussard & Northup, 1995,
Note. The numbers in parentheses indicate studies that in- 1997), disruptive classroom behavior of stu-
cluded one specific topography in the analysis contingency
class. dents with emotional disabilities (DePaepe,
Shores, Jack, & Denny, 1996), finger suck-
ing by children in the home (Ellingson et
variables for SIB (179 or 64.6%; see Table al., 2000), inappropriate classroom behavior
3). Aggression (113 or 40.8%) and disrup- of an elementary student (Lewis & Sugai,
tion (53 or 19.1%) were, respectively, the 1996), reluctant speech of an elementary
second and third most common topogra-
school student (Mace & West, 1986), off-
phies of problem behavior evaluated. A large
task behavior of elementary school children
percentage of studies (85.0%) included only
with mild learning disabilities (Meyer,
topographically similar responses (e.g., only
1999), and classroom disruptive behaviors of
self-injurious head hitting or only pica) in
children with attention deficit disorder
the class of responses for which consequenc-
es were programmed. A fairly large percent- (Northup et al., 1995; Umbreit, 1995b).
age of studies (27.8%) included two or more These studies represent a first step towards
(typically more) topographies in the target extending functional analysis methodology
response class in at least one of the func- to the problem behavior of typically devel-
tional analyses. oping children, and the continued extension
Although the majority of functional anal- and refinement of these methods represents
ysis studies have examined some form of an exciting and important area for future
SIB, aggression, or disruption, the method- work.
ology has been extended to a variety of other In contrast to recent demonstrations with
problem behaviors including reluctant or bi- typically developing children, functional
zarre vocalizations (e.g., Durand & Crim- analysis methodology has not yet been ex-
mins, 1987; Mace & West, 1986), vocal tics tended to behavior problems (e.g., nail bit-
(J. E. Carr, Taylor, Wallander, & Reiss, ing, complaining, smoking, drug abuse,
1996), stereotypy (e.g., Mace, Browder, & overeating, or problem behaviors associated
Lin, 1987), mouthing (Goh et al., 1995), with mental illnesses such as depression, bu-
breath holding (Kern, Mauk, Marder, & limia, or anorexia) exhibited by adults with-
Mace, 1995), pica (Mace & Knight, 1986; out disabilities. This too represents an im-
Piazza, Hanley, & Fisher, 1996), hair pulling portant area for the systematic extension of
(Miltenberger, Long, Rapp, Lumley, & El- functional analysis methodology (see addi-
liot, 1998), noncompliance (Reimers et al., tional discussions by Axelrod, 1991; Haynes
156 GREGORY P. HANLEY et al.

& OBrien, 1990; McManus & Waller, Table 4


1995). Methodological Characteristics of
Functional Analysis
The extension of functional analysis
methods to novel topographies of problem Number Percentage
behavior has also necessitated procedural of studies of sample
variations. For instance, analyses of pica (Pi-
Model type
azza et al., 1998), mouthing (Goh et al., ABC model 241 87.0
1995), and property destruction (Fisher, AB model 56 20.2
Lindauer, Alterson, & Thompson, 1998) re- Both models 20 7.2
quired the selection and arrangement of safe Supplementary assessments
materials to consume or destroy; the analysis Descriptive or indirect 29 10.5
of elopement (Piazza, Hanley, et al., 1997) Descriptive 23 8.3
Indirect 12 4.3
required that the participant could be re- Descriptive and indirect 7 2.5
trieved at least several times so that multiple Condition types
opportunities to respond were available; Social-negative reinforcement 247 89.2
analyses of aggression required that a person Social-positive reinforcement 237 85.6
be within reach and protected; and analyses Attention 229 82.7
of noncompliance (Reimers et al., 1993) ne- Tangible 96 34.7
Automatic reinforcement 165 59.6
cessitated the delivery of a consistent num-
ber of instructions. Other topographies of Number of test conditions
Multiple 248 89.5
problem behavior (e.g., out of seat, crying) Single 51 18.4
have been included in functional analyses;
Assessment length
however, these particular responses were part Full 229 82.7
of a larger group of topographically distinct Brief 36 13.0
behaviors. Therefore, the utility of function- Unknown 14 5.0
al analysis methods remains undemonstrated Session duration
for these and other unique, or possibly com- 5 min 31 11.1
mon, problem behaviors. 10 min 144 52.0
15 min 78 28.2
Other 12 4.3
Type of Functional Analysis Unknown 22 7.9
Functional analysis models. As described Experimental designs
earlier, one general model of functional anal- Multielement 225 81.2
ysis in current use involves the exclusive ma- Reversal 43 15.5
Pairwise 7 2.5
nipulation of antecedent events (an AB Combination 7 2.5
model; see E. G. Carr & Durand, 1985). Unknown 16 5.8
The second involves manipulation of all as- Data presentation
pects of the three-term contingency (i.e., an- Session values 208 75.1
tecedent and consequent events; an ABC Condition means 74 26.7
Within-session values 3 1.1
model; see Iwata et al., 1982/1994). The
ABC model was incorporated in 241 or
87.0% of the functional analysis studies,
whereas the AB model was used in 56 or ther with the same participant or across dif-
20.2% of the studies (see Table 4). Most of ferent participants.
the 277 studies employed one model to the Supplementary functional assessments. De-
exclusion of the other; however, 20 studies scriptive data (i.e., data collected through di-
(7.2%) included both types of assessment ei- rect observation of behavior in the absence
FUNCTIONAL ANALYSIS 157

of experimental manipulation) were includ- (1986); however, R. M. Day, Rea, Schussler,


ed in 23 studies (8.3%), data generated from Larsen, and Johnson (1988) published the
indirect means (i.e., questionnaires, rating earliest demonstration of behavioral main-
scales) were included in 12 studies (4.3%), tenance by access to tangible items. Since
either a descriptive or indirect assessment then, tests for behavioral sensitivity to tan-
was included in 29 studies (10.5%), and gible forms of reinforcement have often been
both descriptive and indirect functional as- included in functional analyses (47.0% of
sessment data in addition to that derived functional analyses in the last 5 years in-
from a functional analysis of problem be- cluded a tangible test condition). The ma-
havior were included in 7 studies (2.5%). jority of tests for maintenance via social-pos-
Many studies described the use of prelimi- itive reinforcement arranged a contingency
nary assessments as part of the process of between problem behavior and access to at-
identifying variables influencing problem be- tention or tangible items (94.9%); a smaller
havior. However, studies in which it was percentage of studies (6.7%) exclusively ma-
noted that observations, interviews, and so nipulated antecedent events (e.g., altered the
forth were conducted prior to a functional percentage of time in which attention or
analysis of problem behavior, but in which tangible items were available) and inferred
data from these supplementary assessments behavioral function from the resulting data.
were not reported, were not included in Social-negative reinforcement. Most func-
these calculations. tional analysis studies involved tests for be-
havioral maintenance by escape or avoidance
Condition Types contingencies (89.2%) and typically referred
The basis for identifying environmental to this arrangement as the demand or escape
influences in a functional analysis lies in a condition. In most cases (88.3% of those
comparison of behavior under test and con- testing for a negative reinforcement rela-
trol conditions. Test conditions involve some tion), a brief break from ongoing instruc-
potentially relevant independent variable tions (or social interaction, e.g., Vollmer et
(e.g., contingency between problem behavior al., 1998) was provided following problem
and access to toys), whereas control condi- behavior such that a direct test of the effect
tions are generally constructed so that the of a negative reinforcement contingency was
same independent variable is absent (e.g., evaluated. In a smaller percentage of the
toys are not provided following problem be- studies (18.6%), consequences were not ma-
havior). nipulated (and in many instances not spec-
Social-positive reinforcement. Most func- ified or controlled); instead, various charac-
tional analysis studies involved tests for be- teristics of the antecedent environment (e.g.,
havior maintained by social-positive rein- task difficulty) were altered.
forcement (85.6%), with the majority of Automatic reinforcement. More than half
these studies (96.6%) specifically assessing (59.6%) of the functional analysis studies in-
the effects of attention on problem behavior. cluded conditions to test for maintenance by
A smaller percentage of studies assessed the automatic reinforcement. These tests are
effects of other forms of social-positive re- necessarily indirect because the delivery of
inforcement such as foods, toys, or other automatic reinforcement typically cannot be
tangible items (38.3%) on the occurrence of controlled or directly manipulated by others.
problem behaviors. Assessment of the effect Therefore, the test for this sort of relation
of tangible reinforcement on problem be- relies on a strategy in which the influence of
havior was first reported by Mace and West social reinforcement is removed by observing
158 GREGORY P. HANLEY et al.

behavior under relatively barren conditions and escape) or indirectly (automatic rein-
(this test is typically referred to as the alone forcement) are assessed by comparing the
or ignore condition). If problem behavior relative rates of behavior in these test con-
persists under these conditions, in which no ditions to those observed in a single control
social reinforcement contingencies are pro- condition. This general type of control con-
grammed and ambient stimulation that may dition was used in 91.7% of the 241 func-
occasion escape-maintained behavior is ab- tional analysis studies that employed the
sent, evidence of maintenance via automatic ABC model. The remaining 8.3% relied on
reinforcement is provided. In an attempt to a test condition for one function serving as
decrease the possibility that bursts of socially a control for another test condition (e.g., no
mediated behavior are misdiagnosed as au- demands are presented in the attention con-
tomatically reinforced behavior, several re- dition, and contingent attention is unavail-
searchers have included either extended (of able in the demand condition). This strategy
longer duration) or repeated (consecutive was often employed in brief functional anal-
sessions) observations in the alone condition yses (e.g., Northup, 1991) in which practical
(e.g., Vollmer, Marcus, Ringdahl, & Roane, constraints limited the number of sessions
1995). Persistence under these conditions that could have been conducted. The main
provides further support for the nonsocial limitation of these and other studies that did
mediation of problem behaviors. not arrange a deliberate control condition is
Control conditions. Of the 56 functional the inability to discriminate between multi-
analysis studies employing the AB model, 40 ply controlled and undifferentiated (or ex-
(71.4%) included a control condition to de- clusively automatically reinforced) respond-
termine the influence of one of two anteced- ing (see E. G. Carr, Yarbrough, & Langdon,
ent variables (attention and task difficulty) 1997, for examples).
on problem behavior. The remaining studies Number of test conditions. Early research
in this group simply excluded a relevant an- on the functional analysis of problem behav-
tecedent event (i.e., the event present in the ior (i.e., studies published prior to 1982)
test condition) from the control condition. evaluated the effects of a single source of re-
The typical control condition in ABC func- inforcement. However, the majority of stud-
tional analyses (described originally by Iwata ies in the current review (89.5%) evaluated
et al., 1982/1994, as the play condition) the effects of multiple sources of influence
also controls for multiple sources of influ- through the inclusion of two or more test
ence. More specifically, no demands are pre- conditions in each functional analysis. In
sented, attention is withheld for problem be- other words, most functional analysis studies
havior and is available either freely or for included test conditions to delineate control
appropriate behavior, and access to alterna- by positive versus negative reinforcement or
tive forms of stimulation is continuously social versus automatic sources of reinforce-
available (i.e., free access to toys is arranged). ment. Clearly, the comprehensive approach
Thus, the EOs (deprivation from attention represents a refinement in assessment strat-
or stimulation, or the presentation of de- egy because it (a) identifies important con-
mands) for the three sources of reinforce- trolling relations, (b) rules out competing re-
ment, as well as the contingencies for the lations, (c) allows one to select a treatment
two sources of social reinforcement, are matched to the function of behavior, and (d)
eliminated or at least minimized in this con- avoids programming changes that will not
dition. From this, the effects of several con- affect the occurrence of problem behavior
tingencies arranged either directly (attention (or will be contraindicated). In addition,
FUNCTIONAL ANALYSIS 159

comprehensive functional analyses may functional analyses (i.e., two or fewer obser-
identify (or rule out) multiple sources of vations in each test condition). The length
control (H. M. Day, Horner, & ONeill, of assessment was not described and is there-
1994). fore unknown in 14 studies (5.0%), and full
analyses (three of more observations per con-
Session Duration dition) were included in the majority of
Data from published studies indicate that studies (229 or 82.7%).
most functional analysis sessions are 10 min
(52.0%), 15 min (28.2%), or 5 min Experimental Design
(11.1%) in duration. However, a few studies Most single-subject experimental designs
have programmed sessions as brief as 1 min involve observation of several features of be-
(Sigafoos & Meikle, 1996; Sigafoos & Sag- havior (i.e., level, trend, and stability) across
gers, 1995) or as long as 30 min (e.g., Arn- two or more conditions in which relevant
dorfer, Miltenberger, Woster, Rortvedt, & stimuli are either present (test conditions) or
Gaffaney, 1994; Reese, 1997). absent (control conditions). The design
most commonly used in functional analysis
Assessment Duration studies was the multielement (225 studies or
Assessment duration refers to the number 81.2%), which is characterized by the rapid
of sessions that comprise a functional anal- alternation of the experimental conditions.
ysis. Most functional analyses are conducted This design is attractive for functional anal-
until stability is achieved (i.e., no a priori yses because it is an efficient way to examine
criteria are used for terminating the analysis; the effects of several independent variables
rather, the analysis is concluded when useful (e.g., social-positive, social-negative, or au-
information has been obtained). This is con- tomatic reinforcement). In addition, organ-
sistent with the general strategies of single- ismic or other extraneous variables (e.g., al-
subject research (Sidman, 1960). However, lergies, medication changes) should affect
the exigencies of clinical practice (e.g., time behavior similarly across all conditions be-
limitations) often compromise attempts at cause the individual is exposed to the alter-
thorough assessment, thus necessitating the nating conditions within a relatively short
use of either alternative functional assess- period.
ment tools (e.g., indirect methods) or mod- The second most common experimental
ifications to functional analysis methodolo- design was the reversal or ABAB design
gy. In a significant contribution to the lit- characterized by repeated observations of be-
erature, Northup et al. (1991) illustrated the havior under a single condition, followed by
latter strategy through development of what the introduction, withdrawal, and reintro-
has come to be called the brief functional duction of an experimental variable. The re-
analysis. It was designed to accommodate a versal design was used in 43 functional anal-
90-min outpatient evaluation, thereby cir- ysis studies (15.5%) and was more common
cumventing limitations posed by the use of in studies evaluating a single source of influ-
indirect assessment while addressing the ence on behavior or in those employing an
practical limitations posed by more lengthy AB model. Although the reversal design is a
functional analysis. Basically, one or two ses- time-consuming strategy for evaluating mul-
sions were conducted under various test con- tiple sources of behavioral control, Vollmer,
ditions to determine the function of the tar- Iwata, Duncan, and Lerman (1993) provid-
get behavior. The current review identified ed evidence that reversal designs may be
36 studies (13.0%) that employed brief helpful if the rapidly alternating conditions
160 GREGORY P. HANLEY et al.

of the multielement designs produce inter- (Hopkins, Cole, & Mason, 1998; Huitema,
action effects (Higgins Hains & Baer, 1989). 1986; Michael, 1974). Thus, it is not sur-
Iwata, Duncan, Zarcone, Lerman, and prising that most functional analysis data
Shore (1994) described a method for con- have been depicted in line charts displaying
ducting functional analyses that combined individual session values (208 studies or
features of the reversal and multielement de- 75.1%). However, 74 functional analysis
signs. Test conditions were implemented se- studies (26.7%) reported condition means
quentially (as in a reversal design); however, only (in text or bar charts), which is some-
each test condition was alternated with a what troublesome because it eliminates ac-
control condition in a multielement format. cess to aspects of the data (changes in level,
The sequential testcontrol (or pairwise) de- trend, or stability) that could influence con-
sign was intended to minimize interaction clusions about behavioral function.
effects while decreasing the number of re- In contrast to displaying data as whole-
versals required to demonstrate a functional session values, some authors have binned
relation and was shown to result in differ- data within smaller time intervals (e.g., 1
entiated outcomes for 2 participants whose min) from each session to view more fine-
prior multielement analyses yielded unclear grained trends (Kahng & Iwata, 1999; Voll-
results. mer, Iwata, Zarcone, Smith, & Mazaleski,
The pairwise design has appeared in six 1993a, 1993b; Vollmer et al., 1995). It is
subsequent functional analysis studies. Three possible that differences in responding across
studies used the design from the outset of conditions may go undetected if data are
the assessment (Fisher, Kuhn, & Thompson, collapsed as session means (Roane, Lerman,
1998; Lalli, Casey, & Kates, 1995; Shirley, Kelley, & Van Camp, 1999); however, the
Iwata, Kahng, Mazaleski, & Lerman, 1997), utility of viewing within-session patterns has
and all analyses yielded a clear behavioral not been established because the same con-
function. Two studies provided systematic clusions typically can be drawn regardless of
replications of the strategy described by Iwa- whether data are portrayed as sessions means
ta, Duncan, Zarcone, Lerman, and Shore or within-session patterns. A notable excep-
(1994), in which pairwise analyses yielded tion was reported by Vollmer et al. (1993b),
clear results following initially unclear mul- who showed that 1 participants SIB, which
tielement outcomes (Piazza, Fisher, et al., was high and undifferentiated across six
1997; Piazza, Hanley, et al., 1997). functional analysis sessions, revealed extinc-
tion effects in the play and alone conditions
Data Display and Analysis and maintenance of SIB in the attention
Researchers in behavior analysis have long condition. These effects were apparent only
relied on visual inspection of data for draw- when within-session data were examined.
ing conclusions about the effects of experi-
mental variables. Visual analysis of data is an STIMULUS PARAMETERS
attractive tool because it allows researchers Researchers typically arrange similar an-
(a) to view much of the raw data, (b) to tecedent conditions (i.e., low levels of atten-
detect interesting changes in behavior (e.g., tion, presentation of instructions) and, if in-
extinction bursts), (c) to analyze data on a corporated, similar consequent events (i.e.,
continual basis (as opposed to waiting until attention, escape from instructions) during
all of the data have been collected), and (d) the conditions of a functional analysis. How-
to assess effects of experimental variables ever, some researchers have evaluated specific
without relying on inferential statistics aspects of these antecedent and consequent
FUNCTIONAL ANALYSIS 161

events or have incorporated unusual varia- Pace, et al. (1994) described a fixed cycle of
tions of these events in their functional anal- condition presentation (alone, attention,
yses. These methodological variations, cate- play, demand) that maximized EOs during
gorized as either antecedent or consequent assessment. For example, if problem behav-
variables, are described below. ior were maintained by attention, the alone
condition would provide presession depri-
Variations in Antecedent Events vation from attention, whereas the noncon-
Most of the antecedent events manipulat- tingent attention delivered during the play
ed in functional analyses (e.g., amount of condition should eliminate attention depri-
attention) may best be conceptualized as vation (or at least minimize carryover from
EOs that influence behavior by altering the the preceding attention session). Presession
reinforcing effectiveness of some conse- variables other than assessment sessions per
quence (Michael, 1982). This is in contrast se that may exist immediately prior to as-
to SDs, which influence behavior through sessment observation also have been dem-
their correlation with the differential avail- onstrated to influence responding during
ability of reinforcement. Both can be con- functional analyses (Berg et al., 2000;
trasted with other descriptive labels for an- OReilly, 1999; OReilly & Carey, 1996).
tecedent variables (setting events, contextual More specifically, OReilly showed that levels
variables) that do not specify a particular of SIB were higher during a contingent at-
source of influence in any responsereinforc- tention test condition when presession atten-
er relation (see Iwata, 1994, and Smith & tion was withheld compared to when pre-
Iwata, 1997, for discussions). Although pro- session attention was available on a rich
cedural labels have often been used in the schedule. These results suggest that certain
literature, antecedent manipulations can be condition sequences may facilitate rapid re-
discussed in the context of particular types sponse differentiation, especially during out-
of reinforcement relations (e.g., social-posi- patient evaluations in which sessions are
tive, social-negative, automatic). The evoca- conducted with little time between sessions.
tive effects of low levels of attention for at- However, this strategy may exert little influ-
tention-maintained problem behavior, low ence if extended periods of time expire be-
levels of ambient stimulation for automati- tween sessions, in which case researchers
cally reinforced behavior, and the presenta- should consider the establishing effects of
tion of instructions for escape-maintained specific presession variables and program
behavior have been repeatedly demonstrated conditions that maximize the effects of as-
in the functional analysis literature. How- sessment contingencies.
ever, some authors have suggested strategies During the attention test condition in
for either increasing the influence exerted by ABC functional analyses, the antecedent
these typical antecedent events or demon- event typically involves having the therapist
strating functional control of qualitatively engage in a solitary activity. Mace, Page,
different antecedent events. Ivancic, and OBrien (1986) introduced an
Variations for social-positive reinforcement interesting variation of the attention condi-
relations. Responding in a particular test tion, referred to as divided attention, in
condition may be influenced by the extent which the therapist attends to another per-
to which the putative reinforcer is available son in the room, and the utility of this novel
prior to the actual test session. Taking ad- arrangement was demonstrated in subse-
vantage of the establishing (and abolishing) quent studies (Fisher, Kuhn, & Thompson,
effects of functional analysis sessions, Iwata, 1998; OReilly, Lancioni, King, Lally, &
162 GREGORY P. HANLEY et al.

Dhomhnaill, 2000; J. C. Taylor, Sisson, 1993; Lee, Sugai, & Horner, 1999; J. T.
McKelvey, & Trefelner, 1993). For example, Taylor, Ekdahl, Romanczyk, & Miller, 1994;
Taylor et al. showed that the functional re- Vaughn & Horner, 1997; Weeks & Gay-
lation between problem behavior and con- lord-Ross, 1981); however, the functional
tingent attention was dependent on the ther- basis for these influences has been difficult
apist attending to another person by show- to specify in the absence of reinforcement
ing zero or near-zero rates of problem be- contingencies (Smith & Iwata, 1997).
havior during a typical attention condition By contrast, Smith et al. (1995) evaluated
and high rates of problem behavior during a EOs in the presence of an escape contingen-
divided-attention condition. Other interest- cy for problem behavior and showed that
ing antecedent variations that have been several aspects of the demand situation (task
shown to influence responding include ar- novelty, duration of instructional session,
ranging for the therapist to leave the room and rate of task presentation) altered the ef-
following the delivery of reinforcement fects of negative reinforcement in different
(Vollmer et al., 1998) or situating the client ways across individuals. The strategy of
in particular positions (sitting in a wheel- maintaining a negative reinforcement con-
chair as opposed to sitting on a mat; Adelin- tingency while manipulating aspects of the
is, Piazza, Fisher, & Hanley, 1997). Al- antecedent condition to identify idiosyncrat-
though these variations and their results may ic EOs was used in several studies in which
seem idiosyncratic, they illustrate the general otitis media (OReilly, 1997), sleep depri-
strategy described by Smith, Iwata, Goh, vation (OReilly, 1995), amount of attention
and Shore (1995) of assessing antecedent in- or instruction during prior classroom con-
fluences by holding a reinforcement contin- ditions (OReilly & Carey, 1996), or various
gency constant while manipulating the an- instructional procedures (McComas, Hoch,
tecedent event of interest. These studies also Paone, & El-Roy, 2000) were shown to af-
show that the influence of antecedent events fect levels of problem behavior in demand
may be best understood in the context of conditions. Collectively, these studies are ex-
contingencies and indicate that the influence emplary in demonstrating the effects of an-
of idiosyncratic antecedent events should be tecedent events (temporally proximate or
considered when typical analyses fail to un- distant) on the occurrence of negatively re-
cover functional relations. inforced problem behavior.
Variations for social-negative reinforcement Although task instructions are typically
relations. In most tests for social-negative re- programmed in tests for negatively rein-
inforcement relations, some form of task de- forced problem behavior, researchers have
mand is presented as a means of establishing demonstrated the evocative effects of other
the reinforcing efficacy of escape and, con- types of EOs such as medical examinations
sequently, of evoking escape-maintained (Iwata, Pace, Kalsher, Cowdery, & Cataldo,
problem behavior. The identification of idi- 1990), noise or other auditory stimulation
osyncratic antecedent events (i.e., task diffi- (Derby et al., 1994; OReilly, 1997; Smith
culty, lack of choice among tasks, curricular et al., 1995), and social interaction (Frea &
influences, social interaction) has been Hughes, 1997; Vollmer et al., 1998). Con-
prominent in studies employing an AB func- tinued improvements in the assessment and
tional analysis model (e.g., DePaepe et al., treatment of negatively reinforced problem
1996; Dunlap, Kern-Dunlap, Clarke, & behavior may be realized by additional re-
Robbins, 1991; E. G. Durand & Carr, search that documents the influence of (a)
1991; Kennedy, 1994; Kennedy & Itkonen, different classes of EOs (e.g., loud noises),
FUNCTIONAL ANALYSIS 163

(b) temporally proximate events that occur illustrates a method for assessing the influ-
within demand conditions (e.g., pace of in- ence of temporally distant antecedent events
struction), and (c) temporally distant events on the occurrence of behavior (in this case,
(e.g., illness) that culminate in establishing automatically reinforced SIB) during func-
the value of escape. tional analyses. Caution should be taken (as
Variations for automatic reinforcement re- OReilly suggested) in concluding that re-
lations. The influence of antecedent events spite care was an EO for this individuals
on problem behavior maintained by auto- behavior because, although respite care and
matic reinforcement (beyond those typically the occurrence of SIB were correlated, the
manipulated, such as low levels of ambient functional relation was unknown. The next
stimulation during the alone condition) has step to be taken in such analyses is to iden-
not often been examined in the functional tify the critical events that are correlated
analysis literature. Several studies (e.g., Fish- with respite care (deprivation from stimula-
er, Lindauer, Alterson, & Thompson, 1998; tion in this case?) that may have greater
Goh et al., 1995; Piazza, Adelinis, Hanley, functional significance to the maintenance of
Goh, & Delia, 2000) have shown that access SIB. From these extended analyses, treat-
to certain leisure materials may compete ments of greater scope and effectiveness may
with the stimulation produced by problem be derived.
behavior, thereby diminishing its reinforcing
effects. Manipulations of this sort are typi- Variations in Consequent Events
cally involved in functional analyses. For in- Consequence manipulations in functional
stance, a variety of toys are included in the analysis research are organized and discussed
control (i.e., play) and the attention condi- according to their properties of quality, type,
tions but typically are absent in the test con- duration, and schedule.
dition for automatic reinforcement. Al- Quality or type. The qualitative aspects of
though antecedent events are not typically attention, delivered as positive reinforce-
altered prior to or during test conditions for ment, often are described only briefly: Re-
automatically reinforced problem behaviors, searchers typically note that reprimands
Van Camp et al. (2000) showed that unusual (e.g., dont do that, you might hurt your-
antecedent events (a specific toy, social in- self ) and brief physical contact such as a
teraction) evoked 2 childrens stereotypic be- pat on the back or a touch to the shoulder
havior that persisted in the absence of social are provided by adults on a contingent basis.
contingencies. However, several studies have shown that the
OReilly (1996) described another notable source of attention may be an important fac-
exception in which an individuals SIB did tor. For instance, the problem behavior of
not occur during a functional analysis for 35 some students has been shown to be sensi-
out of 40 days; however, SIB persisted across tive to attention provided by peers but not
all conditions (including an alone condition) by adults (Broussard & Northup, 1997;
during the 5 days that were preceded by Lewis & Sugai, 1996; Northup et al., 1995,
nights spent at a respite care facility. The 1997). Although the critical variables re-
effect of spending a night at the respite fa- sponsible for observed differences were pre-
cility was then systematically manipulated, sumably qualitative (e.g., form or intensity
and the results showed that high rates of SIB of attention) or historical (e.g., children may
were observed only following nights at the have customarily attended to the problem
respite facility (as opposed to nights spent at behavior of the participants), these factors
home). This study is exemplary in that it were not directly evaluated and therefore
164 GREGORY P. HANLEY et al.

represent an interesting area of future re- It may be difficult to determine behavior-


search. al function if the reinforcing value of events
Fisher, Ninness, Piazza, and Owen-De- presented during assessment varies across
Schryver (1996) published the initial study time. An example of this phenomenon was
showing some forms of attention (repri- described by Bowman, Fisher, Thompson,
mands) functioned as reinforcement for and Piazza (1997), whose initial functional
problem behavior, whereas others (state- analyses yielded unclear results for 2 partic-
ments unrelated to the problem behavior) ipants. Informal observations suggested that
did not. Richman and Hagopian (1999) and problem behavior was occasioned by paren-
Piazza et al. (1999) also demonstrated im- tal noncompliance with each childs mands
portant qualitative differences in the rein- (i.e., when the parent did not deliver or re-
forcing effectiveness of attention. Initial move something the child had requested, the
functional analyses in the Richman and Ha- child engaged in problem behavior). During
gopian study showed undifferentiated pat- assessment, problem behavior was observed
terns of responding. Parental interview and at its highest rates when the therapist com-
informal observation suggested that the form plied with the childs mands following oc-
of the attention used in the initial analyses currences of problem behavior. Subsequent
(i.e., verbal reprimands) was different than analyses showed that when mands were im-
that provided by caregivers who either deliv- mediately reinforced, zero or near-zero rates
ered exaggerated vocal (high level of voice of problem behavior occurred. Two things
intonation and dramatic description of the were unique about the relations described by
problem behavior) or physical attention (be- Bowman et al. First, problem behavior did
ing picked up and held) following problem not appear to be maintained by access to any
behaviors. Incorporation of these idiosyn- one particular type of reinforcement. Sec-
cratic types of attention into subsequent ond, the event that evoked problem behavior
functional analyses yielded differentiated re- was specified by the participant prior to its
sults. occurrence (via manding). The generality of
Tests for behavioral maintenance by atten- this relation between mands and multiple re-
tion as well as by food, toys, and particular inforcement contingencies for problem be-
activities have been demonstrated often havior has yet to be determined, but the
within functional analyses (e.g., Durand & strategy described by Bowman et al. could
Crimmins, 1988; Vollmer et al., 1995). A be a promising way to identify variables that
unique form of social-positive reinforcement influence problem behaviors exhibited by
that has been shown to support SIB is that persons with verbal repertoires. A second ex-
of restraint. Smith, Lerman, and Iwata ample of problem behavior that appeared to
(1996) showed that access to self-restraint be motivated by different events across time
(self-initiated confinement) functioned as re- was described by Fisher, Adelinis, Thomp-
inforcement for SIB in a woman with pro- son, Worsdell, and Zarcone (1998). Follow-
found mental retardation, and Vollmer and ing initial undifferentiated functional analy-
Vorndran (1998) replicated these findings. ses, several subsequent analyses showed that
Although SIB and self-restraint may be re- problem behavior was evoked by instruc-
lated in several different ways (see Fisher & tions to engage in behavior that interfered
Iwata, 1996), these studies demonstrate that with an individuals current ongoing activity,
idiosyncratic and mundane materials (e.g., and that problem behavior was maintained
particular clothing types) may serve as pos- by resumption of that activity. These results
itive reinforcers for severe problem behavior. suggest that, although instructions may
FUNCTIONAL ANALYSIS 165

evoke problem behavior, it is possible that unequal across the tests for attention, tan-
the target response could be maintained by gible, and escape conditions, with attention
positive reinforcement (resumption of a pre- delivery being brief and tangible and escape
ferred activity) rather than by negative re- consequences lasting for 30 s. Durations of
inforcement (termination of an aversive reinforcement were equal (escape, tangible,
event). and attention consequences were all 30 s) in
By contrast, Adelinis and Hagopian the second and fourth phases. Rates of prob-
(1999) found that the type of instruction lem behavior were higher in the three test
that interrupted the preferred activity (do conditions relative to the play (control) con-
vs. dont requests) was influential in evok- dition in all phases. However, markedly
ing problem behavior. Specifically, they higher rates of responding were observed in
found that dont requests that interrupted the attention condition when relative rein-
an activity (e.g., dont lie on the floor) forcer durations were unequal, whereas sim-
evoked problem behavior, whereas symmet- ilar levels of responding across the three test
rical do requests (e.g., sit in a chair) did conditions were observed when reinforcer
not. These results suggest that the form of durations were equal. These data suggest
the instruction, in addition to the context in that the relative duration of reinforcement
which it is delivered, may contribute to the (and related EOs) should be considered
control of problem behavior for some indi- when designing or interpreting functional
viduals. Additional studies in this area may analyses. In other words, responding was
help to clarify the respective roles of partic- higher in the attention condition of the typ-
ular forms of instructions and the type of ical functional analysis, not because atten-
relations involved. Nevertheless, these stud- tion was a more potent reinforcer or that
ies provide sound experimental evidence of problem behavior was primarily sensitive to
complex behavior relations that, once dis- attention as reinforcement, but, rather, be-
covered, lead to effective interventions. cause there was simply more opportunity to
Duration. Consequences used in function- respond under relevant deprivation (i.e., ab-
al analysis conditions are typically delivered sence of the reinforcing event). One may
briefly, which permits repeated contact be- wish to equate reinforcement duration (and
tween problem behavior and the pro- the length of exposure to the EOs) during
grammed contingency within a session. Al- functional analyses to avoid interpretative
though the duration of reinforcement has difficulties. Alternatively, it may be best (a)
varied across conditions within a study and to include a condition that controls for the
within the same condition across studies, at- effects of the contingencies in the test con-
tention typically is delivered for 5 to 10 s, ditions and (b) always to compare rates of
tangible items are delivered for 30 s, and problem behavior in each test condition to
escape is provided for either 30 s or the re- that of the control (as opposed to rates of
mainder the intertrial interval (ranging from responding in other test conditions) when
1 to 29 s). determining behavioral function.
The influence of reinforcer (and related Schedule. Most studies arrange conse-
EO) duration on the outcomes of functional quences on a continuous reinforcement
analyses was examined by Fisher, Piazza, and (CRF) schedule during functional analyses
Chiang (1996) in an ABAB design. During such that each occurrence of problem be-
the first and third phases, which typified havior results in a programmed reinforcer
most ABC multielement functional analyses, (216 of the 241 studies [89.6%] employing
the relative durations of reinforcement were an ABC model incorporated CRF schedules
166 GREGORY P. HANLEY et al.

Table 5
Functional Analysis Outcome Summary

Undifferen- Differen-
Topography tiated tiated Escape Attention Tangible Automatic Multiple

Self-injury 13 222 65 59 28 55 15
Aggression 2 50 24 9 6 1 10
Property destruction 0 2 0 0 2 0 0
Pica 0 6 0 1 0 3 2
Disruption 0 16 11 3 1 1 0
Vocalizations 1 14 6 3 1 0 4
Noncompliance 0 8 1 2 1 0 4
Elopement 0 3 0 0 0 0 3
Stereotypy 1 30 6 0 0 19 5
Tantrums 0 6 2 1 1 0 2
Other 0 13 4 5 0 1 3
Aberrant 5 144 57 47 12 1 27
Total number 22 514 176 130 52 81 75
Percentage of sample 4.1 95.9 34.2 25.3 10.1 15.8 14.6

in the analysis). Intermittent schedules were reinforcing contingency on a sufficient num-


used in 10 studies (4.1%), and the type of ber of occasions early in the functional anal-
schedule was not clear in 15 studies. Studies ysis and thus may necessitate more lengthy
that incorporated intermittent schedules analyses. (d) Higher rates or intense bursts
(e.g., Kern, Carberry, & Haidara, 1997; Lal- of behavior may be engendered by intermit-
li & Casey, 1996; Mace et al., 1986; Paisey, tent reinforcement relative to a CRF sched-
Whitney, & Hislop, 1991; Sturmey, Carlsen, ule. A direct comparison of the types of per-
Crisp, & Newton, 1988) typically based the formances and outcomes generated by CRF
schedule on data generated from a descrip- and intermittent schedules in functional
tive assessment in which a caregiver was ob- analyses is an interesting and necessary fu-
served to deliver the putative reinforcers in- ture area of research.
termittently. The ecological validity of the
functional analysis may be enhanced SUMMARY OF FUNCTIONAL
through such an approach, and the potential ANALYSIS OUTCOMES
of problem behavior entering into novel re- Through 2000, 536 graphed individual
sponsereinforcer relations may be decreased data sets (with at least one data point per
with the use of leaner schedules. However, observation session) have been published de-
the use of intermittent schedules in func- picting the results of functional analyses (see
tional analyses may pose certain problems: Table 5). The majority of these graphs (514
(a) Lengthy descriptive assessments would be or 95.9%) were interpreted by their authors
required to identify the parameters of the as representing differentiated outcomes.
intermittent schedule. (b) Intermittent Large proportions of differentiated function-
schedules, although derived from descriptive al analyses showed behavioral maintenance
assessment, do not represent the actual through social-negative (34.2%) and social-
schedule that generated or maintains the positive reinforcement (35.4%). More spe-
problem behavior outside the environment cifically, 25.3% showed maintenance via at-
in which specific interactions were observed. tention and 10.1% via access to tangible
(c) Problem behavior may not contact the items. Automatic reinforcement was impli-
FUNCTIONAL ANALYSIS 167

cated in 15.8% of cases. Finally, multiple re- dent, such that function cannot be predicted
inforcement contingencies were identified in by the topography of problem behavior.
14.6% of cases. A small proportion of cases Less than 5% of the studies have included
(4.1%) were interpreted as undifferentiated undifferentiated functional analysis results
by their authors. such that an assessment-based course of
Conclusions regarding the multiple con- treatment could not be identified. Given
trol of aberrant behavior are somewhat trou- that publication contingencies generally fa-
blesome given that the aberrant behavior vor positive findings, this low percentage of
category is comprised of multiple response undifferentiated results may not be represen-
topographies. Therefore, the data beg the tative of the actual failure rate in clinical set-
question as to whether each topography of tings. Although publication of only assess-
problem behavior was sensitive to multiple ment failures is rare, many studies included
reinforcers or whether different behavioral in the present review described initially un-
topographies served single (but different) clear results that were clarified by one of sev-
behavioral functions. The studies in which eral strategies: (a) inclusion of idiosyncratic
one response topography was analyzed may antecedent and consequent variables during
provide a more accurate estimate of the prev- subsequent functional analyses (Bowman et
alence of multiply controlled behavior. In al., 1997; Fisher, Lindauer, Alterson, &
addition, more analyses such as those con- Thompson, 1998; Thompson, Fisher, Piaz-
ducted by Smith, Iwata, Vollmer, and Zar- za, & Kuhn, 1998), (b) altering the experi-
cone (1993), in which different function- mental design (e.g., Iwata, Duncan, Zarco-
based treatments are assessed as a means of ne, Lerman, & Shore, 1994; Piazza, Fisher,
supporting conclusions of multiple control, et al., 1997) or aspects of the experimental
may provide more rigorous demonstrations arrangement (Conners et al., 2000) to facil-
of multiply controlled behavioral phenome- itate discrimination across conditions, or (c)
na. including assessments of the efficacy of the
As evident in the epidemiological study putative reinforcers for problem behavior to
conducted by Iwata, Pace, et al. (1994) and strengthen alternative behaviors (e.g., Steege,
the present analysis, the function of SIB Wacker, Berg, Cigrand, & Cooper, 1989).
varies across individuals and necessitates in- In addition, Vollmer et al. (1995) described
dividualized assessment. Other topographies a methodology that progressed from rela-
show trends that may suggest a particular tively brief assessments to more extended
function for a given topography of problem analyses that resulted in clear and replicable
behavior. For instance, an overwhelming response patterns for 85% of participants.
majority of functional analyses identified es-
cape as the reinforcer for aggression and au-
tomatic reinforcement as supporting stereo- DISCUSSION
typy. However, there have been exceptions In considering historical and current func-
to the predominant function for both to- tional analysis research, two final areas of
pographies, and a relatively small number of discussion seem warranted: (a) experimental
functional analyses have been conducted ex- integrity and (b) the ecological validity of
clusively for each topography. Even consid- functional analysis. The importance of spe-
ering the trends in the summary of function cific issues related to these broad topics will
across topography, it appears that behavioral be presented, as will future directions for re-
function and topography remain indepen- search and suggestions for best practices.
168 GREGORY P. HANLEY et al.

EXPERIMENTAL INTEGRITY tomatic reinforcement) would persist in


Functional Analysis Models these assessments. One possible explanation
is that observed rates of problem behavior
Both general models of functional analysis may simply reflect an early stage of extinc-
attempt to identify behavioral function so as tion (i.e., elevated rates of responding). Be-
to facilitate development of an effective, cause the majority of studies employing the
function-based treatment; however, the ABC AB model also conduct few observations per
model provides a more rigorous demonstra- condition in a reversal design, the latter stage
tion of causation. The AB model is consid- of the extinction process (i.e., near-zero rates
ered a functional analysis in that a functional of responding) may not be readily apparent
relation is demonstrated between an envi- in the data. An alternative possibility is that
ronmental event and problem behavior, and AB assessments involve responding during
from this, situations in which problem be- extinction that culminates in reinforcement
havior is more likely are clearly identified. (e.g., problem behavior may be followed by
However, because putative reinforcers are attention that is provided in every third in-
not manipulated in the AB analysis, the terval) and therefore the problem behavior is
source of reinforcement for problem behav- maintained even though reinforcement is
ior must be inferred on the basis of the cor- not programmed for problem behavior. In
relation between behavior and the anteced- essence, the processes responsible for behav-
ent conditions under which a contingency is ioral persistence during the AB functional
likely to operate. analysis model may be related to by-prod-
However, patterns of responding observed ucts of extinction processes (Goh & Iwata,
in AB analyses may sometimes lead to er- 1994) or adventitious reinforcement (Voll-
roneous conclusions about behavioral func- mer, Ringdahl, Roane, & Marcus, 1997).
tion. For example, demands may occasion By contrast, the ABC model of functional
problem behavior not because they establish analysis involves a strong contingency be-
escape as negative reinforcement but because tween problem behavior and putative rein-
the prompting may signal that attention forcers (typically, every response results in
(positive reinforcement) is available for the programmed consequence of a given
problem behavior (Vollmer, Iwata, Smith, & condition) in the presence of strong EOs
Rodgers, 1992). Alternatively, high levels of (putative reinforcement is available only fol-
problem behavior observed when low levels lowing the emission of problem behavior).
of antecedent attention are arranged may be Whereas AB assessments rely on inference,
indicative of an attention function or may probabilistic contingencies, or indirect ef-
reflect the evocative effects of a relatively fects of extinction in determining behavioral
barren environment on automatically rein- function, an ABC model demonstrates func-
forced behavior. However, it should be noted tional relations between antecedent events
that this could also be a limitation of ABC and behavior (e.g., low levels of antecedent
assessments when the quality of attention attention are associated with high levels of
delivered contingent on problem behavior problem behavior) as well as the adaptive
does not compete effectively with the puta- function of problem behavior (e.g., the level
tive automatic reinforcer. of problem behavior is high when and only
Because AB assessments do not arrange when attention is arranged as a conse-
social reinforcement for problem behavior, it quence).
is somewhat counterintuitive that target be- Given the apparent advantages of con-
haviors (other than those maintained by au- ducting ABC analyses, it is difficult to de-
FUNCTIONAL ANALYSIS 169

termine why AB analyses continue to be analyses do not appear to be associated with


conducted as a method for assessing behav- greater risk or less efficiency than do AB
ioral function. It cannot be because AB anal- analyses. Moreover, given the greater preci-
yses are more efficientthe delivery of con- sion of the ABC analysis, the best practice
sequences takes hardly any time. It may be recommendation is to include the manipu-
argued that ABC analyses are less attractive lation of consequences in functional analy-
because they contain explicit reinforcement ses.
contingencies for problem behavior (i.e., in-
creased risk to the client), whereas AB anal- Number of Topographies in the
yses do not. However, both assessments are Contingency Class
designed to produce occurrences of problem When several topographies of problem
behavior; therefore, both assessments place behavior are included in the same functional
the client at a similar risk of harm from en- analysis, interpretation based on examina-
gaging in problem behavior. tion of aggregate data may obscure impor-
Finally, some authors (e.g., Martin, Gaf- tant findings. For instance, Derby et al.
fan, & Williams, 1999; Sturmey, 1995) have (1994, 2000) showed that different forms of
suggested that arranging explicit contingen- problem behavior (e.g., aggression and ste-
cies may result in new learning during as- reotypy) appeared to have different functions
sessment. However, the contingent events when graphed separately. A more rigorous
manipulated in functional analyses are either strategy for determining if multiple topog-
mundane events (e.g., attention or escape) raphies belong to the same response class
that are typically delivered for problem be- (i.e., are maintained by the same reinforce-
haviors (Thompson & Iwata, 2001) or are ment) is to arrange extinction contingencies
derived from clinical interview or observa- for the predominant responses and then to
tion (Fisher, Kuhn, & Thompson, 1998), so observe reinforcement effects in the remain-
the likelihood of establishing new response ing responses (e.g., Lalli, Mace, Wohn, &
reinforcer relations seems low. Incidental Livezey, 1995; Richman, Wacker, Asmus,
maintenance of problem behavior has been Casey, & Andelman, 1999). Magee and Ellis
observed rarely during a functional analysis (2000) arranged contingencies for multiple
and only when highly preferred tangible topographies of behavior (out of seat, yell-
items were presented following problem be- ing, inappropriate language, and destruction
havior in the absence of information sug- for 1 participant; object mouthing, destruc-
gesting that a contingency may have existed tive and aggressive behavior for a 2nd par-
(Shirley, Iwata, & Kahng, 1999). But, this ticipant), and only one of the topographies
finding underscores the importance of clin- was observed consistently in a single test
ical interview or observation prior to arrang- condition for each participant. The authors
ing contingencies between problem behavior then implemented extinction for the most
and atypical and potentially potent tangible frequent response and observed a decrease in
reinforcers during functional analyses. In ad- those responses and increases in other re-
dition, identifying new contingencies that sponses. This procedure was continued
support problem behavior is not necessarily across the remaining topographies, and sim-
unimportant. It may lead to measures that ilar effects were observed, thereby demon-
prevent the development of new reinforce- strating functional relations for each distinct
ment contingencies in the natural environ- topography.
ment (i.e., redirection to preferred items If additional graphing solutions (e.g.,
may be strongly discouraged). In sum, ABC Derby et al., 1994, 2000) or extinction as-
170 GREGORY P. HANLEY et al.

sessments (Magee & Ellis, 2000) are not em- and rigor of some of the control conditions
ployed, some functional relations may go varied.
unnoticed when multiple topographies of Although the play condition described by
behavior are included in the same contin- Iwata et al. (1982/1994) has functioned as
gency class. In such cases, separate analyses an effective control condition for the major-
for single topographies may increase the like- ity of functional analyses, several variations
lihood of developing effective treatments have been described. The first involves the
(see Thompson et al., 1998, for an example manner in which attention is delivered: (a)
of this phenomenon). Practical consider- according to a fixed-time 30-s schedule, (b)
ations are probably the strongest justification in conjunction with a brief omission contin-
for including more than one topography in gency, or (c) continuously available. These
a functional analysis. This approach may not arrangements differ in the extent to which
be deleterious if the multiple topographies the EO or contingency involving attention
of behavior do indeed serve the same func- is removed. The second variation involves
tion; however, this cannot be known prior the type of leisure materials provided. Most
to conducting a functional analysis. In ad- studies do not specify the manner in which
dition, as the number of different topogra- items were selected for inclusion, but several
phies of behavior included in the contingen- strategies are available: (a) Toys are selected
cy class increases, so does the likelihood that that are thought to be sufficiently stimulat-
some target behaviors will serve different ing to compete with automatically rein-
functions (resulting in an undifferentiated forced behavior, (b) the most highly pre-
assessment outcome). From this, the rec- ferred items based on a systematic preference
ommendation for best practice is to mini- assessment are selected, (c) toys are selected
mize the number of different topographies from those that caregivers report can be
of problem behavior included in a single found in the natural environment, or, if a
functional analysis. If multiple topographies test for behavioral sensitivity to tangible re-
are included in the analysis, then each to- inforcement is arranged, (d) the same items
pographical class of behavior should be sub- as used in the tangible test condition are se-
jected to extinction or separate topographies lected. To date, no systematic evaluation of
these strategies has been undertaken, yet this
of behavior should be graphically analyzed
is an important area of future research be-
to verify membership in the reinforced class.
cause without effective control conditions,
Parameters of Control Conditions determination of behavioral function is un-
likely.
A distinguishing feature of applied behav- In contrast to the various ways of con-
ior analysis is its emphasis on single-subject trolling for the effects of positive reinforce-
experimental designs, in which each partic- ment during functional analyses, the effects
ipant serves as his or her own control in eval- of negative reinforcement are typically con-
uating the effects of independent variables. trolled by providing continuous escape (i.e.,
Control conditions are based on the general absence of instructions) in the play condi-
strategy of retaining all features of the ex- tion. However, Kahng and Iwata (1998) sug-
perimental condition with the exception of gested that the alone condition might be a
the contingency of interest (Barlow & Her- better control for the evaluation of negative
sen, 1984; Rescorla, 1967). All of the func- reinforcement on problem behavior (that
tional analysis studies included some type of does not require the presence of another per-
control condition; however, the specific type son), because the SD (a person who may
FUNCTIONAL ANALYSIS 171

have a history of delivering instructions) as its and limitations of using a single control
well as the EO (instructions) and contingen- condition to assess the effects of several test
cy (escape following problem behavior) are conditions versus arranging a specific control
absent from the alone condition. The au- condition (with only one altered variable)
thors provided supporting data, in that for each test condition. Best practice rec-
greater differences between alone and de- ommendations regarding the selection of
mand conditions were observed for most control conditions are highly dependent on
participants relative to the differences ob- time constraints. If efficiency is required, de-
served between play and demand conditions. signing a single control condition that pro-
A strength of the control conditions vides high levels of noncontingent access to
(whether play or alone) in ABC functional all reinforcers to be tested while eliminating
analyses is the efficiency with which infor- the contingency between reinforcers and
mation regarding behavioral function can be problem behavior is recommended. If the
obtainedmultiple sources of influence can situation calls for the most thorough analy-
be assessed through comparison with a single sis, designing individual control conditions
condition. In addition, large discrepancies matched to each test condition is recom-
between test and control conditions are ar- mended.
ranged (i.e., strong contingencies and EOs
are present in the test conditions, whereas Tests for Automatic Reinforcement
these same events are typically absent from Slightly more than 40% of the functional
the control condition), which presumably analysis studies did not include a test for be-
result in more rapid differentiation between havioral persistence in the absence of social
rates of problem behavior across test and contingencies. Unfortunately, the omission
control conditions. However, these same as- of conditions controlling for the effect of so-
pects may compromise the experimental in- cial events may lead to erroneous conclu-
tegrity of the analysis in that multiple fea- sions about behavioral function. For exam-
tures of the environment are altered across ple, high rates of behavior observed in a con-
test and control conditions (i.e., specifica- dition in which low levels of attention are
tion of the primary sources of behavioral in- scheduled may reflect the effects of depri-
fluence may be difficult given the multiple vation from attention (social reinforcement);
differences between test and control condi- alternatively, the behavior could be auto-
tions). The extent to which this may be a matically reinforced and observed only un-
problem has not been clearly demonstrated. der conditions of low stimulation (Daven-
Nevertheless, examples of analyses that ex- port & Berkson, 1963).
clusively controlled for a contingency ar- The type of problem behavior examined
ranged in test conditions were provided by (e.g., aggression) in many studies may imply
Fisher, Kuhn, and Thompson (1998), Siga- maintenance through social reinforcement,
foos and Meikle (1996), and Sigafoos and suggesting that a test for automatic rein-
Saggers (1995); these authors arranged for forcement may not be warranted. This also
the reinforcer used as a consequence in the may be problematic. For example, Thomp-
test condition to be available continuously son et al. (1998) showed that some forms of
and noncontingently in the control condi- a young boys aggression (e.g., hitting, kick-
tion while all other features of the environ- ing) were socially mediated; whereas another
ment remained unchanged across test and (chin grinding) was automatically reinforced
control conditions. Future research could be by the stimulation directly produced by the
directed towards evaluating the relative mer- response. The authors provided further sup-
172 GREGORY P. HANLEY et al.

port for these conclusions by demonstrating by stimuli common across assessment con-
the effectiveness of separate treatments ditions (e.g., people).
matched to the specific function of each Because interventions for problem behav-
form of aggression. iors maintained by either social or automatic
Although the test for automatically rein- reinforcement involve the manipulation of
forced problem behavior typically consists of very different contingencies and environ-
observing the persistence of responding in a mental events, it is important to be able to
condition in which the individual is alone, distinguish between the two types of main-
some authors have used an ignore condition taining contingencies. Therefore, it is rec-
in which a person is present but does not ommended that functional analyses include
provide social consequences for responding. tests for automatic reinforcement that min-
This modification is included in analyses be- imize all features of the environment that
cause some behavior cannot occur in the ab- may occasion socially mediated responding.
sence of another person (e.g., aggression), Consistent with strategies described by Goh
particularly intense SIB may require block- et al. (1995) and Piazza et al. (1998), future
ing if it exceeds a particular frequency, or the research should proceed beyond demonstrat-
assessment environment does not allow un- ing behavioral persistence in the absence of
obtrusive observation. A potential problem social consequences and continue analyses to
with the ignore condition as a test for au- identify the specific features of the environ-
tomatically reinforced behavior is that an SD ment that serve to maintain automatically
for socially mediated behavior is present. Be- reinforced problem behavior.
cause social reinforcement is not delivered
following occurrences of problem behavior Alternative Methods of Data Analysis
during ignore conditions, socially mediated Interpretation of results from functional
problem behavior should be extinguished. analyses typically is done through visual in-
However, Hanley, Piazza, Fisher, and Adelin- spection of graphed data, a process that is
is (1997) showed that the presence of an- somewhat informal. Several authors have
other person might exert powerful stimulus suggested using either explicit, structured
control over responding to the extent that criteria during visual inspection (Hagopian
their participants attention-maintained et al., 1997; Toogood & Timlin, 1996) or
problem behavior persisted under extinction more formal statistical analysis of the data
(when the person was present) for 50 con- (Martin et al., 1999). Hagopian et al. de-
secutive sessions (problem behavior was rap- veloped a set of formal criteria for visual in-
idly extinguished in the absence of the per- spection of multielement functional analysis
son). These data suggest that elevated re- data based on consensus by experts, which
sponse rates across all assessment conditions resulted in higher interrater agreement than
(suggestive of automatically maintained re- that resulting from unstructured visual in-
sponding or of undifferentiated responding) spection. These results suggested that (a) de-
may be under the discriminative control of terminations of behavioral function may be
the presence of a person. Conducting true less reliable than generally assumed, (b) rules
alone sessions in which the SD (a person) for used by experts to determine behavioral
social reinforcement is absent or creating function can be operationalized, and (c) in-
novel discriminative control by correlating dividuals with a limited history of interpret-
conditions with different cues (Conners et ing functional analysis data can be trained
al., 2000) may help to clarify undifferenti- to apply these rules to improve accuracy and
ated analyses that result from control exerted consistency. A strength of the interpretative
FUNCTIONAL ANALYSIS 173

strategy described by Hagopian et al. is that analysis (or which functional assessment
it established a consistent basis for evaluating strategy) is more accurate is that there is no
complex data arrays (e.g., the relation be- universal standard for comparison. Compar-
tween trends and levels of responding in par- ing the efficacy of different courses of action
ticular test and control conditions) that typ- (e.g., treatments) suggested by either differ-
ically requires a more extensive history with ent interpretative methods or different types
functional analysis data. of assessment may be helpful on a practical
Martin et al. (1999) compared a modified level, but this would not directly address the
version of the structured criteria described issue of accuracy. Although an apparent need
by Hagopian et al. (1997; 50% or more ses- for more studies on the interrater agreement
sions in the highest condition had to be 1 and accuracy of visual inspection or statisti-
SD above the mean rates in the control con- cal strategies for interpreting functional anal-
dition for a determination of behavioral ysis data is suggested by the Hagopian et al.
function) with another set of criteria pro- (1997), Martin et al. (1999), and Toogood
posed by Toogood and Timlin (1996; 50% and Timlin (1996) studies, an alternative
or more sessions in the highest condition line of research might attempt to identify the
had to be at least 50% higher than the over- historical and methodological variables that
all assessment mean to ascribe behavioral give rise to noisy and difficult-to-interpret
function) and with a probability-based sta- functional analysis outcomes, thereby reduc-
tistical procedure (modified z score) for in- ing the need for more subtle (statistical) in-
terpreting functional analysis results. The terpretive strategies. That is, tighter control
authors found generally low agreement over influential variables would lead to more
across the three interpretive strategies. They easily interpretable, differentiated functional
also evaluated the validity of each strategy analyses. Therefore, a best practice recom-
by examining the number of assessments for mendation is to continue to refine and in-
which each interpretive strategy yielded a be- dividualize the various components of func-
havioral function, and found that the prob- tional analysis methodology until a clear (vi-
ability-based procedure resulted in the larg- sual) determination of behavioral function
est number of identified functions. However, can be made.
the fact that an analytic strategy yields a de-
termination of behavior function does not ECOLOGICAL VALIDITY
imply that it would be the basis for effective
Assessment Settings and Therapists
therapeutic action because the behavioral
functions identified via the probability-based Although the settings in which functional
statistical procedure may have represented analyses are conducted may vary (e.g.,
false-positive outcomes. school, hospital, etc.), most are conducted
This same difficulty is present in other under well-controlled conditions that may
studies that compared outcomes of various not closely resemble the settings in which
functional assessments (indirect assessments, the problem behavior of interest typically oc-
descriptive or functional analyses) by com- curs. For example, a functional analysis in
paring the number of assessments for which the school may be conducted in the corner
a determination of behavioral function was of the classroom away from the other stu-
achieved (e.g., Toogood & Timlin, 1996). In dents (e.g., J. Taylor & Miller, 1997). Thus,
other words, simply guessing will always functional analyses usually are conducted in
yield a behavioral function. The main diffi- settings that are neutral with respect to be-
culty in determining which method of data havioral history. The advantage of arranging
174 GREGORY P. HANLEY et al.

controlled conditions away from the natural able only following aggression; in the second
environment is that changes in rates of be- minute, the same reinforcer was available
havior across conditions can be attributed to continuously. Using these brief and intermit-
the variables explicitly manipulated by the tent test-control trials, the authors observed
experimenter; thus, conclusions regarding differential responding for both participants.
behavioral function can be derived with con- This approach was replicated (Sigafoos &
fidence. Meikle, 1996) with 2 boys with autism who
Nevertheless, several authors have ques- engaged in multiple problem behaviors, and
tioned the ecological validity of functional successful function-based treatments were
analysis methodology based on the fact that again prescribed. Just as the brief functional
assessment takes place outside the natural analysis (Northup et al., 1991) serves as a
environment (these authors typically refer to practical and effective substitute when more
functional analyses as analogue assessments), thorough analyses cannot be conducted, the
therefore rendering conclusions about the approach described by Sigafoos and col-
function of behavior somewhat questionable leagues may be an attractive option for en-
(e.g., Conroy, Fox, Crain, Jenkins, & Bel- hancing the ecological validity of behavioral
cher, 1996; Martin et al., 1999; Sturmey, assessment. Strategies such as these may lead
1995). Ecological validity is used here to in- to an understanding of the natural conditions
dicate the extent to which functional rela- under which problem behaviors occur more
tions tested in the analysis are consistent readily than indirect or descriptive assessment
with those that operate in the natural envi- methods, which omit methodological re-
ronment. In other words, the functional quirements (e.g., direct observation and ma-
analysis may identify a responsereinforcer nipulation) for isolating behavioral phenom-
relation that is not necessarily the same as ena. However, a potential limitation of this
the one that maintains problem behavior ei- approach is that the complex nature of nat-
ther at home or at school. Several researchers urally occurring events may compromise pro-
have seemingly circumvented this issue by cedural integrity (i.e., lack of control over the
(a) conducting sessions during a childs typ- type, quality, schedule, and duration of the
ical routine in the home (e.g., Arndorfer et programmed consequences). This may lead
al., 1994; H. M. Day et al., 1994; Ellingson to extended assessment durations or may pro-
et al., 2000), (b) embedding sessions in a hibit entirely the accurate identification of
childs normal mealtime routine (Paisey et specific variables that influence problem be-
al., 1991), or (c) conducting sessions during havior.
typical classroom activities (e.g., Lalli, Brow- Another strategy for increasing the eco-
der, Mace, & Brown, 1993; Northup et al., logical validity of functional analyses in-
1995, 1997; Sasso et al., 1992; Umbreit, volves incorporating into assessment sessions
1995a, 1995b). individuals who have a previous history of
An interesting approach to conducting interaction with the person exhibiting prob-
functional analyses under naturalistic condi- lem behavior, such as parents or other family
tions was described by Sigafoos and Saggers members (e.g., Reimers et al., 1993; Um-
(1995), who assessed the effects of contingent breit, 1996; Vollmer et al., 1996), teachers
attention, tangible items, or escape from task- (e.g., Mace, Yankanich, & West, 1989; Wat-
related instructions during 20 2-min trials son, Ray, Turner, & Logan, 1999), or class-
distributed at various times across the childs room peers (e.g., Broussard & Northup,
school day. Each trial consisted of two parts. 1995; J. E. Carr et al., 1996). Results of a
In the first minute, the reinforcer was avail- preliminary study by Ringdahl and Sellers
FUNCTIONAL ANALYSIS 175

(2000) suggest that having caregivers deliver ically and without supplementary informa-
programmed consequences during function- tion.
al analyses may facilitate clearer outcomes; One general criticism found in several
however, additional research is needed to de- commentaries on functional analysis meth-
termine the utility of this practice. In other odology that appeared in the 1994 special
words, the extent to which conducting ses- issue of JABA was that functional analyses
sions in the natural environment during typ- do not adequately sample all relevant aspects
ical routines or having caregivers or teachers of the controlling environment. Some of
conduct assessment sessions will improve the these aspects include (a) physiological or in-
efficiency or accuracy of a functional analysis ternal states related to illness or drugs (E. G.
is still not well known. Large-scale direct Carr, 1994), (b) structural aspects of rein-
comparisons of ecologically valid and more forcers (Carr) or EOs (Horner, 1994), or (c)
tightly controlled analyses are needed to temporally distant events that may influence
identify both the benefits and limitations of the occurrence of problem behavior (Repp,
incorporating natural features of the clients 1994). The authors referred to these poten-
environment into analysis conditions. With tially important antecedent events as either
respect to best practice, it seems reasonable setting events (Carr; Repp) or contextual
to incorporate as many features of the cli- variables (Carr; Horner) and lamented their
ents natural environment into the assess- lack of consideration in functional analysis
ment conditions as possible, as long as the methodology. They also suggested that the
integrity of the experimental arrangements is identification of these contextual variables
monitored to insure that procedures are im- would result in more accurate behavioral as-
plemented as specified. sessment and ultimately more effective and
durable treatments. Relevant to the present
Supplemental Assessments discussion, several authors suggested that a
Many studies described the use of prelim- greater emphasis on either indirect or de-
inary assessments (indirect or descriptive scriptive assessments might resolve this par-
procedures) to facilitate the functional anal- ticular limitation of functional analysis
ysis process. Although these types of assess- methodology (Carr; Horner; Mace, 1994;
ment may be helpful in structuring more Repp).
precise functional analyses of problem be- Mace (1994) suggested that descriptive as-
havior, the data derived from these supple- sessments might be useful in identifying id-
mentary assessments have been presented iosyncratic reinforcers or schedules, which
rarely, and the relevance of the supplemen- could be programmed subsequently in a
tary assessment data in the studies that did functional analysis. However, the extent to
include them is not well known (Iwata, which complex descriptive analyses would
1994). In other words, it is possible that improve the efficiency or accuracy of a func-
analysis outcomes would be the same wheth- tional analysis has not been demonstrated
er or not results from additional assessments empirically. In fact, indirect and descriptive
were used as a basis for structuring the func- assessment strategies (e.g., Durand & Crim-
tional analysis. The definitive study yet to mins, 1988; Mace & Lalli, 1991) do not
be conducted would involve conducting two necessarily sample a wider range of EOs or
simultaneous but procedurally different reinforcers. Instead, they seem to represent a
functional analyses, one of which would be different way of gathering information about
based on results of a supplementary assess- the same variables manipulated in a func-
ment and the other would be structured typ- tional analysis, and, as a result, would not
176 GREGORY P. HANLEY et al.

enhance the accuracy or ecological validity tired or ill, the impact of these EOs can be
of the functional analysis. observed by conducting functional analysis
By contrast, results from several studies conditions before and following exercise,
have suggested that unstructured observa- nights of disturbed sleep, or bouts of illness.
tions conducted outside the assessment con- Unique and temporally proximal or distant
text sometimes have been helpful in identi- EOs may exert control over behavior main-
fying unusual events that, once incorporated tained by a given contingency (e.g., negative
into a functional analysis, resulted in the reinforcement); however, these effects can
identification of functional relations (see only be verified in the context of experimen-
Bowman et al., 1997; Fisher, Adelinis, tal manipulation. In other words, progress
Thompson, Worsdell, & Zarcone, 1998; from general classes of maintaining contin-
Fisher, Kuhn, & Thompson, 1998; Fisher, gencies (e.g., positive vs. negative reinforce-
Lindauer, Alterson, & Thompson, 1998; ment or social vs. automatic reinforcement)
Richman & Hagopian, 1999; Thompson et to more individualized environmentbehav-
al., 1998). The authors of these studies did ior relations (i.e., increased precision) might
not use formal descriptive analysis tech- be best accomplished through more careful
niques, questionnaires, or rating scales; in- experimental manipulation (e.g., Smith et
stead, they merely had observers document al., 1995; Thompson et al., 1998) rather
unusual aspects of caregiverchild interac- than through more extensive indirect or de-
tions or environmental conditions that were scriptive strategies.
present when problem behavior occurred. To summarize, all forms of functional as-
This process may be best characterized as an sessment are limited in that potentially im-
open-ended descriptive assessment in which portant, idiosyncratic reinforcers and EOs
the antecedent and consequent events are may go unnoticed. Identifying idiosyncratic
not specified prior to the observation. It is a events prior to conducting a functional anal-
process that is more akin to the narrative ysis is not always essential for determining
recording technique described by Bijou et al. behavioral function and designing effective
(1968), which permits the identification of intervention (e.g., sleep deprivation may act
idiosyncratic types of EOs (e.g., talking to as an EO and increase escape-reinforced
another person) and consequences (e.g., spe- problem behavior, but escape extinction
cific type of attention) that typically would should be effective even if the relation be-
not be included in a functional analysis and tween sleep and problem behavior goes un-
also would not be captured via questionnaire detected). However, observing idiosyncratic
or descriptive analysis. As illustrated in the features of the natural environment when
studies cited, these events, once identified, functional analyses yield undifferentiated re-
may be incorporated into a functional anal- sults is highly recommended. Future re-
ysis when initial analyses yield uninformative search should establish guidelines for exam-
results. ining natural environments for idiosyncratic
If idiosyncratic events that may influence features that may be relevant to the main-
behavior are identified via supplemental as- tenance of problem behavior.
sessment, their functional role as EOs or re-
inforcers still requires demonstration by way PRACTICAL ENHANCEMENTS
of a functional analysis. For instance, if the
hypothesis derived from clinical observation Tests for Single Sources of Influence
is that SIB is maintained by escape from in- Although most functional analysis re-
structions but only when the individual is search includes tests for multiple sources of
FUNCTIONAL ANALYSIS 177

control, assessment of single responserein- observation of problem behavior under mul-


forcer relations continues to a lesser extent tiple test and control conditions.
(at least two studies each year since 1990
have tested for single sources of control). As- Discriminative Stimuli
sessment strategies that focus on single re- Conners et al. (2000) attempted to min-
sponsereinforcer relations may be beneficial imize interaction effects within functional
under some conditions. For example, if an- analyses by increasing the discriminability of
ecdotal information or descriptive assess- assessment conditions. Multielement func-
ment data are strongly suggestive of a par- tional analyses were conducted on the SIB
ticular source of influence, that variable can or aggression of 8 adults with mental retar-
be evaluated quickly, and treatment matched dation in which each condition initially was
to that function can be initiated prior to or correlated with a specific therapist and room
concurrent with tests of other sources of in- color. In subsequent analyses, the pro-
fluence. grammed SDs were removed (i.e., all condi-
tions were conducted by the same therapist
Brief Session Duration in the same room). Results indicated that
Wallace and Iwata (1999) examined the the inclusion of distinctive visual SDs facili-
extent to which variations in session dura- tated differential responding during multi-
tion (5, 10, or 15 min) affected the out- element functional analyses in half of the
comes of functional analyses. All of the 10- participants. Strengthening the stimulus
control exerted in each analysis condition
min data sets yielded interpretations identi-
may be especially helpful when conducting
cal to those based on the 15-min data sets
(a) brief assessments (e.g., Northup et al.,
(i.e., 100% agreement), whereas three of the
1991), (b) sessions of short duration (e.g., 5
5-min data sets yielded interpretations dif-
min), or (c) assessments in the same setting
ferent than those based on the 15-min data
(e.g., classroom). Future research could iden-
sets (i.e., 93.5% agreement). Their results
tify a range of stimuli that facilitate discrim-
suggested that assessment efficiency could be
ination under different situations (e.g., par-
improved with little or no loss in clarity by
ticular seating positions in the classroom) or
conducting brief (5- or 10-min) sessions. by different individuals (e.g., scents with
Brief Assessment Duration persons who are blind), so that therapists
can construct efficient assessment procedures
Kahng and Iwata (1999) compared data in spite of practical constraints imposed by
sets from 50 full functional analyses (35 of time, setting, or personnel.
which showed clear response patterns and 15
of which were undifferentiated) with those Proceeding from Undifferentiated Analyses
from brief assessments that were constructed Although not commonly reported, undif-
by isolating the first session of each condi- ferentiated analysis outcomes occur, most of-
tion from the full analyses. The outcomes of ten when problem behavior occurs sporadi-
the brief assessments corresponded with cally or at very low rates. As noted previ-
those of the full analyses in 66% of cases. ously, the inclusion of information gathered
The results of this study, along with numer- through other sources (e.g., about idiosyn-
ous replications of the procedures described cratic EOs, qualitative aspects of reinforce-
by Northup et al. (1991), suggest that the ment) in functional analyses may clarify ini-
brief functional analysis may be adequate tially undifferentiated results (see E. G. Carr
when circumstances do not permit repeated et al., 1997, for an example). Including
178 GREGORY P. HANLEY et al.

stimuli from the clients natural environment tings. Systematic growth in the use of func-
or conducting the analysis in the environ- tional analysis methodology as a primary
ment in which the problem behavior occurs method of behavioral assessment and, more
(e.g., Sigafoos & Saggers, 1995) also may generally, as a means of studying environ-
increase the likelihood of a differentiated as- mentbehavior relations is evident in the
sessment outcome (although this has not sharply increasing trend in the publication
been empirically demonstrated). Minimizing rates of functional analysis research (see Fig-
the number of response topographies in the ure 1).
contingency class (e.g., Thompson et al., At the present time, functional analysis re-
1998) or graphing response topographies search has not yielded an established set of
separately (Derby et al., 1994, 2000) may rules for conducting an assessment; however,
also yield clear assessment outcomes. When best practices are beginning to emerge. As
functional analyses based on brief session noted in the Discussion, these practices in-
duration (5 to 15 min) yield undifferenti- clude (a) limiting response classes to one or
ated results, observing the effects of contin- a few behavior topographies, (b) program-
gencies over longer periods (hours, days, or ming consequences for the occurrence of tar-
weeks) may allow relevant EOs to operate get behaviors, (c) incorporating EO influ-
for a sufficient amount of time to evoke ences before and during assessment, (d) in-
problem behavior and yield clear assessment cluding SDs to facilitate discrimination of
results (e.g., Arndorfer et al., 1994; Reese, test conditions, (e) conducting relatively
1997). Finally, arranging contingencies to brief (e.g., 10-min) sessions, (f ) including
follow reported precursors to the target re- tests to identify behavior maintained by au-
sponses or even arbitrary responses (see Pi- tomatic reinforcement, (g) considering rela-
azza, Hanley, et al., 1997, Grace, Thomp- tive reinforcement durations when interpret-
son, & Fisher, 1996) may demonstrate be- ing analysis results, (h) testing for functional
havioral sensitivity to particular forms of re- relations between problem behavior and tan-
inforcement that can then be incorporated gible reinforcement only when preliminary
into treatments for problem behavior (this assessment information suggests a relation
strategy may be an improvement over arbi- might exist, (i) starting brief and simple (i.e.,
trarily selecting an intervention, but it does arranging common test conditions) and pro-
not determine the actual function of prob- gressing to more lengthy or complex assess-
lem behavior). ments as needed, and (j) using other sources
of information (e.g., open-ended interviews
and observations) as adjuncts to structure
CONCLUSION the more complex analyses.
Several researchers have suggested that Although functional analysis has been re-
there has been a lack of systematic extension peatedly shown to be a powerful behavioral
of functional analysis methodology (e.g., assessment tool for prescribing effective
Gable, 1996; Gresham et al., 1999; Sturmey, treatments, concerns have been raised about
1995). However, it is apparent from the the feasibility of conducting functional anal-
present review that comprehensive function- yses in typical service settings (e.g., class-
al analysis models (E. G. Carr & Durand, rooms) due to either time requirements (Ap-
1985; Iwata et al., 1982/1994) have gener- plegate, Matson, & Cherry, 1999; Pyles,
ated a voluminous database of extensions as Riordan, & Bailey, 1997) or the level of
well as replications across a wide range of training and clinical expertise needed to in-
client populations, target behaviors, and set- sure procedural fidelity (Crawford, Brockel,
FUNCTIONAL ANALYSIS 179

Schauss, & Miltenberger, 1992; Sturmey, extension to the evaluation of environment


1994). However, as noted earlier, over 80 behavior relations that influence clinical
functional analysis studies have been con- phenomena constitutes a clear improvement
ducted in school settings. In addition, Iwata over alternative strategies.
et al. (2000) noted that the time commit-
ment appears no greater than that required
by a descriptive assessment and presented a REFERENCES
2-hr strategy for teaching the basic skills nec- Adelinis, J. D., & Hagopian, L. P. (1999). The use
essary for conducting a functional analysis. of symmetrical do and dont requests to inter-
Combining these instructional procedures rupt ongoing activities. Journal of Applied Behavior
Analysis, 32, 519523.
with information regarding best practice in Adelinis, J. D., Piazza, C. C., Fisher, W. W., & Han-
conducting functional analyses and analyz- ley, G. P. (1997). The establishing effects of client
ing their results should produce a set of basic location on self-injurious behavior. Research in De-
velopmental Disabilities, 18, 383391.
competencies in clinicians and educators Applegate, H., Matson, J. L., & Cherry, K. E. (1999).
who are charged with assessing and treating An evaluation of functional variables affecting se-
problem behaviors. vere problem behaviors in adults with mental re-
As with all behavioral assessment strate- tardation by using the questions about behavioral
function scale (QABF). Research in Developmental
gies, it is important to consider ethical issues Disabilities, 20, 229237.
when implementing functional analysis Arndorfer, R. E., Miltenberger, R. G., Woster, S. H.,
methodology. Questions arise concerning Rortvedt, A. K., & Gaffaney, T. (1994). Home-
based descriptive and experimental analysis of
the necessity of a rigorous functional analysis problem behaviors in children. Topics in Early
and the potential harm to the individual that Childhood Special Education, 14, 6487.
may be incurred during assessment. Wheth- Axelrod, S. (1991). Smoking cessation through func-
tional analysis. Journal of Applied Behavior Analy-
er a problem behavior necessitates such rig- sis, 24, 717718.
orous analysis may be partially addressed by Bachman, J. A. (1972). Self-injurious behavior: A be-
considering the short- and long-term cost havioral analysis. Journal of Abnormal Psychology,
(or danger) of ineffective treatment versus 80, 211224.
Baer, D. M., Wolf, M. M., & Risley, T. R. (1968).
the potential harm that may be engendered Some current dimensions of applied behavior
during the course of a functional analysis. analysis. Journal of Applied Behavior Analysis, 1,
Consideration of brief assessments, careful 9197.
Barlow, D. H., & Hersen, M. (1984). Single case ex-
structuring of analyses, and evaluations of perimental designs (2nd ed.). Needham Heights,
within-session patterns may be appropriate MA: Allyn & Bacon.
measures to avoid lengthy behavioral assess- Berg, W. K., Peck, S., Wacker, D. P., Harding, J.,
McComas, J., Richman, D., et al. (2000). The
ment. effects of presession exposure to attention on the
Although a number of permutations of results of assessments of attention as a reinforcer.
functional analysis procedures have been de- Journal of Applied Behavior Analysis, 33, 463477.
scribed, further refinements are still needed Bijou, S. W., Peterson, R. F., & Ault, M. H. (1968).
A method to integrate descriptive and experimen-
to improve the efficiency and generality of tal field studies at the level of data and empirical
functional analysis methodology while its concepts. Journal of Applied Behavior Analysis, 1,
precision is maintained. Developing practi- 175191.
Bowman, L. G., Fisher, W. W., Thompson, R. H., &
cal yet precise assessment strategies may be Piazza, C. C. (1997). On the relation of mands
an elusive goal, but its realization can be and the function of destructive behavior. Journal
achieved only through systematic refinement of Applied Behavior Analysis, 30, 251265.
based on sound research. Experimental Broussard, C. D., & Northup, J. (1995). An ap-
proach to functional assessment and analysis of
methods have led to a greater understanding disruptive behavior in regular education class-
of a wide range of phenomena, and their rooms. School Psychology Quarterly, 10, 151164.
180 GREGORY P. HANLEY et al.

Broussard, C., & Northup, J. (1997). The use of ment and intervention. Journal of Applied Behavior
functional analysis to develop peer interventions Analysis, 27, 279289.
for disruptive classroom behavior. School Psychol- Day, R. M., Rea, J. A., Schussler, N. G., Larsen, S.
ogy Quarterly, 12, 6576. E., & Johnson, W. L. (1988). A functionally
Carr, E. G. (1977). The motivation of self-injurious based approach to the treatment of self-injurious
behavior: A review of some hypotheses. Psycholog- behavior. Behavior Modification, 12, 565589.
ical Bulletin, 84, 800816. DePaepe, P. A., Shores, R. E., Jack, S. L., & Denny,
Carr, E. G. (1994). Emerging themes in the func- R. K. (1996). Effects of task difficulty on the
tional analysis of problem behavior. Journal of Ap- disruptive and on-task behavior of students with
plied Behavior Analysis, 27, 393399. severe behavior disorders. Behavioral Disorders, 21,
Carr, E. G., & Durand, V. M. (1985). Reducing be- 216225.
havior problems through functional communica- Derby, K. M., Hagopian, L., Fisher, W. W., Richman,
tion training. Journal of Applied Behavior Analysis, D., Augustine, M., Fahs, A., et al. (2000). Func-
18, 111126. tional analysis of aberrant behavior through mea-
Carr, E. G., Newsom, C. D., & Binkoff, J. A. (1976). surement of separate response topographies. Jour-
Stimulus control of self-destructive behavior in a nal of Applied Behavior Analysis, 33, 113117.
psychotic child. Journal of Abnormal Child Psy- Derby, K. M., Wacker, D. P., Peck, S., Sasso, G.,
chology, 4, 139153. DeRaad, A., Berg, W., et al. (1994). Functional
Carr, E. G., Newsom, C. D., & Binkoff, J. A. (1980). analysis of separate topographies of aberrant be-
Escape as a factor in the aggressive behavior of havior. Journal of Applied Behavior Analysis, 27,
two retarded children. Journal of Applied Behavior 267278.
Analysis, 13, 101117. Derby, K. M., Wacker, D. P., Sasso, G., Steege, M.,
Carr, E. G., Yarbrough, S. C., & Langdon, N. A. Northup, J., Cigrand, K., et al. (1992). Brief
(1997). Effects of idiosyncratic stimulus variables functional assessment techniques to evaluate ab-
errant behavior in an outpatient setting: A sum-
on functional analysis outcomes. Journal of Ap-
mary of 79 cases. Journal of Applied Behavior Anal-
plied Behavior Analysis, 30, 673686.
ysis, 25, 713721.
Carr, J. E., Taylor, C. C., Wallander, R. J., & Reiss,
Dunlap, G., Kern-Dunlap, L., Clarke, S., & Robbins,
M. L. (1996). A functional-analytic approach to
F. R. (1991). Functional assessment, curricular
the diagnosis of a transient tic disorder. Journal of revision, and severe behavior problems. Journal of
Behavior Therapy and Experimental Psychiatry, 27, Applied Behavior Analysis, 24, 387397.
291297. Durand, V. M., & Carr, E. G. (1991). Functional
Chapman, S., Fisher, W., Piazza, C. C., & Kurtz, P. communication training to reduce challenging be-
F. (1993). Functional assessment and treatment havior: Maintenance and application in new set-
of life-threatening drug ingestion in a dually di- tings. Journal of Applied Behavior Analysis, 24,
agnosed youth. Journal of Applied Behavior Anal- 251264.
ysis, 26, 255256. Durand, V. M., & Crimmins, D. B. (1987). Assess-
Conners, J., Iwata, B. A., Kahng, S., Hanley, G. P., ment and treatment of psychotic speech in an au-
Worsdell, A. S., & Thompson, R. H. (2000). tistic child. Journal of Autism and Developmental
Differential responding in the presence and ab- Disorders, 17, 1728.
sence of discriminative stimuli during multiele- Durand, V. M., & Crimmins, D. B. (1988). Identi-
ment functional analyses. Journal of Applied Be- fying the variables maintaining self-injurious be-
havior Analysis, 33, 299308. havior. Journal of Autism and Developmental Dis-
Conroy, M., Fox, J., Crain, L., Jenkins, A., & Belcher, orders, 18, 99117.
K. (1996). Evaluating the social and ecological Ellingson, S. A., Miltenberger, R. G., Stricker, J. M.,
validity of analog assessment procedures for chal- Garlinghouse, M. A., Roberts, J., & Galensky, T.
lenging behaviors in young children. Education L. (2000). Analysis and treatment of finger suck-
and Treatment of Children, 19, 233256. ing. Journal of Applied Behavior Analysis, 33, 41
Crawford, J., Brockel, B., Schauss, S., & Miltenberger, 52.
R. G. (1992). A comparison of methods for the Fisher, W. W., Adelinis, J. D., Thompson, R. H.,
functional assessment of stereotypic behavior. Worsdell, A. S., & Zarcone, J. R. (1998). Func-
Journal of the Association for Persons with Severe tional analysis and treatment of destructive behav-
Handicaps, 17, 7786. ior maintained by termination of dont (and
Davenport, R. K., & Berkson, G. (1963). Stereo- symmetrical do) requests. Journal of Applied Be-
typed movements in mental defectives: II. Effects havior Analysis, 31, 339356.
of novel objects. American Journal of Mental De- Fisher, W. W., & Iwata, B. A. (1996). On the func-
ficiency, 67, 879882. tion of self-restraint and its relationship to self-
Day, H. M., Horner, R. H., & ONeill, R. E. (1994). injury. Journal of Applied Behavior Analysis, 29,
Multiple functions of problem behaviors: Assess- 9398.
FUNCTIONAL ANALYSIS 181

Fisher, W. W., Kuhn, D. E., & Thompson, R. H. to extinction in attention-maintained SIB. Re-
(1998). Establishing discriminative control of re- search in Developmental Disabilities, 18, 251260.
sponding using functional and alternative rein- Haynes, S. N., & OBrien, W. H. (1990). Functional
forcers during functional communication training. analysis in behavior therapy. Clinical Psychology
Journal of Applied Behavior Analysis, 31, 543560. Review, 10, 649668.
Fisher, W. W., Lindauer, S. E., Alterson, C. J., & Higgins Hains, A. H., & Baer, D. M. (1989). Inter-
Thompson, R. H. (1998). Assessment and treat- action effects in multielement designs: Inevitable,
ment of destructive behavior maintained by ste- desirable, and ignorable. Journal of Applied Behav-
reotypic object manipulation. Journal of Applied ior Analysis, 22, 5769.
Behavior Analysis, 31, 513527. Hopkins, B. L., Cole, B. L., & Mason, T. L. (1998).
Fisher, W. W., Ninness, H. A. C., Piazza, C. C., & A critique of the usefulness of inferential statistics
Owen-DeSchryver, J. S. (1996). On the reinforc- in applied behavior analysis. The Behavior Analyst,
ing effects of the content of verbal attention. Jour- 21, 125137.
nal of Applied Behavior Analysis, 29, 235238. Horner, R. H. (1994). Functional assessment: Con-
Fisher, W. W., Piazza, C. C., & Chiang, C. L. (1996). tributions and future directions. Journal of Applied
Effects of equal and unequal reinforcer duration Behavior Analysis, 27, 401404.
during functional analysis. Journal of Applied Be- Huitema, B. E. (1986). Statistical analysis and single-
havior Analysis, 29, 117120. subject designs: Some misunderstandings. In A.
Fisher, W. W., Piazza, C. C., & Hanley, G. P. (1998). Poling & R. W. Fuqua (Eds.), Research methods in
Informing readers of the presence of data com- applied behavior analysis (pp. 209232). New
mon to multiple investigations. Journal of Applied York: Plenum.
Behavior Analysis, 31, 703704. Iwata, B. A. (1994). Functional analysis methodolo-
Frea, W. D., & Hughes, C. (1997). Functional anal- gy: Some closing comments. Journal of Applied Be-
ysis and treatment of social-communicative be- havior Analysis, 27, 413418.
havior of adolescents with developmental disabil- Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K.
ities. Journal of Applied Behavior Analysis, 30, 701 E., & Richman, G. S. (1994). Toward a func-
704. tional analysis of self-injury. Journal of Applied Be-
Gable, R. A. (1996). A critical analysis of functional havior Analysis, 27, 197209. (Reprinted from
assessment: Issues for researchers and practition- Analysis and Intervention in Developmental Dis-
ers. Behavioral Disorders, 22, 3640. abilities, 2, 320, 1982)
Goh, H., & Iwata, B. A. (1994). Behavioral persis- Iwata, B. A., Duncan, B. A., Zarcone, J. R., Lerman,
tence and variability during extinction of self-in- D. C., & Shore, B. A. (1994). A sequential, test-
jury maintained by escape. Journal of Applied Be- control methodology for conducting functional
havior Analysis, 27, 173174. analyses of self-injurious behavior. Behavior Mod-
Goh, H., Iwata, B. A., Shore, B. A., DeLeon, I. G., ification, 18, 289306.
Lerman, D. C., Ulrich, S. M., et al. (1995). An Iwata, B. A., Kahng, S., Wallace, M. D., & Lindberg,
analysis of the reinforcing properties of hand J. S. (2000). The functional analysis model of
mouthing. Journal of Applied Behavior Analysis, behavioral assessment. In J. Austin & J. E. Carr
28, 269283. (Eds.), Handbook of applied behavior analysis (pp.
Grace, N. C., Thompson, R., & Fisher, W. W. 6190). Reno, NV: Context Press.
(1996). The treatment of covert self-injury Iwata, B. A., Pace, G. M., Dorsey, M. F., Zarcone, J.
through contingencies on response products. Jour- R., Vollmer, T. R., Smith, R. G., et al. (1994).
nal of Applied Behavior Analysis, 29, 239242. The functions of self-injurious behavior: An ex-
Gresham, F. M., Quinn, M. M., & Restori, A. perimental-epidemiological analysis. Journal of Ap-
(1999). Methodological issues in functional anal- plied Behavior Analysis, 27, 215240.
ysis: Generalizability to other disability groups. Iwata, B. A., Pace, G. M., Kalsher, M. J., Cowdery,
Behavioral Disorders, 24, 180182. G. E., & Cataldo, M. F. (1990). Experimental
Groden, G. (1989). A guide for conducting a com- analysis and extinction of self-injurious escape be-
prehensive behavioral analysis of a target behavior. havior. Journal of Applied Behavior Analysis, 23,
Journal of Behavior Therapy and Experimental Psy- 1127.
chiatry, 20, 163169. Iwata, B. A., Wallace, M. D., Kahng, S., Lindberg, J.
Hagopian, L. P., Fisher, W. W., Thompson, R. H., S., Roscoe, E. M., Conners, J., et al. (2000). Skill
Owen-DeSchryver, J., Iwata, B. A., & Wacker, D. acquisition in the implementation of functional
P. (1997). Toward the development of structured analysis methodology. Journal of Applied Behavior
criteria for interpretation of functional analysis Analysis, 33, 181194.
data. Journal of Applied Behavior Analysis, 30, Kahng, S., & Iwata, B. A. (1998). Play versus alone
313326. conditions as controls during functional analyses
Hanley, G. P., Piazza, C. C., Fisher, W. W., & Adelin- of self-injurious escape behavior. Journal of Applied
is, J. D. (1997). Stimulus control and resistance Behavior Analysis, 31, 669672.
182 GREGORY P. HANLEY et al.

Kahng, S., & Iwata, B. A. (1999). Correspondence Mace, F. C. (1994). The significance and future of
between outcomes of brief and extended func- functional analysis methodologies. Journal of Ap-
tional analyses. Journal of Applied Behavior Anal- plied Behavior Analysis, 27, 385392.
ysis, 32, 149159. Mace, F. C., Browder, D. M., & Lin, Y. (1987). Anal-
Kennedy, C. H. (1994). Manipulating antecedent ysis of demand conditions associated with stereo-
conditions to alter the stimulus control of prob- typy. Journal of Behavior Therapy and Experimental
lem behavior. Journal of Applied Behavior Analysis, Psychiatry, 18, 2531.
27, 161170. Mace, F. C., & Knight, D. (1986). Functional anal-
Kennedy, C. H., & Itkonen, T. (1993). Effects of ysis and the treatment of severe pica. Journal of
setting events on the problem behavior of students Applied Behavior Analysis, 19, 411416.
with severe disabilities. Journal of Applied Behavior Mace, F. C., & Lalli, J. S. (1991). Linking descriptive
Analysis, 26, 321327. and experimental analyses in the treatment of bi-
Kern, L., Carberry, N., & Haidara, C. (1997). Anal- zarre speech. Journal of Applied Behavior Analysis,
ysis and intervention with two topographies of 24, 553562.
challenging behavior exhibited by a young woman Mace, F. C., Page, T. J., Ivancic, M. T., & OBrien,
with autism. Research in Developmental Disabili- S. (1986). Analysis of environmental determi-
ties, 18, 275287. nants of aggression and disruption in mentally re-
Kern, L., Mauk, J. E., Marder, T. J., & Mace, F. C. tarded children. Applied Research in Mental Retar-
(1995). Functional analysis and intervention for dation, 7, 203221.
breath holding. Journal of Applied Behavior Anal- Mace, F. C., & West, B. J. (1986). Analysis of de-
ysis, 28, 339340. mand conditions associated with reluctant speech.
Lalli, J. S., Browder, D. M., Mace, F. C., & Brown, Journal of Behavior Therapy and Experimental Psy-
D. K. (1993). Teacher use of descriptive analysis chiatry, 17, 285294.
data to implement interventions to decrease stu- Mace, F. C., Yankanich, M. A., & West, B. J. (1989).
dents problem behaviors. Journal of Applied Be-
Toward a methodology of experimental analysis
havior Analysis, 26, 227238.
and treatment of aberrant classroom behaviors.
Lalli, J. S., & Casey, S. D. (1996). Treatment of mul-
Special Services in the Schools, 4, 7188.
tiply controlled problem behavior. Journal of Ap-
Magee, S. K., & Ellis, J. (2000). Extinction effects
plied Behavior Analysis, 29, 391395.
Lalli, J. S., Casey, S., & Kates, K. (1995). Reducing during the assessment of multiple problem behav-
escape behavior and increasing task completion iors. Journal of Applied Behavior Analysis, 33, 313
with functional communication training, extinc- 316.
tion, and response chaining. Journal of Applied Be- Martin, N. T., Gaffan, E. A., & Williams, T. (1999).
havior Analysis, 28, 261268. Experimental functional analyses for challenging
Lalli, J. S., Mace, F. C., Wohn, T., & Livezey, K. behavior: A study of validity and reliability. Re-
(1995). Identification and modification of a re- search in Developmental Disabilities, 20, 125146.
sponse-class hierarchy. Journal of Applied Behavior Matson, J. L., Bamburg, J. W., Cherry, K. E., & Pa-
Analysis, 28, 551559. clawskyj, T. R. (1999). A validity study on the
Lee, Y., Sugai, G., & Horner, R. H. (1999). Using questions about behavioral function (QABF)
an instructional intervention to reduce problem scale: Predicting treatment success for self-injury,
and off-task behaviors. Journal of Positive Behavior aggression, and stereotypies. Research in Develop-
Interventions, 1, 195204. mental Disabilities, 20, 163176.
Lerman, D. C., & Iwata, B. A. (1993). Descriptive McComas, J., Hoch, H., Paone, D., & El-Roy, D.
and experimental analyses of variables maintaining (2000). Escape behavior during academic tasks: A
self-injurious behavior. Journal of Applied Behavior preliminary analysis of idiosyncratic establishing
Analysis, 26, 293319. operations. Journal of Applied Behavior Analysis,
Lewis, T. J., & Sugai, G. (1996). Descriptive and 33, 479493.
experimental analysis of teacher and peer attention McManus, F., & Waller, G. (1995). A functional
and the use of assessment-based intervention to analysis of binge-eating. Clinical Psychology Re-
improve pro-social behavior. Journal of Behavioral view, 15, 845863.
Education, 6, 724. Meyer, K. A. (1999). Functional analysis and treat-
Lovaas, O. I., Freitag, G., Gold, V. J., & Kassorla, I. ment of problem behavior exhibited by elemen-
C. (1965). Experimental studies in childhood tary school children. Journal of Applied Behavior
schizophrenia: Analysis of self-destructive behav- Analysis, 32, 229232.
ior. Journal of Experimental Child Psychology, 2, Michael, J. (1974). Statistical inference for individual
6784. organism research: Mixed blessing or curse? Jour-
Lovaas, O. I., & Simmons, J. Q. (1969). Manipula- nal of Applied Behavior Analysis, 7, 647653.
tion of self-destruction in three retarded children. Michael, J. (1982). Distinguishing between discrim-
Journal of Applied Behavior Analysis, 2, 143157. inative and motivational functions of stimuli.
FUNCTIONAL ANALYSIS 183

Journal of the Experimental Analysis of Behavior, (1999). The impact of functional analysis meth-
37, 149155. odology on treatment choice for self-injurious and
Miltenberger, R. G., Long, E. S., Rapp, J. T., Lumley, aggressive behavior. Journal of Applied Behavior
V., & Elliot, A. J. (1998). Evaluating the function Analysis, 32, 185195.
of hair pulling: A preliminary investigation. Be- Piazza, C. C., Adelinis, J. D., Hanley, G. P., Goh, H.,
havior Therapy, 29, 211219. & Delia, M. D. (2000). An evaluation of the
Northup, J., Broussard, C., Jones, K., George, T., Voll- effects of matched stimuli on behaviors main-
mer, T. R., & Herring, M. (1995). The differ- tained by automatic reinforcement. Journal of Ap-
ential effects of teacher and peer attention on the plied Behavior Analysis, 33, 1327.
disruptive classroom behavior of three children Piazza, C. C., Bowman, L. G., Contrucci, S. A., Delia,
with a diagnosis of attention deficit hyperactivity M. D., Adelinis, J. D., & Goh, H. (1999). An
disorder. Journal of Applied Behavior Analysis, 28, evaluation of the properties of attention as rein-
227228. forcement for destructive and appropriate behav-
Northup, J., Jones, K., Broussard, C., DiGiovanni, G., ior. Journal of Applied Behavior Analysis, 32, 437
Herring, M., Fusilier, I., et al. (1997). A prelim- 449.
inary analysis of interactive effects between com- Piazza, C. C., Fisher, W. W., Hanley, G. P., LeBlanc,
mon classroom contingencies and methylpheni- L. A., Worsdell, A. S., Lindauer, S. E., et al.
date. Journal of Applied Behavior Analysis, 30, (1998). Treatment of pica through multiple anal-
121125. yses of its reinforcing functions. Journal of Applied
Northup, J., Wacker, D., Sasso, G., Steege, M., Cig- Behavior Analysis, 31, 165189.
rand, K., Cook, J., et al. (1991). A brief func- Piazza, C. C., Fisher, W. W., Hanley, G. P., Remick,
tional analysis of aggressive and alternative behav- M. L., Contrucci, S. A., & Aitken, T. L. (1997).
ior in an outclinic setting. Journal of Applied Be- The use of positive and negative reinforcement in
havior Analysis, 24, 509522. the treatment of escape-maintained destructive be-
ONeill, R. E., Horner, R. H., Albin, R. W., Storey, havior. Journal of Applied Behavior Analysis, 30,
K., & Sprague, J. R. (1990). Functional analysis 279298.
of problem behavior: A practical guide. Sycamore, Piazza, C. C., Hanley, G. P., Bowman, L. G., Ruyter,
IL: Sycamore Press. J. M., Lindauer, S. E., & Saiontz, D. M. (1997).
OReilly, M. F. (1995). Functional analysis and treat- Functional analysis and treatment of elopement.
ment of escape-maintained aggression correlated Journal of Applied Behavior Analysis, 30, 653672.
with sleep deprivation. Journal of Applied Behavior Piazza, C. C., Hanley, G. P., & Fisher, W. W. (1996).
Analysis, 28, 225226. Functional analysis and treatment of cigarette
OReilly, M. F. (1996). Assessment and treatment of pica. Journal of Applied Behavior Analysis, 29,
episodic self-injury: A case study. Research in De- 437450.
velopmental Disabilities, 17, 349361. Pinkston, E. M., Reese, N. M., LeBlanc, J. M., &
OReilly, M. F. (1997). Functional analysis of episod- Baer, D. M. (1973). Independent control of a
ic self-injury correlated with recurrent otitis me- preschool childs aggression and peer interaction
dia. Journal of Applied Behavior Analysis, 30, 165 by contingent teacher attention. Journal of Applied
167. Behavior Analysis, 6, 115124.
OReilly, M. F. (1999). Effects of presession attention Pyles, D. A. M., Riordan, M. M., & Bailey, J. S.
on the frequency of attention-maintained behav- (1997). The stereotypy analysis: An instrument
ior. Journal of Applied Behavior Analysis, 32, 371 for examining environmental variables associated
374. with differential rates of stereotypic behavior. Re-
OReilly, M. F., & Carey, Y. (1996). A preliminary search in Developmental Disabilities, 18, 1138.
analysis of the effects of prior classroom condi- Reese, R. M. (1997). Biobehavior analysis of self-in-
tions on performance under analogue analysis jurious behavior in a person with profound hand-
conditions. Journal of Applied Behavior Analysis, icaps. Focus on Autism and Other Developmental
29, 581584. Disabilities, 12, 8794.
OReilly, M. F., Lancioni, G. E., King, L., Lally, G., Reimers, T. M., Wacker, D. P., Cooper, L. J., Sasso,
& Dhomhnaill, O. N. (2000). Using brief as- G. M., Berg, W. K., & Steege, M. W. (1993).
sessments to evaluate aberrant behavior main- Assessing the functional properties of noncompli-
tained by attention. Journal of Applied Behavior ant behavior in an outpatient setting. Child and
Analysis, 33, 109112. Family Behavior Therapy, 15, 115.
Paisey, T. J. H., Whitney, R. B., & Hislop, P. M. Repp, A. (1994). Comments on functional analysis
(1991). Brief report: Non-intrusive operant anal- procedures for school-based behavior problems.
ysis of aggressive behavior in persons with mental Journal of Applied Behavior Analysis, 27, 409411.
retardation. Behavioral Residential Treatment, 6, Rescorla, R. A. (1967). Pavlovian conditioning and
5461. its proper control procedures. Psychological Review,
Pelios, L., Morren, J., Tesch, D., & Axelrod, S. 74, 7180.
184 GREGORY P. HANLEY et al.

Richman, D. M., & Hagopian, L. P. (1999). On the J. R. (1993). Experimental analysis and treatment
effects of quality of attention in the functional of multiply controlled self-injury. Journal of Ap-
analysis of destructive behavior. Research in De- plied Behavior Analysis, 26, 183196.
velopmental Disabilities, 20, 5162. Smith, R. G., Lerman, D. C., & Iwata, B. A. (1996).
Richman, D. M., Wacker, D. P., Asmus, J. M., Casey, Self-restraint as positive reinforcement for self-in-
S. D., & Andelman, M. (1999). Further analysis jurious behavior. Journal of Applied Behavior Anal-
of problem behavior in response class hierarchies. ysis, 29, 99102.
Journal of Applied Behavior Analysis, 32, 269283. Smolev, S. R. (1971). Use of operant techniques for
Ringdahl, J. E., & Sellers, J. A. (2000). The effects the modification of self-injurious behavior. Amer-
of different adults as therapists during functional ican Journal of Mental Deficiency, 76, 295305.
analyses. Journal of Applied Behavior Analysis, 33, Steege, M. W., Wacker, D. P., Berg, W. K., Cigrand,
247250. K. K., & Cooper, L. J. (1989). The use of be-
Roane, H. S., Lerman, D. C., Kelley, M. E., & Van havioral assessment to prescribe and evaluate treat-
Camp, C. M. (1999). Within-session patterns of ments for severely handicapped children. Journal
responding during functional analyses: The role of Applied Behavior Analysis, 22, 2333.
of establishing operations in clarifying behavioral Sturmey, P. (1995). Analog baselines: A critical review
function. Research in Developmental Disabilities, of the methodology. Research in Developmental
20, 7389. Disabilities, 16, 269284.
Sailor, W., Guess, D., Rutherford, G., & Baer, D. M. Sturmey, P., Carlsen, A., Crisp, A. G., & Newton, J.
(1968). Control of tantrum behavior by operant T. (1988). A functional analysis of multiple ab-
techniques during experimental verbal training. errant responses: A refinement and extension of
Journal of Applied Behavior Analysis, 1, 237243. Iwata et al.s (1982) methodology. Journal of Men-
Sasso, G. M., Reimers, T. M., Cooper, L. J., Wacker, tal Deficiency Research, 32, 3146.
D., Berg, W., Steege, M., et al. (1992). Use of Taylor, J., & Miller, M. (1997). When timeout works
descriptive and experimental analyses to identify some of the time: The importance of treatment
the functional properties of aberrant behavior in integrity and functional assessment. School Psy-
school settings. Journal of Applied Behavior Anal- chology Quarterly, 12, 422.
ysis, 25, 809822. Taylor, J. C., Sisson, L. A., McKelvey, J. L., & Tre-
Shirley, M. J., Iwata, B. A., & Kahng, S. (1999). felner, M. F. (1993). Situation specificity in at-
False-positive maintenance of self-injurious behav- tention-seeking problem behavior. Behavior Mod-
ior by access to tangible reinforcers. Journal of Ap- ification, 17, 474497.
plied Behavior Analysis, 32, 201204. Taylor, J. T., Ekdahl, M. M., Romanczyk, R. G., &
Shirley, M. J., Iwata, B. A., Kahng, S., Mazaleski, J. Miller, M. L. (1994). Escape behavior in task sit-
L., & Lerman, D. C. (1997). Does functional uations: Task versus social antecedents. Journal of
communication training compete with ongoing Autism and Developmental Disorders, 24, 331
contingencies of reinforcement? An analysis dur- 344.
ing response acquisition and maintenance. Journal Thomas, D. R., Becker, W. C., & Armstrong, M.
of Applied Behavior Analysis, 30, 93104. (1968). Production and elimination of disruptive
Sidman, M. (1960). Tactics of scientific research. New classroom behavior by systematically varying
York: Basic Books. teachers behavior. Journal of Applied Behavior
Sigafoos, J., & Meikle, B. (1996). Functional com- Analysis, 1, 3545.
munication training for the treatment of multiply Thompson, R. H., Fisher, W. W., Piazza, C. C., &
determined challenging behavior in two boys with Kuhn, D. E. (1998). The evaluation and treat-
autism. Behavior Modification, 20, 6084. ment of aggression maintained by attention and
Sigafoos, J., & Saggers, E. (1995). A discrete-trial ap- automatic reinforcement. Journal of Applied Be-
proach to the functional analysis of aggressive be- havior Analysis, 31, 103116.
haviour in two boys with autism. Australia & New Thompson, R. H., & Iwata, B. A. (2001). A descrip-
Zealand Journal of Developmental Disabilities, 20, tive assessment of social consequences following
287297. problem behavior. Journal of Applied Behavior
Skinner, B. F. (1953). Science and human behavior. Analysis, 34, 169178.
New York: Macmillan. Toogood, S., & Timlin, K. (1996). The functional
Smith, R. G., & Iwata, B. A. (1997). Antecedent assessment of challenging behavior: A comparison
influences on behavior disorders. Journal of Ap- of informant-based, experimental and descriptive
plied Behavior Analysis, 30, 343375. methods. Journal of Applied Research in Intellectual
Smith, R. G., Iwata, B. A., Goh, H., & Shore, B. A. Disabilities, 9, 206222.
(1995). Analysis of establishing operations for Touchette, P. E., MacDonald, R. F., & Langer, S. N.
self-injury maintained by escape. Journal of Ap- (1985). A scatter plot for identifying stimulus
plied Behavior Analysis, 28, 515535. control of problem behavior. Journal of Applied
Smith, R. G., Iwata, B. A., Vollmer, T. R., & Zarcone, Behavior Analysis, 18, 343351.
FUNCTIONAL ANALYSIS 185

Ulman, J. D., & Sulzer-Azaroff, B. (1975). Multiel- G., & Mazaleski, J. L. (1993b). Within-session
ement baseline design in educational research. In patterns of self-injury as indicators of behavioral
E. Ramp & G. Semb (Eds.), Behavior analysis: function. Research in Developmental Disabilities,
Areas of research and application (pp. 377391). 14, 479492.
Englewood Cliffs, NJ: Prentice Hall. Vollmer, T. R., Marcus, B. A., Ringdahl, J. E., & Ro-
Umbreit, J. (1995a). Functional analysis of disruptive ane, H. S. (1995). Progressing from brief assess-
behavior in an inclusive classroom. Journal of Early ments to extended experimental analyses in the
Interventions, 20, 1829. evaluation of aberrant behavior. Journal of Applied
Umbreit, J. (1995b). Functional assessment and in- Behavior Analysis, 28, 561576.
tervention in a regular classroom setting for the Vollmer, T. R., Northup, J., Ringdahl, J. E., LeBlanc,
disruptive behavior of a student with attention def- L. A., & Chauvin, T. M. (1996). Functional
icit hyperactivity disorder. Behavioral Disorders, analysis of severe tantrums displayed by children
20, 267278. with language delays: An outclinic assessment. Be-
Umbreit, J. (1996). Assessment and intervention for havior Modification, 20, 97115.
the problem behaviors of an adult at home. Jour- Vollmer, T. R., Progar, P. R., Lalli, J. S., Van Camp,
nal of the Association for Persons with Severe Hand- C. M., Sierp, B. J., Wright, C. S., et al. (1998).
icaps, 21, 3138. Fixed-time schedules attenuate extinction-induced
Van Camp, C. M., Lerman, D. C., Kelley, M. E., phenomena in the treatment of severe aberrant
Roane, H. S., Contrucci, S. A., & Vorndran, C. behavior. Journal of Applied Behavior Analysis, 31,
M. (2000). Further analysis of idiosyncratic an- 529542.
tecedent influences during the assessment and Vollmer, T. R., Ringdahl, J. E., Roane, H. S., & Mar-
cus, B. A. (1997). Negative side effects of non-
treatment of problem behavior. Journal of Applied
contingent reinforcement. Journal of Applied Be-
Behavior Analysis, 33, 207221.
havior Analysis, 30, 161164.
Vaughn, B. J., & Horner, R. H. (1997). Identifying Vollmer, T. R., & Vorndran, C. M. (1998). Assess-
instructional tasks that occasion problem behav- ment of self-injurious behavior maintained by ac-
iors and assessing the effects of student versus cess to self-restraint materials. Journal of Applied
teacher choice among these tasks. Journal of Ap- Behavior Analysis, 31, 647650.
plied Behavior Analysis, 30, 299312. Wallace, M. D., & Iwata, B. A. (1999). Effects of
Vollmer, T. R., Iwata, B. A., Duncan, B. A., & Ler- session duration on functional analysis outcomes.
man, D. C. (1993). Extensions of multielement Journal of Applied Behavior Analysis, 32, 175183.
functional analysis using reversal-type designs. Watson, T. S., Ray, K. P., Turner, H. S., & Logan, P.
Journal of Developmental and Physical Disabilities, (1999). Teacher-implemented functional analysis
5, 311325. and treatment: A method for linking assessment
Vollmer, T. R., Iwata, B. A., Smith, R. G., & Rodgers, to intervention. School Psychology Review, 28,
T. A. (1992). Reduction of multiple aberrant be- 292302.
haviors and concurrent development of self-care Weeks, M., & Gaylord-Ross, R. (1981). Task diffi-
skills with differential reinforcement. Research in culty and aberrant behavior in severely handi-
Developmental Disabilities, 13, 287299. capped students. Journal of Applied Behavior Anal-
Vollmer, T. R., Iwata, B. A., Zarcone, J. R., Smith, R. ysis, 14, 449463.
G., & Mazaleski, J. L. (1993a). The role of at- Weiseler, N. A., Hanson, R. H., Chamberlain, T. P.,
tention in the treatment of attention-maintained & Thompson, T. (1985). Functional taxonomy
self-injurious behavior: Noncontingent reinforce- of stereotypic and self-injurious behavior. Mental
ment and differential reinforcement of other be- Retardation, 23, 230234.
havior. Journal of Applied Behavior Analysis, 26, 9 Received March 8, 2002
21. Final acceptance February 20, 2003
Vollmer, T. R., Iwata, B. A., Zarcone, J. R., Smith, R. Action Editor, Dorothea Lerman