AN EVALUATION OF ROUTINES ANALYSES WITHIN
FUNCTIONAL BEHAVIOR ASSESSMENT
by
AARON C. BARNES
A DISSERTATION
Presented to the Department of Special Education
and Clinical Sciences
and the Graduate School of the University of Oregon
in partial fulfillment of the requirements
for the degree of
Doctor of Philosophy
December 2009
11
University of Oregon Graduate School
Confirmation of Approval and Acceptance of Dissertation prepared by:
Aaron Barnes
Title:
"An Evaluation of Routines Analyses within Functional Behavior Assessment"
This dissertation has been accepted and approved in partial fulfillment of the requirements for
the Doctor of Philosophy degree in the Department of Special Education and Clinical Sciences
by:
Cynthia Anderson, Chairperson, Special Education and Clinical Sciences
Robert Horner, Member, Special Education and Clinical Sciences
Richard Albin, Member, Special Education and Clinical Sciences
Jean Stockard, Outside Member, Planning Public Policy & Mgmt
and Richard Linton, Vice President for Research and Graduate Studies/Dean of the Graduate
School for the University of Oregon.
December 12,2009
Original approval signatures are on file with the Graduate School and the University of Oregon
Libraries.
© 2009 Aaron C. Barnes
III
An Abstract of the Dissertation of
Aaron C. Barnes for the degree of
in the Department of Special Education and Clinical Sciences
to be taken
IV
Doctor of Philosophy
December 2009
Title: AN EVALUATION OF ROUTINES ANALYSES WITHIN FUNCTIONAL
BEHAVIOR ASSESSMENT
Approved: _
Dr. Cynthia M. Anderson
Procedures for direct observation as part of functional behavior assessment (FBA)
in natural settings continue to be an important area of inquiry and evaluation in the field
of education. Spread across a continuum of control and rigor, various direct FBA
methods involve a variety of strengths and limitations. The purpose of this study was to
evaluate the treatment utility of routines analysis when applied to direct observation as
part of the function-based assessment and intervention process in general education
classrooms. Central to this procedure is the use of routines analysis during the FBA
interview to inform and develop direct observation conditions. This procedure was
evaluated across 3 students in grades K-6. Data collected via this procedure showed
utility when compared to traditional ABC observation methods such that clearer
vindications of a hypothesized function of behavior were obtained. Interventions
developed from the assessment data resulted in an observed decrease in problem behavior
for each participant. Results of this study suggest the importance of routines analysis as a
possible way to increase the efficiency and effectiveness of the FBA process.
VI
CURRICULUM VITAE
NAME OF AUTHOR: Aaron C. Barnes
PLACE OF BIRTH: Iowa City, IA
DATE OF BIRTH: Apri124th, 1981
GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED:
University of Oregon; Eugene
Minnesota State University, Mankato; Mankato
Luther College; Decorah, IA
DEGREES AWARDED:
Doctor of Philosophy, School Psychology, 2009, University of Oregon
Master of Arts, Clinical Psychology, 2006, Minnesota State University, Mankato
Bachelor of Arts, Psychology, 2003, Luther College
AREAS OF SPECIAL INTEREST:
Function-based Assessment and Intervention
Systems-level Changes in Schools
School-wide Models of Support
Applied Behavior Analysis
PROFESSIONAL EXPERIENCE:
Assistant Professor, University of Wisconsin - Stout, Menomonie WI,
Aug 2009-present
Social Behavior Collaborative Planner, St. Croix River Education District,
Rush City, MN, 2008-2009
Teaching Assistant,
University of Oregon, 2005-2008
Minnesota State University, Mankato 2003-2005
Vll
Research Team Member, Education and Community Supports, University of
Oregon, 2005-2008
GRANTS, AWARDS AND HONORS:
School Psychology Student-Faculty Representative, University of Oregon,
2007-2008
School Psychology Candidates Interview Weekend Student Coordinator,
University of Oregon, 2008
Magna cum Laude, Minnesota State University, Mankato
Regent's Scholar, Luther College
PUBLICATIONS:
Barnes, A. C., & Harlacher, J. E. (2008). Response-to-Intervention as a set of
principles: Clearing the confusion. Education & Treatment o.fChildren,
31, 41 7-431.
Cisler, J., Barnes, A. c., Farnsworth, D. & Sifer, S. (2007). Reporting practices of
psychological research using a wait-list control: Current state and
suggestions for improvement, International Journal o.fMethods in
Psychiatric Research, 16, 34-42.
Vlll
ACKNOWLEDGMENTS
I wish to express my sincere gratitude and appreciation first to Professor
Anderson for her assistance and support throughout this process. She provided sound
advice and professional judgment in formulating this dissertation from the beginning to
end. Her support helped to see this project through and I am thankful for her
commitment.
Thanks also go out to Dr. B. Stiller and the teachers who participated in the study.
Dr. Stiller's aid in recruitment, consultation, and collaboration was a major boon to this
project. Thank you to Carla Zimmennan, Denise Ermatinger, Dianne Gilland and Amy
Burrow for allowing me to conduct a study within your respective classrooms. Your
commitment to the study provided valuable data for your students, schools, and the
academic community at large.
A special thank you to fellow school psychology students who participated in data
collection: Shelley Mullen, Courtney Coughlin, Billie Jo Rodriguez, and Brad Cohn.
Immense gratitude is reserved for R. Justin Boyd. Your assistance was invaluable in
keeping this project moving towards completion during its final year.
For my wife Emily. You have shared in the work and sacrifice. Now you
share in the accomplishment.
IX
TABLE OF CONTENTS
Chapter Page
1. INTRODUCTION AND LITERATURE REVIEW............................................ 1
Functional Behavior Assessment 2
Indirect Functional Behavior Assessment 3
Direct Functional Behavior Assessment 7
Descriptive Methods '" 7
Experimental Analysis 9
Structural Analysis. 11
Statement of the Problem 14
II. PHASE I: FUNCTIONAL BEHAVIOR ASSESSMENT 16
Method 16
Participants and Setting 16
Measure, Response Definitions and Interobserver Agreement 18
Procedure.................................................................................................. 23
FACTS Interview 23
Teacher Training 24
Comparison of Direct FBA Observations................................................. 24
Integrity Measurement 26
Data Analysis 29
Results 33
Andre 33
Mark 34
Renee 38
x
Chapter
Xl
Page
III. PHASE II: INTERVENTION ANALYSIS 41
Method 41
Participants and Setting 41
Response Definitions and Interobserver Agreement 41
Procedure 42
Results 46
Problem Behavior and Academic Engagement 46
Contextual Fit 50
IV. DISCUSSION 53
Summary of Findings 53
Study Limitations 56
Implications for Research and Practice 58
APPENDICES 62
A. CONSENT AND ASSENT FORMS 62
B. FUNCTIONAL ASSESSMENT CHECKLIST FOR TEACHERS AND
STAFF (FACTS) 68
C. SELF ASSESSMENT OF CONTEXTUAL FIT 71
REFERENCES 76
XlI
LIST OF FIGURES
Figure Page
1. Comparison of Phase I Observation Conditions for Andre 35
2. Comparison of Phase I Observation Conditions for Mark 37
3. Comparison of Phase I Observation Conditions for Renee 39
4. Intervention Evaluation Data for Andre . 48
5. Intervention Evaluation Data for Mark 49
6. Intervention Evaluation Data for Renee 51
xiii
LIST OF TABLES
Table Page
1. Average Percent (range) ofInterobserver Agreement for Phase 1................. 22
2. Direct Observation Conditions and Description for Each Participant 27
3. Percentage of Session Time Scored with Antecedent Variables 31
1CHAPTER I
INTRODUCTION AND LITERATURE REVIEW
Functional behavior assessment (FBA) may be one of the most effective resources
available for improving outcomes for students exhibiting problem behavior in today's
schools. Generally speaking, FBA is a collection of techniques used to develop
hypotheses about the environmental variables of which behavior is a function. This
information can be used to design behavior support plans to increase desired behaviors
and decrease problem behaviors. Functional behavior assessment, along with a
functionally-derived intervention, can be described collectively as function-based
supports. The utility of function-based supports has been well documented in clinical
settings and other non-school settings (e.g., Carr & Durrand, 1985; Carr et al., 1999;
Didden, Duker, & Korzilius, 1997; Iwata et al., 1982/1994;) as well as in schools (e.g.,
Dunlap, Kern-Dunlap, Clarke, and Robbins, 1991; Erickson, Stage, & Nelson, 2006;
Roberts, Marshal, Nelson, Albers, 2001)
Although IDEA reauthorization in 1997 often requires an FBA as part of
individualized education plans (IEPs) for students exhibiting serious problem behaviors,
educational professionals often find themselves without adequate training or tools
necessary to access the potential benefits of function-based supports. Researchers in the
area of function-based suppOlis have not done enough to develop, evaluate, and
2disseminate methods of functional behavior assessment appropriate for use in schools. As
a result of this failure to bridge the research-to-practice gap, the degree to which
function-based supports are accurately and efficiently utilized in school settings continues
to suffer.
In this chapter various methods for conducting FBA are described; special
attention is given to FBA research in school settings. The chapter concludes with a
summary statement of the current problem and an introduction to the purpose of the
study.
Functional Behavior Assessment
The purpose of FBA is to generate information on the events preceding and
following a target behavior; and to determine which events are reliably associated with
the occurrence of the behavior. Events that occur before a target behavior are called
antecedents, and those events that occur after the target behavior are called consequences.
Types of antecedents can be further broken down into discriminative stimuli, or events
that signal an increased likelihood that the target behavior will be followed by a specific
consequence (i.e., reinforcement or punishment); and establishing operations, which are
events that increase the potency of a particular reinforcer at a particular time.
Consequences may be classified as either positive or negative reinforcement or positive
or negative punishment. The use ofpositive or negative refers to the respective
introduction or removal of environmental stimuli, while reinforcement or punishment
refers to either increasing or decreasing the future probability of the target behavior
occurrmg.
3Types of FBA are often delineated based upon the methods used to collect
information. These methods can be either indirect or direct. Both types are desctibed
below, along with examples, advantages, and limitations.
Indirect Functional Behavior Assessment
Indirect assessments involve gatheting information about behaviors of interest
from the perspective of another - either the target individual, or someone who is familiar
with the target individual (e.g. parent, teacher, or mentor). Interview or rating scale
formats may be used for this type of assessment. Examples of functional assessment
interviews include the Functional Assessment Interview for Teachers and Staff (FACTS;
March et aI., 2000), the Functional Assessment Interview (FAI; O'Neill, Homer, Albin,
Sprague, Storey, & Newton, 1997), the Preliminary Functional Assessment Survery
(Dunlap et aI., 1993), the Problem Identification Interview (PH; Bergan & Kratochwill
1990), and the Student-Guided Functional Assessment Interview (Reed, Thomas,
Sprague, & Homer, 1997). Commonly cited rating scales and checklists include the
Problem Behavior Questionnaire (PBQ; Lewis, Scott, and Sugai, 1994), the Motivational
Assessement Scale (MAS; Durand & Crimmins, 1998), and the Teacher Functional
Behavioral Assessment Checklist (TFBAC; Stage, Cheney, Walker, & LaRocue, 2002).
Each gathers information in order to develop hypotheses regarding the function of the
problem behavior. Differences remain, however, in the amount and type of information
collected. Some interviews and rating scales focus solely on a limited number of potential
consequences for problem behavior (e.g. sensory stimulation, escape, attention, access to
tangible rewards), while others collect information regarding potential antecedent events,
4descriptions of problem behavior and/or appropriate behavior, or the likelihood of
problem behavior occurring during a particular time or in a particular place.
Indirect assessments have the potential to provide large amounts of infonnation in
relatively small amounts of time, are typically recommended practice during the early
stages of a multimethod FBA process (Floyd, Phaneuf, & Wilezynski, 2005), and
individuals from a wide range of behavioral expertise can be trained to administer (but
perhaps not score) these tools. In addition, they are useful for behaviors that occur only
infrequently or are covert-thus rendering direct observation difficult. Studies focusing
on empirical evidence for the use of indirect methods have produced mixed results; some
studies demonstrate good reliability and validity whereas others report limitations. Four
major areas of technical adequacy have been identified in the literature. These areas,
interrater reliability, convergent validity between sources, convergent validity with other
indirect measures, and convergent validity with direct observations, are discussed below.
Examining interrater reliability and convergent validity between sources involves
administering the indirect measure to multiple infonnants who may have varying degrees
of familiarity with the target student. Ratings can be examined for agreement on
identifying problem behaviors, setting events or antecedents, and consequences or
functions ofbehaviof. Studies have found initial evidence ranging from low to high
interrater reliability between multiple teachers as well as between teachers and students
(Barton-Arwood, Wehby, Gunter, & Lane, 2003; Borgrneier, 2003; Kinch, Lewis-Palmer,
Hagan-Burke, & Sugai, 2001; Murdock, O'Neill & Cunningham, 2005; Reed, Thomas,
Sprague, & Homer, 1997; Zarcone, Rogers, Iwata, Rourke, & Dorsey, 1991). Agreement
5between raters is generally highest regarding problem behaviors and consequences, and
lowest regarding setting events. Interviews appear to show the strongest convergent
agreement between students and teachers (Kinch et aI.).
Evaluating convergent validity with other indirect measures involves
administering distinct interviews or rating scales to one or more individuals and
comparing the results. Much of the evidence for convergent validity in this area is poor,
whether examining agreement between interviews and rating scales (Alter, Conroy,
Mancil, & Hayden, 2008; Kwak, Ervin, Anderson, & Austin, 2004), or between different
rating scales (Barton-Arwood et aI., 2003; Stage et aI., 2006).
To further demonstrate the validity of indirect methods, researchers and
practitioners often attempt to provide evidence that the functional relation identified by
the interview or rating scale coincides with the functional relation identified by another,
demonstrably accurate, measure (Shriver, Anderson, & Proctor, 2001). Studies making
these types of comparisons often utilize systematic, direct observation methods
(described next) as a means of comparison. Several studies have found limited evidence
of convergent agreement between indirect and direct measures (Alter et ai. 2008; Kwak et
aI., 2004; Murdock, O'Neill, & Cunningham, 2005; Stage, Cheney, Walker, & La
Rocque, 2002; Stage et aI., 2006). Agreement regarding maintaining consequences of a
behavior has been particularly low in many of these comparisons.
Overall, poor, inconsistent, or limited evidence for technical adequacy of indirect
FBA measures in the areas of interrater reliability, convergent validity between sources,
convergent validity with other indirect measures, and convergent validity with direct
6observations, justify further exploration of these measures. Because of these concerns, the
use of indirect methods often is encouraged in conjunction with other direct methods
(Carr et aI., 1999). Depending on a wide range of factors (e.g. detail of tools, skill of
assessor, skill of respondent as an observer and reporter, attention given to routines and
events associated with problem and desired behaviors, etc), the utility of indirect
assessment results for planning behavioral interventions can vary greatly. Information
provided by these measures may not be sufficient to develop adequate supports for
behaviors of increased intensity and in contexts of increased complexity.
To the extent that reliability and validity of indirect methods are a concern, there
may be risk of either Type 1 error (concluding a particular function of behavior when, in
fact, the behavior is not primarily maintained by that function) and Type 2 error (failing
to identify the primarily maintaining consequence of a particular behavior) obscuring the
assessment conclusions.
The FACTS interview is unique amongst indirect methods ofFBA because it
involves first identifying routines (e.g., academic tasks, unstructured times) during which
problem behavior most often occurs and then guides the interviewer to identify relevant
environmental variables for those specific routines. This is a significant strength of the
FACTS as topography is not the same as function; the same response (e.g., hitting others)
can occur in different situations but be maintained by different consequences. For
example, a child might hit peers during math class because she routinely is sent to the
office and hence avoids working on math problems (negative reinforcement) but hitting
on the playground results in attention from her friends (positive reinforcement). A review
7of recent studies using the FACTS interview suggests that this indirect assessment
produces similar hypotheses to more direct methods ofFBA (described below) and is
useful for developing effective interventions (McIntosh et al 2008).
Direct Functional Behavior Assessments
Next different types of direct functional assessment are reviewed. These methods
are characterized by information gathering via direct observation as opposed to the report
of another person. The various direct methods may be differentiated by the degree of
control associated with their use. Descriptive methods are reviewed first followed by
experimental methods as these represent two ends of the continuum of control. Finally,
structural analyses are discussed as they incorporate aspects of both descriptive and
experimental methods.
Descriptive Methods
On the lower end of the control continuum lie descriptive assessments. These
methods involve recording instances of target behaviors as well as the environmental
events that immediately precede or follow those behaviors (e.g., Erickson, Stage, &
Nelson, 2006; Lerman & Iwata, 1993; Newcomer & Lewis, 2004; Penno, Frank, &
Wacker, 2000; Roberts, Marshal, Nelson, Albers, 2001). Descriptive assessments most
often are carried out in the setting in which problem behavior reportedly occurs, and
environmental events are not manipulated or altered in any way. For example, Roberts et
al. worked with three elementary students and conducted observations during various
school activities while measuring off-task behavior and associated environmental
variables. More specifically, task type (academic or non-academic) and difficulty
8(instructional or frustrational) were recorded as antecedent stimuli and escape, teacher
attention, or peer attention were recorded as consequences. Results were used to form
hypotheses regarding events that preceded and maintained off-task behavior. For all
participants, problem behavior occurred more often during frustrational or difficult tasks
than during instructional or non-academic tasks for all three students. Escape from
difficult tasks was hypothesized as a maintaining consequence. As shown in this
example, environmental events associated with problem behavior may be observed and
noted during descriptive assessements, particularly if the observation is conducted in the
general setting where problem behavior has been repOlied to occur. Such assessments are
typically conducted, however, without explicit efforts to ensure the presence of variables
that are likely to co-occur with problem behavior.
Because descriptive assessments are conducted in natural settings and because
environmental variables are not manipulated, they may allow for observation of naturally
occurring relations between environmental variables and an individual's behavior. The
extent to which these relations are clearly identified has significant implications for the
ease and effectiveness of using the assessment results to design and implement behavior
supports. Because events are not manipulated systematically, however, hypotheses cannot
be verified (Iwata, Volmer, & Zarcone, 1990). Another potential limitation to descriptive
assessments stemming from a lack of environmental control is that if environmental
events of interest occur infrequently or if multiple events tend to occur in close temporal
proximity to the target behavior, it may be difficult to develop hypotheses about
functional relations between environmental variables and the target response. Finally,
9simply documenting that an event often precedes or follows the response does not
demonstrate that the event is functionally-related to the event.
Experimental Analysis
On the "high-control" end of the continuum lie experimental methods.
Experimental methods of functional assessment are called functional analyses and use
appropriate methods of single-subject research design to demonstrate causal relations
between environmental events and problem behavior. These methods involve
systematically manipulating environmental variables that are hypothesized to evoke and
maintain problem behavior (e.g., Carr & Durand, 1985; Hall, 2005; Harding et aI., 1999;
Iwata, Dorsey, Silfer, Bauman, & Richman, 1982/1994; Peck, Sasso, Stambaugh (1998).
Arguably the "gold standard" of functional analysis is the analog functional
analysis developed by Iwata et ai. (1982/1994). The analog typically is run in a controlled
setting by trained experimenters and consists of three to five test conditions each
designed to test a different hypothesis about events that evoke and maintain problem
behavior. Commonly evaluated antecedents include attention deprivation, task
presentation, and removal of preferred items. Consequences tested include attention
delivery, removal of items, and access to preferred items. The analog functional analysis
has been used to identify environment-behavior relations and has successfully been used
to develop interventions in a large number of-largely clinic-based-studies. This
method has also been used in more basic research to study environment-behavior
relations (cf Lennan, 2003).
10
Other methods of functional analysis exist as well including "antecedent only"
methods and brief functional analysis. In antecedent only functional analysis, antecedent
stimuli are manipulated while no programmed consequences are delivered for problem
behavior (e.g., Carr and Durand, 1985) Recent attention has been given to ways the
analog functional analysis might be shortened and a commonly used approach is the brief
functional analysis (e.g. Northrup et aI., 1991; Watson & Sterling,1998 ; Wilder, Chen,
Atwell, Pritchard, & Wienstein, 2006). In brief functional analyses, a shortened
experimental design is used; for example only 1-2 sessions per condition may be
conducted or sessions or run for brief periods (e.g., 5 min). For example, Northrup et al.
examined the behavior of three individuals exhibiting self-injurious and aggressive
behavior. Pmiicipants were systematically presented with one to two trials of two or four
conditions. Each condition lasted from five to ten minutes. These conditions resembled
those from a traditional analog functional assessment. A specific condition was
associated with elevated rates of problem behavior for each participant. To demonstrate
functional control, a contingency reversal condition was also run for each participant, in
which the consequence associated with highest levels of problem behavior (e.g.
tangible/activity or attention) was only presented after a predetermined appropriate
behavior such as signing "please," or saying, "please come here." Levels of problem
behavior were reduced in contingency reversal conditions for each participant.Because
experimental methods focus on manipulating variables in a systematic and controlled
manner, they are useful for demonstrating clear functional relations between
environmental events and problem behavior. Limitations to these methods include the
11
high level of training needed to carry them out (Peck, Sasso, & Stambaugh, 1998), and
also that their treatment utility in natural settings-especially when the functional
analysis is conducted in an controlled setting-may be low (Anderson & Long, 2002;
English & Anderson, 2006; Carr, Yarbrough, & Langdon, 1997; Van Camp et aI., 2000).
For example, English and Anderson compared caregiver and experimenter-conducted
analog functional analysis in a laboratory setting and a specific type of structural analysis
(described next) conducted in the natural setting with three children diagnosed with
developmental disabilities exhibiting severe problem behavior (e.g., self-injury,
aggression). Different patterns of responding were observed across methods of functional
assessment. English and Anderson went on to evaluate interventions based on each
method of assessment in participant's natural settings. For all participants, interventions
based on the structural analysis-conducted in the natural setting-were more effective
then interventions based on the functional analyses.
Structural Analysis
Structural analysis is similar to descriptive assessment in that observations take
place in the natural setting and typical caregivers are involved, but in structural analysis
some environmental variables are controlled systematically. All structural analyses
involve manipulating antecedent variables however different approaches to consequent
stimuli have been taken (e.g., Anderson & Long 2002; English & Anderson, 2006; Peck,
Sasso, Stambaugh, 1998; Stichter, Sasso, & Jolivette, 2004). In one type of structural
analysis developed by (Peck and Sasso 1997; Stichter et aI., 2004; Stitchter & Conroy,
2005) antecedent variables are selected for manipulation based upon a service-provider's
12
recommendation and unstructured observations. Environmental variables associated with
either the occurrence or the nonoccurrence of problem behavior may be manipulated.
Importantly, in this type of structural analysis, consequent stimuli are not recorded;
instead the focus is on the proportion ofproblem behavior that occurs in the presence of
different antecedent stimuli. Most often, the maintaining consequence (e.g., escape as a
reinforcer in the presence of difficult academic tasks) is assumed. Studies conducted
using this method have demonstrated potential for identifying and verifying relevant
antecedent variables and developing effective interventions in schools. For example, a
teacher working with Stichter et al. indicated that a student's problem behavior was most
likely to occur when there was significant background noise in the classroom, and that
problem behavior was much less likely to occur when students not present or making
noise in the classroom. The student was systematically exposed to repeated conditions of
high and low levels of noise, structure, and social interaction, and those conditions
associated with lower levels of problem behavior (particularly high structure) were noted.
When the student was presented with tasks involving high structure and moderate levels
of noise and social interaction, researchers observed fewer instances of problem behavior
compared to a control condition involving moderate levels of structure, noise, and social
interaction.
Although structural analyses such as those conducted by Stichter and her
colleagues are useful for identifying putative establishing operations and discriminative
stimuli, they provide little information about consequences that might reinforce problem
behavior. In contrast, an alternative method of structural analysis, the structured
13
descriptive assessment (SDA) developed by Anderson and her colleagues (Anderson &
Long, 2002; Anderson, English, & Hendrick, 2006; English & Anderson, 2006) can be
used to develop hypotheses about the entire three- or four-term contingency. During the
SDA, pre-determined antecedent variables are systematically manipulated, using a multi-
element design while problem behavior and consequences for problem behavior are not
altered. Data are collected using real time scoring on a variety of environmental events-
regardless of whether they precede or follow problem behavior. Events typically scored
include delivery or removal of attention, delivery or removal of requests to complete
tasks, and delivery or removal of preferred activities. Data are analyzed to detennine the
proportion of problem behavior evoked by various events and as well the proportion of
problem behavior followed by a given event. For example, data analysis might reveal
that, when attention from adults is not available, 75% of all problem behavior is followed
by adult attention. In contrast, problem behavior never occurred when adult attention was
available. The SDA has been used in a number of studies and found to be useful for
identifying functional relations and building effective interventions (e.g., Anderson &
Long, 2002; English & Anderson, 2006). A limitation ofthe SDA is that, to date,
conditions have been conducted using fairly standard antecedent variables-this method
may be more useful for typically developing individuals ifmore idiographic stimuli were
chosen and manipulated.
Relative to both unstructured observations (in which no variables are
manipulated) and functional analyses (typically conducted in controlled settings and
involving standardized environmental events) structural analyses may provide an
14
increased likelihood to observe relevant behavior-consequence relations, and therefore
lead to more conclusive and accurate hypotheses regarding environment-behavior
relations. Another advantage of using structural analyses in natural settings may be the
comparatively low training and expertise requirements required. Training and fluency in
behavior assessment may be required to design conditions and interpret results, but
service providers and peers are often required only to "respond as you normally would"
to problem behaviors. Formal functional analyses, by comparison, require the precise
creation and manipulation of environmental stimuli. Of course, a limitation of structural
analyses is that functional control is not demonstrated. In addition, if-in the case of the
SDA-problem behavior is followed by a variety of consequences, it may be difficult to
develop a hypothesis about which event or events actually maintain the response.
Statement of the Problem
As described above, a variety of methods of FBA exist, each with strengths and
limitations. Analog and other experimental methods of functional analysis are valued
because they allow for a demonstration of functional control but such methods often are
unrealistic for use in schools (Peck, Sasso, Stambaugh, 2005; Stichter & Conroy, 1998).
Structural methods of functional analysis may be useful but no clear guidelines exist for
identifying potentially relevant antecedent conditions under-which observations should
be conducted. For researchers and those with a great deal of expertise in behavior
analysis, this is less problematic however many practitioners may struggle to identify
environmental conditions to test. Although indirect methods ofFBA often are criticized
as unreliable, the FACTS interview stands out as a potentially very useful tool as (a) it
15
takes only a short amount of time to complete and (b) because it focuses on relevant
routines, it may be more useful for developing hypotheses about environment-behavior
relations.
The purpose of this study is to build on existing research on the FACTS interview
and subsequent direct FBA methods by evaluating the treatment utility of conducting a
FACTS interview to identify relevant routines within which observations (direct FBA)
could be conducted. After completion of the FACTS interview, an alternating treatments
design was used to compare results of observations during the specific activity suggested
by the FACTS to contain relevant antecedent variables to the more general context in
which a typical ABC might be conducted. Finally, interventions were implemented to
assess the treatment utility of the routines-analysis of the FACTS. To summarize, the
present study aims to contribute to existing literature on FBA methodology and utility by
addressing the following research questions: (a) whether observed levels of problem
behavior differ between the routines-based, traditional ABC, and control conditions; (b)
whether suggested functions of behavior provided by the FACTS interview and direct
observation data were in agreement; and (c) whether there existed a functional relation
between an intervention based upon the hypothesized function of problem behavior and a
decrease in problem behavior?
The present study consisted of two phases, the functional behavior assessment
phase and the intervention analysis phase. They are described below, beginning with the
methods and results of Phase I, followed by the methods and results of Phase II.
16
CHAPTER II
PHASE I: FUNCTIONAL BEHAVIOR ASSESSMENT
Method
In Phase I the FBA was conducted. A FACTS interview was conducted with the
teacher of each participant. The interview was followed by a comparison of the routines-
based observation to a traditional ABC observation.
Participants and Setting
Three typically developing children exhibiting problem behavior in school settings
participated. Participants were recruited from public schools in the Pacific Northwest. All
three schools were implementing School Wide Positive Behavior Supports (SWPBS)
across a three-tiered model as documented by scoring 80% or higher on the School-wide
Evaluation Tool (SET; Homer et aI., 2004). Scoring 80% or higher on this tool is
suggested as a criterion for meaningful outcomes by the authors. Procedures associated
with SWPBS (e.g., established behavior support teams, student-referral processes) may
have aided in the recruitment and willingness of school personnel to participate.
Participants were recruited in several steps. First, researchers were contacted by school
behavior support teams as they identified students and teachers that seemed appropriate
and expressed a willingness to participate. Researchers then contacted the classroom
teacher to provide information and answer questions. If teachers provided their informed
17
consent to participate, they contacted parents to assess parental interest and, if the parent
was interested, sent a consent fonn home to be signed. When teacher and parent consent
were obtained, the teacher ananged a meeting between a researcher and the student,
during which the student had opportunity to provide assent. Teacher, parent, and student
consent/assent fonns can be found in Appendix A.
Andre was a kindergarten student who received 100% of his instruction in the
general education setting. He was refened to the study due to disruptive behavior.
Andre's teacher indicated that he was most likely to engage in disruptive behavior during
independent reading and writing tasks. More specifically, she reported that Andre
engaged in crying, whining, and refusal (e.g. saying "I can't do it," "I'm dumb," or "I
don't know how.") All observations for Andre took place during reading and language
arts instruction where students were expected to cycle between large-group, small-group,
and independent work tasks.
Mark was a 2nd grade student who received 100% of his instruction in the general
education setting. He was refened to the study due to disruptive behavior. Mark's teacher
indicated that he was most likely to engage in talking with or touching, poking, or
kicking peers during large group reading instruction. All observations for Mark took
place during moming reading instruction, which was primarily delivered in a whole-
class, large-group fonnat.
Renee was a 6th grade student who received 100% of her instruction in the general
education setting. She was refened to the study due to disruptive and off-task behavior.
Renee's teacher indicated that she was most likely to engage in off-topic comments
18
directed to peers and adults, and to be out-of-seat during large-group instruction. All
observations for Renee took place in a general education classroom incorporating lecture,
discussion and independent work tasks.
Measure, Response Definitions and Interobserver Agreement
The Functional Assessment Checklist for Teachers and Staff (FACTS, March et
al. 2000) was administered to all teachers prior to completing the comparison of direct
observation conditions. Using the FACTS, the researcher spent 20-25 minutes
interviewing the participant's teacher to identify the problem behavior, develop an
operational definition, identify routines in which the problem behavior often occurred,
and developed a hypothesis statement specifying events that evoked and maintained
problem behavior. A copy of the FACTS is included in Appendix B.
During direct observation, data were collected on pre-defined student and teacher
behaviors and on relevant contextual variables. All observation data were collected using
a real-time data collection system on laptop computers.
For students, data were collected on problem behavior and one or more relevant
appropriate responses. Responses were identified and defined in the FACTS interview.
Prior to conducting the comparison of direct observation conditions, an initial observation
was conducted to verify the definitions. Problem behavior for both Andre and Mark was
identified as disruption defined as any vocalization or physical interaction that was not
related to the current academic task. For Andre, disruption tended to include crying,
whining, and refusal (e.g. saying "I can't do it," "I'm dumb," or "I don't know how.")
For Mark, disruption entailed behaviors such as talking at or touching, poking, or kicking
19
peers. For Renee, the problem behavior identified was talk-outs defined as any verbal
statement made during instruction when the expectation was to remain quiet or when the
student made statements without raising a hand or without being prompted (e.g., when
the teacher had not asked for comments or answers from the group or the target student).
Follow-up with the classroom teacher confirmed talk-outs as behaviors of concern.
During data collection, disruption was measured on a partial interval system, and talk-
outs were measured via frequency count. Because each student's targeted behavior was
most likely to occur during academic work, academic engagement was selected as an
appropriate behavior. Academic engagement was defined as (1) following teacher
requests within 10 s (2) eyes oriented toward teacher or relevant materials for academic
task, and (3) completing in-class tasks as requested by the teacher. Academic engagement
was scored using duration recording. In addition, there was a 3-s delay for scoring the
onset and offset of academic engagement to control for discrete instances of behavior
(e.g., briefly looking at the teacher or relevant materials) that were scored only if they
continued beyond the 3-s delay.
Data were collected as well on teacher responses, peer responses (as relevant) and
on key contextual events. Teacher responses were scored as a partial-interval measure
across consecutive 5-s intervals. Data on the following teacher responses were collected:
request, scored when a teacher provided an academically-related instruction or prompt to
the student using verbal statements or physical guidance; escape allowed, scored when, in
the absence of academic engagement, requests were not delivered for one complete
interval; teacher attention, scored when the teacher interacted with a student either
20
verbally or physically, in a non-instructional manner (e.g. high five, pat on the back,
saying "good job Although we initially coded positive/neutral attention separately from
negative teacher attention, these codes were combined in order to facilitate the analysis of
environment-behavior relations. Data were collected as well on peer attention (also
scored separately as positive and negative but combined for analysis of environment-
behavior relations). Peer attention was scored whenever a peer interacted with the student
verbally or physically, in a manner other than what was related to instruction (e.g. "way
to go," "How was your weekend?," high-five, "stop it," or "be quiet").
Interobserver agreement data were collected for 31 % of all sessions for each
participant. Agreement for frequency measures was calculated by dividing sessions into
consecutive 5-s intervals, comparing observers' records for each interval, dividing the
smaller number of recorded responses by the larger, and averaging the proportions across
the session and multiplying by 100 to obtain an agreement coefficient. For all responses
coded as partial interval measures, total agreement, OCCUlTence agreement, and
nonOCCUlTence agreement were calculated. Total agreement was calculated by dividing
the number of intervals in which both observers agreed on the occurrence or
nonOCCUlTence of the response by the total number of intervals and multiplying by 100 to
obtain a percentage of agreement. OCCUlTence only agreement was calculated by dividing
the total number of intervals both observers agreed a response occulTed by the number of
intervals either observer scored a response and multiplying by 100. Non-occulTence
agreement was calculated by dividing the total number of intervals both observers agreed
21
a response did not occur by the total number of intervals either observer did not score a
response and multiplying by 100.
Prior to beginning data collection, each data coder participated in three two-hour
training sessions and a final one-hour training session. The first session involved
reviewing the measures and procedures. The second and third sessions involved
practicing observations using videos and calculating interobserver agreement. Before
baseline data were collected, each observer conducted a minimum of two observations in
a classroom setting with the researcher. Practice sessions continued until agreement
coefficients were 80% or higher for all target responses during two consecutive practice
sessions. If total agreement fell below 80% for three consecutive sessions at any time
during the study, the data collector ceased data collection and was re-trained until the
80% cli terion was reached.
For problem behavior, total agreement averaged 96% (range = 91 % - 98%),
occurrence only averaged 88% (range = 0% - 100%), and nonoccurrence only averaged
96% (range = 91 % - 100%). For academic engagement, total agreement averaged 93%
(range = 82% - 100%), occurrence only averaged 91 % (range = 82% - 100%), and
nonoccurrence only averaged 93% (range = 83% - 100%). For adult attention, total
agreement averaged 98% (range = 90% - 99%), occurrence only averaged 91 % (0%-
98%), and nonoccurrence averaged 98% (range = 91 % - 100%). For peer attention, total
agreement averaged 93% (range = 80% - 100%), occurrence only averaged 82% (range =
0% - 100%), and nonoccurrence averaged 93% (range 83% - 100%). Interobserver
agreement data are shown for each participant in Table 1.
22
Table 1. Average Percent (range) ofInterobserver Agreement for Phase I
Total Agreement Occurrence Only Non-occurrence Only
Andre
Problem Behavior 96% .95% 97%
(91 % - 100%) (89%-100%) (91%-100%)
Academic Engagement 92% 90% 97%
(89%- 94%) (85%-97%) (93%-100%)
Adult Attention 98% 98% 98%
(94%-99%) (97%-98%) (94% - 100%)
Peer Attention 82% 75% 85%
(80%- 98%) (74%- 96%) (83% -100%)
Mark
Problem Behavior 96% 93% 97%
(92%-100%) (89% - 100%) (94%-100%)
Academic Engagement .89% 86% 91%
(82%- 99%) (82%- 99%) (83% - 100%)
Adult Attention .98% 91% 98%
(90%- 99%) (86%-96%) (91%-99%)
Peer Attention 94% 91% 94%
(90%- 98%) (86%- 99%) (93%- 96%)
23
Table 1. (Continued).
Total Agreement Occurrence Only Non-occurrence Only
Renee
Problem Behavior 98% 00% 98%
(98% - 98%) (00%- 00%) (98% ~ 98%)
Academic Engagement 99% 98% 92%
(98% - 100%) (98% - 100%) (84% - 100%)
Adult Attention 98% 58% 98%
(97%-99%) (00%-67%) (97%-99%)
Peer Attention 99% 50% 99%
(98% - 100%) (00% - 100%) (98%-100%)
Procedure
The FACTS interview was conducted first with teachers. Next, teacher training
occurred and then direct observations were conducted.
FACTS Interview
The FACTS interview was administered to all participating teachers by the
primary researcher for all participants except Renee; her FACTS was conducted by a
doctoral student in the school psychology program with extensive expertise in function-
based supp0l1s. The interview was administered in an area away from other students or
adults-to maintain confidentiality-at a pre-arranged time selected by the teacher.
When the FACTS interview was completed, a summary hypothesis statement was
developed for each routine and presented to the teacher to ensure that he or she agreed
24
with the relevant antecedent and consequent stimuli. All teachers agreed with the
hypothesis statement generated. The hypothesis statements generated for each participant
were as follows. During independent reading and writing tasks, Andre engages in
disruption including crying, whining, yelling (e.g., "I can't do this," "I don't know how.")
maintained by adult attention. During large group reading instruction, Mark engages in
disruption including talking to, touching, poking, or kicking peers maintained by peer
attention. During large-group instruction in reading/writing, Renee talks out of tum and
this behavior is maintained by adult attention
Teacher Training
Prior to conducting direct observations, written information was provided
regarding the purpose ofthe various assessment conditions. In addition, once the
assessment began, verbal or written feedback was provided following each session, and
teachers were given opportunity to contact the researcher with any questions.
During sessions, prompts to reestablish relevant antecedent conditions were
provided as described below.
Comparison a.!Direct FBA Observations
A multi-element design was used to compare outcomes of observations conducted in the
relevant routine identified in the FACTS interview to observations conducted without
regard to the presence or absence of that routine-as typically occurs during ABC
methods ofFBA. The assessment was conducted during the activity (e.g., math) in which
problem behaviors most often occurred, as identified during the teacher interview and
three conditions were conducted with each participant. Data collection occurred during
25
similar times of day across all sessions for each participant. For Andre, the activity during
which the assessment was conducted was independent tasks during reading and writing;
for Mark it was large group instruction during reading, and for Renee it was large group
instruction during writing. A routines-based condition, a traditional ABC condition, and a
control condition were developed for each participant. The order of conditions was
determined randomly except that no one condition occurred more than twice in
succession. Conditions consisted of 10-min sessions and sessions were repeated until
differentiated responding was observed via visual inspection. A minimum of three
sessions were conducted per condition.
In all conditions, teachers were given specific instructions-described next-
regarding how to set up the antecedent conditions but were asked to work with the
student and respond to problem behavior as they typically would, as if observers were not
present.
In the routines-based condition, the specific antecedent stimulus hypothesized to
evoke problem behavior on the FACTS was present. For all participants the antecedent
stimulus involved presentation of instructional activities. Teachers were asked to present
the activity as they typically would and to respond to problem behavior as usual. If, in the
absence of academic engagement or problem behavior, the teacher did not re-establish
the presence of independent reading/writing, large group reading, or large-group writing
for Andre, Mark, or Renee, respectively, within 2 minutes, the teacher was prompted by a
researcher to re-establish that relevant antecedent stimulus. For example, an observer
might say to the teacher, "could you please ask Andre to begin working on independent
26
work again," or, "could you please ask Mark/Renee to attend to the group discussion or
lecture once again?"
During the traditional ABC condition, observations were conducted during the
problematic activity indicated on the FACTS, however the presence of the specific
antecedent identified as a discriminative stimulus or establishing operation was not
controlled. For each participant, the traditional ABC condition was conducted during
reading/writing, reading, and writing for Andre, Mark, and Renee respectively. During
this condition, any stimuli that typically were presented (e.g., difficult assignments, group
work, "easy" assignments) might have occurred. As in the routines-based condition, the
general activity was re-established after two consecutive minutes if, in the absence
academic engagement or problem behavior, the teacher did not prompt the student to
return to the currently expected task.
During the control condition the general routine was present however the variable
identified as a SD or EO was absent. Thus, for Andre, independent work did not occur,
and for Mark and Renee, large group instruction did not occur. In this condition, if the
antecedent stimulus manipulated in the routines-based condition was present in this
condition for more than 2 min, the teacher was asked to please cease the activity. The
conditions conducted with each participant are summarized in Table 2.
Integrity Measurement
The primary issue of integrity during Phase I was the manipulation of antecedent
variables. Particularly during the routines-based and control conditions, relevant,
predefined environmental variables (e.g. difficult or easy math tasks, respectively) would
Table 2. Direct Observation Conditions and Description for Each Participant
27
Condition Description
Andre
Routines-based condition Independent task with reading/writing
content.
Traditional ABC condition Reading/writing content (e.g., large-group,
small-group, independent work).
Control condition Small-group reading/writing instruction.
Mark
Routines-based condition Large-group reading instruction.
Traditional ABC condition Reading instruction (e.g., large-group,
small group, independent work, partner
reading).
Control condition Partner Reading.
Renee
Routines-based condition Large-group writing instruction
Traditional ABC condition
Control condition
Writing content (e.g., large-group, small-
group, independent work)
Independent writing task
28
occur frequently in their respective conditions, but would occur rarely, if ever, in the
opposing condition. Further, a variety of antecedent stimuli (e.g., difficult math, group
work, easy math) would occur during the traditional ABC condition however the general
routine (math) should be present. The fidelity of implementation of each condition was
assessed by calculating the percent of session time during which predefined
environmental variables were scored. For the routines-based condition, any session in
which the specific antecedent stimulus (e.g., difficult math) did not occur during at least
50% of intervals was not included in the data analysis. During the traditional ABC
condition, any session in which the general context (e.g., math content of some type) did
not occur at least 50% of intervals was not included in the data analysis. Finally, for the
control condition, any session in which the key variable for that condition (e.g., easy
math) did not occur during at least 50% of intervals was not included in the data analysis.
The presence of the various antecedent variables for each participant and session can be
seen in Table 3. In the table, the conditions conducted and the relevant environmental
variables for each condition are in the left column. The top row depicts possible
antecedent variables. Throughout the table, percentages in each row illustrate the mean
amount of session time scored with a particular antecedent across the various conditions.
For example, for Andre, during the routines-based condition, large group, small group or
partner instruction did not occur. Fmiher, independent work, the antecedent controlled for
in this condition, occurred an average of 94% of session time with a standard deviation
across the 5 sessions of 6.27%. Across pmiicipants, the putative SD/EO occurred most
often during the routines-based condition and rarely or not at all during the control
29
condition. The general context was present during the majority of all traditional ABC
observations. These data suggest the routines-based, traditional ABC and control
conditions were implemented with fidelity.
Data Analysis
When observations were completed for each student, the data were examined to
develop hypotheses about environment-behavior relations. Line graphs were developed to
evaluate trends within and across conditions. In addition, contingency space analyses
(Martens et aI, 2008) were calculated to evaluate the relation between environmental
variables and student behavior. To facilitate calculations, all responses scored as
frequency measures were converted to partial interval data by dividing each observation
session into continuous 5 s intervals. For these calculations, a response was coded as
being followed by a particular event each time the event was scored within the current or
subsequent interval. For example, an interval scored with disruption would be scored as
being followed by adult attention if that event was recorded in the same or subsequent
interval. If adult attention was recorded in the same interval but before disruption was
recorded, those events were not scored as an instance of disruption followed by adult
attention. If problem behavior was recorded in three consecutive intervals, and the third
instance of problem behavior was followed by adult attention within the third or
subsequent interval, only the third instance of problem behavior was scored as being
followed by adult attention.
To facilitate analyses of the data, contingency space analysis (CSA) (Martens et
aI, 2008) was conducted for each observation session from Phase 1. This involved
30
calculating and graphing the difference between the probability of a consequence given a
behavior, and the probability of a consequence given the absence of a behavior. Dividing
the number of intervals in which a behavior was followed by a consequence by the total
number of intervals in which the behavior was scored provided values between 0 and 1
for the probability of a consequence given a particular behavior. For example, if
disruption was scored in 30 intervals and was followed by attention in 25 of those 30
intervals, then the resulting proportion would be .83. Dividing the number of times a
particular consequence was scored in the absence of problem behavior by the number
intervals during which the behavior was not scored provided a value between 0 and 1 for
the probability of a consequence given the absence of a behavior (i.e., base rates of a
consequence). For example if attention was scored in the absence of disruption in 6
intervals and there were 42 intervals dming which disruption was not scored, the
resulting proportion would be .14 (indicating a low base rate of attention; attention rarely
occurred in the absence of problem behavior). Resulting proportions were summed across
sessions and divided by the number of sessions per condition to glean mean proportions
for each condition. These were graphed in coordinate space such that the probability of a
given response given the presence or absence of a consequence could be viewed. On a
contingency space analysis graph, values were shown in relation to each other and to a
diagonal line drawn through the graph, showing where two points would be plotted if the
values were equal (i.e., a pm1icular consequence was as likely to be delivered following a
behavior as it was delivered when the behavior had not occurred).
31
Table 3. Percentage ofSession Time Scored with Antecedent Variables
Mean Percent (SD)
Small
Condition Large Group Independent Group Partner
Unstructured
and key variable Instruction Work Instruction Work
Andre
Routi nes-based
Independent seat-
work 94% (6.27) 6% (6.26)
Traditional ABC
28% 39% 16%
General
reading/writing (33.75) (35.86) 16% (6.76) (13.46)
Control
Small group
instruction 2% (3.36) 98% (3.36)
Mark
Routi nes-based
Large-group
reading 97% (5.25) 3% (5.25)
Traditional ABC
General reading 67% (44.5) 33% (44.50)
Control 93%
Partner reading 6% (11) 1% (2) (10.30)
Note. Dashes indicate that a particular antecedent variable was not observed during a
particular condition.
32
Table 3. (Continued)
Mean Percent (SD)
Small
Condition Large Group Independent Group Partner
Unstructured
and key variable Instruction Work Instruction Work
Renee
Routines-based
Lecture
Reading/writing 97% (4.76)
Traditional ABC
General
reading/writing 29% (35.79) 61 % (30.39) 10% (20)
Control
Independent seat-
work 100% (0)
Note. Dashes indicate that a particular antecedent variable was not observed during a
particular condition.
For example, consider a problem behavior such as disruptive contact, defined as any peer
to peer contact occurring during large-group instruction. During observations, disruptive
contact occurred in 35 intervals and was followed by peer attention 18 times. This would
result in a proportion of .51, where problem behavior was followed by peer attention 51 %
of the time. Ifpeer attention was delivered when no problem behavior had occurred 3
times throughout the observation, and there were 85 intervals total scored without
33
problem behavior, this would yield a proportion of .04, where peer attention was
delivered in the absence of problem behavior 4% of the time. Data such as these might
suggest peer attention as a maintaining function of the problem behavior, and would also
be compared to calculations for other potential maintaining variables.
Results
Phase I of the present study investigated (1) whether observed levels of problem
behavior differed between the routines-based, traditional ABC, and control conditions,
and (2) whether functional relations suggested by the FACTS interview, traditional ABC
observation, and routines-based structural analysis were in agreement. First, data are
presented regarding levels of problem behavior across sessions of the functional behavior
assessment. Second, results on the CSA for each pmiicipant's assessment data are
presented.
Results for problem behavior across conditions of the functional assessment are
depicted in Figure 1 for Andre, Figure 2 for Mark, and Figure 3 for Renee. For all
figures, the top panel depicts the percent of intervals the antecedent variable most
strongly associated with problem behavior was present. The second panel from the top
shows the percent of intervals scored with problem behavior or the number of instances
of problem behavior recorded across sessions. The bottom three panels depict the CSA
results for routines-based, traditional ABC, and control conditions.
Andre
For Andre, the putative relevant antecedent variable, independent reading or
writing, occurred far more often during the routines-based condition then either the
34
traditional ABC or control condition (top panel, Figure 1). As depicted in the second
panel of Figure 1, problem behavior was scored most often during independent seat-work
during reading/writing. For the routines-based, traditional ABC, and control conditions
respectively, problem behavior occurred during an average of32%, (range 26%- 37%),
10% (range = 5% - 13%), and 10% (range = 5%-18%) of intervals.
The contingency space analysis shown in the third and fourth panels ofFigure 1
indicates that during the routines-based condition, teacher attention was positively
contingent on disruption and was more likely to occur following problem behavior than
either peer attention or escape. Adult attention was delivered following problem behavior
approximately 40% of the time, and was delivered only 7% ofthe time when no problem
behavior had occurred. During the traditional ABC condition, both adult attention and
peer attention were only slightly more likely to be delivered following disruption than to
occur during other times. No dependencies were found during the control condition, as
shown in the fifth panel where all points are plotted near to or on the diagonal line.
Taken together, the results of the alternating treatments design for Andre support
the utility of the routines-based analysis as problem behavior occurred more often during
this condition, allowing for more frequent observation of behavior-consequence relations.
Further, the contingency between problem behavior and teacher attention was stronger in
the routines-based analysis.
Mark
Results obtained with Mark are in Figure 2. As is shown in the top panel of the
Figure, although the putative SD/EO, large-group instruction, occurred most consistently
..c 100% ~ 35
+J
...... ..'3
....
"0 80%Q) ..><: ~Routines-based
....
....
0 0
u 5 60% ~Traditional ABCVl
c +JC0 Q)
'U;
"0 40%U'> cQ) Q)Vl a.
"I- Q) 20%0 "0
+J C
C
Q)
u 0%....
Q)
Cl... 0 2 4 6 8 10 12 14 16
...-----~------~
.;!1 .Ql00% ~
rn ~ 80% ->
.... ..c
Q) Q) I
+J CD 60% ic
"l- E I
0 Q) 40% -,
+J ..c
20% Jc 0Q) ....u Cl...
0% -I
....
..cQ)
Cl... +J
'3
0 2 4 6 Session 8
~Routines-based
~Traditional ABC
"",*"""Control
10 12 14 16
Routines-based
<> Peer Attention
D Adult Attention
A Escape
0.2 0.4 0.6 0.8 1
p(consequence/no problem behavior)
1 ,
i
i
E 0.8 -I
<IJ !
:c .
K9.,..6 J
1
___ L
<IJ 0
~"S: I
<IJ ® -: II
1:' I·
o ~"....-~~--~-~---~---
o cons~~encefM probl~tIi behaR6lr) 1
1
E
=* 0.8
o5._
~~
c >
<IJ ~
5-~
~
c
80.2
c:
Control
Traditional ABC
<> Peer Attention
D Adult Attention
A Escape
<> Peer Attention
D Adult Attention
AEscape
Figure 1. Comparison of Phase I Observation Conditions for Andre.
36
during the routines-based condition, it was present for the majority of intervals during
three sessions of the general condition as well. Perhaps as a result of this, results were
somewhat undifferentiated between the routines-based and traditional ABC conditions
(second panel Figure 2). Importantly, although intervals scored with disruptive behavior
were relatively stable during the routines-based condition (when the presence oflarge
group instruction was stable), during the traditional ABC condition, problem behavior
was scored most often during sessions when large group instruction occurred and less
often when that variable was absent problem behavior occurred during an average of33%
(range 38%-23%), 32 % (range 13%-60%), and 13% (range 5%-24%) in the routines-
based, traditional ABC, and control conditions respectively.
The CSA of Mark's data, shown in the bottom panels of Figure 2, indicate that
across the routines-based, traditional ABC, and control conditions, peer attention was
positively contingent upon disruption and was more likely to occur following problem
behavior than either adult attention or escape. Further, these contingencies were stronger
during the routines-based condition where peer attention was delivered approximately
42% of the time following disruption and only approximately 1% of the time when
disruption had not occurred. During the traditional ABC condition, peer attention was
delivered approximately 34% of the time following disruption and only 1% of the time
when disruption had not occurred. During the control condition-partner reading-peer
attention was likely to follow problem behavior 39% of the time, but was also likely to be
delivered 23% of the time when disruption had not occurred.
2--A1r--*.....~J--,-----,-------,-------,
~Control
~Traditional ABC
37
~Routines based
12 1410864
100% •
80%
60% J
40% I
20% l
0% ~-----,---­
o
-cw
u
:;;
a..
u 100% I
w (;(; I
u ';;: 80%
J
V'l I1l
'"
.r:
I1l w
1: CD
2 E 60% I
c w
:0
4- ~0 40%
-
a..
c ~w
u '~:;; 20% ~a..
~Routines based
~Traditional ABC
~Control
0% +-----,--------,--------,-------,-------,----~'-----
o 2 4 6 Session 8 10 12 14
1
Peer Attention
A Escape
IE] Teacher Attention
Traditional ABC
0.2 0.4 0.6 0.8
p(conseqence/no problem behavior)
o
1
Peer Attention
A Escape
IE] Teacher Attention
Routines-Based
0.2 0.4 °16 0.8p(consequence/no prOD em behavior)o
1 l
.ID.8 -1
~l~
c I1l I
w.r: '~.4t
g I
u ~
CU.2 I
o -1',---r-,-----,------,---,----,
Control
1
E
:to.8
o
0.-
...... ~
W;~
C I1l
W .r:
J w
3P~
c
o
~.2
Teacher Attention
A Escape
10.2 0.4 / O.h l b n.8.pfconsequence no prOD em e'Fiavror
o -!lIXlII-----.~7""_-,_--___,---__,__,_____,
o
Figure 2. Comparison of Phase I Observation Conditions for Mark.
38
Taken together, the results of the alternating treatments design for Mark support
the utility of the routines-based analysis as problem behavior occurred more often during
this condition and because problem behavior occurred during the traditional ABC
condition only when that variable was present. The contingency between problem
behavior and peer attention was indicated across all three conditions, but was strongest in
the routines-based analysis.
Renee
Results obtained with Renee are in Figure 3. As is shown in the top panel, the
putative SD/EO (large group writing) occurred most consistently during the routines-
based condition although it occurred during the majority of intervals of two sessions of
the traditional ABC condition as well. Intervals scored with problem behavior are in the
second panel. Problem behavior occurred most consistently during the routines-based
condition and, with the exception ofthe first session, never during the control condition.
In contrast, problem behavior occurred somewhat variably during the traditional ABC
condition and comparison of the first two panels does not reveal a clear relation between
the presence oflarge group writing and problem behavior in this condition.
The CSA of Renee's data, shown in the bottom panels of Figure 3, indicate that,
across conditions, teacher attention was positively contingent on problem behavior
although this relation was strongest during the routines-based condition where adult
attention followed problem behavior 38% of the time problem behavior was scored and
only 4% of all adult attention was scored independent of problem behavior. Interestingly,
39
14
-EJ-Traditional ABC
--$-Routines based
........Control
121086
_------4~~--.$-----.C.::J 1I:
19 <>=OJ 0.8[>D
ro ~
--' c
-5 OJ 0.6'".~ ~
0-
C 0.00 c 0.4'v; :f:
'"OJ 3:Vl
..... 0.20
~
c
OJ 0t'
OJ
0-
0 2 4
--$-Routines based
-EJ-Traditional ABC
4
~
::J
? 3-"
ro
I-
.....
0 2
'"OJ
U
c
2 1
'"E
0
0 2 4 6 8 10
........Control
12 14
Session
1
!!J.8
OJ
:n
~.~
-... 0~ 'S;
C ro
f'~ I]]
c
~.2
Routines-Based Condition
• Peer Attention
L1Teacher
Attention
A Escape
Traditional ABC
• Peer Attention
L1Teacher
Attention
A Escape
o 0.2 0.4 0.6 0.8
p(consequence/no problem behavior)
1 o 0.2 0./4 0.6 b h 0.8)p(consequence no pr05'lem e aVlor 1
1 Control
E 0.8
OJ
:n
L.Q·6
QJ a
~ 's;
~ -!l.4
0'.0
~
g 0.2
u
c:
<>. eer
Attention
L1Teacher
Attention
A Escape
01A----."L----~------_
o p(consequence/ Rci~roblem behavior) 1
Figure 3. Comparison of Phase I Observation Conditions for Renee.
40
peer attention was positively contingent on problem behavior during the
traditional ABC and control conditions but no contingency was observed during the
routines-based condition as peer attention was never scored. Finally, escape was
positively contingent on problem behavior only during the traditional ABC condition;
escape was not scored during the other two conditions.
Results obtained with Renee support the utility of the routines-based observations
relative to a traditional ABC analysis only tentatively. Problem behavior did occur more
in the routines analysis condition however different consequences were observed across
conditions and the validity of these consequences as reinforcers is unknown.
In sum, across two of three participants, the routines-based analysis resulted in the
clearest patterns of responding and hypotheses about response-consequence relations.
The validity of these hypotheses is unknown however. Thus, the purpose of Phase II was
to assess validity indirectly, via an evaluation of the treatment utility of hypotheses
gleaned from each assessment.
41
CHAPTER III
PHASE II: INTERVENTION ANALYSIS
Method
The purpose of this phase was to evaluate interventions developed based upon the
results of Phase 1. This allows for an indirect assessment of the validity of the various
observation methods by focusing on treatment utility. Interventions were matched to the
hypothesized function of problem behavior (as suggested by Phase I results).
Interventions were implemented and evaluated under the same conditions as the routines-
based assessment condition as this allowed for interventions to be evaluated during
conditions in which problem behavior was most likely to occur. Upon this study's
completion, multi-component interventions were developed and implemented for each
participant as appropriate.
Participants and Setting
Participants and settings were as per Phase 1. Interventions were carried out
during the routines-based condition identified in Phase 1. Across all Phase II sessions, the
routines-based variable associated with occurrence of problem behavior was present for
Andre, Mark, and Renee an average of 98%, 96% and 99% of session time respectively.
Response Definitions and Interobserver Agreement
Definitions of problem behavior, contextual variables and teacher/peer responses
were as per Phase 1. Interobserver agreement data were collected for 30% of sessions
42
using the same procedure as in Phase I. For problem behavior, total agreement averaged
98% (range = 91 % - 100%), occurrence only averaged 90% (range = 0% - 100%), and
nonoccurrence only averaged 97% (range = 91 % - 100%). For academic engagement,
total agreement averaged 96% (range = 85% - 100%), occurrence only averaged 91 %
(range = 82% - 100%), and nonoccurrence only averaged 88% (range = 0% - 100%). For
adult attention, total agreement averaged 97% (range = 0% - 100%), occurrence only
averaged 81 % (0% -100%), and nonoccurrence averaged 97% (range = 96% - 100%).
For peer attention, total agreement averaged 97% (range = 82% - 99%), occurrence only
averaged 82% (range = 0% - 99%), and nonoccurrence averaged 98% (range 94%-
100%).
Procedure
Antecedent conditions were established as per Phase I and intervention sessions
were 10 min in duration. All intervention phases were conducted until stability was
observed via visual inspection. A minimum of three sessions were conducted for each
phase of the intervention. Appropriate single subject designs were used to assess
functional control. Each intervention was matched to the hypothesis statement suggested
in Phase I and was designed to assess the treatment utility of consequences identified as
potentially reinforcing. For Andre, both the routines-based and traditional ABC
conditions suggested that problem behavior might be maintained by adult attention and,
to a lesser extent, peer attention. For Mark, a positive contingency between problem
behavior and peer attention was observed during all three conditions of the alternating
treatments design and for Renee, the routines-based condition suggested only teacher
43
attention to be reinforcing whereas the traditional ABC condition suggested that teacher
attention, peer attention, and escape might be reinforcing.
For Andre, the intervention analysis was conducted during independent work. To
test the hypothesis that problem behavior was maintained by adult attention. Andre was
taught to approach and check in with the teacher after completion of a teacher selected
portion (that should take approximately 3-5 min) of work. Prior to the OCCUlTence of
independent work, Andre's teacher, reviewed the material, checked for understanding,
and told Andre to complete a portion (approximately 3-5 min) of the task then check in
with the teacher to share his work. When Andre checked in, his teacher reviewed the
work and provided verbal feedback for between 20 and 40 s and told Andre to complete
the next segment of the task. The classroom teacher was instructed to ignore disruptive
behavior whenever possible and to provide brief re-direction in a neutral tone of voice
only if Andre's behavior presented a safety risk or was prohibiting the learning of his
peers. Across intervention sessions, adult attention was delivered for the appropriate
duration following successful completion of a segment of work for 94% of opportunities.
This represents approximately 8% of intervals being scored with adult attention following
appropriate behavior compared to 1% of intervals scored with adult attention following
problem behavior (see Table 4). In order to assess for functional control, Andre's
intervention was evaluated using an ABAB reversal design.
The results of the FBA for Mark suggested that disruptive behavior occulTed most
often during large-group reading instruction and was maintained by peer attention.
Because the instructional context was not conducive to increasing the amount of peer
44
Table 4. Data Concerning Fidelity ofImplementation for Phase II Interventions
Andre
Mark
Renee
Pal1icipant Intervention Fidelity Data
Adult attention delivered following 94% of successful
self-recruitment procedures.
Without intervention: peer attention delivered following
problem behavior 7.5% ofthe time.
During intervention: peer attention delivered following
problem behavior 3.5% of the time.
During intervention: adult attention delivered following
problem behavior 0% of the time.
interaction, an intervention was developed with the goal of decreasing the amount of peer
attention that followed disruptive behavior during group instruction. At the beginning of
each intervention session, the classroom teacher held up a glass jar and reviewed the
classroom expectations during large group instruction. The class was directed to respond
to peers' disruptive behavior by modeling appropriate behavior and otherwise ignoring
peer disruption. The classroom teacher told students that as she observed them meeting
classroom expectations - and especially if she observed them modeling appropriate
behavior during peer disruption - she would place marbles into the jar. When marbles in
the jar reached certain levels, the class would earn rewards (e.g. extra recess, end-of-day
45
party, etc.). Across intervention sessions in which peers were directed to ignore problem
behavior during large-group instruction, peer attention following problem
behavior was delivered to Mark during 2.3% of intervals compared to during 7.5% of
intervals whenever the intervention was not in place (see Table 4). An alternating
treatments design was used to assess functional control for Mark's intervention.
For Renee, the routines-based condition suggested only teacher attention to be
reinforcing whereas the traditional ABC condition suggested that teacher attention, peer
attention, and escape might be reinforcing. An examination of the antecedent stimuli
present during problem behavior in the traditional ABC condition showed that each talk-
out followed by peer attention or escape occurred during independent work, while 75% of
talk-outs followed by adult attention occurred during large group instruction. Although
variability was observed in the level of observed problem behavior during large group
instruction during the alternating treatments design of Phase I as shown on the two upper
panels of Figure 3, anecdotal report from the classroom teacher confinned large-group
instruction as a primary routine of concern. The intervention was designed accounting for
adult attention as the primary maintaining reinforcer for Renee's talk-out behavior. The
intervention implemented (described next) allowed for flexibility to implement
intervention components based on peer attention or escape following completion of the
study if necessary. In order to facilitate Renee's receiving adult attention contingent upon
attending to lecture and refraining from talk-outs, an intervention was developed based
on the school's already existing PRIDE card intervention. At the beginning of the writing
class, Renee would check-in with the teacher and was reminded of the expectation that
46
she remain academically engaged and refrain from engaging in talk-out behaviors. At the
end of the period, the teacher would review the card, providing larger amounts of points
and verbal feedback if Renee had successfully met the expectation and fewer points and
minimal verbal feedback if Renee had failed to meet the expectations. Because the card
was part of a larger intervention in the school, Renee did receive points and feedback for
behaviors other than those targeted in the current study, but during check-in, teachers
would emphasize the importance of academic engagement and talk-outs during reading
and writing class. Adults were directed to ignore any talk-outs during intervention
sessions. A copy of Renee's PRIDE card can be found in Appendix C. Across
intervention sessions in which adults were directed to ignore problem behavior during
large-group instruction and points and attention were delivered at the end of class
contingent upon academic engagement, adults delivered attention to Renee 0 times
following problem behavior (see Table 4). An ABAB reversal design was used to assess
functional control for Renee's intervention.
Results
In this section, data are presented regarding problem behavior and academic
engagement recorded during the intervention evaluation phase.
Problem Behavior and Academic Engagement
Results for problem behavior and academic engagement across the assessment
and intervention phases are depicted in Figure 4 for Andre, Figure 5 for Mark, and Figure
6 for Renee. All data for this phase were collected during conditions matching the
routines-based structural analysis for that participant. Because all teachers expressed
47
desire for a timely intervention, new baseline data were not collected, instead the
routines-based sessions were used as baseline.
As shown in Figure 4, an ABAB reversal design was used to assess functional
control for Andre's intervention. Upon implementation of the intervention, a 73%
reduction in intervals scored with problem behavior was observed (left panel). Following
a return to baseline to establish functional control, the intervention was re-instutited. In
this final phase, an 84% reduction in problem behavior (relative to the initial baseline)
was observed.
The intervention resulted in changes in academic engagement for Andre as well
(bottom panel, Figure 4). Although initially variable, academic engagement increased
steadily upon implementation of the intervention. Following a return to baseline,
academic engagement again increased and a 257% increase in academic engagement
relative to the initial baseline was observed.
Mark's intervention was evaluated using an alternating treatments design in which
baseline conditions (no marble jar) were compared to the intervention (Figure 5). During
baseline, the teacher had the students divided into four groups and students earned or lost
tick marks contingent on appropriate and inappropriate behavior. There was no system
for redeeming tick marks for additional rewards. While the intervention was in place, a
77% reduction in disruption was observed (top panel, Figure 5).
Mark's observed levels of academic engagement increased during intervention as
well (bottom panel, Figure 5). Although initially variable, following initial baseline
48
BL Self - recruit BL Self-recruit
100
c
0 90
'';:::;
C.
::l 80....
VI
0
_Disruption
~ 70
......
'3 60
"C
Q)
.... 500
u
Vl
~ 40
ro ~i>....Q) 30......C
4- 200
......
c 10Q)
u
....
Q)
0- 0
0 5 10 15 20
Session
I
100 II
~
I
90 Iro I
u I
'E 80 IIQ) I
"C I
ro 70 IU I
« I
Q) 60 IE II
i= "C IQ) 50 IC tlO
../:0 ro'';:::; tlO 40ro c
> L.U
....
Q) 30VI
.0
0 204-
0
_Academic Engagement
...... 10c
Q)
u
.... 0Q)
0-
0 5 10 15 20
Session
Figure 4. Intervention Evaluation Data for Andre.
49
100
c
0 90
'';:;
0..
:::J 80....
Vl
.......... MarbleJar0
.c 70
-l-'
'3 60
-c
OJ
.... 50 Baseline (Tick Marks)0
u
Vl 40~
ro
> 30....OJ
-l-'
C 20
4-
0
-l-' 10c
OJ
u 0....OJ
0-
0 2 4 6 8 10 12 14 16
Session
.......... MarbleJar
161412
Baseline (Tick Marks)
106 8
Session
42
100
.:?:
ro 90
u
E 80OJ
-c
ro 70u
«
OJ 60E
i= -cOJ 50c tlD
0 ro
'';:; tlD
ro C 40
> UJ
....
OJ 30Vl
.0
0
4- 200
-l-'
C 10OJ
u
....
OJ 00-
0
Figure 5. Intervention Evaluation Data for Mark.
50
observed levels of academic engagement were 39% higher than during baseline
conditions.
As shown in Figure 6, an ABAB reversal design was used to assess functional
control for Renee's intervention. Observed talk-outs gradually increased during baseline.
Upon implementation of the intervention, observed talk-outs increased for one session
before decreasing (top panel, Figure 6). During the first intervention phase, a 50%
reduction in talk-outs was observed. Following a return to baseline to establish functional
control, the intervention was re-instutited. In this final phase, an 89% reduction in
problem behavior (relative to the initial baseline) was observed.
Changes in academic engagement were observed during Renee's intervention as
well (bottom panel, Figure 6). Although initially decreasing, academic engagement levels
were variable across the first intervention session and a return to baseline. Although
functional control was not demonstrated, academic engagement increased upon
reinstating the intervention and a 22% increase in academic engagement relative to the
initial baseline was observed.
Contextual Fit
Contextual fit was assessed after data collection had ended for each participant
using the Self-Assessment of Contextual Fit in Schools (Horner, Salentine and Albin,
2003). Classroom teachers were asked to complete the 16-item questionnaire as it applied
to the specific intervention developed for each participant. The contextual fit
questionnaire can be found in Appendix C
51
Modified Modified
BL Point BL Point
6 Card Card
5
V1
......
::::l
0 4I~
co
I-
.....
0
V1 3(])
u
c
co
......
V1
C
"'Cl 2
(])
C
(]) Talk-outsV1
.0
0 1
0 ,,-- .......,...." ... _-.
0 5 10 15 20 25
Session
BL BL Modified
100% j't'"tCard......c
(])
E 80%(])
tlO
co
AcademictlOc
UJ Engagement
u
'E 60%(])
"'Cl
co
u
«
..c.
......
40%'~
c
0
'Vi
V1
(])
Ul 20%.....
0
......
c
(])
u
...
(])
0- 0%
0 5 10 15 20 25
Session
Figure 6, Intervention Evaluation Data for Renee.
52
Overall, classroom teachers rated the interventions high on contextual fit (M =
4.96). All items were rated as "5" or higher, except for "I have received any training I
need to be able to implement this support plan," which Mark's teacher rated as a "2", and
"Implementing this behavior plan will not be stressful" which Andre's teacher rated as a
"4".
53
CHAPTER IV
DISCUSSION
The goal of this study was to evaluate the utility of conducting a routines-analysis
to guide the conduct of direct observations. Specifically, this study examined whether
direct observations conducted during a specific routine identified in a routines-analysis
conducted via interview would yield more conclusive results regarding environrnent-
behavior relations. This was assessed via an alternating treatments design in which
observations were conducted in the presence of that relevant routine and during a more
general condition in which the general context was similar but the presence of the
relevant routine was not controlled for. A control condition-during which the specific
routine was absent-was conducted as well. Finally, interventions based on hypotheses
from the routines-based observations were implemented to assess the intervention utility
of routines-based direct observations.
Summary of Findings
In this section, results are summarized and discussed in relation the research
questions.
Whether observed levels ofproblem behavior d(ftered between the routines-based,
traditional ABC, and control conditions? Whether greater levels 0.[problem behavior
54
were observed when the relevant routine identifiedfrom the FACTS interview was
present?
Phase I of this study utilized an alternating treatments design to compare the
observed levels of problem behavior across the routines-based, traditional ABC, and
control conditions. Overall, problem behavior was recorded more often during the
routines-based condition than during traditional ABC or control conditions. This was
most clear while observing Andre, and the relevant routine identified from the FACTS,
independent reading/writing tasks, rarely occurred during traditional ABC observations.
For Mark's data, there was considerable overlap between observed levels of
problem behavior, but a closer look at the data revealed that during each traditional ABC
observation showing high levels problem behavior, the relevant routine-large-group
reading instruction-was present for 80% or more of the session.
Differences across observation conditions were less clear when looking at
Renee's data. Across four sessions of the routines-based condition, during which large-
group instruction was present for a minimum of90% of the time, observed instances of
problem behavior gradually increased. There was, however, overlap between the
frequency of talk-outs observed during the routines-based and traditional ABC
conditions. Because the traditional ABC condition may have presented a variety of
antecedent variables within the general context of reading and writing, it was important to
determine what specific routines coincided with problem behavior during those sessions.
During the first traditional ABC session, problem behavior was observed only in the
presence of the relevant routine, and during the second session one instance of problem
55
behavior was observed during the absence of large group instruction. The third traditional
ABC session showed 3 instances of problem behavior occurring when large group
instruction was not present. During the fourth session, no instances of problem behavior
were recorded. The relatively low frequency of Renee's target behavior may have
contributed to the lack of clarity in these data.
Are the suggested/unctions o.fbehavior provided by the FACTS interview, traditional
ABC observation, and routines-based observation in agreement?
This question was evaluated by comparing the primary hypothesized function
from the FACTS interview to the contingency space analyses for each participant.
Overall, the data analyzed from the routines-based condition provided the clearest
indication of hypothesized function. For each participant the suggested function of
problem behavior from the FACTS was supported by direct observation data.
Is there afimctional relation between an intervention based upon the hypothesized
.fimction o.fproblem behavior and a decrease in problem behavior?
This question was evaluated using either an ABAB reversal design or an
alternating treatments design. Overall interventions based upon the hypothesized function
of problem behavior were functionally related to reductions in problem behavior.
Although academic engagement was not the primary dependent variable of
interest in this study, and functional control was not clearly established for all
participants, there was some indication that the interventions implemented may have
increased academic engagement. Data for each participant showed variability and some
overlapping data points between baseline and intervention phases, but mean levels of
56
academic engagement were generally higher while interventions were being
implemented.
Study Limitations
The alternating treatments and reversal designs used in this study allowed for the
comparison of various conditions and to evaluate the utility of routines analyses within
functional behavior assessment. This section describes several limitations of the present
study.
Several limitations for this study relate to the potential benefit of collecting
additional data. One such limitation is the amount of data points collected during the
Phase I alternating treatments designs. Particularly when looking at Mark's and Renee's
data, where variability of observed problem behavior and overlap between routines-based
and traditional ABC observation data necessitated a careful analysis of which routines
coincided with problem behavior during the traditional ABC conditions, clarity of visual
and quantitative analyses may have improved if the alternating treatments design had
been extended.
The length of the Phase II intervention analysis is another limitation that may
have been mitigated if additional data had been collected. Due to time constraints, Phase I
data were used for initial baseline, and phases of ABAB reversal designs were generally
transitioned as soon as reasonable arguments could be made regarding functional
control. Although observation conditions during Phase II matched those of the routines-
based condition for Phase I, a separate baseline for intervention analysis would have
allowed further demonstration of the stability of behavior before interventions were
57
implemented. Similarly, additional data collected during second baseline and intervention
phases may have strengthened demonstration of the functional relations between the
interventions and problem behavior.
Finally, the lack of explicit data collection for variables relevant to integrity
during intervention is another limitation of the current study. During intervention for
Andre, this study was able to compare the percentage of intervals scored with adult
attention following the successful completion of a segment of independent work with the
percentage of intervals scored with adult attention following problem behavior.
Intervention fidelity may have been better documented if further data had been collected
regarding the details of Andre's behavior during the self-recruitment sequence.
Specifically, the use of a fidelity check-list with a task analysis of the appropriate
sequence may have simplified the task of assessing the fidelity of this intervention as
opposed to relying upon session logs and raw data. Additionally, because adults were
directed to provide attention if Andre's behavior represented a risk to safety or a serious
disruption, it may have been beneficial to collect further data on the problem behavior
that resulted in adult attention.
During intervention for Mark, we were able to compare overall instances of adult
attention delivered following problem behavior with those delivered at other times.
Intervention fidelity would have been better documented had we recorded when Mark's
disruption specifically involved peers and whether or not peer attention was delivered
following those disruptions.
58
During intervention for Renee, intervention fidelity would have been better
documented had we collected data for student and adult completion of the PRIDE point
card, and specifically the degree to which attention was delivered contingent upon Renee
meeting the expectations to remain academically engaged and refrain from talk-outs.
Finally, two of the participants were primary school-aged males, while one
participant was a middle school-aged female. In is unknown whether similar results could
be expected with students of different age, sex, or ethnicity. Additionally, as the only
female in the study, and showing problem behavior at relatively low rate, conclusions
drawn from Renee's data may be tentative even when applied to other females her age.
Implications for Research and Practice
In this section, implications tor practice regarding routines-based functional assessment,
including indirect methods and direct observation, will be discussed. In addition,
implications for future research will be discussed.
Implications for Practice
The present study documented the utility of identifying relevant routines during
an interview to guide direct observations. A routines-based direct observation can be a
useful part of FBA in the classroom setting. When direct observations were focused on
specific routines identified by the FACTS, higher rates of problem behavior were
observed, and clearer indications of possible function of problem behavior were
generated through CSA. As part of a multi-source FBA process, an indirect assessment
like the FACTS can be used to gain valuable infonnation about specific routines
associated with higher levels of problem behavior. This infonnation can be used to design
59
direct observation conditions that may lead to more effective and efficient direct
observations.
In addition to making direct observations more efficient, the routines-based
analysis procedures utilized in this study may hold benefits for classroom teachers. This
study suggested that teachers were able to implement the observation conditions we
created with fidelity. They were involved with the direct observation portion ofthe FBA
instead of being interviewed and eventually given an intervention plan to try. All teachers
in this study rated their knowledge and awareness of the Phase II interventions highly on
the contextual fit survey. Because the interventions developed are directly related to the
data collected during the routines-based assessment condition, it may be that teacher
participation during the direct observation contributed to their perceived knowledge and
familiarity with the intervention plan.
Implications for Research
This section will outline potential areas for future research, based on the results
and limitations of the present study. First, research regarding whether or not routines-
focused FBA can lead to more effective and efficient direct observation will be discussed.
Second, future studies on the agreement between various FBA methods will be discussed.
The present study contributed to the literature supporting the use of routines-
based observation conditions as part of FBA in a school setting. Many questions remain,
however, regarding the extent to which routines-based observation may serve as part of
an efficient and effective FBA process. One question in this area is whether or not higher
rates of problem behaviors are observed during routines-based conditions compared to
60
other observation conditions. This was generally found to be true in the current study, but
without looking deeper into the specific routines present during the traditional ABC
condition, this was not always clearly indicated. Future research should include routines-
based observation where a specific routine associated with problem behavior is present
and can be compared to control conditions where the relevant routine is not present.
Furthermore, future research involving direct observation similar to the traditional ABC
condition should document the presence of specific routines. Repeated comparison of
observed problem behavior in the presence and the absence of relevant routines may
clarify the answer to this question. Another question that should be explored in future
research is the consistency of suggested functions of behavior across various assessment
methods. The present study generally found agreement between the FACTS and the
various direct observation conditions. When agreement was not entirely clear between
conditions, this was due to inconclusive or mixed results from data collected during
traditional ABC or control conditions. Future research should repeat these comparisons in
educational settings, and also examine data collected by observing a wider range of
teachers, participant demographics, and target behaviors.
Another question related to agreement between assessment methods is whether or
not data collected during the routines-based condition offered clearer evidence of a
suggested function of behavior when compared to the traditional ABC condition. This
study generally found this to be the case via CSA calculations. Review of Mark's data
suggested peer attention as the function of his disruptive behavior across all observation
methods, but CSA results were stronger during the routines-based conditions.
61
Another area of for future research involves whether or not interventions based
upon suggested functions are successful. Future research should make note of when
separate or dual functions of behavior are indicated by the various observation methods
and then compare the effectiveness of distinct interventions based upon those competing
hypotheses. For example, if data from the routines based condition suggested adult
attention as a primary function, and data from the traditional ABC condition suggested
both adult and peer attention as functions of problem behavior, then Phase II of the study
could incorporate a component analysis of an adult/peer attention-based intervention.
Such a comparison could evaluate the relative effectiveness of a multi-function
intervention compared to a single-function intervention, thus informing the relative
intervention utility of the observation methods upon which they were based.
Another limitation of the current study related to intervention evaluation was the
amount of time interventions were in place. Phase II intervention and reversal data were
only collected during 10-minute sessions. Data were not always collected on consecutive
days, some retum-to-baseline phases were shOliened due to classroom scheduling
constraints, and no long-term follow-up data were collected due in part to the fact that the
school year was ending during intervention data collection. Future studies should address
these limitations by collecting additional data over longer periods of time on a more
consistent schedule, and by collecting follow up data.
APPENDIX A
CONSENT AND ASSENT FORMS
62
63
Informed Consent to Participate in a Research Study
ParentiGuardian/Family/Student Consent
The Impact of Functional Behavior Assessment Methods on
Promoting Desired Behaviors and Decreasing Problem Behaviors
Your son/daughter is invited to participate in a study conducted through the University of
Oregon designed to evaluate behavior support methods used in schools. The study will be
conducted by Aaron Barnes, under the supervision of Cynthia Anderson, both from the
University of Oregon's College of Education. The purpose of the study is to examine
how different assessment methods can aid educators in developing effective behavior
supports for students. Your son/daughter was selected as a possible participant in this
study because staff at his or her school believes he or she may benefit from additional
behavior supports. The study will begin in September 2007 and end in June 2009.
To provide students with effective and efficient behavior supports, participation by your
son or daughter would involve:
• Behaving as they normally do when presented with typical teacher requests (like
completing a math assignment or transitioning from one activity to another).
• Providing their own input or feedback regarding what a support plan might entail
or what rewards they might like to work toward.
• Participating in a designed support plan with knowledge of what behaviors are
expected, which behaviors are inappropriate, and what consequences will follow
appropriate and problem behaviors.
To tailor the behavior supports to your child's needs, the teacher and school personnel
would complete the following activities:
• Review your child's academic and behavioral school records, social strengths and
weaknesses, and attendance and discipline referral patterns, if applicable.
• Complete an interview detailing your child's behavior in various settings and
during various activities.
To conduct the study, researchers from the University of Oregon will complete the
following activities:
• collaborate with school personnel to collect and review data
• conduct direct observations of your child in his or her classroom to collect data on
social behavior and/or academic behavior.
64
Your child will not be identified in written or professional presentations of the results of
this study. Every effort will be made to organize information using altered names, and
professional presentations will never refer to your child by name. In addition, all
information will be kept in a lockable location and destroyed after the study and holding
period are complete. There remains, however, a small risk that your student may be
identified as a participant in this study.
There is a distinct likelihood that your student may benefit from participation in the
study. Function-based behavioral suppolis have shown promising results in previous
studies where social and academic gains were documented for participating students.
Your consent to your child's participation in the study is voluntary. Your decision
whether or not to allow your child to participate will not affect your relationship with the
school district or the instruction your child receives in his or her school. If you allow your
child to participate, you are free to withdraw your consent and terminate your child's
participation in the study at any time without penalty.
Prior to your child's participation in the study, we will also ask your son/daughter ifhe or
she give their assent to participate. Their assent will be necessary for participation in the
study.
If you have any questions, please feel free to contact Aaron Barnes at (541) 285-1077 or
Cynthia Anderson at the University of Oregon (346-2671). If you have questions
regarding your rights as a research subject, contact the Office of Human Subjects
Compliance, University of Oregon, Eugene, OR 97403, (541) 346-2510. You have been
given a copy of this form to keep.
Your signature indicates that you have read and understand the information provided
above, that you willingly agree to participate, that you may withdraw your consent at any
time and discontinue participation without penalty, that you have received a copy of this
form, and that you are not waiving any legal claims, rights or remedies.
Parent/Legal Guardian _
Signature _
Name of Child
--------------
Date
--------
65
Informed Consent to Participate in a Research Study
Teacher Consent
The Impact of Functional Behavior Assessment Methods on
Promoting Desired Behaviors and Decreasing Problem Behaviors
You are invited to participate in a research study conducted by Aaron Barnes, under the
supervision of Cynthia Anderson, both in the University of Oregon's College of
Education. The purpose of the study is to examine how different assessment methods can
aid educators in developing effective behavior supports for students. The study will begin
in September 2007 and end in June 2009. You were selected as a possible participant,
because you currently teach students who might benefit from additional behavioral
support.
The Function-based supports in this study will generally involve:
• Gathering detailed information about appropriate behaviors and problem
behaviors.
• Gathering detailed information about situations or settings in which these
behaviors of interest are likely to occur.
• Developing hypotheses for the function of targeted behaviors and developing
interventions or behavior support plans based on those hypotheses.
• Monitoring levels of appropriate and problem behaviors in order to evaluate the
effects of the intervention.
• Direct observation in your classroom by a researcher.
If you choose to participate in the study, researchers will ask you to complete the
following activities to maximize the benefit for the student:
• Complete the Functional Assessment Checklist for Teachers and Staff (FACTS),
a 20-25 minute interview with a researcher. This information will be used to
identify the function of students' problem behavior.
• Present the student with various tasks or requests (directed by the researcher) and
responding to the student's behavior as you normally would.
• Consult with a researcher to develop and implement behavior supports for the
student.
• Complete the Contextual Fit Checklist to provide your feedback regarding the
behavior support plan and how well it matches with your skills, values, resources,
and administrative support. This form typically takes about 10 minutes to
complete.
66
Researchers from the University of Oregon will conduct direct observations of
participating students in their classrooms to collect data on social and academic
behaviors.
Neither you nor the student will be identified in written or professional presentations
concerning this study. Every effort will be made to organize information using altered
names, and professional presentations will never refer to you or your student by name. In
addition, all information will be kept in a lockable location, and destroyed after the study
and holding period are complete. There remains, however, a small risk that you may be
identified as a participant in this study.
There is a distinct likelihood that your student may benefit from participation in the
study. Empirical evidence suggests that function-based supports lead to students' social
and academic gains.
Your participation is voluntary. Your decision whether or not to participate will not affect
your relationship with the school district. If you decide to participate, you are free to
withdraw your consent and discontinue participation at any time without penalty.
If you have any questions, please feel free to contact Aaron Barnes at (541) 285-1077 or
Cynthia Anderson at the University of Oregon (346-2671). If you have questions
regarding your rights as a research subject, contact the Office of Human Subjects
Compliance, University of Oregon, Eugene, OR 97403, (541) 346-2510. You have been
given a copy of this form to keep.
Your signature indicates that you have read and understand the information provided
above, that you willingly agree to participate, that you may withdraw your consent at any
time and discontinue participation without penalty, that you have received a copy of this
form, and that you are not waiving any legal claims, rights or remedies.
Print Name
----------------------------
Signature _
Date
------------
67
Student Assent
The Impact of Functional Behavior Assessment Methods on
Promoting Desired Behaviors and Decreasing Problem Behaviors
We want to ask you if you want to be part of a research study. Your (parents, guardian,
etc) has told us it is ok for you to be part of this project, but we want to ask you if you are
willing to help us by being part of the study.
If you are part ofthe study you will do function-based supports which means you will
work with your teachers, parents, and a researcher to learn some ways to be more
successful in school. We will have people come to watch you and the other students to
see if the program is helpful, but you will work primarily with your teacher.
If you choose to be part ofthe study, we will not use your name when we share with
other people how you do in school. Nobody except the people you work with will know
who you are.
We expect to work with you for several weeks, and in the end you should have skills that
will help you in school.
If you choose to be part of the study, you can always change your mind, and if you
choose not to be part of the study, it will not affect anything else about what you do at
school.
Do you have any questions?
If you are willing to be pmi of the study we would ask you to sign this form.
Student Signature: _
APPENDIXB
FUNCTIONAL ASSESSMENT CHECKLIST FOR TEACHERS AND STAFF
(FACTS)
68
69
Functional Assessment Checklist for Teachers and Staff (FACTS-Part A)
Student/ Grade: _
Interviewer:__~ _
Date: _
Respondent(s): ~ _
Student Profile: Please identify at least three strengths or contributions the student brings to
school:
Problem Behavior(s): Identify problem behaviors
_Tardy
_ Unresponsive
Withdrawn
_ Fight/physical Aggression
_ Inappropriate Language
Verbal Harassment
_Disruptive
Insubordination
Work not done
Theft
Vandalism
Other
_Verbally Inappropriate
Describe problem behavior:
_ Self-injury
Identifying Routines: Where, When and With Whom Problem Behaviors are Most Likely.
Schedule
Activity Likelihood of Problem Behavior Specific Problem Behavior
(Times)
Low High
I 2 3 4 5 6
I 2 3 4 5 6
I 2 3 4 5 6
I 2 3 4 5 6
I 2 3 4 5 6
I 2 3 4 5 6
I 2 3 4 5 6
I 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
Select 1-3 Routines for fUl'ther assessment: Select routines based on (a) similarity of activities (conditions) with ratings of 4,5
or 6 and (b) similarity of problem behavior(s). Complete the FACTS-Part B for each rontine identified.
March. Homer. Lewis-Palmer, Brown. Crone. Todd & CmT (2000) 1/19/05
70
Functional Assessment Checklist for Teachers & Staff (FACTS-Part B)
Student/ Grade: Date: _
Interviewer: Respondent(s): _
Routine/Activities/Context: Which routine (only one) from the FACTS-Part A is assessed?
Routine/Activities/Context Problem Behavior(s)
Provide more detail about the problem behavior(s):
What does the problem behavior(s) look like?
How often does the problem behavior(s) occur?
How long does the problem behavior(s) last when it does occur?
What is the intensityllevel of danger of the problem behavior(s)?
What are the events that predict when the problem behavior(s) will occur? (Predictors)
Related Issues (setting events) Environmental Features
illness Other: _ reprimand/correction
-
Other
-
_drug use
_structured activity _with peers
_ negative social
conflict at home _ physical demands
- unstructured time _ activity too longAcademic failure -
~
_ socially isolated _ tasks too boring
-
tasks too difficult
What consequences appear most likely to maintain the problem behavior(s)?
Things that are Obtained Things Avoided or Escaped From
-
adult attention Other:
-
hard tasks Other:
--peer attention _reprimands
_ preferred activity ~eer negatives
-
money/things ~hysical effort
-
adultattention
Identify the summary that will be used to build a plan of behaVIOr support.
Setting Events & Predictors Problem Behavior(s) Maintaining
Consequence(s)
How confident are you that the Summary of BehaVIOr is accurate?
Not very confident
Confident
1 2 3 4 5
Very
6
What current efforts have been used to control the problem behavior?
Strategies for responding to problem
Strategies for preventing problem behavior behavior
_ schedule change Other: __ reprimand Other:
_ seating change
-
office referral
_ curriculum change
-
detention
APPENDIXC
SELF-ASSESSMENT OF CONTEXTUAL FIT
71
72
Self-Assessment of Contextual Fit in Schools
Homer, Salentine, & Albin, 2003
The purpose of this interview is to assess the extent to which the elements of a behavior
support plan fit the contextual features of your school environment. The interview asks
you to rate (a) your knowledge ofthe elements of the plan, (b) your perception of the
extent to which the elements of the behavior support plan are consistent with your
personal values, and skills, and (c) the school's ability to support implementation of the
plan. This information will be used to design practical procedures that will help school
personnel support children with problem behaviors. The information you provide will be
maintained and reported in a confidential manner consistent with the standards of the
American Psychological Association. You will never be identified.
Please read the attached behavior support plan, and provide your perceptions of the
specific elements in this plan. Thank you for your contribution and assistance.
Name ofInterviewee: Role:
--------------- --------
Support plan reviewed: _
Knowledge of elements in the Behavior Support Plan.
1. I am aware of the elements ofthis behavior support plan.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
2. I know what I am expected to do to implement this behavior support plan.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
Skills needed to implement the Behavior Support Plan
3. I have the skills needed to implement this behavior support plan.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
73
4. I have received any training that I need to be able to implement this behavior support
plan.
No training needed _
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
Values are consistent with elements ofthe behavior support plan
5. I am comfortable implementing the elements of this behavior support plan
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
6. The elements of this behavior support plan are consistent with the way I believe
students should be treated.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
Resources available to implement the plan
7. My school provides the faculty/staff time needed to implement this behavior support
plan.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
8. My school provides the funding, materials, and spaced needed to implement this
behavior support plan.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
74
Administrative Support
9. My school provides the supervision support needed for effective implementation of
this behavior support plan.
I
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
10. My school administration is committed to investing in effective design and
implementation of behavior support plans.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
Effectiveness ofBehavior Support Plan
11. I believe the behavior support plan will be (or is being) effective in achieving targeted
outcomes.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
12. I believe the behavior support plan will help prevent future occurrence of problem
behaviors for this child.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
Behavior Support Plan is in the best interest ofthe student
13. I believe this behavior support plan is in the best interest of the student.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
75
14. This behavior support plan is likely to assist the child to be more successful in school.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
The Behavior Support Plan is efficient to implement
15. Implementing this behavior support plan will not be stressful.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
16. The amount of time, money and energy needed to implement this behavior support
plan is reasonable.
1
Strongly
Disagree
2
Moderately
Disagree
3
Barely
Disagree
4
Barely
Agree
5
Moderately
Agree
6
Strongly
Agree
76
REFERENCES
Alter, P. J., Conroy, M. A., Mancil, G. R., & Haydon, T. (2008). A comparison of
functional behavior assessment methodologies with young children: Descriptive
Methods and Functional Analysis. Journal ofBehavioral Education, 17,200-219.
Anderson, C. M., English, C. L., & Hendrick, T. M. (2006). Use of structured descriptive
assessment with typically developing children. Behavior Mod~fication, 30, 352-
378.
Anderson, C. M. & Long, E. S. (2002). Use of structured descriptive assessment
methodology to identify variables affecting problem behavior. Journal ofApplied
Behavior Analysis, 35, 137-154.
Barton-Atwood, S. M., Wehby, J. H., Gunter, P. L., & Lane, K. L. (2003). Functional
Behavior assessment rating scales: Intrarater reliability with students with
emotional and behavioral disorders. Behavioral Disorders, 28, 386-400.
Bergan, J. R., & Kratochwill, T. R. (1990). Behavioral consultation and therapy. New
York: Plenum Press.
Borgmeier, C. (2003). An evaluation ofinformant cOl~fidence ratings as a predictive
measure ofthe accuracy ofhypotheses.fromfitnctional assessment interviews.
Unpublished doctoral dissertation, University of Oregon.
Carr, E. G., & Durand, V. M. (1985). Reducing behavior problems through functional
communication training. Journal ofApplied Behavior Analysis, 18, 111-126.
Carr, E. G., Horner, R. H., Turnbull, A. P., Marquis, J. G., McLaughlin, D. M., McAtee,
M. L., et al. (1999). Positive behavior support for people with developmental
disabilities: A research synthesis. Washington, DC: American Association on
Mental Retardation.
Carr, E. G., Yarbrough, S. c., & Langdon, N. A. (1997). Effects of idiosyncratic stimulus
variables on functional analysis outcomes. Journal ofApplied Behavior Analysis,
30, 673-686.
77
Didden, R., Duker, P. c., & Korzilius, H. (1997). Meta-analytic study on treatment
effectiveness for problem behaviors with individuals who have mental retardation.
American Journal ofMen tal Retardation, 101, 389-399.
Dunlap, G., Kem-Dunlap, L., Clarke, S., & Robbins, F. R. (1991). Functional assessment,
curricular revision, and severe behavior problems. Journal ofApplied Behavior
Analysis, 24, 387-397.
Durrand, V. M., & Crimmins, D. B. (1988). The Motivation Assessment Scale: An
administration manual. Albany: State University of New York.
English, C. L. & Anderson, C. M. (2006). Evaluation of the treatment utility of the
analog functional analysis and the structured descriptive assessment. Journal of
Positive Behavior Interventions, 8, 212-229.
Erickson, M. J., Stage, S. A, & Nelson, R. J. (2006). Naturalistic study of the behavior of
students with EBD referred fro functional behavioral assessment. Journal of
Educational and Behavioral Disorders, 14, 31-40.
Floyd, R. G., Phaneuf, R. L., & Wilczynski, S. M. (2005). Measurement properties of
indirect assessment methods for functional behavioral assessment: A review of
research. School Psychology Review, 34, 58-73.
Hall, S. S. (2005). Comparing descriptive, experimental and informant-based assessments
ofproblem behaviors. Research in Developmental Disabilities, 26, 514-526.
Harding, J., Wacker, D. P., Cooper, L. J., Asmus, J., Jensen-Kovalan, P., & Grisolano, L.
A (1999). Combining descriptive and experimental analysis of young children
with behavior problems in preschool settings. Behavior Modification, 23, 316-
333.
Homer, R. H., Todd, A. W., Lewis-Palmer, T. Irvin, L. K., Sugai, G., & Boland, J. B.
(2004). The School-wide evaluation tool (SET): A research instrument for
assessing school-wide positive behavior support. Journal ofPositive Behavior
Interventions, 6, 3-12.
Iwata, B. A, Pace, G. M., Dorsey, M. F., Zarcone, J. R., Vollmer, T. R., Smith, R. G., et
al. (1994). Toward a functional analysis of self-injury. Journal 0.[Applied
Behavior Ana(ysis, 27, 197-209.
Iwata, B. A" Vollmer, T. R. & Zarcone, J. R. (1990). The experimental (functional)
analysis of behavior disorders: Methodology, applications, and limitations. In A
C. Repp & N. N. Singh (Eds.). Perspectives 0/1 the use o.[nonaversive and
aversive intenJentionsfor persons with developmental disabilities (pp. 301-330).
Sycamore, IL: Sycamore.
78
Kinch, C., Lewis-Palmer, T., Hagan-Burke, S., & Sugai, G. (2001). A comparison of
teacher and student functional behavior assessment interview information from
low-risk and high-risk classrooms. Education and Treatment o.[Children, 24, 480-
494.
Kwak, M. M., Ervin, R. A., Anderson, M. Z., & Austin, J. (2004). Aggreement of
function across methods used in school-based functional assessment with
preadolescent and adolescent students. Behavior Mod(fication, 28, 375-401.
Lerman, D. C. (2003). From the laboratory to community application: translational
research in behavior analysis, Journal 0/Applied Behavior Analysis, 36, 415-419
Lerman, D. C. & Iwata, B. A. (2003). Descriptive and experimental analyses of variables
maintaining self-injurious behavior. Journal 0.[Applied Behavior Analysis, 26,
293-319.
Lewis, T. J., Scott, T. M. & Sugai, G. (1994). The problem behavior questionnaire: A
teacher-based instrument to develop functional hypotheses of problem behavior in
general education classrooms. Diagnostique, 19, 103-115.
McIntosh, K., Borgmeier, C. J., Anderson, C. M., Homer, R. H., Rodriguez,
B. & Tobin, T. (2008). Technical adequacy of the Functional Assessment
Checklist-Teachers and Staff. Journal of Positive Behavior Interventions.
March, R., Homer, R. H., Lewis-Palmer, L., Brown, D., Crone, D., Todd, A. W., & Carr,
E. (2000). Functional Assessment Checklist/or Teachers and Staff(FACTS).
Eugene: Department of Educational and Community Supports, University of
Oregon.
Martens, B. K., DiGennaro, F. D., Reed, D. D., Szczech, F. M., & Rosentthal, B. D.
(2008). Contingency space analysis: an alternative method for identifying
contingent relations from observational data. Journal 0.[Applied Behavior
Analysis, 41, 69-81.
Murdock, S. G., O'Neill, R. E., & Cunningham, E. (2005). A comparison of results and
acceptability of functional behavioral assessment procedures with a group of
middle school students with emotional/behavioral disorders. Journal 0.[
Behavioral Education, 14, 5-18.
Newcomer, L. L. & Lewis, T. J. (2004). Functional behavioral assessment: An
investigation of assessment reliability and effectiveness of function-based
interventions. Journal 0.[Emotional and Behavioral Disorders, 12, 168-181.
79
Northup, 1., Wacker, D., Sasso, G., Steege, M., Cigrand, K, Cook, J. et al. (1991). A
brief functional analysis of aggressive and alternative behavior in an outclinic
setting. Journal ofApplied Behavior Analysis, 24, 509-522.
O'Neill, R. E., Homer, R. H., Albin, R. W., Sprauge, 1. R., Storey, K., & Newton, J. S.
(1997). Functional assessment and program developmentfor problem behaviors:
A practical handbook. New York: Brooks/Cole.
Peck 1. & Sasso, G. M. (1997). Use of the structural analysis hypothesis testing model to
improve social interactions via peer-mediated intervention. Focus on Autism &
Other Developmental Disabilities, 12, 219-231.
Peck, J., Sasso, G. M., & Stambaugh, M. (1998). Functional analyses in the classroom:
Gaining reliability without sacrificing validity. Preventing School Failure, 43, 14-
18.
Penno, D. A., Frank, A. R., & Wacker, D. P. (2000). Instructional accommodations for
adolescent students with severe emotional or behavioral disorders. Behavioral
Disorders, 25, 325-343.
Reed, H., Thomas, E., Sprague, J. R., & Homer, R. H. (1997). The student guided
functional assessment interview: An analysis of student and teacher agreement.
Journal o.fBehavioral Education, 7, 33-49.
Roberts, M. L., Marshall, J., Nelson, J. R., Albers, C. A. (2001). Curriculum-based
assessment procedures embedded within functional behavioral assessments:
Identifying escape-motivated behaviors in a general education classroom. School
Psychology Review, 30, 264-277.
Stage, S. A., Cheney, D., Walker, B., & LaRocque, M. (2002). A preliminary
discriminant and convergent validity study of the teacher functional behavioral
assessment checklist. School Psychology Review, 31, 71-93.
Stage, S. A., Jackson, H. G., Moscovitz, K., Erickson, M. J., Thurman, S. 0., Wyeth, J.,
et al. (2006). Using multimethod-multisource functional behavioral assessment
for students with disabilities. School Psychology Review, 35, 451-471.
Stichter, J. & Conroy, M. (2005). Using structural analysis in natural settings: A
responsive functional assessment strategy. Journal of Behavioral Education, 14
(1), 19-34.
Stichter, J. P., Sasso, G. M., & Jolivette, K. (2004). Structural analysis and intervention in
a school setting: Effects on problem behavior for a student with an
emotional/behavioral disorder.
80
Van Camp, C. M., Lerman, D. C., Kelley, M. E., Roane, H. S., Contrucci, S. A., &
Vorndran, C. M. (2000). Further analysis of idiosyncratic antecedent influences
during the assessment and treatment of problem behavior. Journal ofApplied
Behavior Analysis, 33, 207-221.
Watson, T. S. & Sterling H. E. (1998). Brief functional analysis and treatment of a vocal
tic. Journal ofApplied Behavior Analysis, 3], 471-474.
Wilder, D. A., Chen, L., Atwell, J., Pritchard, J., & Weinstein, P. (2006). Brief functional
analysis and treatment of tantrums associated with transitions in preschool
children. Journal ofApplied Behavior Analysis, 39, 103-107.
Zarcone, J. R., Rodgers, T. A., Iwata, B. A., Rourke, D. A., & Dorsey, M. F. (1991).
Reliability analysis of the Motivation assessment scale: A failure to replicate.
Research in Developmental Disabilities,]2, 349-360.