JOURNAL OF APPLIED BEHAVIOR ANALYSIS 1997, 30, 653–672 NUMBER 4 (WINTER 1997)

FUNCTIONAL ANALYSIS AND TREATMENT

OF ELOPEMENT
CATHLEEN C. PIAZZA, GREGORY P. HANLEY, LYNN G. BOWMAN,
JOHN M. RUYTER, STEVEN E. LINDAUER, AND DEBORAH M. SAIONTZ
KENNEDY KRIEGER INSTITUTE AND
JOHNS HOPKINS UNIVERSITY SCHOOL OF MEDICINE

Elopement is a dangerous behavior because children who run away may encounter life-
threatening situations (e.g., traffic). We conducted functional analyses of the elopement
of 3 children who had been diagnosed with developmental disabilities. The results iden-
tified a maintaining reinforcer for the elopement of 1 child, but the data were difficult
to interpret for 2 of the children. Subsequent reinforcer assessments were used to help
to clarify the reinforcers for elopement for these 2 children. Results of the functional
analyses and reinforcer assessments then were used to develop successful treatments to
reduce elopement. The findings are discussed in terms of (a) the application of functional
analysis methodology to elopement, (b) the use of reinforcer assessments to identify
potential reinforcers when standard functional analyses are undifferentiated, and (c) the
utility of assessment-based treatments for elopement.
DESCRIPTORS: elopement, developmental disabilities, functional analysis, rein-
forcer assessments, concurrent operants

Elopement is typically defined as repeated ment and treatment of this behavior prob-
attempts to leave designated areas without lem. In most studies on elopement, individ-
permission or supervision (Bodfish, 1992). uals were treated with multiple interventions
Elopement may interfere with instructional such as reinforcement for the absence of
activities and hinder skill acquisition in elopement, time-out for elopement, and
classroom settings (Chambers, Sanok, & graduated levels systems in which the indi-
Striefel, 1980), as well as expose an individ- vidual gained access to less restrictive envi-
ual to dangerous situations (e.g., traffic). Ac- ronments and greater access to reinforcers
cording to Garner (1991), individuals who contingent upon the absence of elopement
elope often are placed in more restrictive set- (Chambers et al., 1980; Garner, 1991).
tings to maintain their safety. Garner (1991) treated the elopement of 1
Jacobson (1982) estimated the prevalence child with profound mental retardation who
of elopement to be 4.9% in a population of resided in a group home. The treatment pro-
over 30,000 individuals receiving develop- gram involved increased amounts of freedom
mental disabilities services. Despite this rel- of movement within the group home (i.e.,
atively high prevalence of elopement among from restricting him to rooms without exits
persons with developmental disabilities,
in the home to allowing him unrestricted
there is a paucity of research on the assess-
access to all areas in the home) contingent
This investigation was supported in part by Grant on the absence of elopement, and included
MCJ249149-02 from the Maternal and Child Health response interruption (teaching the partici-
Service of the U.S. Department of Health and Human
Services. We thank Rachel Thompson for her helpful pant to respond to the command ‘‘stop’’
comments on this manuscript. when he began to run away), time-out fol-
Requests for reprints should be sent to Cathleen C. lowing episodes of elopement, and reinforce-
Piazza, Neurobehavioral Unit, The Kennedy Krieger
Institute, 707 N. Broadway, Baltimore, Maryland ment for appropriate communication to
21205. leave the group home. A similar program

653
654 CATHLEEN C. PIAZZA et al.

was described by Chambers et al. (1980) for disabilities. Subsequent reinforcer assess-
an individual who was described as ‘‘ungov- ments were used to identify reinforcers for 2
ernable.’’ Treatment was implemented with- of the children. The results of the functional
in a self-contained classroom and consisted analyses and reinforcer assessments then
of time-out contingent upon elopement and were used to develop treatments to reduce
a levels program in which the participant elopement. The treatments were altered
gained increasing access to classroom activi- (e.g., schedules of reinforcement were
ties and movement within the classroom thinned) to make them more practical for
contingent upon the absence of elopement. caregivers and were extended from analogue
These studies are limited in that functional conditions to more natural settings (e.g., the
control of the treatment was not demon- community).
strated, and each study involved only 1 par-
ticipant.
The standard of practice for reducing de- GENERAL METHOD
structive behavior is to prescribe treatments This is a three-experiment study with data
based on the results of behavioral assess- for each participant presented individually in
ments. For example, Iwata, Dorsey, Slifer, each experiment. First, a functional analysis
Bauman, and Richman (1982/1994) showed of elopement was conducted with each par-
that the consequences that maintain self-in- ticipant. If the results of the functional anal-
jurious behavior (SIB) could be identified ysis were unclear, reinforcer assessments were
using functional analysis. Vollmer, Iwata, conducted. The results of these behavioral
Zarcone, Smith, and Mazaleski (1993) assessments then were integrated into treat-
showed that the results of functional analyses ment packages.
could be used to prescribe treatments to re-
duce SIB. The functional analysis method- Participants
ology was developed originally for the as- Three individuals were admitted to an in-
sessment of SIB but has been modified to patient unit specializing in the assessment
assess and treat a variety of behavior prob- and treatment of destructive behavior. All 3
lems such as aggression (Fisher et al., 1993; participants were admitted primarily for the
Piazza et al., 1997), pica (Piazza, Hanley, & assessment and treatment of elopement, and
Fisher, 1996), tantrums (Carr & Newsom, they also engaged in aggressive and disrup-
1985), and psychotic speech (Fisher, Piazza, tive behaviors. Owen was a 10-year-old boy
& Page, 1989; Mace & Lalli, 1991). In cases who had been diagnosed with moderate
in which results of a functional analysis are mental retardation, autism, attention deficit
equivocal or suggest that behavior is main- hyperactivity disorder (ADHD), and a sei-
tained independent of the social environ- zure disorder. Owen could follow one-step
ment (i.e., maintained by automatic rein- instructions and communicated through ges-
forcement), results of reinforcer assessments tures. Owen’s caregivers reported that he
have been used to prescribe treatments (Fish- would commonly elope from rooms and
er et al., 1994; Steege, Wacker, Berg, Cig- then engage in dangerous behaviors such as
rand, & Cooper, 1989). These same strate- touching electrical cords and climbing on
gies could be applied to the assessment and furniture and windowsills.
treatment of elopement. Ray was an 11-year-old boy who had been
In the current investigation, we modified diagnosed with severe mental retardation,
the functional analysis method to assess the autism, bipolar disorder, and ADHD. Ray
elopement of 3 children with developmental was independent with his daily living skills,
 ELOPEMENT 655

and had good receptive language skills (e.g., situations in which the participants ran
he could follow two-step instructions) but away.
limited expressive language skills (e.g., he Demand sessions were conducted to de-
used a few rote expressions in a variety of termine whether the participants engaged in
contexts). Ray’s caregiver reported that he elopement to escape tasks. Attention sessions
frequently ran away when out in the com- were conducted to determine whether the
munity. participants engaged in elopement to gain
Ty was a 4-year-old boy who had been access to adult attention. Tangible sessions
diagnosed with cerebral palsy, a seizure dis- were conducted to evaluate whether the par-
order, and learning and speech delays (his ticipants engaged in elopement to gain ac-
cognitive level had not been evaluated suc- cess to tangible items. Ignore sessions were
cessfully in the past because of his destruc- conducted to determine whether the partic-
tive behavior). Ty followed complex instruc- ipants engaged in elopement in the absence
tions (e.g., ‘‘Go to your room and get your of social consequences. Toy play or control
football’’), initiated conversation, and spoke sessions were conducted to evaluate the rates
in complete sentences. All 3 participants re- of elopement in a condition in which the
quired constant supervision because of the absence of elopement resulted in differential
severity and dangerous nature of their elope- reinforcement.
ment. Target Behaviors, Data Collection, and
The functional analyses for elopement Interrater Agreement
were based on procedures described by Iwata
Elopement was defined as any part of the
et al. (1982/1994). Sessions were 10 min in
participant’s body passing through the door-
length and were modified as follows. First,
way (Owen and Ray) and moving or at-
we attempted to simulate the setting in
tempting to move 3 m (or more) away from
which the participant typically engaged in
the therapist (for Ty during all assessments
elopement in the natural environment. Sec- and for Ray during treatment extension
ond, because elopement is a dangerous be- only) during the functional analyses and
havior, the participants could not be allowed treatment assessments. During the treatment
to run away without eventually retrieving analyses for Ty, attempts to elope were
them (i.e., the behavior could not be ig- scored as elopement. During the reinforcer
nored), and all families reported going after assessments, card touches were defined as any
their children when they ran away. Retriev- part of the participant’s hand touching a
ing the child also permitted multiple oppor- card. For Owen, the cards were a green rect-
tunities to observe the behavior and for the angle (7 cm by 18 cm), a yellow triangle (13
child to experience the consequences of cm by 13 cm by 13 cm), and a red square
elopement that were prescribed in the func- (13 cm by 13 cm). For Ty, blue, orange, and
tional analysis. Therefore, across all condi- white index cards (7 cm by 13 cm) were
tions, when the participant eloped, he was used. The duration of appropriate walking,
retrieved by the therapist on a fixed-time defined as the participant remaining within
schedule as described below. No differential 3 m (or less) of the therapist, was recorded
consequence occurred for the participants’ during Ty’s treatment analyses. Trained ob-
other destructive behavior (e.g., aggression) servers used laptop computers to record each
across all conditions. The inclusion or exclu- occurrence of elopement and card touches
sion of some functional analysis conditions and duration of appropriate walking (Ty
was based on caregiver report regarding the only). The percentage of the session with ap-
656 CATHLEEN C. PIAZZA et al.

propriate walking then was calculated by di- EXPERIMENT 1:

viding the total duration of appropriate RAY
walking by the total session time. METHOD
Two observers simultaneously but inde-
pendently scored target responses during Functional Analysis
44%, 40%, and 42% of the functional anal- The initial functional analysis was con-
ysis sessions and 40%, 59%, and 38% of the ducted within a multielement design in
elopement treatment sessions for Owen, which four conditions (tangible, ignore, at-
Ray, and Ty, respectively. Interrater agree- tention, and toy play) were alternated ac-
ment was assessed during 35% and 56% of cording to a random schedule. Because the
the reinforcer assessment sessions for Owen rates of elopement were variable during the
and Ty, respectively. Interrater agreement ignore and attention sessions of the multiel-
was assessed during the schedule thinning ement analysis for Ray, a sequential pairwise
for Ray and Ty and during the extension of analysis was conducted in which the atten-
treatment to different settings and caregivers tion and ignore conditions were each com-
for Owen and Ray during 60%, 36%, and pared to the toy play condition in separate
50% of sessions for Owen, Ray, and Ty, re- phases (Iwata, Duncan, Zarcone, Lerman, &
spectively. Agreement coefficients for elope- Shore, 1994) to further evaluate the effects
ment, card touches, and appropriate walking of attention and low stimulation.
were calculated by partitioning each session Ray’s mother reported that he typically
ran away to obtain food (e.g., chips) and
into 10-s intervals and dividing the number
that she often gave him food following
of exact agreements by the sum of agree-
elopement to ‘‘calm him down.’’ She also re-
ments plus disagreements and multiplying
ported that Ray did not run away during
by 100%.
instructional activities (i.e., school work or
During the functional analyses, mean
tasks). Therefore, attention, tangible, ignore,
agreement for elopement was 99% (range, and toy play conditions were assessed with
97% to 100%), 99% (range, 93% to 100%), Ray, but no demand sessions were conduct-
and 89% (range, 82% to 98%) for Owen, ed.
Ray, and Ty, respectively. Mean agreement for Ray’s sessions were conducted in an ex-
card touches was 97% (range, 84% to 100%) perimental area consisting of two rooms (4
and 99% (range, 85% to 100%) for Owen m by 5 m) connected by a single doorway.
and Ty, respectively. During the treatment as- Each session began with Ray situated in
sessments, mean agreement for elopement Room A. Room A contained two chairs and
was 99% (range, 97% to 100%), 100%, and a table, and Room B contained a table.
95% (range, 67% to 100%) for Owen, Ray, A tone was sounded every 40 s in all con-
and Ty, respectively. During the treatment ditions as a signal to the therapist to physi-
analyses for Ty, mean agreement for appro- cally guide Ray to sit in a chair in Room A.
priate walking was 85% (range, 41% to If he then eloped to Room B, he was guided
100%). During the schedule thinning and back to Room A to sit in a chair. If Ray was
treatment extension, mean agreement for already seated in a chair in Room A or had
elopement was 100%, 99% (range, 96% to returned to Room A independently, he was
99%), and 98% (range, 93% to 100%), for physically guided to sit in a different chair
Owen, Ray, and Ty, respectively, and mean in Room A. Ray was guided to sit in a chair
agreement for appropriate walking was 96% so that physical interaction would occur in-
(range, 93% to 100%) for Ty. dependent of elopement in an attempt to
 ELOPEMENT 657

control for physical interaction as a differ- Treatment Assessment

ential consequence. That is, we wanted to The effects of treatment were evaluated
isolate escape, attention, and access to tan- using an ABAB design. The baseline (A
gible items that were the programmed con- phase) was followed by implementation of
sequences for elopement during the func- treatment (B phase) followed by a return to
tional analysis. The tone was also used as a baseline (A) and a return to treatment (B).
prompt to the therapist to initiate or discon- Sessions were 10 min in length and were
tinue contingencies specific to each func- conducted in the same rooms (A and B)
tional analysis condition, as described below. used in the functional analysis. Across all
During attention sessions, Ray was given conditions, a tone signaled the therapist to
toys and was asked to play quietly. If he en- physically guide Ray to sit in a chair in
gaged in elopement, the therapist followed Room A every 40 s. If Ray was already seat-
him into Room B and provided mild verbal ed in a chair in Room A or had returned to
reprimands (e.g., ‘‘don’t run away’’) until the Room A independently, he was physically
40-s interval elapsed (i.e., when the tone guided to sit in a different chair in Room A.
sounded) or until the child returned to Prior to each session, Ray was given access
Room A. All other responses were ignored. to attention and chips for 1 min.
During toy play sessions, the therapist The baseline condition was similar to the
and child were in Room A, and preferred tangible and attention conditions of the
toys and food were available. The therapist functional analysis (i.e., contingent on
interacted with Ray (e.g., threw a ball back elopement, he was given access to chips and
and forth) and provided praise or food every attention in Room B). We combined the
40 s contingent upon the first 5-s period in contingencies of tangible and attention dur-
which elopement did not occur. Otherwise, ing baseline because the results of the func-
no differential consequence was provided for tional analysis indicated that Ray’s elope-
elopement or other destructive behavior. ment was maintained by both sources of re-
During tangible sessions, a bag of chips inforcement. The treatment condition was
was placed in Room B. Ray was allowed to similar to the baseline condition in that the
eat chips in Room B for 1 min prior to the tangible items were located in Room B.
start of the session. When the session began, During treatment, access to 20 s of attention
the therapist guided Ray into Room A. Con- or to 5 s of chips was delivered every 50 s
tingent on elopement, Ray was allowed to contingent on the nonoccurrence of elope-
eat chips in Room B for the remainder of ment (i.e., a differential-reinforcement-of-
the 40-s interval. Small amounts of chips other-behavior [DRO] schedule). Attention
were given to Ray by the therapist to prevent consisted of verbal and physical interaction
him from consuming large quantities of (e.g., saying, ‘‘You’re doing a great job,’’
chips in a short period of time and to pre- while patting him on the back). Small pieces
vent him from bringing chips back into of chips were given to Ray by the therapist
Room A. each time he consumed the previous chip.
During the ignore condition, the therapist Ray was allowed to choose between the two
and Ray were in Room A. No other mate- reinforcers by touching one of two colored
rials were present. No differential conse- index cards (12 cm by 7 cm) that corre-
quence occurred for elopement. The thera- sponded to attention (green card) or chips
pist observed Ray from the doorway of (blue card). If Ray touched a card, the cor-
Room A when he eloped. responding reinforcer was delivered. At-
658 CATHLEEN C. PIAZZA et al.

Figure 1. Number of elopements per minute during the functional analysis (top panel) and during the
treatment assessment for Ray (bottom panel).

tempts to touch both cards were blocked. ducted under this arrangement. Subsequent-
Elopement (leaving Room A) resulted in the ly, treatment was extended to various places,
resetting of the DRO timer, and no atten- including hospital vending areas, the hospi-
tion or chips were delivered (the therapist tal cafeteria, and restaurants in the com-
blocked his attempts to eat the chips in munity.
Room B). The DRO interval resumed when
Ray reentered Room A. These consequences RESULTS AND DISCUSSION
were implemented because Ray’s mother During Ray’s functional analysis (top pan-
could not block or prevent Ray from run- el of Figure 1), high levels of elopement were
ning away because of his size, but she agreed consistently observed in the tangible condi-
not to give him food items (e.g., chips) when tion relative to the toy play condition. Ray
he ran away. engaged in more variable levels of elopement
Following the treatment analysis, the in the ignore and attention conditions rela-
schedule of reinforcement was thinned and tive to the toy play condition. Pairwise com-
the treatment was extended across different parisons (attention and ignore) and control
settings. The thinning of the schedule oc- (toy play) conditions were conducted be-
curred over eight sessions, and the final cause rates of elopement were similar across
schedule arrangement consisted of access to these two test conditions. When attention
55 s of attention or 25 s of chips delivered was compared to toy play, Ray engaged in
contingent on remaining in Room A for 5 high levels of elopement in the attention
min. Eight additional sessions were con- condition, suggesting that attention served
 ELOPEMENT 659

as reinforcement for elopement. When ig- EXPERIMENT 2:

nore was compared to toy play, Ray engaged OWEN
in near-zero levels of elopement in both con-
ditions, suggesting that Ray’s elopement was METHOD
not maintained independent of social rein- Functional Analysis
forcement. The reason for the similar rates
of elopement during attention and ignore Owen’s mother reported that he ran away
conditions of the multielement analysis was in all situations (e.g., in the community, at
not clear. As Vollmer, Marcus, Ringdahl, and school) and noted that he often attempted
Roane (1995) suggested, it is possible that to obtain stringy items (e.g., pieces of a
some individuals may have difficulty dis- mop) when he eloped. She typically allowed
criminating between the rapidly alternating him to play with these objects following
contingencies during the multielement func- elopement. Five functional analysis condi-
tional analysis. It appears that this may have tions (attention, demand, ignore, tangible,
been the case with Ray, who may not have and toy play) were assessed using a multiel-
discriminated between the attention and ig- ement design.
nore conditions when they were rapidly al- The setting and descriptions of conditions
ternated during the multielement analysis for Owen’s functional analysis were similar
because of the similarity of the two condi- to those described for Ray, with the follow-
tions (i.e., the only difference was the verbal ing exceptions. During demand sessions, the
reprimand in the attention condition). In therapist issued instructions using sequential
addition, behavior may not have been extin- verbal, gestural, and physical prompts on a
guished in the ignore condition because, at fixed-time (FT) 40-s schedule. If Owen
times, the same therapist conducted both completed the task following the verbal or
conditions. The differences between condi- gestural prompt, he received praise from the
tions may have been easier to discriminate therapist and a break until the beginning of
when only two conditions (test and control) the next 40-s interval. If Owen eloped, the
were compared during the pairwise analysis. therapist removed the materials and turned
However, this hypothesis is speculative. away from him until the start of the next
The results for the treatment assessment 40-s interval (i.e., he was permitted to escape
are depicted in the bottom panel of Figure from the task). If Owen eloped between in-
1. In the baseline condition (first and third structions, he was guided back to the chair
phases) during the treatment assessment in at the next 40-s interval and given a new
which elopement produced access to adult instruction.
attention and chips, Ray engaged in high During tangible sessions, a string (56 cm
levels of elopement. During the treatment in length and 2 cm in diameter) was placed
condition (second and fourth phases) in in Room B. Owen was allowed to play with
which the DRO schedule was implemented, the string in Room B for 2 min prior to the
Ray engaged in near-zero levels of elope- start of the session. When the session began,
ment. In addition, near-zero levels of elope- the therapist guided Owen into Room A.
ment were maintained during schedule thin- Contingent on elopement, Owen was al-
ning. Although elopement was somewhat lowed to play with the string in Room B
variable when treatment was extended to until the 40-s interval elapsed. The string
other settings, it remained substantially low- was tied to a table in Room B to prevent it
er than baseline. from being brought back into Room A.
660 CATHLEEN C. PIAZZA et al.

Reinforcer Assessment touch. Training trials were discontinued

During reinforcer assessment, a concur- when Owen independently engaged in card
rent-operants design (Catania, 1963; Herrn- touching for 80% of three consecutive
stein, 1970) was used to evaluate the rein- blocks of 10 trials. Before each individual
forcing effects of attention and access to a session of the reinforcer assessment, training
tangible item in three phases. The order in trials were conducted to expose Owen to the
which the two variables were assessed was consequences for touching each card in the
randomly determined and occurred as fol- session (descriptions of the phases and the
stimuli assessed are presented below). Trials
lows: attention versus control in the first
ended for each individual session when he
phase; access to the tangible item versus con-
independently chose any card except the
trol in the second phase; access to the tan-
control card for three consecutive trials.
gible item versus access to adult attention
Sessions were 10 min in length and were
versus control in the final phase.
conducted in a room (3 m by 5 m). Owen
During the functional analysis, the high-
was seated in a chair at a table and was ver-
est rates of elopement occurred in the tan-
bally prompted to touch a card at the start
gible condition, followed by the attention
of the session. Touching a card resulted in
condition. Therefore, we began our assess- the contingencies described above for 30 s
ment by evaluating the reinforcing effects of on a fixed-ratio (FR) 1 schedule (e.g., if
these two stimuli (tangible item and atten- Owen touched the yellow triangle, he re-
tion). Laminated cardboard shapes corre- ceived 30 s of adult attention). When Owen
sponded to the different consequences avail- touched a card, the cards were removed. The
able for card touching throughout each as- location of the cards was randomly rear-
sessment. After Owen touched a card, all ranged during the reinforcement interval.
cards were removed and the consequence as- Following the reinforcement interval, Owen
sociated with that card was available for 30 was repositioned in front of the cards and
s. Touching a green rectangle resulted in ac- was verbally prompted to touch a card. He
cess to the tangible item (string), touching a was blocked from leaving the chair until he
yellow triangle resulted in access to attention touched a card. Otherwise, if Owen did not
(e.g., saying, ‘‘You’re doing a great job,’’ touch a card, no differential consequence oc-
while patting him on the back), and touch- curred.
ing a red square resulted in no differential
consequence (control). Treatment Assessment
Prior to the start of the reinforcer assess- The treatment assessment was conducted
ment, training trials were conducted in using an ABAB design. The baseline phase
which Owen was allowed approximately 5 s (A) was followed by the introduction of
to independently touch a card placed on a treatment (B), a return to the baseline phase
table in front of him. After 5 s, if Owen did (A), and a reintroduction of treatment (B).
not touch a card, sequential verbal, gestural, Sessions were 10 min in length and were
and physical prompts were used to prompt conducted in the same rooms (A and B)
him to touch one of the cards (randomly used in the functional analysis and reinforcer
determined) to allow him to contact the cor- assessment. Treatment was evaluated in the
responding consequence. Attempts to touch tangible condition of the functional analysis
more than one card simultaneously were because (a) rates of elopement were highest
blocked. Access to the chosen consequence in the tangible condition during the func-
was delivered immediately following a card tional analysis, (b) the reinforcer assessment
 ELOPEMENT 661

demonstrated that the tangible items were his teacher. During all sessions, the mother
more effective reinforcers than attention was, and teacher implemented the treatment con-
and (c) the antecedent conditions for the tingencies described above.
tangible sessions (i.e., the absence of mate-
rials, adult attention, and tasks) were iden- RESULTS AND DISCUSSION
tical to those of the attention, ignore, and During Owen’s functional analysis (top
toy play conditions. panel of Figure 2), the highest rates of elope-
In all sessions, the tangible item (string) ment occurred in the tangible condition fol-
was located in Room B, and, prior to each lowed by the attention condition suggesting
session, Owen was given access to string for that his elopement may have been main-
2 min. The baseline and treatment condi- tained by access to tangible items (i.e., access
tions were similar to the tangible condition to stringy items). However, we could not
in the functional analysis (i.e., contingent on rule out the possibility that Owen’s elope-
elopement, Owen gained access to the string ment was sensitive to multiple sources of re-
in Room B). In addition, Owen was given inforcement, because rates of elopement
continuous noncontingent (NCR) access to were relatively similar across three (attention,
string-like items in Room A during treat- demand, ignore) of the four test conditions
ment. The string-like items included ropes, relative to the toy play condition. It was pos-
bungee cords, and shoelaces and were placed sible that the lack of differentiated results for
on the table in Room A. Owen was pre- Owen were due to his failure to discriminate
vented from taking the strings from Room between conditions, as suggested for Ray.
A to Room B. Elopements were ignored by However, with Owen, rates of elopement
the therapist (i.e., the therapist provided no were equivalent across conditions that were
differential consequence); however, Owen highly dissimilar (i.e., demand vs. ignore);
was able to obtain the string in Room B therefore, it seemed less likely that failure to
when he eloped. discriminate between session contingencies
Following the treatment analysis, the contributed to the lack of differentiation be-
treatment was extended across different set- tween conditions. Therefore, rather than
tings and caregivers. Owen’s mother and conduct additional functional analysis ses-
teacher were trained to implement the treat- sions, we decided to evaluate whether a re-
ment contingencies (i.e., to provide Owen inforcer assessment could be used as an in-
with noncontingent access to string and no direct method for identifying potential re-
differential consequence for elopement). inforcers for use in the treatment of elope-
Training was conducted first by the therapist ment.
modeling the treatment contingencies for The results of the reinforcer assessment
the mother or teacher. Next, the mother or are depicted in Figure 2. Owen touched the
teacher role played implementation of the card that produced attention more often
treatment with a confederate (a therapist). than he touched the control card in the first
Finally, the mother or teacher implemented phase of the reinforcer assessment, suggest-
the treatment with Owen while the therapist ing that attention functioned as a reinforcer.
observed and provided feedback. Nine ses- He touched the card that produced access to
sions were conducted in the rooms described the tangible item more often than he
above with Owen’s mother as therapist. touched the control card in the second
Three additional sessions were conducted in phase, suggesting that access to the tangible
Owen’s home by his mother, and three ses- item functioned as a reinforcer. During the
sions were conducted in Owen’s school by third phase, Owen touched the card to ob-
662 CATHLEEN C. PIAZZA et al.

Figure 2. Number of elopements per minute during the functional analysis (top panel), number of card
touches per minute during reinforcer assessment (middle panel), and number of elopements per minute during
the treatment assessment for Owen (bottom panel).

tain the tangible item at a high rate and item, Owen engaged in high levels of elope-
more often than he touched the attention ment. During treatment (second and fourth
and control cards, suggesting that access to phases), in which Owen was given contin-
the string was a more effective reinforcer uous noncontingent access to tangible items
than was adult attention. while elopement also resulted in access to
The results of the treatment assessment the tangible item, he engaged in near-zero
are also depicted in Figure 2. In the baseline levels of elopement. From a conceptual
condition (first and third phases), in which standpoint, the paradigm used during treat-
elopement resulted in access to the tangible ment was similar to a concurrent-operants
 ELOPEMENT 663

arrangement. In this type of arrangement, ran away. Therefore, attention, demand, ig-
two or more responses are available at the nore, and control sessions were conducted
same time, and each is correlated with a dif- with Ty. The functional analysis was con-
ferent schedule or type of reinforcement ducted using a multielement design in which
(Catania, 1992). In this case, remaining in four conditions (attention, ignore, demand,
Room A and eloping to Room B were si- and toy play) were alternated according to a
multaneously available, and each was corre- random schedule.
lated with different reinforcers (a variety of The procedures used in Ty’s functional
strings in Room A and a single string in analysis were different from those described
Room B). In addition, the effort associated for Ray and Owen because the analysis was
with each response was different (eloping to conducted in an open common area (8 m
Room B involved a higher response effort by 24 m) to more closely approximate the
than remaining in Room A). Consistent conditions under which Ty eloped in the
with studies on concurrent-operants arrange- natural environment (i.e., he typically ran
ments, Owen allocated his time almost ex- away when he was with an adult in an open
clusively to the behavior (remaining in area). Located within the common area were
Room A) that resulted in a higher quality of four doorways and numerous chairs and so-
reinforcement (Miller, 1976) and was asso- fas lining the walls. Ty could not open the
ciated with a lower response effort (Horner doors independently because of his size. Pri-
& Day, 1991). or to each session, the therapist would ex-
At home, Owen’s mother gave him access plain the session contingencies to Ty. In all
to stringy items when she could not engage conditions, the therapist prompted Ty to
him in a structured activity (e.g., when she walk next to him or her on an FT 30-s
was preparing dinner). The school provided schedule across all sessions by briefly guiding
Owen with stringy items following comple- Ty to the therapist’s side while saying, ‘‘Stay
tion of tasks. When the treatment was ex- next to me.’’ After this brief prompt, the
tended to different caregivers (i.e., mother therapist had no further physical contact
and teacher) and environments (i.e., home with Ty until the next 30-s interval.
and school), treatment gains were main- The demand session consisted of instruct-
tained. Owen did not engage in elopement ing Ty to walk to a designated work area and
when his mother or teacher implemented then to complete a task (putting crumpled
the treatment. pieces of paper into a bin) in that area. The
therapist issued the instruction, ‘‘Walk with
me’’ and, when Ty and the therapist reached
EXPERIMENT 3: the work area, the therapist issued instruc-
TY tions to Ty using sequential verbal, gestural,
METHOD and physical prompts. If Ty completed the
task, he received verbal praise from the ther-
Functional Analysis apist and then was prompted to walk toward
Ty’s mother reported that when he ran the next work area. If Ty eloped during any
away, he would often stop, turn around and part of the transition or during the instruc-
laugh, and see whether anyone was following tional sequence in the work area, the thera-
or paying attention to him. She usually pist stopped walking and issuing instructions
chased after him and reprimanded him for for the remainder of the 30-s interval (i.e.,
running away. She indicated that he did not Ty was permitted to escape from the task).
attempt to obtain tangible items when he During attention sessions, the therapist
664 CATHLEEN C. PIAZZA et al.

delivered attention in the form of a brief ver- may have been maintained by the conse-
bal reprimand (e.g., ‘‘Don’t run away’’) while quences produced by running. We attempt-
running after him contingent on elopement. ed to evaluate the reinforcing effects of run-
The therapist continued to run after Ty until ning by assessing whether it would strength-
the end of the 30-s interval. If Ty walked en another response (card touches). A com-
appropriately (remained within 3 m of the parison of the potential positive reinforcers
therapist), no differential consequence was (attention and running) was conducted.
delivered. Index cards corresponded to the different
During the control condition, elopement consequences available for card touching
resulted in no differential consequence. In throughout each assessment. After Ty
addition, Ty received brief verbal attention touched a card, all cards were removed and
every 30 s contingent on 5 s of no elope- the consequences associated with that card
ment. were available for 30 s. Touching a blue card
In the ignore condition, no differential resulted in access to running, touching an
consequence was provided for appropriate orange card resulted in access to attention,
walking or elopement. That is, the therapist and touching a white card resulted in no
continued walking in the same general di- differential consequence (control). Running
rection as he or she had been walking before consisted of the therapist and Ty running
elopement occurred, and no verbal or phys- side by side in the hallway (the therapist
ical attention was provided to Ty following held Ty’s hand so that Ty remained in the
elopement. hallway). The therapist did not otherwise in-
teract with Ty while running. Attention con-
Reinforcer Assessment sisted of verbal praise and physical contact
The reinforcing effects of running versus (e.g., saying, ‘‘You’re doing a great job,’’
those associated with the control condition while patting him on the back).
were assessed in an ABAB design. Next, a Prior to the start of the reinforcer assess-
concurrent-operants paradigm was used to ment, training trials were conducted in
compare attention to a control condition. A which Ty was allowed approximately 5 s to
concurrent-operants paradigm then was used touch a card placed on a table in front of
to compare the reinforcing effects of running him. After 5 s, if Ty did not touch a card
and attention to a control. independently, sequential verbal, gestural,
Rates of elopement were equivalent across and physical prompts were used to prompt
all conditions in Ty’s functional analysis, him to touch one of the cards (randomly
which suggested that (a) elopement or run- determined) to allow him to contact the cor-
ning away produced its own reinforcement, responding consequence. Attempts to touch
or (b) the behavior was maintained by mul- more than one card simultaneously were
tiple sources of reinforcement (i.e., access to blocked. Access to the chosen consequence
adult attention, escape from demands, and was delivered immediately following a card
the automatic reinforcement derived from touch. Training trials were discontinued
running away). Methods for identifying the when Ty independently engaged in card
specific reinforcement directly produced by touching for 80% of three consecutive
a behavior have not been well developed blocks of 10 trials. Before each individual
(Vollmer, 1994). We speculated that because session of the reinforcer assessment, training
Ty’s elopement consisted of running away trials were conducted to expose Ty to the
and persisted in the absence of social con- consequences for touching each card in the
sequences (ignore condition), elopement session. Trials ended for each individual ses-
 ELOPEMENT 665

sion when he independently chose any card That is, the antecedent conditions for the
except the control card for three consecutive ignore session (i.e., the absence of materials,
trials. adult attention, and tasks) were identical to
Sessions were 10 min in length and were those of the attention and control condi-
conducted in a hallway. Ty stood in front of tions. However, the presence of demands in
a table and was verbally prompted to touch the demand condition necessitated that a
a card at the start of the session. Touching treatment evaluation be conducted under
a card resulted in the contingencies de- those specific stimulus conditions.
scribed above for 30 s on an FR 1 schedule The baseline condition for the first treat-
(e.g., if Ty touched the orange card, he ment assessment was identical to the ignore
would receive 30 s of adult attention). When condition in the functional analysis (i.e., no
Ty touched a card, the cards were removed. differential consequences were provided for
The location of the cards was randomly re- either elopement or appropriate walking).
arranged during the reinforcement interval. Treatment consisted of DRA and blocking.
Following the reinforcement interval, Ty was Reinforcement was delivered for 5 s of ap-
repositioned in front of the cards and was propriate walking and consisted of providing
verbally prompted to touch a card. He was Ty with a choice between 30 s of attention
blocked from leaving the table until he or 20 s of access to running. Attention was
touched a card. Otherwise, if Ty did not verbal and physical interaction (e.g., saying,
touch a card, no differential consequence oc- ‘‘Great walking next to me,’’ while giving
curred. him a ‘‘high five’’). Running consisted of the
therapist and Ty running side by side (the
Treatment Assessment therapist held Ty’s hand to prevent falling).
Treatment was assessed in two different The therapist did not otherwise interact with
functional analysis conditions, ignore and Ty while running. Ty made choices between
demand. A multielement analysis was used reinforcers by touching one of two index
to compare the ignore baseline condition cards that corresponded to attention (green
with differential reinforcement of appropri- card) or running (blue card). The cards were
ate behavior (DRA) plus blocking. Treat- attached to a board that was placed in front
ment for Ty’s elopement in the demand con- of Ty by the therapist when Ty reached cri-
dition was evaluated using an ABAB design. terion for reinforcement. If Ty touched a
The A phase was the demand baseline con- card, the corresponding reinforcer was deliv-
dition of the functional analysis, and the B ered. Attempts to touch both cards were
phase was DRA plus blocking. blocked. All attempts to elope were blocked.
All sessions were 10 min in duration and Blocking was used so that Ty would have
were conducted in the same area as the func- access to running only as a consequence for
tional analysis. Because elopement persisted appropriate walking rather than as a conse-
across all functional analysis conditions, two quence for elopement.
treatment assessments were conducted, one Following this treatment analysis, the
in the ignore condition and a second in the schedule of reinforcement was thinned. A
demand condition. These two conditions token economy (Kazdin & Bootzin, 1972)
were chosen because the planned treatment was introduced so that delivery of attention
(providing access to attention and running or running could occur at times and in set-
contingent upon appropriate walking) tings that were convenient to the mother.
would have been implemented similarly for The value of each token was equivalent to
the ignore, attention, and toy play sessions. 15 s of attention or 15 s of running. The
666 CATHLEEN C. PIAZZA et al.

schedule of reinforcement was thinned over sions conducted later in each session block.
six sessions, during which the time when to- Thus, it appeared that access to running was
kens could be exchanged was also extended. a reinforcer; however, its effects were greatly
The final arrangement consisted of token de- attenuated after repeated exposure to the re-
livery contingent upon 30 s of appropriate inforcer (i.e., satiation).
walking, with token exchange occurring af- The results of the reinforcer assessment of
ter 5 min. attention appear in the bottom left panel of
Treatment for Ty’s elopement in the sec- Figure 3. Ty touched the card associated
ond treatment assessment (demand condi- with adult attention more often than he
tion) was conducted in an ABAB design. touched the control card, suggesting that
The baseline condition was identical to the adult attention functioned as a reinforcer.
demand condition in the functional analysis The results of the simultaneous assessment
in which the therapist and Ty walked be- of running, attention, and control appear in
tween two work areas. Instructions were de- the bottom right panel of Figure 3. When
livered to complete a task in the work areas, all three cards were simultaneously available,
and compliance resulted in praise. Elope- Ty touched the attention card at the highest
ment resulted in a 30-s break from the task. rate, followed by the running card and the
The treatment condition consisted of differ- control card. These results suggested that at-
ential reinforcement and blocking for elope- tention served as a more potent reinforcer
ment. Reinforcement consisted of a choice than access to running.
between 30-s access to attention or 20-s ac- The results of the functional analysis and
cess to running, as described above, contin- reinforcer assessments were integrated into a
gent on 5 s of appropriate walking or staying treatment package designed to reduce Ty’s
in the task area. Compliance continued to elopement and increase his appropriate
result in verbal praise. Following treatment walking (Figure 4) during the first treatment
in the demand condition, the schedule of assessment. Baseline was associated with
reinforcement was thinned as described for high levels of elopement and low levels of
the ignore condition, and the treatment was appropriate walking. Treatment, consisting
extended to different settings in the hospital of differential reinforcement of appropriate
during the final four sessions. walking and blocking attempts to elope, re-
sulted in near-zero levels of elopement and
RESULTS AND DISCUSSION high levels of appropriate walking. It should
Ty consistently engaged in elopement in be noted that due to the blocking compo-
all assessment conditions, suggesting that his nent, Ty never successfully eloped during the
elopement was maintained either indepen- analysis. All of the responding during treat-
dent of the social environment (i.e., auto- ment represents attempts to elope. If Ty at-
matic reinforcement) or by multiple sources tempted to elope, the therapist blocked
of reinforcement. The middle panel of Fig- elopement, which resulted in Ty staying
ure 3 depicts the results for the reinforcer within 3 m of the therapist and thereby
assessment of running. Ty touched a card to maintaining high levels of appropriate walk-
gain access to running more frequently than ing. In addition, near-zero levels of elope-
he touched a control card. The high peaks ment and attempts to elope and high levels
in the data during the second and fourth of appropriate walking were maintained dur-
phases correspond to the first session con- ing the thinning of the reinforcement sched-
ducted in each session block, and the low ule.
levels of card touches correspond to the ses- Treatment for Ty’s elopement was also
 ELOPEMENT 667

Figure 3. Number of elopements per minute during the functional analysis (top panel) and number of
card touches per minute during reinforcer assessments for Ty (middle and bottom panels).

evaluated in a second treatment assessment also maintained during the thinning of the
(demand condition). The data for this anal- reinforcement schedule and extension of
ysis appear in the third and fourth panels of treatment to other settings.
Figure 4. During baseline (first and third
phases), Ty engaged in high levels of elope-
ment and a low percentage of appropriate GENERAL DISCUSSION
walking. During treatment (second and
fourth phases), in which attempts to elope Functional analysis methodology, origi-
were blocked and appropriate walking re- nally developed for the assessment of SIB,
ceived differential reinforcement, Ty engaged has been extended and modified to identify
in near-zero levels of elopement and at- reinforcers for a variety of destructive behav-
tempts to elope and high levels of appropri- iors. However, its extension to the behavior
ate walking. Near-zero levels of elopement of elopement, as was conducted in the cur-
and high levels of appropriate walking were rent investigation, posed some particular
668 CATHLEEN C. PIAZZA et al.

Figure 4. Number of elopements per minute (top panel) and percentage of session with appropriate walking
(second panel) during the treatment evaluated in the ignore condition, and the number of elopements per
minute (third panel) and percentage of session with appropriate walking (fourth panel) during the treatment
evaluated in the demand condition for Ty.

methodological problems that may have af- sion) cannot be easily regulated because they
fected interpretation of the results. create dangerous situations (e.g., getting hit by
First, some of the typical consequences of a car). Therefore, if the assessment of elope-
elopement (e.g., escape from adult supervi- ment is conducted in the natural environ-
 ELOPEMENT 669

ment, some preventive measures (e.g., retriev- Response differentiation was observed in
ing the child from an unsafe area) must be the functional analysis for Ray and Owen de-
applied when the behavior occurs, which may spite the existence of the retrieval contingency.
interfere with the internal validity of the as- Furthermore, treatments that were based on
sessment. If the behavior is assessed in an an- the results of functional analyses were effective
alogue environment (e.g., one or more in reducing elopement for these participants.
rooms), preventive consequences may not be The influence of the retrieval contingency is
necessary, but the external validity of the as- most problematic for Ty, whose functional
sessment may be compromised because the analysis was marked by undifferentiated re-
naturalistic consequences are unavailable. sponding. During the reinforcer assessments,
Second, assessment of elopement presents Ty responded for access to attention and run-
some challenges relative to measurement. Pre- ning (which also may have involved a social
vious investigations of elopement were con- component because Ty ran with a therapist).
ducted in group homes or residential facilities, Both treatments that resulted in a reduction
where elopement was measured across the day in Ty’s elopement also involved social rein-
over extended periods of time (e.g., 1 year; forcement components. Therefore, the undif-
Chambers et al., 1980; Garner, 1991). There- ferentiated responding in the functional anal-
fore, even if the behavior occurred only once ysis that involved a retrieval contingency,
per day, the lengthy time interval (e.g., along with the behavior observed in the re-
months) resulted in a sufficient baseline from inforcer and treatment assessments, make the
which to evaluate treatment outcome. Partic- interpretation of Ty’s data and identification
ipants in the current investigation were inpa- of behavioral function difficult. It is possible
tients in a pediatric hospital in which oppor- that Ty engaged in elopement to produce ac-
tunities for elopement were minimal because cess to the reinforcers of attention and run-
of high patient-to-staff ratio and restricted ac- ning. However, it is also plausible that elope-
cess to or from particular locations. Therefore, ment was simply maintained by attention, and
opportunities for responding were created by high undifferentiated levels of elopement were
returning the participant to a designated area observed in the functional analysis because the
or position (Room A for Owen and Ray, next retrieval procedure functioned as adventitious
to the therapist for Ty) on a fixed-time sched- social reinforcement.
ule after elopement occurred throughout each A third consideration when functional anal-
session. This procedure allowed multiple epi- yses are modified is the extent to which the
sodes of elopement to occur within each ses- consequences in the analogue condition match
sion and also allowed participants to contact those in the natural environment. Caregivers
the prescribed consequences for elopement re- of participants in the current investigation re-
peatedly. The potential confounding variable ported that they always retrieved the partici-
(i.e., physical attention) introduced by retriev- pants when they ran away. By retrieving the
ing and returning participants to designated participants, we more closely approximated
areas or positions was controlled by imple- these natural contingencies. However, as men-
menting this contingency as a constant during tioned previously, the consequences of retrieval
all assessment sessions. In addition, we further could have served as reinforcement for elope-
attempted to control for the effects of physical ment in addition to or independent of the
attention by physically guiding the partici- other programmed consequences (attention,
pants to a different chair (Owen and Ray) or escape, access to tangible items).
next to the therapist (Ty) throughout all ses- Because of the methodological limitations
sions independent of elopement. of the functional analysis of elopement, we
670 CATHLEEN C. PIAZZA et al.

conducted reinforcer assessments to identify elopement, there is increased confidence in

reinforcers for 2 participants. Vollmer, Mar- the utility of the assessments.
cus, and LeBlanc (1994) showed that the re- A final limitation of the current investi-
sults of choice assessments could be used to gation was that the active components of
identify stimuli for use in enriched environ- multiple treatments were not isolated for
ment treatments that were effective in re- Ray and Ty. Ray’s treatment included access
ducing SIB when the results of functional to two reinforcers (attention and tangible
analyses were equivocal. Similarly, Grace, items) contingent on the nonoccurrence of
Thompson, and Fisher (1996) showed that elopement, and no access to reinforcement
the results of a reinforcer assessment could for elopement. Ty’s treatment consisted of
be used to treat SIB when a functional anal- access to two reinforcers (attention and run-
ysis was difficult to conduct. ning) and blocking elopement. In both
Reductions in elopement during treat- cases, the relative contributions of the rein-
ment suggested that the reinforcer assess- forcement and extinction or blocking com-
ments may have been useful in identifying ponents were not identified. Further, the
either the functional reinforcers for elope- mechanism by which blocking may have re-
ment or those that served as effective sub- duced elopement is unclear. It is possible
stitutes. That is, when reinforcement was that the blocking procedure functioned as
provided noncontingently (Owen), in the extinction (i.e., disrupted the response–re-
absence of elopement (Ray), or for an ap- inforcer relation), or the presentation of
propriate alternative behavior (Ty), elope- blocking following an elopement attempt
ment was reduced. Smith, Iwata, Vollmer, may have functioned as punishment for
and Zarcone (1993) suggested that treat- elopement (see Lerman & Iwata, 1996). The
ment outcome is one method of verifying consequence for elopement was the same as
behavioral function. They conducted func- the consequence for remaining in the appro-
tional analyses with 3 participants with SIB, priate room (i.e., access to string) for Owen,
and found that SIB was multiply controlled. so it appeared that providing access to the
Smith et al. then tested the validity of the functional reinforcer was effective in reduc-
functional analysis results by implementing ing his elopement. Future research should
treatments that either matched an identified incorporate component analyses (Cooper et
behavioral function (e.g., a noncontingent al., 1995; Wacker et al., 1990) to identify
toy treatment when SIB was hypothesized to the active variables that were responsible for
be maintained by automatic reinforcement) treatment effectiveness.
or did not match an identified behavioral In the current investigation, we attempted
function (e.g., a noncontingent toy treat- to address issues of both internal and exter-
ment when SIB was hypothesized to be nal validity. Sessions were conducted in rel-
maintained by attention). If the matched atively controlled environments (two adja-
treatment resulted in reductions in SIB, cent rooms for Owen and Ray and an open
Smith et al. concluded that the identified common area for Ty) in which the partici-
function was confirmed. However, if the pants could elope without concerns for their
matched treatment did not result in reduc- safety. Experimental control was enhanced
tions in SIB, Smith et al. concluded that the further by prescribing the consequences for
identified function was a spurious finding. elopement (i.e., attention, access to tangible
The same logic could be applied to the pres- items, escape from demands) as opposed to
ent results. That is, because the assessment- allowing consequences to occur in an un-
derived treatments were effective in reducing controlled manner. We attempted to increase
 ELOPEMENT 671

the external validity of the study by having (1996). The treatment of covert self-injury
through contingencies on response products. Jour-
parents and teachers conduct the treatment nal of Applied Behavior Analysis, 29, 239–242.
in different settings (school, home, cafete- Herrnstein, R. J. (1970). On the law of effect. Journal
rias, hospital lobby), and we demonstrated of the Experimental Analysis of Behavior, 13, 243–
that treatment was successful in maintaining 266.
Horner, R. H., & Day, M. D. (1991). The effects of
low levels of elopement when it was extend- response efficiency on functionally equivalent
ed to these more naturalistic conditions. competing behaviors. Journal of Applied Behavior
Analysis, 24, 719–732.
Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K.
REFERENCES E., & Richman, G. S. (1994). Toward a func-
tional analysis of self-injury. Journal of Applied Be-
Bodfish, J. W. (1992). AWOL behavior. In E. A. havior Analysis, 27, 197–209. (Reprinted from
Konarski, J. E. Favell, & J. E. Favell (Eds.), Man- Analysis and Intervention in Developmental Dis-
ual for the assessment and treatment of the behavior abilities, 2, 3–20, 1982)
disorders of people with mental retardation (Tab Iwata, B. A., Duncan, B. A., Zarcone, J. R., Lerman,
BD17, pp. 1–8). Morganton, NC: Western Car- D. C., & Shore, B. A. (1994). A sequential, test-
olina Center Foundation. control methodology for conducting functional
Carr, T. E., & Newsom, C. (1985). Demand-related analyses of self-injurious behavior. Behavior Mod-
tantrums: Conceptualization and treatment. Be- ification, 18, 289–306.
havior Modification, 9, 403–426. Jacobson, J. W. (1982). Problem behavior and psy-
Catania, A. C. (1963). Concurrent performances: A chiatric impairment within a developmentally dis-
baseline for the study of reinforcement magni- abled population I: Behavior frequency. Applied
tude. Journal of the Experimental Analysis of Be- Research in Mental Retardation, 3, 121–139.
havior, 6, 299–300. Kazdin, A. E., & Bootzin, R. R. (1972). The token
Catania, A. C. (1992). Learning. Englewood Cliffs, economy: An evaluative review. Journal of Applied
NJ: Prentice Hall.
Behavior Analysis, 5, 343–372.
Chambers, J. H., Sanok, R. L., & Striefel, S. (1980).
Lerman, D. C., & Iwata, B. A. (1996). A method-
Using contingent decreased freedom-of-move-
ology for distinguishing between extinction and
ment to eliminate classroom running away: A case
punishment effects associated with response
study. Education and Treatment of Children, 3,
blocking. Journal of Applied Behavior Analysis, 29,
123–132.
231–233.
Cooper, L. J., Wacker, D. P., McComas, J. J., Brown,
K., Peck, S. M., Richman, D., Drew, J., Fris- Mace, F. C., & Lalli, J. S. (1991). Linking descriptive
chmeyer, P., & Millard, T. (1995). Use of com- and experimental analyses in the treatment of bi-
ponent analyses to identify active variables in zarre speech. Journal of Applied Behavior Analysis,
treatment packages for children with feeding dis- 24, 553–562.
orders. Journal of Applied Behavior Analysis, 28, Miller, H. L. (1976). Matching-based hedonic scaling
139–153. in the pigeon. Journal of the Experimental Analysis
Fisher, W. W., Piazza, C. C., Bowman, L. G., Kurtz, of Behavior, 26, 335–347.
P. F., Sherer, M. R., & Lachman, S. R. (1994). Piazza, C. C., Fisher, W. W., Hanley, G. P., Remick,
A preliminary evaluation of empirically derived M. A., Contrucci, S. A., & Aitken, T. (1997).
consequences for the treatment of pica. Journal of The use of positive and negative reinforcement in
Applied Behavior Analysis, 27, 447–457. the treatment of escape-maintained destructive be-
Fisher, W., Piazza, C., Cataldo, M., Harrell, R., Jef- havior. Journal of Applied Behavior Analysis, 30,
ferson, G., & Conner, R. (1993). Functional 279–297.
communication training with and without extinc- Piazza, C. C., Hanley, G. P., & Fisher, W. W. (1996).
tion and punishment. Journal of Applied Behavior Functional analysis and treatment of cigarette
Analysis, 26, 23–36. pica. Journal of Applied Behavior Analysis, 29,
Fisher, W., Piazza, C. C., & Page, T. J. (1989). As- 437–450.
sessing independent and interactive effects of be- Smith, R. G., Iwata, B. A., Vollmer, T. R., & Zarcone,
havioral and pharmacologic interventions for a cli- J. R. (1993). Experimental analysis and treatment
ent with dual diagnoses. Journal of Behavior Ther- of multiply controlled self-injury. Journal of Ap-
apy and Experimental Psychiatry, 20, 241–250. plied Behavior Analysis, 26, 183–196.
Garner, M. M. (1991). The treatment of elopement Steege, M. W., Wacker, D. P., Berg, W. K., Cigrand,
behavior in a retardate using a graduated levels K. K., & Cooper, L. J. (1989). The use of be-
program. Journal of Behavior Therapy and Experi- havioral assessment to prescribe and evaluate treat-
mental Psychiatry, 21, 277–280. ments for severely handicapped children. Journal
Grace, N. C., Thompson, R., & Fisher, W. W. of Applied Behavior Analysis, 22, 23–33.
672 CATHLEEN C. PIAZZA et al.

Vollmer, T. R. (1994). The concept of automatic re- Roane, H. S. (1995). Progressing from brief as-
inforcement: Implications for behavioral research sessments to extended experimental analyses in the
in developmental disabilities. Research in Devel- evaluation of aberrant behavior. Journal of Applied
opmental Disabilities, 15, 187–207. Behavior Analysis, 28, 561–576.
Vollmer, T. R., Iwata, B. A., Zarcone, J. R., Smith, R. Wacker, D. P., Steege, M. W., Northup, J., Sasso, G.,
G., & Mazaleski, J. L. (1993). The role of atten- Berg, W., Reimers, T., Cooper, L., Cigrand, K.,
tion-maintained self-injurious behavior: Noncon- & Donn, L. (1990). A component analysis of
tingent reinforcement and differential reinforce- functional communication training across three
ment of other behavior. Journal of Applied Behav- topographies of severe behavior problems. Journal
ior Analysis, 26, 9–21. of Applied Behavior Analysis, 23, 417–429.
Vollmer, T. R., Marcus, B. A., & LeBlanc, L. (1994).
Treatment of self-injury and hand mouthing fol- Received April 15, 1997
lowing inconclusive functional analyses. Journal of Initial editorial decision June 30, 1997
Applied Behavior Analysis, 27, 331–344. Final acceptance July 24, 1997
Vollmer, T. R., Marcus, B. A., Ringdahl, J. E., & Action Editor, Brian A. Iwata

STUDY QUESTIONS

1. What is elopement, why is it considered dangerous, and how was it operationally defined
in this study?

2. What was the uniform consequence for elopement during all functional analysis conditions,
and what were the rationales for using this consequence?

3. What data from the functional analysis suggested that food or attention maintained Ray’s
elopement, and why did the authors conduct a subsequent pairwise comparison?

4. Describe the key features of Ray’s treatment.

5. It has been suggested that noncontingent reinforcement (NCR) may produce behavioral
suppression through satiation or extinction. What feature of Owen’s treatment and data
partially addresses this issue?

6. Given the results of Ty’s additional assessments, how might the consequence for elopement
during his initial functional analysis have affected its outcome?

7. What feature of Ty’s treatment presents interpretive problems?

8. The authors developed a variety of effective interventions (NCR, DRO, DRA) for elopement
that was maintained by positive reinforcement. Suggest some strategies for reducing the
frequency of elopement that is maintained by negative reinforcement (escape).

Questions prepared by Eileen Roscoe and Michele Wallace, The University of Florida