Sie sind auf Seite 1von 17

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2019, 52, 422–438 NUMBER 2 (SPRING)

Response latency as a measure of behavior in the assessment of


elopement
MICHELE R. TRAUB AND TIMOTHY R. VOLLMER
UNIVERSITY OF FLORIDA

Elopement is a common form of problem behavior but is relatively underrepresented in the


functional analysis literature. One barrier to assessing elopement experimentally is the need to
retrieve the subject following an instance of elopement. This retrieval confounds programmed
session contingencies when the goal is to obtain repeated measurement of free-operant behavior.
The current study evaluated latency to elopement as an alternative to free-operant measurement.
We first compared response latency to allocation in 5-min sessions and then measured latency
alone in a trial-based format. The identified reinforcers matched across both data analysis modal-
ities in the session-based assessments, and the trial-based functional analysis showed a significant
time savings in identifying the function of behavior over a session-based assessment. Results
indicated that elopement serves idiosyncratic social functions in young children with intellectual
and developmental disabilities, and that a latency-based assessment saves time while yielding
equally clear results.
Key words: elopement, functional analysis, latency

Elopement is commonly defined as leaving a factor to higher death rates among individuals
designated area without supervision or permis- with autism than in the overall population
sion (Jacobson, 1982), and is a common behav- (Mouridsen, Bronnum-Hansen, Rich, & Isager,
ioral problem among individuals with 2008; Shavelle, Strauss, & Pickett, 2001).
intellectual and developmental disabilities Despite the prevalence and risks of elope-
(IDD). Recent reports indicate that 50% or ment among individuals with IDD, few studies
more of children with autism have eloped a sig- involving functional analysis (FA), treatment,
nificant distance or for an extended period of or systems management of this behavior have
time (Anderson et al., 2012; Jang, Dixon, Tar- been published. In a review of elopement, Lang
box, & Granpeesheh, 2011), and about 33% et al. (2009) identified only 10 peer-reviewed
of adults with autism have eloped from residen- studies published in behavior-analytic journals,
tial or day treatment care (Matson & Rivet, only five of which were experimentally con-
2008). Elopement, and specifically the trolled. Furthermore, 6 of the 10 studies had
increased risk for accident and injury while only one participant, and only one had more
unsupervised, has been cited as a contributing than three participants, thus limiting the num-
ber of total published assessments. One reason
This research was completed in partial fulfillment of for this lack of publication may be the practical
the first author’s requirements for the doctoral degree in
Psychology at the University of Florida. Special thanks to
difficulties of conducting FAs of elopement.
Brian Iwata, Darragh Devine, and Jennifer Elder for their A typical FA, based upon the procedures
input on this manuscript, and to Catherine Simms and described by Iwata, Dorsey, Slifer, Bauman,
Kara Wunderlich for their contributions to the early stages
of the research.
and Richman (1982/1994), involves delivering
Address correspondence to: Michele R. Traub, Depart- social consequences such as attention or escape
ment of Community Psychology, Counseling, and Family from demands contingent upon the occurrence
Therapy, St. Cloud State University, 720 4th Avenue South,
St. Cloud, MN 56301 (Email: mtraub@stcloudstate.edu) of a problem behavior during individual test
doi: 10.1002/jaba.541 sessions, and withholding those same
© 2019 Society for the Experimental Analysis of Behavior
422
RESPONSE LATENCY IN ELOPEMENT 423

consequences during other test sessions. For differentiated results for one participant; addi-
instance, in a test for escape from academic tional analyses were required for two other par-
demands as a reinforcer for self-injury, a thera- ticipants. Second, although this assessment
pist would deliver academic demands and programmed noncontingent “retrieval” across
remove those demands following the occur- all experimental conditions, it was impossible
rence of self-injury. Attention to the problem to retrieve the participant following an instance
behavior is specifically withheld during these of elopement without providing attention con-
sessions to avoid confounding task removal and tingent on, and directly related to, the target
attention delivery. At the end of the reinforce- behavior. The authors also noted that not
ment interval, the therapist would then rein- retrieving the individual presented a barrier to
state demands until the next instance of self- assessment; specifically, only one opportunity
injury. However, in the case of elopement, if to observe elopement would be present in each
the participant eloped from the experimental session.
area to initiate an escape interval, the therapist Lehardy, Lerman, Evans, O’Connor, and
would have to retrieve the participant to return LeSage (2013) described an alternative FA of
him or her to the experimental area in order to elopement that addressed concerns raised by
resume the analysis. Retrieval requires the deliv- Piazza et al. (1997). Specifically, this assessment
ery of physical interaction, which would con- model was conducted in a single room sepa-
found the delivery of escape with attention, rated in half by tape across the floor. The thera-
making the results of the assessment difficult to pist presented the establishing operation (EO)
interpret when trying to identify an isolated on one side of the room and delivered the rein-
reinforcement contingency. forcer when the participant eloped by crossing
Piazza et al. (1997) partially addressed these to the other side of the room. Rather than
concerns with the “standard” model of FA with retrieve the participant, the therapist simply
three individuals who engaged in frequent presented the EO on the side of the room
elopement. These experimenters conducted where the participant was at the end of a rein-
assessment sessions in two adjoining session forcement interval (i.e., relocating contingencies
rooms. The participant experienced a relevant instead of retrieving the participant). The
evocative event in one room (e.g., demand pre- results of this assessment corresponded to those
sentation) and accessed a putative reinforcer of assessments conducted across two rooms
(e.g., demand termination) contingent upon (similar to the procedures used by Piazza et al.,
eloping to the other room. To eliminate the 1997), and resulted in effective function-based
confounds of retrieval, the therapist noncontin- interventions in each of four applications. This
gently “retrieved” participants on a fixed time model would be more replicable in settings
schedule in both rooms. That is, every 40 s the without a second therapy room, and was
experimenters guided the participant to sit in a reported as easier to implement without requir-
chair in the initial room. A multielement com- ing timed retrieval intervals.
parison of tests for positive and negative social One promising advance in FA methodology
reinforcement resulted in differentiated out- that may further inform the development of
comes with one participant, and subsequent elopement assessments involves the measure-
pairwise (single test–control) comparisons ment of response latency (a measure of time
yielded differentiated outcomes for the other from some event until the first instance of a
two participants. Although promising, this behavior) as a primary dependent measure in
assessment model was limited in a few regards. lieu of response rate. For problem behaviors
First, the multielement assessment only yielded that cannot occur in repeated bursts
424 MICHELE R. TRAUB and TIMOTHY R. VOLLMER

(e.g., vomiting) or that require a therapist to The purpose of the current study was two-
reset for the next opportunity to occur fold. Thomason-Sassi et al. (2011) compared
(e.g., disrobing), latency measurement may pro- latency-based analysis to a standard repeated-
vide a practical alternative to response rate. In a measures analysis in published data sets, none
retrospective analysis of existing FA data, of which included elopement as a target behav-
Thomason-Sassi, Iwata, Neidert, and Roscoe ior. Therefore, we first evaluated latency-based
(2011) found that latency and response rate measures of elopement by assessing their corre-
indicated the same behavioral function in spondence with more traditional continuous
33 out of 38 cases. These authors then con- measures of behavior, in this case allocation to
ducted FAs with 10 new subjects, initially ter- opposing contingencies. Second, we evaluated
minating sessions after the first instance of the utility of latency measures in lieu of
problem behavior (a latency analysis) and then repeated occasioning of problem behavior dur-
following with 10-min sessions in which the ing FAs of elopement. Unlike the analyses con-
problem behavior was occasioned repeatedly ducted by Davis et al. (2013) and Neidert et al.
(a response-repetition or standard analysis). (2013), we informed our assessment environ-
The results of the latency-based analysis corre- ment by the strategies offered by Piazza et al.
sponded with those of the standard analysis in (1997) and Lehardy et al. (2013). Specifically,
9 of 10 instances, while occasioning fewer we conducted our assessment in two physically
instances of problem behavior. separated areas (either two joined rooms or a
Neidert, Iwata, Dempsey, and Thomason- single room with a divider). Elopement
Sassi (2013) were the first to conduct latency- occurred when a participant traveled from one
based FAs of elopement with two students with area of the room to the other, thus allowing us
profound intellectual disabilities. In this assess- to maintain control over the contingencies
ment, evocative events were presented to the delivered for elopement. Further, rather than
participant in an empty classroom. The student adopt multielement experimental designs, we
eloped by exiting the doorway of his classroom, incorporated test–control comparisons of each
and the assessment included tests for attention, potential reinforcer.
tangibles, and escape from demands, alternated
with a control session in a multielement design.
METHOD
The results of these analyses were undifferen-
tiated across both participants but were clear Subjects, Setting, and Materials
after the authors compared conditions in a Nine children participated in the study,
reversal design or a pairwise design. The pro- ranging in age from 3 years, 2 months to
grammed contingencies in this study may have 6 years, 2 months. All participants had a diag-
been confounded by the consequences the par- nosis of IDD and were referred to the study by
ticipants experienced when leaving the room parents or teachers due to concerns about
(e.g., peer attention), and still required partici- elopement. Sessions took place at the partici-
pant retrieval by the experimenters. Davis et al. pants’ school or clinic, and the specific arrange-
(2013) also conducted a trial-based FA of ment of the room depended on the available
elopement with an 8-year-old boy. As in Nei- space at the facility. For Kieran, Tanya, Mikey,
dert et al., uncontrolled reinforcers in the post- Ivan, Jasper, and Violet, sessions were con-
elopement environment may have confounded ducted in a 2.1 × 2.7-m therapy room adjoin-
assessment results. Thus, the utility of latency- ing an empty classroom. The therapy room
based FAs in the assessment and treatment of contained a single door leading to the class-
elopement has not yet been established. room and a window for viewing. A camping
RESPONSE LATENCY IN ELOPEMENT 425

tent measuring 2.4 × 2.4 × 1.5 m was placed will refer to as location A and location B. One
immediately outside the therapy room to of those areas was associated with the free avail-
approximate the two-room setting of Piazza ability of a social reinforcer (noncontingent
et al. (1997). The tent had two opposing zip- reinforcement, or NCR, condition); the other
pered entrances, one of which was left perma- area was associated with extinction (EXT) or
nently open with flaps pulled back and was the nonavailability of that reinforcer. Test ses-
placed in the open doorway of the room. The sions were those in which the participant began
zippered entrance on the opposite side of the the session in the EXT area (which was
tent was kept closed during sessions, but the designed to establish the value of the putative
zipper was freely accessible to the participants reinforcer) and could gain access to reinforce-
and therapists. The tent had a fabric roof and ment by eloping into the NCR area. Control
sides with clear plastic or mesh windows. Dur- sessions were those in which the participant
ing experimental sessions, data collectors and began in the NCR area; eloping to the extinc-
observers stood along the three sides of the tent tion area would result in the termination of
not adjacent to the room to prevent partici- reinforcement. Test and control sessions were
pants from eloping beyond the confines of the alternated in a multielement design, and we
tent, though this rarely occurred. For Ned and sequentially evaluated attention, tangible, and
Ace, sessions were conducted in two 2.4 × escape from demands across all participants
3.6-m rooms that were joined by a 0.6-m-wide (the order of exposure was counterbalanced
doorway. There was a single door in one of the across participants). We also included a tangible
rooms that exited to the hallway; this door was condition without attention for Violet when we
kept closed during sessions. Both rooms were suspected that therapist attention was oversha-
empty of all furniture and contained only the dowing any functional value of tangibles, and
materials needed for sessions (i.e., toys during we included an escape-to-tangible condition—
tangible sessions). For Liam, sessions were con- similar to the escape-to-attention condition
ducted in a single 3 × 6-m room with two win- described by Mueller, Sterling-Turner, and
dows, two doors that were kept closed during Moore (2005)—with Jasper when the individ-
sessions, and a partial-wall divider. The divider ual tangible and escape assessments showed
split the room into equal 3 × 3-m halves with weak functional control over his elopement.
a 0.6-m-wide opening in the divider to create a The role of the therapist (i.e., which location
doorway. The divider was the only furniture any particular therapist was associated with)
present in the room. and the location of NCR and EXT areas were
Data collection took place using handheld randomized across sessions.
computers or smart phones, and some sessions Attention test. In the EXT area, the therapist
were videotaped using a handheld video camera withheld attention from the participant. In the
to facilitate data collection, procedural integrity NCR area, a second therapist provided contin-
checks, and interobserver agreement (IOA). uous attention in the form of general com-
Additional materials specific to session types ments, reciprocal conversation, light physical
were present as needed and are described contact (e.g., bounces, tickles), or some combi-
below. nation. The specific form of attention provided
to each participant was based on caregiver
report and observation. No leisure items were
Experimental Design present in either area.
Similar to Lehardy et al. (2013), we divided Tangible test. Prior to each session, the par-
our assessment space into two areas, which we ticipant was brought to the toy closet and asked
426 MICHELE R. TRAUB and TIMOTHY R. VOLLMER

to choose a toy to play with. The therapist in independently, which was determined by obser-
the NCR area provided continuous access to vation and teacher reports. Second, demands
the selected toy as well as social interaction. could not involve any items. Thus, all demands
The therapist in the EXT area provided contin- targeted verbal and motor skills, such as “touch
uous attention but no access to the selected your nose” or “give me a high-five.”
toy. We delivered attention in both areas to Escape-to-tangibles test (Jasper only). In the
prevent confounding tangible delivery with EXT area, the therapist delivered demands
attention, and we varied the therapists used in identical to the escape conditions described
each location to minimize the effect of a pre- above. In the NCR area, the therapist provided
ferred (or nonpreferred) therapist in any one continuous attention and interactive play with
session. If the subject attempted to carry a toy the toys. The attention delivered by the NCR
from the NCR area to the EXT area, the thera- therapist was matched in magnitude and voice
pist from the EXT area removed the toy with- tone to the demand therapist in an attempt to
out comment and returned it to the NCR area. equate the value of the attention in both envi-
Tangible test with no attention (Violet only). ronments. All other session aspects were identi-
These sessions were identical to the tangible cal to escape sessions.
sessions described above, with the exception Automatic test (Liam only). For test sessions,
that no attention was provided by therapists in both locations constituted an EXT environ-
either area. Because attention was found to be a ment; no toys were available and both thera-
potent reinforcer for Violet, we suspected that pists ignored Liam.
the attention available in both locations during For control sessions, both locations consti-
the tangible assessment was overshadowing any tuted an NCR environment in which a similar
possible preference for tangible items; thus, we toy was freely available and therapists provided
removed the attention in both locations to continuous attention. Elopement from one
examine her preference for toys in the absence location to the other resulted in access to the
of a more potent reinforcer, namely, attention. identical contingency (implemented by the
Escape test. In the EXT area, the therapist other therapist).
presented continuous demands using a gradu-
ated, three-step (verbal-model-physical)
prompting sequence. That is, the therapist Procedure and Response Measurement
delivered an initial verbal demand Session-based assessment. Kieran, Tanya,
(e.g., “Touch your nose”). If the participant Mikey, Ivan, Jasper, and Violet participated in
did not respond within 3 s of the initial verbal session-based assessments. In this assessment,
demand, the therapist modeled the response all sessions were 5 min and could include mul-
while repeating the demand (e.g., “Touch your tiple instances of elopement. Prior to the first
nose, like this”). If the participant did not session of the day, participants received 15-s
respond within 3 s of the model prompt, the exposure to the contingencies in both locations,
therapist physically prompted compliance with after which they were brought to the starting
the demand. In the NCR area, the therapist location for the session. The participant could
did not deliver demands, nor otherwise interact move freely between the two locations for the
with the participant. No toys or tangible items duration of each session; therapists delivered or
were available in either location. withheld reinforcement according to where the
We chose demands for each participant participant was located.
based on two criteria. First, the child had to be Two primary dependent measures were
capable of exhibiting the instructed behavior scored for session-based assessments. Allocation
RESPONSE LATENCY IN ELOPEMENT 427

by location (A vs. B) was scored using continu- considered measures to “not match” when only
ous duration measures. A change in allocation one measure identified a reinforcement effect
was scored at the moment any part of the par- within an assessment.
ticipant’s foot crossed the threshold (a natural Trial-based assessment. Ned, Ace, and Liam
break in flooring visible to participants, thera- participated in trial-based assessments. As in
pists, and data collectors), regardless of how the session-based assessments, participants
much of the participant’s body actually crossed received 15-s exposure to the contingencies in
into the other area. No participant ever both locations prior to the first session each
remained on the threshold; each instance of day, and were then brought to the starting loca-
crossing from one area to the other involved tion for the first session. Unlike the session-
complete movement of the body across the based assessments, though, only one instance of
threshold. We then translated these data as elopement could occur in the trial-based assess-
being on the “test” area or “control” area based ment. After an instance of elopement, partici-
upon the arranged contingencies in each ses- pants remained in the new area for 15 s and
sion. We also scored elopement, which we sessions were then terminated. Contingencies
defined as the moment in which the data col- were then rearranged across the two locations,
lector scored a change of allocation (i.e., the the participant was brought to the new starting
moment the participant crossed from location location, and a new trial began. Scoring during
A to location B or the reverse.) These data were sessions was identical to that of the session-
then reported as a latency in seconds from the based assessment; however, only data on the
start of the session to the first instance of latency to elopement were reported and
elopement. analyzed.
Decisions regarding stability and function
were made using visual analysis of the latency
data alone. When a determination was made Interobserver Agreement
for a pairwise comparison, either because A second, independent observer scored
responding in test and control conditions was responding on an average of 44% of sessions
sufficiently differentiated to show a func- (range, 26% to 54% across participants).
tional relation, or because responding was Interobserver agreement was then calculated
undifferentiated enough to indicate the lack for allocation (session-based only) and latency
of a functional relation, the condition was (both session- and trial-based) measures.
terminated and a new pairwise comparison Allocation IOA was calculated by determin-
began. ing the percentage of the session scored in
Each comparison was interpreted as either the test condition by each observer, dividing
demonstrating a reinforcement effect according the smaller value by the larger value, and
to both the allocation measure (consistently ele- converting the quotient into a percentage.
vated response allocation to the NCR area rela- Allocation IOA was 95.9% (range, 93.2% to
tive to the EXT area) and the latency measure 97.5%) for all participants. Latency IOA was
(consistently shorter latencies in test relative to calculated by comparing both observers’ time
control conditions), or not demonstrating a stamps of the first elopement of the session,
reinforcement effect. We considered the mea- dividing the shorter measure by the longer
sures to “match” when both allocation and measure, and converting the quotient into a
latency identified a reinforcement effect within percentage agreement. Mean IOA for latency
an assessment or both identified no reinforce- was 96.5% (range, 90.5% to 98.3%) across
ment effect within the same assessment. We participants.
428 MICHELE R. TRAUB and TIMOTHY R. VOLLMER

RESULTS break location (during the escape assessment)


Figure 1 shows Kieran’s data. Kieran’s and the toy location (during the tangible assess-
latency to elopement during attention sessions ment). As no similar pattern of allocation was
observed in the attention assessment, we con-
was variable across both test and control condi-
cluded that the two assessment modalities
tions, whereas in both escape and tangible ses-
matched in all three cases and that Kieran’s
sions he eloped quickly in the test condition elopement was maintained by escape from tasks
and, after some variability in initial sessions, and access to tangibles.
rarely eloped at all in control sessions. His allo- Figure 2 shows Tanya’s data. Tanya’s latency
cation data reflected a similar pattern, with to elopement and session allocation was vari-
consistent allocation of session time to the able throughout all phases of the assessment;

Figure 1. Results of Kieran’s assessment. Latency to elopement is in the left column, and percent of session allocated
to each location is in the right column. Data points above the dotted line represent sessions in which no elopement
occurred.
RESPONSE LATENCY IN ELOPEMENT 429

Figure 2. Results of Tanya’s assessment.

however, some differentiation was seen toward and control sessions in Mikey’s escape (middle
the end of her tangible and in her escape assess- panel) or attention (bottom panel) assessments.
ments, with consistently low latencies to elope- These results were confirmed by his allocation
ment during test sessions and overall higher data, which showed no consistent allocation of
latencies during control sessions. Her allocation session time to either location. In Mikey’s tan-
of session time also showed a preference for the gible assessment, however, some differentiation
toy and the break, respectively, suggesting that was seen in the final 12 sessions, with variable
her elopement was maintained by escape from latencies to elopement in test sessions but no
tasks and access to tangibles. elopement in control sessions. This suggested
Mikey’s data are shown in Figure 3. No dif- that access to tangibles may have been a weak
ferentiation in latency was seen between test or partial reinforcer for Mikey’s elopement, at
430 MICHELE R. TRAUB and TIMOTHY R. VOLLMER

Figure 3. Results of Mikey’s assessment.

least to the extent that having access to a tangi- elopement, which was borne out by his consis-
ble item prevented elopement. A similar weak tent allocation of session time to the break loca-
differentiation was seen in his allocation data. tion. His tangible assessment (bottom panel of
Figure 4 shows Ivan’s data. In Ivan’s escape Figure 2) was even more clearly differentiated,
assessment he eloped quickly in all test sessions, whereas no differentiation was seen in his
whereas latency to elopement was more variable attention assessment. We concluded that all
in control sessions. This indicated that escape latency and allocation results matched for Ivan
from tasks was a reinforcer for Ivan’s and that his elopement was maintained
RESPONSE LATENCY IN ELOPEMENT 431

Figure 4. Results of Ivan’s assessment.

primarily by access to tangibles and secondarily differentiated as the assessment continued. His
by escape from tasks. tangible assessment (third panel) also showed
Jasper’s data are shown in Figure 5. Laten- some differentiation, though he did not elope
cies to elopement during escape sessions (top at all during two tangible test sessions (5 &
panel) were initially differentiated, but latency 15), and his allocation was variable despite
to elopement decreased in later control ses- favoring the toy location. Based on these
sions; this was consistent with Jasper’s alloca- results, we tested a hypothesis that Jasper’s
tion to the demand and break locations, which elopement was reinforced by escape from
initially favored the break but became less demands only when that escape also produced
432 MICHELE R. TRAUB and TIMOTHY R. VOLLMER

Escape
100
300 Jasper
80
200
60

40
100 Test
Control 20 Demand
0 Break
0
5 10 15 20
5 10 15 20

Attention
100
Latency to Elopement (s)

300

Percent of Session
80

200 60

40
100 Test
20 EXT
Control
0 NCR
0

5 10 15 20 5 10 15 20

Tangible 100
300
80
Test
200 Control 60
No Toy
40 Toy
100
20

0 0

5 10 15 20 5 10 15 20

Escape to Tangible
100
300
80

200 60

40
100
Test 20 Demand
Control Toy
0 0

5 10 15 20 5 10 15 20
Sessions Sessions

Figure 5. Results of Jasper’s assessment.

access to a more preferred (i.e., tangible- elopement was indeed reinforced by escaping
enriched) environment. Clear differentiation demands and accessing tangibles.
was seen in both latency to elopement and ses- Figure 6 shows Violet’s data. Clear differen-
sion allocation during the final escape-to- tiation was seen in Violet’s attention assessment
tangibles condition, suggesting that Jasper’s (middle panel), with low latencies to elopement
RESPONSE LATENCY IN ELOPEMENT 433

Figure 6. Results of Violet’s assessment.

in all test sessions and high latencies (when concurrent availability of attention and tangi-
elopement occurred) in control sessions. Vio- bles overshadowed any clear pattern of
let’s escape assessment showed lower latencies responding for tangible access. We therefore
to elopement in control sessions, further sug- concluded that Violet’s elopement was rein-
gesting that attention was a reinforcer for Vio- forced by access to attention.
let (as attention was only available in the Figure 7 shows results from the trial-based
demand location). A possible tangible function assessments for Ned, Ace, and Liam. Ned (left
was also seen when tangibles were tested in column) eloped with similar latencies across
the absence of attention; however, the test and control sessions of the escape
434 MICHELE R. TRAUB and TIMOTHY R. VOLLMER

Ned - Escape Ace - Attention Liam - Tangible


Latency to Elopement (s)

300 300 300

200 Test 200 200


Control
100 100 100

0 0 0

5 10 15 20 25 5 10 15 20 25 5 10 15

Attention
Escape Attention
300 300 300

200 200 200

100 100 100

0 0 0

5 10 15 20 25 5 10 15 20 25 5 10 15
Sessions
Tangible Escape
300 300

200 200

100 100

0 0

5 10 15 20 25 5 10 15
Sessions
Automatic
300

200

100

0
5 10 15
Sessions

Figure 7. Results of trial-based functional analyses for Ned (left), Ace (center), and Liam (right). Data points above
the dotted line represent sessions in which no elopement occurred.

assessment (suggesting escape was not a rein- tangible assessment (suggesting sensitivity to
forcer), and with shorter latencies in the test tangible reinforcement). For Liam (right col-
condition than the control condition of the umn), we included tests for tangibles, attention,
attention assessment (suggesting attention was escape, and automatic reinforcement. Liam dis-
a reinforcer). For Ace (middle column), we played short and undifferentiated latencies to
included separate assessments for attention, elopement across test and control sessions of all
escape, and tangible items. His elopement conditions. Based on this lack of differentia-
latencies were undifferentiated during attention tion, we tentatively concluded that Liam’s
and escape analyses, but differentiated in the elopement was automatically reinforced.
RESPONSE LATENCY IN ELOPEMENT 435

DISCUSSION Third, the latency-based assessment identi-


We evaluated latency measures as a primary fied sources of reinforcement that corresponded
dependent variable during the FA of elopement with measures of response allocation (see
conducted with nine children with IDD. We Table 1). Traditional psychometricians refer to
first compared latency measures to a more tra- this correspondence as an indicator of conver-
ditional measure of reinforcer strength (alloca- gent validity (i.e., the extent to which a novel
tion). We then conducted three additional FAs assessment corresponds with a more established
in a trial-based format. We were able to iden- assessment; Haynes, 2001).
tify a probable source of reinforcement for each Fourth, the assessment differentiated
participant’s elopement using latency measures between multiple sources of reinforcement,
and without needing to retrieve the participant referred to as discriminative validity (Haynes,
for safety or repeated responding. We were also 2001). That is, if every assessment indicated
able to evaluate the validity of using latency to that elopement was maintained by tangible
responding as a measure of the function of reinforcement, it would be unclear if tangible
behavior, in this case elopement. items were a universal reinforcer for elopement
There are several important considerations in (in which case assessment would be unneces-
evaluating the utility of latency meaures when sary) or if the assessment was incapable of
functionally analyzing elopement. First is the detecting other sources of reinforcement. Our
extent to which a new assessment is capable of assessment identified sensitivity to different
accurately identifying sources of reinforcement, sources of reinforcement across participants
referred to as the sensitivity of the assessment (tangibles for Kieran, Ivan, Tanya, and Mikey;
(Altman & Bland, 1994). That is, if participants escape for Kieran, Ivan, and Tanya; attention
rarely eloped during test conditions or eloped for Violet; and escape to tangibles for Jasper).
inconsistently across all test and control compar- Said another way, this assessment was capable
isons, then the evaluator would have difficulty of showing sensitivity to different reinforcers
identifying reinforcement effects and the assess- across participants.
ment would have limited utility. However, our The final evaluation of the utility of an
results indicated that the latency-based evalua- assessment procedure is the extent to which it
tion identified one or more specific sources of differentially predicts effective treatment,
reinforcement for eight of the nine participants referred to as outcome validity or treatment
and thus was sensitive to reinforcement. utility (Hayes, Nelson, & Jarrett, 1987;
A second important consideration is the
extent to which the assessment is capable of Table 1
ruling out sources of reinforcement, referred to Functions of Elopement Based on Latency and
as the specificity of the assessment (Altman & Allocation Data
Bland, 1994). If every test condition yielded
the same reinforcement effect, then the assess- Function based on Function based on
Name latency allocation
ment would have decreased value. One goal of
Kieran Tangibles & Escape Tangibles & Escape
an FA is to identify those reinforcement rela- Tanya Escape & Tangibles Escape & Tangibles
tions that must be addressed through behav- Mikey Tangibles Tangibles
ioral intervention. In our current data set, Ivan Tangibles & Escape Tangibles & Escape
Jasper Escape to tangibles Escape to tangibles
latency-based assessments demonstrated speci- Violet Attention Attention
ficity by showing insensitivity to one or more Ned Attention ---
Ace Tangible ---
sources of reinforcement for each participant. Liam Automatic ---
436 MICHELE R. TRAUB and TIMOTHY R. VOLLMER

Haynes, 2001). Future research should evaluate environment (as the new physical environment
outcome validity of latency-based FAs in a adopted the contingencies of the original one as
manner similar to Iwata, Pace, Cowdery, and soon as elopement occurred). In the current
Miltenberger (1994b), who showed the differ- study, the consequence of elopement was access
ential efficacy of treatments designed from anal- to a functionally distinct set of events, allowing
ysis results. for the direct comparison of behavior under
Our analysis procedures were similar to those both evocative and abolishing conditions.
described by Piazza et al. (1997) and Lehardy It is worth noting that five of the nine par-
et al. (2013), which included occasioning ticipants’ assessments indicated that elopement
elopement in a confined space to eliminate the was reinforced by access to tangible items. This
confounding interactions associated with partic- result is similar to that of Lehardy et al. (2013),
ipant retrieval. Piazza et al. arranged for elope- who reported that three of four participants’
ment to occur between two adjoining therapy elopement was reinforced by access to tangible
rooms, whereas Lehardy et al. arranged for items. Several authors have raised concern
elopement to occur across a visible room regarding the inclusion of tangible conditions
divider (i.e., tape on the floor). In our study, in functional analyses (Rooker, Iwata, Harper,
we arranged for elopement to occur between Fahmie, & Camp, 2011; Shirley, Iwata, &
two locations separated by a firm, physical Kahng, 1999), warning that the contingent
divider (i.e., the tent, walls, or room divider). delivery of high-preference food or leisure items
Our procedures also differed in that we had may yield novel problem behavior–reinforcer
two therapists conducting each session. relations during the FA, producing false-
Although this increases the staff and physical positive results. Descriptive studies of aggres-
resources necessary to complete this assessment, sion, self-injury, and property destruction indi-
we believe these changes may increase partici- cate that tangible items are rarely delivered
pants’ discrimination of the relevant contingen- when these problem behaviors are exhibited by
cies arranged in each location. Future research adults in a residential institution or by children
is needed to directly compare these procedures in preschool classrooms (McKerchar &
to single-therapist procedures, to determine if Thompson, 2004; Thompson & Iwata, 2001),
the benefits offset the added cost to complete and large scale FAs of problem behavior indi-
the assessment. cate that maintenance by tangible reinforce-
Our procedures also differed from Lehardy ment is relatively uncommon relative to other
et al. (2013) in that we did not alter the con- social and sensory reinforcers (Iwata et al.,
tingencies associated with each location during 1994). Thus, it is possible that the frequent
a session, allowing for the observation of behav- identification of tangible items as a reinforcer
ior in the presence of both a putative EO for elopement is a false positive, but it may also
(in the EXT condition) and a putative abolish- suggest that access to tangibles is a more com-
ing operation (AO) (in the NCR condition). mon consequence for elopement than for other
While the model described by Lehardy et al. forms of problem behavior. Elopement may
(2013) allowed for repeated measures of elope- more commonly produce access to tangibles in
ment in a session, the data obtained were lim- that the locations to which individuals elope
ited to an analysis of evocative events, as the may involve access to leisure items or food
reversal of locations upon elopement (e.g., eloping to a playground, kitchen, or play-
(i.e., moving the putative EO along with the room). Further, when children elope from
participant) did not allow for the direct mea- home into an open space (e.g., by a busy
surement of differential responding to a distinct street), caregivers may attempt to lure them
RESPONSE LATENCY IN ELOPEMENT 437

back to safety with preferred materials. This Elopement remains one of the most danger-
claim is admittedly speculative; additional ous, life-threatening forms of problem behavior
descriptive research on the frequency of parent- exhibited by individuals with IDD, regardless
and teacher-delivered consequences for elope- of age. The most common management strate-
ment is needed. gies for this behavior are antecedent environ-
These potential sources of tangible reinforce- mental manipulations including door alarms,
ment raise a point related to how we typically coded or keyed door locks, closed-circuit cam-
categorize social and nonsocial reinforcement. era monitoring, and other safeguards to prevent
The tangible condition in FAs is typically con- elopement (e.g., Moore, Algase, Powell-Cope,
sidered a test for socially mediated positive rein- Applegarth, & Beattie, 2009), with some pro-
forcement. That is, an adult provides access to fessionals using sedative medication to reduce
an edible or leisure item following each overall activity levels (e.g., Solomon & Lawlor,
instance of a target behavior. In our current 2013). There is an opportunity for functional
evaluation, the delivery of toys (tangibles) was assessment-based intervention to make a signifi-
not mediated by another person; therapists cant impact by reducing the restrictiveness of
were present in both locations but access to the these environments and teaching alternative
reinforcers was a function of presence in the functional skills.
environment, not delivery by the therapist). As
such, the tangible item may be described as an
automatic positive reinforcer. However, in situ- REFERENCES
ations in which a child leaves her home, and Altman, D. G., & Bland, J. M. (1994). Diagnostic tests
the parent provides access to a tangible to lure 1: Sensitivity and specificity. British Medical Journal,
308, 1552.
her home, then this same reinforcer would be Anderson, C., Law, J. K., Daniels, A., Rice, C.,
considered a social positive reinforcer. The dis- Mandell, D. S., Hagopian, L., & Law, P. A. (2012).
tinction between the mediation of these rein- Occurrence and family impact of elopement in chil-
forcers may be important to understanding the dren with autism spectrum disorders. Pediatrics, 130,
870-877. https://doi.org/10.1542/peds.2012-0762
natural consequences of elopement and the Davis, T. N., Durand, S., Bankhead, J., Strickland, E.,
assessment conditions that must be arranged to Blenden, K., Dacus, S.,. . . Machalicek, W. (2013).
capture the contingencies operating in the nat- Brief report: Latency functional analysis of elope-
ment. Behavioral Interventions, 28, 251-259. https://
ural environment. Further, the design of these doi.org/10.1002/bin.1363
procedures should be influenced by parent Hayes, S. C., Nelson, R. O., & Jarrett, R. B. (1987). The
report. If parents suggest that an individual treatment utility of assessment: A functional approach
to evaluating assessment quality. American
often leaves the house to access playground Psychologist, 42, 963-974. https://doi.org/10.1037//
equipment, then the relevant tangible condition 0003-066x.42.11.963
should be incorporated into the assessment Haynes, S. N. (2001). Clinical applications of analogue
rather than conducting a preference assessment behavioral observation: Dimensions of psychometric
evaluation. Psychological Assessment, 13, 73-85.
to identify a high-preference item. The latter https://doi.org/10.1037//1040-3590.13.1.73
seems likely to risk shaping new problem Iwata, B. A., Dorsey, M. F., Slifer, K. J.,
behavior in the manner suggested by Rooker Bauman, K. E., & Richman, G. S. (1994a). Toward
a functional analysis of self-injury. Journal of Applied
et al. (2011). However, should parents identify Behavior Analysis, 27, 197-209. https://doi.org/10.
that they will attempt to lure or tempt children 1901/jaba.1994.27-197 (Reprinted from Analysis
to come home by offering preferred materials, and Intervention in Developmental Disabilities,
than including items identified by a preference 2, 3-20, 1982.)
Iwata, B. A., Pace, G. M., Cowdery, G. E., &
assessment or the items used by the parents Miltenberger, R. G. (1994b). What makes extinction
might be appropriate. work: An analysis of procedural form and function.
438 MICHELE R. TRAUB and TIMOTHY R. VOLLMER

Journal of Applied Behavior Analysis, 27, 131-144. as a variable maintaining problem behavior. School
https://doi.org/10.1901/jaba.1994.27-131 Psychology Review, 34, 425-431.
Jacobson, J. W. (1982). Problem behavior and psychiatric Neidert, P. L., Iwata, B. A., Dempsey, C. M., &
impairment within a developmentally delayed popu- Thomason-Sassi, J. L. (2013). Latency of response
lation I: Behavior frequency. Applied Research in Men- during the functional analysis of elopement. Journal
tal Retardation, 3, 121-139. https://doi.org/10. of Applied Behavior Analysis, 46, 312-316. https://doi.
1016/0270-3092(82)90002-9 org/10.1002/jaba.11
Jang, J., Dixon, D. R., Tarbox, J., & Granpeesheh, D. Piazza, C. C., Hanley, G. P., Bowman, L. G., Ruyter.
(2011). Symptom severity and challenging behavior J. M., Lindauer, S. E., & Saiontz, D. M. (1997).
in children with ASD. Research in Autism Spectrum Functional analysis and treatment of elopement. Jour-
Disorders, 5, 1028-1032. https://doi.org/10.1016/j. nal of Applied Behavior Analysis, 30, 653-672. https://
rasd.2010.11.008 doi.org/10.1901/jaba.1997.30-653
Lang, R., Rispoli, M., Machalicek, W., White, P. J., Rooker, G. W., Iwata, B. A., Harper, J. M.,
Kang, S., Pierce, N.,. . . Lancioni, G. (2009). Treat- Fahmie, T. A., & Camp, E. M. (2011). False-positive
ment of elopement in individuals with developmental tangible outcomes of functional analyses. Journal of
disabilities: A systematic review. Research in Develop- Applied Behavior Analysis, 44, 737-745. https://doi.
mental Disabilities, 30, 670-681. https://doi.org/10. org/10.1901/jaba.2011.44-737
1016/j.ridd.2008.11.003 Shavelle, R. M., Strauss, D. J., & Pickett, J. (2001).
Lehardy, R. K., Lerman, D. C., Evans, L. M., Causes of death in autism. Journal of Autism and
O’Connor, A., & LeSage, D. L. (2013). A simplified Developmental Disorders, 31, 569-576. https://doi.
methodology for identifying the function of elope- org/10.1023/A:1013247011483
ment. Journal of Applied Behavior Analysis, 46, 256- Shirley, M. J., Iwata, B. A., & Kahng, S. (1999). False-
270. https://doi.org/10.1002/jaba.22 positive maintenance of self-injurious behavior by
Matson, J. L., & Rivet, T. T. (2008). Characteristics of access to tangible reinforcers. Journal of Applied
challenging behaviours in adults with autistic disorder, Behavior Analysis, 32, 201-204. https://doi.org/10.
PDD-NOS, and intellectual disability. Journal of Intel- 1901/jaba.1999.32-201
lectual and Developmental Disability, 33, 323-329.
Solomon, O., & Lawlor, M. C. (2013). “And I look
https://doi.org/10.1080/13668250802492600
down and he is gone”: Narrating autism, elopement,
McKerchar, P. M., & Thompson, R. H. (2004). A
and wandering in Los Angeles. Social Science &
descriptive analysis of potential reinforcement contin-
Medicine, 94, 106-114. https://doi.org/10.1016/j.
gencies in the preschool classroom. Journal of Applied
socscimed.2013.06.034
Behavior Analysis, 37, 431-444. https://doi.org/10.
1901/jaba.2004.37-431 Thomason-Sassi, J. L., Iwata, B. A., Neidert, P. L., &
Moore, D. H., Algase, D. L., Powell-Cope, G., Roscoe, E. M. (2011). Response latency as an index
Applegarth, S., & Beattie, E. B. A. (2009). A frame- of response strength during functional analyses of
work for managing wandering and preventing elope- problem behavior. Journal of Applied Behavior
ment. American Journal of Alzheimer’s Disease and Analysis, 44, 51-67. https://doi.org/10.1901/jaba.
Other Dementias, 24, 208-219. https://doi.org/10. 2011.44-51
1177/1533317509332625 Thompson, R. H., & Iwata, B. A. (2001). A descriptive
Mouridsen, S. E., Bronnum-Hansen, H., Rich, B., & analysis of social consequences following problem
Isager, T. (2008). Mortality and causes of death in behavior. Journal of Applied Behavior Analysis, 34,
autism spectrum disorders: An update. Autism, 12, 403- 169-178. https://doi.org/10.1901/jaba.2001.34-169
414. https://doi.org/10.1177/1362361308091653
Mueller, M. M., Sterling-Turner, H., & Moore, J. W. Received January 5, 2017
(2005). Towards developing a classroom-based func- Final acceptance May 8, 2018
tional analysis condition to assess escape-to-attention Action Editor, Jeffrey Tiger

Das könnte Ihnen auch gefallen