-
IMPROVING BEHAVIOR IN A RESIDENTIAL SERVICEFOR YOUTH IN DRUG AND
ALCOHOL
REHABILITATION
Sarah Ann Taylor1,2* and Oliver C. Mudford1
1The University of Auckland, Auckland, New Zealand2Odyssey
House, Auckland, New Zealand
A token reinforcement program, named Thumbs Up, was introduced
in a residential therapeuticcommunity for drug and alcohol
rehabilitation. Sixteen young people aged 1417 years participated
forvarying durations. Participants were able to negotiate target
behaviors and choose backup reinforcers.We evaluated the program
within a multiple baseline design across participants, settings,
and behaviors.Improvements occurred in positive verbal statements
across settings, timely room cleaning, andleadership
responsibilities. Inappropriate statements were also measured and
did not change. One staffmember was trained in the use of Thumbs Up
with positive results. Difculties with implementingbehavioral
interventions in a non-behaviorally oriented treatment service were
identied. Despite needingrenements, it was concluded that a token
reinforcement program can be an effective and acceptablemeans of
improving behavior within a therapeutic community framework.
Copyright 2012 JohnWiley& Sons, Ltd.
In addition to interventions targeted on abusive consumption,
youth with a historyof substance abuse may require behavioral
interventions for various antisocialproblem behaviors; excessive
swearing, vandalistic grafti production, and violentbehavior are
examples. Additionally, they may lack acceptable levels of
positivesocial behavior, including manners, positive statements,
and promptness (Kaminer& Tarter, 2004).Token economies have
been successful in addressing many of these behavioral
excesses and decits in a variety of community and residential
settings for a rangeof populations. A token economy is a behavioral
procedure in which a token isprovided contingent on a desired
target behavior. Tokens can later be exchanged
*Correspondence to: Sarah Ann Taylor, Applied Behavior Analysis
Programme, Psychology Department, TheUniversity of Auckland (Tamaki
Campus), PB 92019, Auckland, New Zealand. E-mail:
[email protected] thesis describing the study fullled the
requirements for an MSc degree in Psychology at The University
ofAuckland.
Copyright 2012 John Wiley & Sons, Ltd.
Behavioral InterventionsBehav. Intervent. 27: 109128
(2012)Published online 14 May 2012 in Wiley Online
Library(wileyonlinelibrary.com) DOI: 10.1002/bin.1342
-
for backup reinforcers, such as highly preferred items or
activities. Tokens, thereby,become conditioned reinforcers.
Populations beneting from token economieshave included adolescents
in residential settings (Liberman, Ferris, Salgado, &Salgado,
1975; Phillips, 1968; Phillips, Phillips, Fixsen, & Wolf, 1971)
and adultsin drug and alcohol rehabilitation programs (Higgins et
al., 1991; Miller, Hersen,& Eisler, 1974; Preston et al., 1997;
Silverman, Chutuape, Bigelow & Stitzer,1996; Silverman, Higgins
et al., 1996; Silverman, Wong et al., 1996). Behaviorstargeted in
token economies have included making aggressive statements
(Phillips,1968), room cleaning (Phillips et al., 1971), positive
interactions (Bowers, Woods,Carlyon, & Friman, 2000), and
reducing alcohol consumption (Miller et al., 1974).Despite this
extensive research on token economies, there does not appear to
beany research on their implementation for adolescents in
residential alcohol and drugrehabilitation facilities.One
residential treatment model for youth with substance abuse problems
is
known as the therapeutic community (TC). TC residents move
through treatmentlevels, including an assessment phase, treatment
phases, and a re-entry phase (basedon De Leon, 2004). A move up the
levels occurs after a resident has met criteria for asignicant
period at their current level.Token reinforcement should t well
within a TC and cohere with many of the TC
principles. Adolescents in residential treatment favor
motivational incentives (pointor token giving) and positive
feedback (Wilner et al., 1977). In the TC, the onlyimmediate
positive reinforcer available is praise. Other reinforcers
available in aTC are delayed, with the most signicant reinforcer
being the opportunity to moveup the treatment levels (offering more
responsibility and privileges). Additionalprivileges in a token
reinforcement include the opportunity to negotiate
individualizedbackup reinforcers.The TCs focus on right living
encourages appropriate social behaviors (De
Leon, 2004), and a token reinforcement may help to improve
behavior required fora move up the treatment levels. Adolescents
with a substance abuse problem enteringa TC are claimed to have an
inability to delay gratication (De Leon, 2004). Tokensare used to
bridge the delay between the appropriate behavior and the
backupreinforcer. Gratication may be further delayed by allowing
participants to savetokens for larger rewards.This study
implemented token reinforcement within a residential TC for
adolescents
with substance abuse problems. It was of interest to see if
tokens would serve toimprove prosocial target behaviors, especially
positive statements. Also, wouldinappropriate statements (including
swearing and complaining) decrease as a result ofrewarding positive
statements? In addition, we wanted to assess if any changes in
targetand non-target behavior would be maintained once the token
reinforcement wasimplemented by other staff at the program.
110 S. A. Taylor and O. C. Mudford
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
GENERAL METHOD
Setting
The study was conducted at a residential community for
adolescents of both sexesaged between 13 and 17 years with a drug
or alcohol problem. The residentialcommunity followed the TC
philosophy. There was a maximum of 11 residents atany time.
Admission was voluntary. The opportunity for adolescents to
formally exitthe program with parental or social worker agreement
was always available, as wasthe opportunity to abscond. Recently
graduated residents had taken between sixand nine months to
complete the treatment program. The residential communityhad a
layout typical of a large modern home. The residents attended a
school, whichwas at a separate site within the TC.
Participants
All participants admitted to the residential TC met criteria
that included an exten-sive substance abuse history, no community
vocational or educational placement, andfailure in community-based
programs. Additionally, many had a diagnosed mentalhealth disorder,
poor or low socialization skills, persistent criminal offending,
andno stable home environment. The youth were admitted only after
initial medicallydirected withdrawal programs had been
successful.Of the 28 residents that participated at some point in
the study, 60% were New
Zealand European, 36% were New Zealand Maori, and 4% were Pacic
Islander.Their ages ranged from 14 to 17 years. Only the most
complete data sets (16 residents)are presented in this paper.The
rst author (researcher) was employed as a youth therapist at the
TC. She had
six months experience before the study commenced. Concurrently
with the study, shewas receiving post-masters degree in supervised
experience in applied behavioranalysis toward professional
credentials.
Materials
Tokens
The tokens were printed pieces of paper measuring 2 4 cm. A
thumbs upsymbol was printed on colored paper not available in the
residential community.Tokens were colored according to the setting
to which they applied. Thumbs upbecame the name of the token
reinforcement at the TC as it signied the positiveorientation of
the intervention.
111Thumbs Up
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
Backup Rewards
Initial backup rewards were inexpensive, required few tokens,
and were easilyconsumable (e.g., candy), so that employing frequent
exchange periods would notdisrupt the overall program schedule.
Once an effect had been demonstrated acrosstwo settings with small
rewards and a resident had been in Thumbs Up for twoweeks, they
were able to choose their own rewards. Residents were advised
thatthe rewards had to be (i) affordable (costing $NZ20 per month
maximum), (ii)relatively easy to provide, and (iii) permitted in
the residential program. Residentscould choose material awards or
activity-based rewards, such as computer time,tutoring, or help to
gain his or her driving license. A notice board on a lounge wallwas
dedicated to Thumbs Up and displayed information regarding the
contingenciesand prices of backup items.
Interobserver Agreement
Interobserver agreement was calculated on sessions where a
second observer,either a staff member or a postgraduate student who
had training, observed.Observations were either direct in vivo or
via audio recording of the behaviorsof interest.
Procedural Fidelity
When the token reinforcement was handed over to another staff
member, theresearcher measured the integrity with which staff gave
praise and tokens. Correctresponses involved giving praise for
target behaviors, giving the correct number oftokens, and giving
corrective feedback if required (e.g., You need to tidy yourdrawers
to get two tokens), and incorrect responses involved not giving
praise,giving tokens for non-target behavior, failing to give
tokens for target behavior, orgiving the wrong number of tokens.
Percentage of correct responses was thencalculated by dividing the
number of correct responses by the sum of correctresponses and
incorrect responses and multiplying by 100.
Social Acceptability
A six-item survey was completed by clients about their
experience with theThumbs Up program. Items concerned whether they
liked and understood theintervention and whether they recommended
its continuation beyond the researchstudy. The participants marked
their responses on a 10-cm visual analog scale. A
112 S. A. Taylor and O. C. Mudford
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
neutral response was 50. Responses toward the negative end of
the scale were values049, and responses toward the positive end of
the scale were values 51100.
EXPERIMENT I: POSITIVE STATEMENTS
Following a meeting with the management, initial target
behaviors were selected inareas that were considered to be
important to the residents while in treatment andafter discharge.
The rst of these target behaviors was positive statements.
Method
Settings
Settings in which data were collected were termed Mealtimes,
Morning Meeting,and Start Group.
Dependent Variables and Measurement
Positive statements were verbal statements that (i). indicated
approval, either inregular English (e.g., She always looks pretty)
or slang (e.g., Sweet as, Thatwas a mean lunch); or (ii) described
a desirable state or mood (e.g., Im so happy);or (iii) showed
concern for another resident, gave thanks, or provided an apology
toanother resident. Statements containing swear words or
immediately followed by anegative (e.g., Youre so cool. . . Not!)
were not scored as positive. These statementswere measured as
frequency within 5-min intervals.Inappropriate statements were
verbal statements that (i) contained swear words; (ii)
included venting, that is, strong disapproval of a person, the
program, or an activity;or (iii) Neg-raving, that is, romanticizing
positively about alcohol and drug use orcriminal behaviors. These
statements weremeasured as frequencywithin 5-min
intervals.Measurement of frequency within 5-min intervals was
facilitated by using an
Invisible Clock (Time Now Corporation, 2004), which was
programmed to signal5-min intervals with a tone or vibration. The
device measured 5.5 4 cm andresembled a pager. When a session
was
-
Procedure
Baseline. Positive statements were recorded in Mealtime, Morning
Meeting, andStart Group. Positive statements were followed by
praise from the researcher inall settings. During Morning Meeting,
the residents were often prompted to saysomething positive in the
Positive Acknowledgements section.Thumbs Up: Start Group. The
researcher verbally gave the participants the
denition of positive statements and role-modeled examples. The
participants wereadvised that they could earn one token per 5-min
interval. Contingent upon a positivestatement, a participant
generally received a token immediately. Labeled verbalpraise was
also delivered with the token. If the researcher was not able to
immedi-ately give the token (e.g., when across the room), she
stated thumbs up orgestured a thumbs up sign to the participant.
New residents who joined the StartGroup during the intervention
remained in baseline for the rst two sessions.Exchange times were
set immediately after each Start Group. Once participants
haddemonstrated an increase in positive statements over three
sessions, they had theopportunity to save their tokens until the
end of the second session to gain a largerreward. Residents who had
chosen individual rewards were able to save until theychose to
exchange their tokens.Thumbs Up: Mealtimes. The procedure for
earning and receiving tokens at
mealtimes was the same as those in Start Group. Exchange times
for the initialrewards were set at about an hour after mealtime
when the participants had nishedtheir post-meal cleaning
jobs.Thumbs Up: Morning Meeting. Tokens were earned in the same
manner as
previous settings. However, the researcher could not provide
tokens immediately asthat would disrupt the ow of the meeting.
Contingent on a positive statement, theresearcher made a thumbs up
gesture or statement and provided labeled praise.Tokens were given
out at the end of the meeting, and exchange times
occurredimmediately after.Generalization: extending Thumbs Up to
other staff. Following training, a staff
member (Heidi) implemented Thumbs Up in Morning Meeting.
Training involvedteaching the staff member to accurately identify
positive statements from a 5-mincriterion record that was created
from audio recordings of Morning Meeting andMealtime settings. In
addition, the staff member was taught (using modeling,
rehearsal,and feedback) how to accurately record the data and
conduct the exchange of tokens.
Experimental Design
The intervention was introduced rst for Jay and Lisa targeting
positive statementsin Start Group once baseline data appeared
stable. When an improvement in their
114 S. A. Taylor and O. C. Mudford
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
behavior was shown, the intervention was introduced to Robyn.
Following this initialdemonstration of apparent treatment effect,
baseline lengths were then minimized inStart Group and Mealtime for
ethical and practical reasons. The sequence of interven-tion for
positive statements met the criteria for multiple baseline across
participantsexperimental design.A return to baseline (time out from
tokens) was implemented for participants
when they returned to the program after being discharged for
disciplinary reasonsor had returned from absconding. Thus, there
were ABA and ABAB withdrawalelements included.The intervention was
introduced in Start Group once stable baseline data had been
gained. Once an increase in positive statements was shown across
residents, theintervention was introduced in Mealtime.
Subsequently, the intervention wasintroduced in Morning Meeting.
This sequence met the criteria for multiple baselineacross
settings.
Results
Figure 1 shows data for positive statements presented as a
non-concurrent multiplebaseline across participants and settings
(Mealtimes, Morning Meeting, and StartGroup) for positive
statements and inappropriate statements. Data paths for
mostparticipants were variable.In baseline, most participants
showed low, stable responding, or a decreasing trend
(e.g., MealtimeBree). Some residents, however, showed increasing
frequenciesduring baseline (e.g., MealtimeTim).The left panel of
Figure 1 shows increases in the frequency of positive
statements
from baseline to intervention for all participants except Tim
(Mealtime), who left theprogram, returned briey, and then left
again. Bree (Mealtime) also left the programand returned to
baseline in her last session.The center panel of Figure 1 shows
increases in the frequency of positive
statements from baseline to intervention for all participants
except for Tim (MorningMeeting), for whom there were limited data.
Some of these residents left the program(Nevil, Lisa, and Mandy)
and returned (Nevil). Although there were not sufcientdata for Jay
(Start Group), there was an increasing trend in his positive
statementsduring the intervention.The right panel of Figure 1 shows
an increase in positive statements from baseline
to intervention for most residents. Limited data were gained for
Hannah (Mealtime)and Bree (Morning Meeting) but showed that the
frequency of positive statementswas decreasing during intervention.
For Jon (Mealtime) in session 31, the intervalwas only 30-s long
and Jon said three positive statements, causing an
inatedextrapolated value. Jon left the program and upon return,
positive statements
115Thumbs Up
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
decreased to zero in a return to baseline in Mealtime and
Morning Meeting and thenincreased again.Before the tokens could be
introduced for subsequent behaviors, data were
analyzed to assess if Thumbs Up had served to increase positive
statements. The datafor positive statements were inspected across
the 16 graphs in Figure 1 that havearrows displayed. The arrows
indicate the session up to which data had been gainedwhen the
graphs were analyzed.Graphs were visually inspected using Millers
(1997) method for visual inspection
of behavioral data or the split-middle line of progress method
(White, 1974). With theuse of these methods, it was determined that
11 graphs showed an increase whencomparing baseline and
intervention phases, four showed no change, and one showeda
decrease. A sign test (Siegel & Castellan, 1988) was performed
with these results.With an alpha level of .05, the benecial effect
of tokens on positive statements wasstatistically signicant (z=
2.89, p= .006).Data for inappropriate statements showed low
frequencies for all participants in
baseline, with the exception of Jay in Start Group. The
frequency of inappropriatestatements remained at near-zero levels
across all participants during the token phaseapart from Jay (Start
Group).
Figure 1. Multiple baseline across participants and settings for
positive (lled points) and inappropriate(unlled points) statements.
The arrows pointing at data points are explained in the text.
116 S. A. Taylor and O. C. Mudford
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
Figure 2 displays the frequency of positive statements across
three participants inMorning Meeting when both the author (lled
points) and Heidi (unlled points)were implementing Thumbs Up. There
was no visually signicant change in thefrequency of positive
statements when Heidi was implementing Thumbs Up whencompared with
the researcher.Heidis average percentage of correct responses for
positive statements was 85%.
She was also 100% correct on three occasions where she and the
researcher role-played token exchange.
Discussion
Positive statements improved across settings for most
participants in the currentstudy, supporting the results of
previous studies that have also selected positivestatements or
interactions as a target behavior (Bowers et al., 2000; Ervin,
Miller, &Friman, 1996; Sanson-Fisher, Seymour, Montgomery,
& Stokes, 1978). However,the increases in positive statements
in the current study were not dramatic. This mayhave been because
the frequency of positive statements was limited by the
contingen-cies. The aim was to increase the frequency of positive
statements to one per 5-mininterval. Having this ceiling may have
prevented further improvements of positivestatements. It may have
been benecial to give tokens for more than one positivestatement
per 5-min or to shorten the intervals to see a more dramatic
behavior change.However, this should be balanced against the
possibility of teaching residents to betoo positive, that is, more
so than their typical age peers.
Annabelle
012345
Rachel
FREQ
UENC
Y PE
R 5-
MIN
UTE
INTE
RVAL
012345
Jon
SESSIONS5 10 15 20 25 30 35 40 45 50 55 60
0
5
30
Baseline
Baseline
Tokens
Tokens
Figure 2. Positive statements data obtained from Morning Meeting
comparing the sessions in whichHeidi (unlled points) and the
researcher (lled points) were implementing Thumbs Up.
117Thumbs Up
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
The levels of inappropriate statements were socially acceptable
for most partici-pants and did not change when Thumbs Up was
introduced. Previous literatureshowed antisocial comments to
decrease as a result of increasing prosocial comments(Sanson-Fisher
et al., 1978). In the current study, inappropriate statements may
havebeen low because of being under aversive control (the presence
of staff in all of thesettings). Petry et al. (1998) found similar
results in an outpatient substance abusetreatment center.
EXPERIMENT II: PROMPTNESS FOR DORM TIGHT
After token reinforcement demonstrated a signicant effect with
positive statements,it was of interest whether similar effects
could be demonstrated with room cleanliness.
Method
Setting
Bedroom areas were checked for tidiness to a set of criteria
30min after allresidents were up and dressed in the mornings.
Breakfast was not announced untilall rooms met all criteria.
Response Denition and Measurement
Promptness for dorm tight was dened as the residents room clean
and tidy to 11Dorm Tight criteria on time. Criteria included oors
being cleared, clothes folded indrawers, and beds made neatly with
sheets tucked in. Residents were given a 2-minwarning before checks
occurred. This is measured as yes/no.A second observer was present
for at least 20% of sessions, and interobserver
agreement was calculated by the number of agreements divided by
the sum of agreementsplus disagreements and multiplied by 100.
Average agreement values for Dorm Tightwere 82% (range 58100%).
Procedure
Baseline. The researcher checked a residents room and assessed
whether itmet all of the Dorm Tight criteria (standards). If it did
not meet all the standards,the researcher pointed out and explained
what needed to be xed. A resident waspraised if his or her room was
up to standard.
118 S. A. Taylor and O. C. Mudford
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
Thumbs Up. A resident received two tokens and verbal praise if
his or her roommet all the standards at Dorm Check time. If one
standard was not met, a resident wasgiven one token with praise for
the standards that had been met. The researcherinformed them what
needed to be performed next time to earn two tokens. If morethan
one standard were not met, corrective feedback was delivered and
sometimesa TC consequence resulted (e.g., loss of a privilege).
Generalization: extending Thumbs Up to other staff. Following
further training,Heidi also implemented Thumbs Up in Dorm Tight.
Further training involvedteaching the staff member to accurately
check dorm rooms.
Experimental Design
A multiple baseline across participants was used to assess the
effects of tokenreinforcement on Dorm Tight.
Results
Figure 3 shows the days in which the participants were able to
achieve the criterionduring baseline and token phases. An
improvement for most participants occurredwhen tokens were
introduced. However, for some participants, baseline levelsappeared
to be increasing before the intervention (Ken, Edward, and
Annabelle).The proportion of days that participants could meet the
standards in token phaseswas higher for all participants when
compared with baseline. Averaged across theparticipants, the
percentage of days that standards were met in baseline was
37%(range 066%) and more than doubled to 80% (range 56100%) in
intervention.However, data for the participants were variable and
limited data were gained forsome participants (Nevil and Mandy).A
return to baseline was implemented for Nevil and Jon upon their
return to the
program and showed that both did not meet Dorm Tight standards.
When returnedto the token phase, Nevil was able to meet the
standards for two days then left theprogram. Jon returned to the
token phase and was able to meet the standards on mostdays.Figure 4
displays data obtained from Dorm Tight across three
participants
including sessions where Heidi was implementing Thumbs Up. For
Annabelle andRachel, there was no visually signicant change in data
when Heidi was implement-ing Thumbs Up when compared with the
researcher. For Jon, there was a reductionin the number of days
where Dorm Tight standards were met when Heidi wasimplementing
Thumbs Up. Heidis average percentage of correct responses for
DormTight was 94%.
119Thumbs Up
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
Discussion
In Dorm Tight, there was an improvement shown for all
participants in thepercentage of days that they were able to meet
all the Dorm Tight standards. Theseresults follow those of previous
studies that have also improved tidiness (Phillips,1968; Phillips
et al., 1971). Improving promptness to Dorm Tight also helped
withthe running of the morning schedule.Despite improvements, the
participants did not consistently meet Dorm Tight
standards when receiving tokens. Data were recorded as yes/no
for Dorm Tight.
Rachel0
1
Jon
DAYS5 10 15 20 25 30 35 40 45 50
0
10
1
Nevil0
1
Annabelle0
1
Ken
DO
RM T
IGHT
STA
NDAR
DS M
ET (1
= YE
S/ 0
= NO)
0
1
Baseline Tokens
Baseline Tokens
MandyBaseline Tokens
Edward0
1
Figure 3. Multiple baseline across participants for Dorm Tight.
This shows the days in which participantsmet criterion during
baseline and token phases.
120 S. A. Taylor and O. C. Mudford
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
The participants often failed on more than one standard during
baseline and improvedto only failing on one standard occasionally
in the token phase. Failing on only one of11 standards was still
recorded as no. Having an all or nothing measure made itimpossible
to see improvements on different areas in the room. It would have
beenbenecial to record the percentage of Dorm Tight standards that
were met. Phillipset al. (1971) had divided room cleaning into 10
areas that were scored according towhether they met specic
criteria.
EXPERIMENT III: ANNOUNCING JOB FUNCTION
Method
Setting
Job function occurred immediately after breakfast. Residents
were required tocomplete rostered domestic activities (e.g.,
mopping) for approximately 20min.
Response Denition and Measurement
Job function was dened as announcing to other residents that it
was time to starttheir domestic chores within 1min of breakfast
ending. The resident announcing Job
Jon
5 10 15 20 25 30 35 40 45 50 55
0
1
Annabelle
DAYS
0
1
Rachel
0
1
Baseline Tokens
Baseline Tokens
DO
RM T
IGHT
STA
NDAR
DS M
ET (1
= YE
S/ 0
= NO)
Figure 4. Dorm Tight data obtained comparing the days in which
Heidi (unlled points) and theresearcher (lled points) were
implementing Thumbs Up.
121Thumbs Up
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
Function must be moving to start his or her own rostered chore
within 1min ofmaking the announcement. This is measured as yes/no.
A second observer waspresent for at least 20% of sessions, and
interobserver agreement was calculated bythe number of agreements
divided by the sum of agreements plus disagreementsand multiplied
by 100. Interobserver agreement was 100%.
Procedure
Baseline. If the resident did not announce Job Function by the
required time, theresearcher prompted them (e.g., Jon, you need to
be getting the others into JobFunction now). If the resident
announced Job Function unprompted, the researchergave praise.Thumbs
Up. Contingent upon announcing Job Function, a resident received
one
token. If they announced Job Function late or announced it but
continued to eat ortalk to residents, they were not awarded a
token. They were also reminded of thecontingencies, Annabelle, you
need to announce Job Function on time if youwant to earn a token.
Initial exchange times occurred after residents had completedtheir
jobs.
Experimental Design
An AB design across two participants was used to assess the
effects of tokenreinforcement on announcing job function. Token
reinforcement was introduced forjob function following intervention
with positive statements and Dorm Tight. Thus,the overall study
design also met the criteria for multiple baseline across
behaviors.
Results
Figure 5 shows the number of days that Jon and Annabelle
announced Job Functionduring baseline and token phases. Tokens
succeeded in increasing the number of daysJon announced Job
Function. Annabelle also showed an improvement in announcingJob
Function. However, neither participant was 100% reliable.
Discussion
Token reinforcement also served to increase the number of days
on which tworesidents announced Job Function. Improving this
behavior ensured that Job Functionstarted on time without prompting
from the researcher. However, neither resident wascompletely
reliable at announcing Job Function. This may have been due to
disruptioncaused by tokens only being available on days when the
researcher was present (four
122 S. A. Taylor and O. C. Mudford
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
days per week) and the residents losing the responsibility when
placed on sanction.There were no data taken on these days. It would
have been of interest to rewardannouncing Job Function in the
evening and other scheduled events (e.g., mealtimesand
meetings).
SOCIAL ACCEPTABILITY
Five participants completed the client satisfaction survey.
Participants gave highratings for Thumbs Up in areas of enjoyment,
(mean 87, range 72100), understandinghow to earn tokens (mean 99,
range 98100), continuing Thumbs Up (mean 89, range46100),
recommending new residents to participate (mean 90, range
63100),improvement of behaviors (mean 94, range 74100), and for
making the stay at theresidential community more enjoyable (mean
65, range 38100).
GENERAL DISCUSSION
In summary, token reinforcement served to improve three target
behaviors for mostparticipants despite poor retention in the
treatment program. Another staff memberwas also able to implement
token reinforcement (known as Thumbs Up in thefacility) in some
settings with positive results. Residents who responded to the
socialvalidity survey found the intervention highly acceptable. The
token reinforcementwas inexpensive and easy to implement for both
the researcher and another
DAYS1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
24AN
NO
UNCE
D JO
B FU
NCTI
ON
(1=YE
S/0=
NO)
0
1
Annabelle0
1
Jon
Baseline Tokens
Figure 5. Multiple baseline across participants for Announcing
Job Function.
123Thumbs Up
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
staff member. This study extends current literature by
implementing token reinforce-ment in a residential rehabilitation
facility for adolescents with substance abuseproblems. Overall, the
study design was multiple baseline across participants,settings,
and behaviors with some treatment withdrawal elements (Barlow,
Hayes,& Nelson, 1984).Another staff member also implemented
Thumbs Up with positive results in
Morning Meeting and Dorm Tight. The staff member showed correct
respondingabove 80% on most training sessions; however, she
occasionally neglected to givetokens out. The researcher reduced
this lapse somewhat by instructing the residentsto request their
tokens if they had not been given out. Training participants to
requestreinforcement has been reported in previous studies (Seymour
& Stokes, 1976).It was aimed to train the staff to implement
Thumbs Up in other settings, however
her role was restructured before the end of the study and she
was no longer present inthese settings.Some residents in the
current study saved their tokens and only exchanged a small
number once before leaving the treatment program. Reasons for
this may have beenthat the tokens themselves may have been
reinforcing: Residents were observed tocompete with their peers in
acquiring and saving their tokens. The praise providedwith tokens
may have been the reinforcer for most participants rather than
theeventual gain of the backup reward. Previous literature has also
shown the tokenalone to reinforce behavior (Hall et al., 1972).A
more probable reason for savings in the current study may be
attributed to the
control over naturally occurring items or activities as backup
rewards. In the currentstudy, television, board games, a weekly
allowance, and basic snacks were freelyavailable at the residential
community and did not need to be purchased with tokens.However,
previous studies have required basic activities such as these to
bepurchased with tokens before any other privilege (e.g., Phillips,
1968; Phillipset al., 1971). Including basic activities (such as
television) as a backup reward mayhave ensured that token spending
was maintained.Enabling participants to change their rewards or
negotiate prices may have helped
to increase the motivation of participants (Barkley, 1987).
Karraker (1977) reported ahigher rate of math performance for
students who chose rewards from a list they hadconstructed than for
students who received a teacher-selected reward. Kamon,Budney, and
Stanger (2005) allowed adolescents to exchange vouchers for
theirchosen appropriate reward.Exchange times in the study were at
set daily times for participants in their rst two
weeks of the program. These new residents were not permitted to
save tokens untilthe next day. This ensured that new participants
came into contact with the backuprewards. For individualized
rewards, exchange periods generally occurred when theresident had
saved the required amount of tokens. Not having a specied
exchange
124 S. A. Taylor and O. C. Mudford
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
time on the schedule may have also resulted in high savings, as
residents were notbeing prompted frequently to purchase backup
rewards.Studies at Achievement Place implemented specied exchange
periods (Wolf
et al., 1976). A new resident rst exchanged points at the end of
the day forprivileges the next day. After demonstrating skills and
self-control, the residentthen exchanged points at the end of the
week for privileges the following week.Similar exchange times may
have improved behavior in the current study. Partici-pants may have
shown higher frequencies of target behaviors if they needed tohave
a set number of tokens by a specied time. It may have also
encouragedregular spending.Because of the difculty with retention,
the researcher did not manage to fade the
token reinforcement for any participants. However, there was an
attempt made toremove Jon from the token reinforcement by electing
him as a manager. This wasonly attempted on a few occasions before
he started a community-based educationalcourse and was no longer
present in the settings in which the token reinforcement
wasoperating. Previous studies have reported the positive results
of peers acting asbehavior modiers (Bailey, Timbers, Phillips,
& Wolf, 1971; Phillips, Phillips, Wolf,& Fixsen, 1973).Data
did not detect generalization across settings for positive
statements or
across behaviors for three residents. Other work duties meant
that the researcherhad limited observation of non-intervention
settings. It would have been benecialto train a staff member to
record positive statements when the researcher could notbe present.
Generalization across behaviors was not expected, as Dorm Tightand
announcing Job Function were not similar to positive statements.
Shortenedbaselines meant that generalization across participants
could not be assessed.Most new participants came into the program,
whereas longer-term residents werealready in the intervention
phase.The rst author continued to lead the administration of Thumbs
Up for one
year following the study and added further target behaviors,
such as completingacademic work and participating in a job skills
training program (Taylor, Mudford,& Phillips, 2011).There were
several difculties in conducting the study. First, the number
of
participants varied greatly during the study. At the beginning
of the study, veresidents consented to being in the study. Numbers
changed as new residents arrivedand other residents left. Three
residents opted not to participate in Thumbs Up at rst,but two
later changed their minds. One other resident who chose not to
participatestayed in the program for two weeks before the
residential community dischargedher because of rule violations. One
resident withdrew from Thumbs Up shortly afterthe intervention
started in Start Groups. Therefore, 29 out of 30 residents at
theresidential community chose to participate in Thumbs Up at some
stage.
125Thumbs Up
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
The number of residents leaving and entering the program limited
the collection ofdata. Therapeutic gains for participants also may
have been lost because of thelimited time spent in intervention.
Gaining informed consent from parents orguardians delayed the
intervention in some cases, and some residents were notexposed to
tokens at all. Poor retention of residents also made individualized
rewardsdifcult. Three residents who initially chose individualized
backup rewards leftwithin days of the researcher purchasing the
items.A second set of difculties concerned changes in rules at the
residential commu-
nity, resulting in restrictions on some popular backup rewards
(e.g., a new rulepreventing residents from using the computer at
the facility). Hobbs and Holt(1976) reported similar organizational
obstacles in implementing a token reinforce-ment. Additionally,
there were days when Thumbs Up was administrativelysuspended
because of participants being placed on sanction. Residents who
wereon sanction in the facility could earn tokens but not exchange
them for backuprewards. This meant that these residents experienced
time out from exchangingtokens, which was not part of Thumbs Up. We
had not experienced these administra-tive hurdles with another
behavioral intervention successfully conducted in the TC(Taylor,
Anderson, & Mudford, 2010). In that study, we taught residents
to preparefor activities outside the TC in a more timely manner. It
is likely that the ThumbsUp study ran into difculties as it
extended over several activities, involved moreresidents, and was
conducted over a longer time.Hobbs and Holt (1976) discussed the
failure of token economies to observe the
change in behavior outside of the treatment facility. In the
current study therewere no data obtained for adolescents who had
left the treatment facility as thiswas against the organizations
policy. Occasionally, graduated residents return tothe treatment
program to visit. It would be of interest to observe behavior
inthis situation.In summary, this study extends the current
literature by implementing a token
reinforcement within a TC framework and with adolescents with a
substance abuseproblem. Results were consistent with those of
previous studies with delinquentadolescents for most of the
participants and target behaviors. Despite some identieddifculties,
the token reinforcement was a successful means of increasing
appropriatebehaviors that generally worked within TC standard
procedures.
ACKNOWLEDGEMENTS
The study was partly funded by a University of Auckland Masters
Scholarship awarded tothe rst author. We thank Kerrie Anderson for
her contributions to the study.
126 S. A. Taylor and O. C. Mudford
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
REFERENCES
Bailey, J. S., Timbers, G. D., Phillips, E. L., & Wolf, M.
M. (1971). Modication of articulation errors ofpre-delinquents by
their peers. Journal of Applied Behavior Analysis, 4, 265281.
Barkley, R. A. (1987). Deant children: A clinicians manual for
parent training. New York: GuilfordPress.
Barlow, D. H., Hayes, S. C., & Nelson, R. O. (1984). The
scientist practitioner. New York: PergamonPress.
Bowers, F. E., Woods, D. W., Carlyon, W. D., & Friman, P. C.
(2000). Using positive peerreporting to improve the social
interactions and acceptance of socially isolated adolescentsin
residential care: A systematic replication. Journal of Applied
Behavior Analysis, 33,239242.
De Leon, G. (2004). Therapeutic communities. In M. Galanter,
& H. D. Kleber (Eds.), The AmericanPsychiatric Publishing
textbook of substance abuse treatment (pp. 485501). Washington DC,
US:American Psychiatric Publishing.
Ervin, R. A., Miller, P. M., & Friman, P. C. (1996). Feed
the hungry bee: Using positive peer reports toimprove the social
interactions and acceptance of a socially rejected girl in
residential care. Journal ofApplied Behavior Analysis, 29,
251253.
Hall, R. V., Axelrod, S., Tyler, S., Grief, L., Jones, E., &
Robertson, R. (1972). Modication of behaviorproblems in the home
with a parent as observer and experimenter. Journal of Applied
BehaviorAnalysis, 5, 5364.
Higgins, S. T., Delaney, D. D., Budney, A. J., Bickel, W. K.,
Hughes, J. R., Foerg, F., et al. (1991). Abehavioral approach to
achieving initial cocaine abstinence. The American Journal of
Psychiatry, 148,12181224.
Hobbs, T. R., & Holt, M. M. (1976). The effects of token
reinforcement on the behavior of delinquentsin cottage settings.
Journal of Applied Behavior Analysis, 9, 189198.
Kaminer, Y., & Tarter, R. E. (2004). Adolescent substance
abuse. In M. Galanter, & H. D. Kleber (Eds.),The American
Psychiatric Publishing textbook of substance abuse treatment (pp.
505517). WashingtonDC, US: American Psychiatric Publishing.
Kamon, J., Budney, A., & Stanger, C. (2005). A contingency
management intervention for adolescentmarijuana abuse and conduct
problems. Journal of the American Academy of Child and
AdolescentPsychiatry, 44, 513521.
Karraker, R. J. (1977). Self versus teacher selected reinforcers
in a token reinforcement. ExceptionalChildren, 43, 454455.
Liberman, R. P., Ferris, C., Salgado, P., & Salgado, J.
(1975). Replication of the Achievement Placemodel in California.
Journal of Applied Behavior Analysis, 8, 287299.
Miller, L. K. (1997). Principles of everyday behavior analysis
(3rd ed.). Pacic Grove, CA.: Brooks/Cole Publishing Co.
Miller, P. M., Hersen, M., & Eisler, R. M. (1974). Relative
effectiveness of instructions, agreements,and reinforcements in
behavioral contracts with alcoholics. Journal of Abnormal
Psychology, 83,548553.
Page, T. J., & Iwata, B. A. (1986). Interobserver agreement:
History, theory and current methods In A.Poling, & R. W. Fuqua
(Eds.), Research methods in applied behavior analysis: Issues and
advances(pp. 99126). New York: Plenum.
Petry, N. M., Bickel, W. K., Tzanis, E., Taylor, R., Kubik, E.,
Foster, M., et al. (1998). A behavioralintervention for improving
verbal behaviors of heroin addicts in a treatment clinic. Journal
of AppliedBehavior Analysis, 31, 291297.
127Thumbs Up
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
Phillips, E. L. (1968). Achievement Place: Token reinforcement
procedures in a home-stylerehabilitation setting for pre-delinquent
boys. Journal of Applied Behavior Analysis, 1,213223.
Phillips, E. L., Phillips, E. A., Fixsen, D. L., & Wolf, M.
M. (1971). Achievement Place: Modication ofthe behaviors of
pre-delinquent boys within a token reinforcement. Journal of
Applied BehaviorAnalysis, 4, 4559.
Phillips, E. L., Phillips, E. A., Wolf, M. M., & Fixsen, D.
L. (1973). Achievement Place: Developmentof the elected manager
system. Journal of Applied Behavior Analysis, 6, 541561.
Preston, K. L., Silverman, K. L., Umbricht-Schneiter, A.,
DeJesus, A., Montoya, I. D., &Schuster, C. R. (1997).
Improvement in naltrexone treatment compliance with
contingencymanagement. In L. S. Harris (Ed.), Problems of drug
dependence, 1996: Proceedings of 58thannual scientic meeting, the
College on Problems of Drug Dependence (pp. 303). NIDAResearch
Monograph 174, NIH Pub No. 974236. Washington, DC: U.S. Government
PrintingOfce.
Sanson-Fisher, B., Seymour, F., Montgomery, W., & Stokes, T.
(1978). Modifying delinquentsconversation using token reinforcement
of self-recorded behavior. Journal of Behavior Therapyand
Experimental Psychiatry, 9, 163168.
Seymour, F., & Stokes, T. F. (1976). Self-recording in
training girls to increase work and evoke staffpraise in an
institution for offenders. Journal of Applied Behavior Analysis, 9,
4154.
Siegel, S., & Castellan, N. J. (1988). Nonparametric
statistics for the behavioral sciences (2nd ed.). NewYork:
McGraw-Hill.
Silverman, K., Chutuape, M. A., Bigelow, G. E., & Stitzer,
M. L. (1996). Voucher-based reinforcementof attendance by
unemployed methadone patients in a job skills training program.
Drug and AlcoholDependence, 41, 197207.
Silverman, K., Higgins, S. T., Brooner, R. K., Montoya, I. D.,
Cone, E. J., Schuster, C. R., &Preston, K. L. (1996). Sustained
cocaine abstinence in methadone maintenance patients
throughvoucher-based reinforcement therapy. Archives of General
Psychiatry, 53, 409415.
Silverman, K., Wong, C. J., Higgins, S. T., Brooner, R. K.,
Montoya, I. D., Contoreggi, C., . . .Preston, K. L. (1996).
Increasing opiate abstinence through voucher-based reinforcement
therapy.Drug and Alcohol Dependence, 41, 157165.
Taylor, S. A., Anderson, K., & Mudford, O. C. (2010).
Effects of textual response prompts foradolescents in a substance
abuse treatment program. Behavioral Interventions, 25, 145155.
Taylor, S. A., Mudford, O. C., & Phillips, K. J. (2011).
Training employment acquisition skills foradolescents with a
substance abuse history. Behavioral Interventions, 26, 282308.
Time Now Corporation. (2004). Invisible clock. San Rafael, CA:
Time Now Corporation.White, O. R. (1974). A manual for calculation
and use of the median slopeA technique of progressestimation and
prediction in the single case. Eugene: University of Oregon,
Regional ResourceCenter for Handicapped Children.
Wilner, A. G., Braukmann, C. J., Kirigin, K. A., Fixsen, D. L.,
Phillips, E. L., & Wolf, M. M. (1977).The training and
validation of youth-preferred social behaviors of child-care
personnel. Journal ofApplied Behavior Analysis, 10, 219230.
Wolf, M. M., Phillips, E. L., Fixsen, D. L., Braukmann, C. J.,
Kirigin, K. A., & Wilner, A. G. (1976).Achievement Place: The
teaching-family model. Child Care Quarterly, 5, 92103.
128 S. A. Taylor and O. C. Mudford
Copyright 2012 John Wiley & Sons, Ltd. Behav. Intervent. 27:
109128 (2012)
DOI: 10.1002/bin
-
Copyright of Behavioral Interventions is the property of John
Wiley & Sons, Inc. and its content may not becopied or emailed
to multiple sites or posted to a listserv without the copyright
holder's express writtenpermission. However, users may print,
download, or email articles for individual use.