This article was downloaded by: [Washington State University Libraries ]

On: 18 November 2014, At: 21:33

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,
UK

Exceptionality: A Special
Education Journal
Publication details, including instructions for authors
and subscription information:
http://www.tandfonline.com/loi/hexc20

Increasing Compliance: A
Quantitative Synthesis of
Applied Research on High-
Probability Request Sequences
David L. Lee
Published online: 08 Jun 2010.

To cite this article: David L. Lee (2005) Increasing Compliance: A Quantitative Synthesis
of Applied Research on High-Probability Request Sequences, Exceptionality: A Special
Education Journal, 13:3, 141-154, DOI: 10.1207/s15327035ex1303_1

To link to this article: http://dx.doi.org/10.1207/s15327035ex1303_1

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the
information (the “Content”) contained in the publications on our platform.
However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or
suitability for any purpose of the Content. Any opinions and views expressed
in this publication are the opinions and views of the authors, and are not the
views of or endorsed by Taylor & Francis. The accuracy of the Content should
not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions,
claims, proceedings, demands, costs, expenses, damages, and other liabilities
whatsoever or howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan, sub-
licensing, systematic supply, or distribution in any form to anyone is expressly
 forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014
 EXCEPTIONALITY, 13(3), 141–154
Copyright © 2005, Lawrence Erlbaum Associates, Inc.

Increasing Compliance: A Quantitative

Synthesis of Applied Research on
High-Probability Request Sequences
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

David L. Lee
Department of Educational and School Psychology and Special Education
Pennsylvania State University

Methods to enhance compliance is a topic of great interest to practitioners serving children

and adults both with and without disabilities. High-probability (high-p) request sequences
are one way practitioners can prevent episodes of noncompliant behavior. Based on the the-
ory of behavioral momentum, high-p request sequences consist of delivering a series of re-
quests to a student that generally results in compliance (high-p) just prior to a request with a
low-probability (low-p) of compliance. The purpose of this meta-analysis was to quantita-
tively synthesize the effects of high-p request sequences on low-p behaviors across different
categories of participants, settings, and procedural variations. Overall, high-p request se-
quences were found to be an effective method to increase compliance. Implications for both
research and practice are discussed.

Dealing with problem behavior is a difficult challenge for practitioners (Westling & Fox,
2000). One type of problem behavior, noncompliance, can have a profound negative im-
pact on vocational, personal, social, and academic success (Hutchinson & Belfiore,
1998; Walker & Walker, 1991). Many interventions have been developed in an effort to
increase compliance among both children and adults across various settings.
Walker and Walker (1991) suggested four approaches to dealing with noncompli-
ance. First, practitioners can change the antecedents that occasion noncompliance. For
example, Kern, Childs, Dunlap, Clarke, and Falk (1994) modified the length, content,
and mode used to perform academic tasks for an 11-year-old student with a history of
academic noncompliance. This multicomponent antecedent intervention resulted in a
decrease in noncompliance. Second, compliance to requests can be increased using
techniques such as differential reinforcement to increase more adaptive behavior while
decreasing noncompliance. Hagopian and Thompson (1999) used reinforcement to
systematically increase the compliance of an 8-year-old boy with cystic fibrosis during

Requests for reprints should be sent to David L. Lee, Department of Educational and School Psychology
and Special Education, Pennsylvania State University, Great Valley, 30 East Swedesford Road, Malvern, PA
19355. E-mail: davidlee@psu.edu
 142 LEE

respiratory treatments. Third, noncompliance can be reduced or eliminated through the

use of punishment. For example, Olmi, Sevier, and Nastasi (1997) used a time-out
procedure to decrease the noncompliance of two children with severe language and
moderate cognitive disabilities in a school setting. Finally, some combination of modi-
fications to antecedents, reinforcement, and punishment can be used. For example, Pi-
azza, Moes, and Fisher (1996) used a combination of differential reinforcement and
extinction to increase compliance of a child with autism during instructional se-
quences.
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

Three of the four methods described by Walker and Walker (1991) are based on deliv-
ering consequences for appropriate and/or inappropriate behavior. Although effective in
many cases, consequence-based strategies are not without drawbacks. For example, pun-
ishing noncompliance using verbal reprimands can work reasonably well if the student is
not seeking attention. If the student is seeking attention, verbal reprimands may inadver-
tently reinforce noncompliance. Similarly, time-out can be effective when the student is
not engaging in inappropriate behavior to avoid or escape from tasks. Removing a stu-
dent from a nonpreferred activity may increase the likelihood of inappropriate behavior
in similar situations in the future. Even with more positive approaches, practitioners
must wait for students to engage in appropriate behavior before reinforcement can occur
(i.e., catch them being good). This may be problematic within the context of a busy class-
room. Ultimately, consequence-based methods rely on practitioners to observe students
and deliver consequences.
A second type of strategy can be implemented prior to the occurrence of non-
compliant behavior and can be used as a preventive technique. Methods that prevent non-
compliance, such as those employed by Kern et al. (1994), may be more efficient and
produce longer lasting results (Walker & Walker, 1991). These techniques have the ad-
vantage of not requiring the inappropriate behavior to occur before the intervention is
implemented. One preventive technique to increase compliance that has received empiri-
cal support is the use of high-probability (high-p) request sequences.
During the past 15 years, much has been written about the use of high-p request se-
quences as a method to enhance compliance (see Davis & Brady, 1993; Killu, 1999,
for reviews). For this intervention, an individual is asked to complete a series of 3 to 4
very brief requests with a high probability of compliance. Immediately after complet-
ing the high-p task sequence, the individual is asked to complete a request with a low
probability of compliance (i.e., low-p). For example, a teacher may deliver a series of
high-p requests (e.g., give me five, touch your nose, clap your hands) just prior to a re-
quest that generally results in noncompliance (e.g., get out your materials). The results
of a series of studies have shown that compliance to low-p requests was enhanced
when those requests were preceded by the completion of several brief, highly pre-
ferred requests/activities (Davis & Reichle, 1996; Horner, Day, Sprague, O’Brian, &
Heathfield, 1991; Mace et al., 1988). Although the mechanism responsible for the ef-
fects of high-p sequences is the subject of debate (see Brandon & Houlihan, 1997;
Houlihan & Brandon, 1996; Nevin, 1996, for discussion), research indicates that the
intervention is effective at increasing compliance across a variety of settings and be-
haviors.
 BEHAVIORAL MOMENTUM 143

High-p request sequences are based on the work of Nevin and colleagues in the area of
behavioral momentum. Nevin, Mandell, and Atak (1983) proposed the theory of behav-
ioral momentum in an effort to explain behavioral persistence under varying environ-
mental conditions. In Newtonian physics, momentum is comprised of an object’s mass
and velocity. Objects with relatively greater mass and velocity are more resistant to envi-
ronmental changes than objects with relatively less mass and velocity. For example, a bus
traveling at 50 mph has more momentum (and is more difficult to stop) than a small car
traveling at the same speed. Nevin (1996) theorized that operant behavior possesses a
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

momentum that is analogous to physical momentum. In behavioral momentum, rate of

responding is analogous to velocity and persistence of behavior (i.e., resistance to
change) is analogous to mass. As the analogy between physical momentum and behavior
suggests, behavior with a high level of momentum is likely to continue over time. In ap-
plied settings (e.g., classrooms, group homes, preschools) where the possibility of non-
compliance is high, techniques based on behavioral momentum can help appropriate be-
havior persist even when competing sources of reinforcement are available in the
environment.
Applied interventions based on the theory of behavioral momentum have focused on
increasing the rate (i.e., velocity) of behavior within a response class. This increase in re-
sponse rate is accomplished by introducing a series of requests with a high probability of
compliance. Changing operant behavior by increasing its velocity (i.e., rate of respond-
ing) has been documented by researchers across a variety of requests, settings, and popu-
lations. For example, Mace et al. (1988) used a high-p request sequence to increase com-
pliance during a showering routine for an individual with severe disabilities. Their
results also indicated that the high-p sequence was more effective than both response
prompts and contingency management in reducing the duration of the shower routine.
Similarly, Sanchez-Fort, Brady, and Davis (1995) increased instances of communication
of two children with moderate to severe disabilities using the high-p procedure.
Although most of the studies that have examined high-p request sequences have fo-
cused on individuals with severe disabilities, work has expanded to more academic set-
tings and tasks. For example, Ardoin, Martens, and Wolfe (1999) used high-p request se-
quences to increase compliance during transition times for second-grade students in a
general education classroom. High-p request sequences have also been embedded into
math worksheets for students with behavior disorders (Belfiore, Lee, Vargas, & Skinner,
1997). These embedded academic high-p requests decreased the latency to initiate
nonpreferred problems.
Overall, the results of individual studies and narrative reviews suggest that high-p re-
quest sequences are an effective method to increase compliance. However, to date there
has been no quantitative synthesis of the high-p literature. In addition to providing an ob-
jective method to evaluate intervention outcomes, quantitative syntheses allow for more
detailed investigation of interactions among key variables (Kavale, 2001). A detailed
analysis of these interactions may be particularly important for an intervention that has
the potential to be applicable across individuals, settings, and behaviors. The purpose of
this meta-analysis is to further examine the effects of high-p request sequences on low-p
behavior across populations, settings, and methodological differences.
 144 LEE

METHODS

Inclusion Criteria and Search Strategy

For a study to be included in the synthesis presented here, its purpose had to pertain to the
use of high-p request sequences in applied settings. More specifically, the intervention
had to consist of presenting a series of brief requests and/or tasks with a high probability
of compliance immediately prior to a request with a low probability of compliance. The
study also had to appear in a refereed journal between the years of 1987 and 2001. This
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

time frame was chosen because the first applied study that examined the use of high-p re-
quest sequences was published in 1987. Finally, data in each study had to be presented
using a single-subject/time-series format. Case studies (AB designs) were excluded from
the review.
The search strategy utilized for this study was a three-stage model identical to that
employed by Schlosser and Lee (2000). First, a computerized search of the literature was
conducted using ERIC and PsycINFO databases with the following key words:
high-probability request/command sequence, high-p requests, interspersed requests, and
behavioral momentum. Second, a hand search was conducted of pertinent journals.1
Finally, an examination of all references of included studies was conducted (i.e., ances-
tral search).

Measure of Intervention Effectiveness

The percentage of nonoverlapping data (PND; Scruggs, Mastropieri, & Casto, 1987a,
1987b) was selected as the metric to assess the effectiveness of high-p request sequences
within each data series. For interventions designed to increase behavior (e. g., compli-
ance) PND is derived by computing the percentage of data that lay above the highest
baseline point. For example, if the highest baseline point in a given series was 15 and the
intervention points were 13, 16, 20, 19, and 21, the PND would be 80%. In the event of a
ceiling effect (e.g., when data were reported as percentages and highest baseline point
was at 100%), PND was calculated using the number of points that were within the high-
est baseline data point (Schlosser & Lee, 2000). Similarly, for interventions designed to
decrease behavior PND is calculated using the lowest baseline point. Scruggs and
Mastropieri (2001) suggested that PNDs above 90 represent very effective interventions,
those between 70 and 90 represent effective interventions, PNDs between 50 and 70 are
questionable, and those below 50 represent ineffective interventions. Although there is
some controversy regarding the PND metric (see Strain, Kohler, & Gresham, 1998), it is
recognized as an efficient method to synthesize data from single-subject research de-
signs (Mathur, Kavale, Quinn, Forness, & Rutherford, 1998).
Several different categories of PND were calculated. Intervention percentage of
nonoverlapping data (IPND) was used to assess the effects of the high-p intervention on
compliance with low-p requests and was calculated using baseline and intervention data
series respectively. Generalization percentage of nonoverlapping data (GPND) was used

1A list of journals hand searched for this synthesis can be obtained from the author.
 BEHAVIORAL MOMENTUM 145

to assess generalization and was calculated using baseline and generalization data series,
respectively. Similarly, maintenance percentage of nonoverlapping data (MPND) was
used to assess maintenance of high-p effects and was calculated using the highest of the
last three points of intervention and maintenance data series.

Coding Categories
All studies were reviewed using the following coding categories:
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

• Classification (mild/moderate mental retardation, severe/profound mental retarda-

tion, multiple disabilities, autism, combined mental retardation and autism, learn-
ing disabilities, behavior disorder, no disability, and other/not specified)
• Setting (group home, institution/residential, family home, segregated classroom/
school, general education/inclusive classroom, community, and other/not speci-
fied)
• Format of intervention (one-to-one, small group, large group, and not specified)
• Intervenor (teachers/instructional assistants, group home staff, experimenters, par-
ents, peers, and other/not specified)
• Dependent variables assessed (compliance to low-p requests, inappropriate behav-
ior, latency to respond, other/not specified)
• Definition of low-p behavior based on: (recommendation of staff only, recommen-
dation and empirical validation, other, and not specified)
• Percentage of compliance during initial low-p behavior assessment
• Category of low-p behavior (vocational, academic, self-care, domestic, communi-
cation/social, other, combination, not specified)
• Basis for development of high-p behavior based on: (recommendation of staff only,
recommendation and empirical validation, other, and not specified)
• Percentage of compliance during initial high-p assessment
• Category of high-p request (topographically related to low-p or unrelated to low-p
request)
• Compliance to high-p sequence assessed during intervention (yes or no)
• Mean percentage of compliance to high-p requests during intervention (estimated
from graphs or taken from text when available)
• Mean number of high-p requests delivered for each trial
• Inter-prompt time between final high-p request and low-p request
• Experimental design (ABA, reversal, multi-treatment, alternating treatments, mul-
tiple-baseline, combination multiple-baseline/reversal; combination alternating
treatments/reversal, other)

RESULTS

Twenty-eight studies met the inclusion criteria for this quantitative analysis. The unit of
analysis for this review was the participant. In the event that multiple measures were used
or data for the participant were collected across settings or behaviors (e.g., a multi-
 146 LEE

TABLE 1
Number of Comparisons, Percentage of Data Series, and Mean IPNDs
Across Participant Classifications

Classifications n % M SD

Severe/profound mental retardation 16 23.50 74.01 36.18

Other/not specified 15 22.10 74.25 27.18
Behavior disorder 10 14.70 89.91 16.42
Mild/moderate mental retardation 10 14.70 64.93 32.63
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

Mental retardation and autism 9 13.20 95.14 11.90

No disability 5 7.40 59.00 40.84
Autism 2 2.90 93.75 8.84
Learning disabilities 1 1.50 — —

Note. IPND = intervention percentage of nonoverlapping data.

ple-baseline design), a mean PND was calculated for that participant. This procedure
yielded 68 participants. Characteristics of participants, interventions, and study designs
were analyzed using one-way analyses of variance (ANOVAs). Main effects were fol-
lowed up using Tukey’s Honestly Significant Difference with an alpha set at .05. Tukey’s
test was selected because it limits comparison-wise Type 1 error to p = .05. Correlational
analyses were also conducted where appropriate.
Interrater agreement for this review was established by having a second coder inde-
pendently code 19% of the data series. The interrater agreement between the author and
the second coder was 97.30%.

Participant and Setting Characteristics

Investigations examining the effects of high-p interventions were conducted mainly with
individuals with severe and profound mental retardation (23.5%; see Table 1). No differ-
ences in IPND were found across classifications of participants, F(7, 56) = 1.49, p = .191.
Twenty-nine percent of the data series involved participants who were of preschool or
kindergarten age (birth–5 years), 30.3% involved children of elementary school age
(5–10 years of age), 24.2% involved children of middle or high school age (10–20 years
of age), and 16.7% involved adults (older than 20). The IPND for each age category (e.g.,
preschool, elementary, middle/high, adult) was subjected to a one-way ANOVA. This
analysis yielded a significant effect of age group, F(3, 58) = 6.06, p = .001. Post hoc tests
revealed higher levels of compliance for the preschool (M = 82.79, SD = 25.33, d =
1.14),2 elementary (M = 87.45, SD = 17.95, d = 1.57), and middle/high school (M =
82.92, SD = 19.76, d = 1.24) groups compared with the adult group (M = 48.39, SD =
39.31).
The majority of interventions were conducted in more segregated settings (27.9% in
segregated classrooms) as opposed to more inclusive settings (general education class-
room = 17.6%, community = 0%; see Table 2). A one-way ANOVA revealed a trend to-

2Standardized mean differences were calculated using pooled standard deviations of the groups being com-

pared (Olejnik & Algina, 2000). Cohen (1988) suggested the following terms for evaluating values of d: small
effect = .20, medium effect = .50, large effect = .80.
 BEHAVIORAL MOMENTUM 147

TABLE 2
Number of Comparisons, Percentage of Data Series, and Mean IPNDs
Across Intervention Settings

Settings n % M SD

Segregated classroom/school 19 27.90 82.36 19.04

Group home 14 20.60 76.33 27.38
General education/inclusive classroom 12 17.60 80.30 28.79
Other/not specified 10 14.70 79.09 29.52
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

Family home 7 10.30 80.12 36.95

Institution/residential 6 8.80 26.47 32.33
Community 0 0 — —

Note. IPND = intervention percentage of nonoverlapping data.

ward differences in compliance among settings, F(5, 58) = 2.29, p = .057. Follow-up
analyses indicated that interventions delivered in segregated classrooms (d = 2.46) and
inclusive classrooms (d = 1.80) were more effective than interventions delivered in resi-
dential placements.

Intervention Characteristics
The majority of intervention sessions (85.3%) were conducted using a one-to-one for-
mat, whereas small (10.3%) and large groups (4.4%) were rarely utilized. No differences
in IPND were found across intervention group formats, F(2, 61) = 1.10, p = .340
(one-to-one M = 76.32, SD = 29.69; small group M = 90.79, SD = 14.75; large group M =
65.00, SD = 26.46).

Low-p request sequences. As shown in Table 3, the majority of low-p requests

were categorized as a combination of requests (30.9%) followed closely by other
(22.1%) and communication requests (20.6%). A one-way ANOVA yielded a trend to-
ward differences among categories of low-p requests, F(6, 57) = 2.24, p = .052. However,
the post hoc analysis failed to reveal differences between specific categories.

TABLE 3
Number of Comparisons, Percentage of Data Series, and Mean IPNDs
Across Category of Low-p Requests

Request n % M SD

Combination 21 30.90 82.41 30.31

Other 15 22.10 75.04 28.72
Communication/social 14 20.60 92.95 11.56
Not specified 7 10.30 73.39 35.28
Academic (e.g., complete math problems) 5 7.40 54.11 19.14
Domestic (e.g., cleaning) 3 4.40 45.78 42.63
Self-care (e.g., showering) 3 4.40 67.97 14.49

Note. IPND = intervention percentage of nonoverlapping data; low-p = low probability.

 148 LEE

Experimenter-delivered consequences (e.g., praise) were delivered for 89.7% of data

series. A difference was observed between participants who were given (M = 78.91, SD =
28.74) and those who were not given (M = 56.61, SD = 22.98) experimenter-delivered
consequences for compliance with low-p requests. However, this difference was not sta-
tistically reliable, F(2, 61) = 1.51, p = .229.

High-p request sequences. High-p requests were most often developed using
recommendations from staff that were followed up by empirical validation (77.9% of
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

participants). For those participants where high-p requests were initially empirically val-
idated, the mean percentage of compliance to high-p requests prior to the intervention
was 84.3% (SD = 9.70). No differences were found for IPND during intervention be-
tween high-p requests that were validated and those that were not validated, F(2, 61) =
.273, p = .762. The high-p sequences utilized for 76.5% of participants were requests that
were unrelated to task completion (e.g., Mace et al., 1988), whereas 20.60% were related
to task completion (e.g., Belfiore et al., 1997). No differences were found between
high-p sequences that were topographically related to the low-p task (M = 75.03, SD =
22.14) and those unrelated to the low-p task (M = 79.52, SD = 28.62), F(1, 61) = .293, p =
.590.
Compliance with high-p requests was monitored throughout intervention phases for
72% of participants resulting in a mean percentage of compliance of 89.50% (SD =
16.57). A moderate correlation was found between compliance with high-p requests dur-
ing intervention phases and IPND (r = .45, p = .002, n = 46). No relationship was found
between compliance with high-p requests during initial assessments and high-p compli-
ance during intervention phases (r = .13, p = .438, n = 39).
For 23.5% of data series, no consequences were delivered by the intervenor for com-
pletion of high-p sequences compared with 73.5% of data series where consequences
were delivered (e.g., verbal praise). An ANOVA revealed differences, F(1, 60) = 4.44, p
= .039, in levels of compliance when intervenors delivered consequences for compliance
with high-p requests (M = 81.36, SD = 28.08) and when they did not (M = 63.84, SD =
27.89; d = .62).
A moderate correlation was found between IPND and time between compliance with
the final high-p request in a sequence and delivery of a low-p request (interprompt time
[IPT]; r = –.28, p = .037, n = 54). To further examine the effects of IPT on compliance,
participants were grouped into three categories based on IPT. Group 1 contained those
participants with an IPT of less than 5 sec (IPND M = 79.26, SD = 28.33). Group 2 con-
tained those participants with IPTs ranging from 5 to 9 sec (IPND M = 89.84, SD =
15.22). Finally, Group 3 contained those participants with IPTs of 10 sec or more (IPND
M = 67.41, SD = 36.55). A trend toward differences was found, F(2, 51) = 3.09, p = .054.
Follow-up analyses yielded differences between Groups 2 and 3, suggesting that the in-
tervention is most effective when the low-p request is delivered within 10 sec of the final
high-p request in a sequence (d = .96).

Design and Measures

Table 4 shows IPND by experimental design. Multiple baseline designs were used most
often (29.4%), followed by a combination of alternating treatments and reversal designs
 BEHAVIORAL MOMENTUM 149

TABLE 4
Number of Comparisons, Percentage of Data Series, and Mean IPNDs
Across Experimental Designs

Designs n % M SD

Multiple baseline/probe 20 29.40 84.19 20.70

Alternating treatments + reversal 13 19.10 92.58 10.16
Reversal 11 16.20 53.45 37.97
Alternating treatments/multielement 9 13.20 63.86 30.61
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

ABA 9 13.20 81.38 24.15

Multiple baseline + reversal 2 2.90 50.00 70.71
Other 4 5.90 87.58 20.63

Note. IPND = intervention percentage of nonoverlapping data.

(19.1%). An ANOVA yielded an effect of design, F(6, 57) = 3.08, p = .011. Follow-up
analyses revealed differences between a combination of alternating treatments/reversal
designs and traditional reversal designs indicating that traditional reversal designs may
be less sensitive to the effects of the high-p intervention (d = 1.72).
The dependent measure utilized most often (70% of participants) was compliance to
low-p requests (IPND M = 80.39, SD = 28.07), followed by a measure of inappropriate
behavior (14.50%; IPND M = 60.08, SD = 38.18), latency to respond to low-p requests
(9.60%; IPND M = 57.00, SD = 29.78), and other (6.00%; IPND M = 86.88, SD = 21.69).
An ANOVA revealed differences in IPND among dependent variables employed, F(3,
79) = 2.91, p = .040. Follow-up analyses, however, failed to reveal any specific differ-
ences between dependent measures.
Procedural integrity data were collected for 76.5% of participants. The mean agree-
ment for procedural integrity reported was 96% (SD = 4.14). Interobserver agreement for
intervention was collected for 100% of data series yielding a mean agreement of 95.9%
(SD = 3.99).

Generalization and Maintenance

Generalization data were collected for 30% of data series. Most generalization locations
were segregated classrooms (37%), followed closely by group homes (32%). The family
home (11%) and inclusive classrooms (16%) were used less frequently as generalization
locations. The mean GPND for the generalization data was 76.13 (SD = 27.69). A posi-
tive correlation between GPND and IPND was found (r = .74, p = .001, n = 17) indicating
that strength of intervention was related to level of generalization. An inferential analysis
could not be conducted on the generalization data because of the relatively few partici-
pants and the high number of possible generalization settings.
Maintenance data were collected for 25% of series. The mean MPND for mainte-
nance data was 71.37 (SD = 36.45), indicating a moderate level of maintenance of treat-
ment effects over time. A moderate correlation between IPND and MPND was found (r =
.41, n = 15). However, that correlation was not reliable (p = .129).
Agreement data were collected for 89% of generalization data series and 60% of
maintenance phase data series with mean agreements of 96.4 (SD = 4.58) and 93.9 (SD =
5.00) for generalization and maintenance, respectively.
 150 LEE

DISCUSSION

The purpose of this article was to synthesize findings from a series of studies that exam-
ined the effects of high-p request sequences as a method to enhance compliance. The
mean IPND across settings, categories of requests, and variations of the high-p proce-
dure of 77.37 (SD = 28.51) indicated that the intervention was effective (Scruggs &
Mastropieri, 2001). However, differential effects of the high-p intervention were found
for age of participants and method of intervention implementation.
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

According to the criteria outlined by Scruggs and Mastropieri (2001), the high-p in-
tervention was effective for preschool, elementary, and middle/high school groups.
However, the intervention was not as effective for adults. There are several factors that
may have contributed to these discrepancies across age groups. First, history of rein-
forcement may have played a role in the persistence of the noncompliant behavior of the
adult group. The adult participants were an average age of 38 years, and the behaviors
targeted for intervention may have been reinforced over a period of many years. The
adults did not begin their respective studies any less compliant than individuals in the
other age groups. However, their behavior was more resistant to change, which may indi-
cate that history of reinforcement may not necessarily increase the rate of responding but
may increase behavioral persistence. Second, the adult participants in these studies pro-
gressed through a school system 20 or more years ago that looked quite different from
schools now in terms of resources and methods to teach positive proactive behavior.
Most of these individuals received programs in group homes or institutional settings
where the protection of laws, such as the Individuals with Disabilities Education Act,
that mandate intensive positive programming do not apply. Clearly, history and schedule
of reinforcement and long-term effects of positive programming are key variables that
must be further examined within the context of increasing compliance.
A second area where differential effects were found was in the way that high-p inter-
ventions were implemented. The high-p request sequence intervention is based on the
premise that increasing the rate of responding within a response class results in enhanced
resistance to change, which allows behavior to persist when disruptions from the envi-
ronment occur (low-p requests). This increase in response rate, and presumably rein-
forcement rate, is accomplished through the use of a series of requests that are likely to
occasion compliance (e.g., high-p requests). Several findings from this quantitative syn-
thesis provide information regarding factors that may affect rate of responding within a
response class and subsequent compliance to low-p requests. First, a relationship was
found between compliance with high-p requests and intervention effectiveness. That is,
for the procedure to work well, the high-p sequence had to occasion compliance through-
out the intervention. Forty percent of data series where the intervention was judged as in-
effective (IPND < 50) had high-p compliance percentages under 80% during interven-
tion phases. Conversely, 90% of data series judged as effective or very effective (IPND >
70) had compliance percentages above 80%.
Second, positive intervenor consequences (e.g., praise) for compliance to high-p re-
quests seemed to enhance the effectiveness of the intervention. Changing the density of
reinforcement associated with a response class through the increased rate of responding
to high-p sequences is a key component to this intervention. The increased level of rein-
 BEHAVIORAL MOMENTUM 151

forcement that results from compliance with high-p requests builds the momentum to a
point whereby responding persists even when changes in the environment, such as the in-
troduction of a low-p request, are evident.
Finally, increases in the time between the last high-p and the low-p request in a series
(IPT) results in decreased compliance to low-p requests. Increasing the IPT may disrupt
the temporal contiguity between high-p and low-p requests and decrease the overall level
of reinforcement during that series of requests (Mace et al., 1988).
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

Implications for Practitioners

Several considerations must be addressed before practitioners can realize the benefits of
high-p request sequences. First, practitioners should be careful to monitor compliance to
high-p requests throughout the intervention. It is possible that some practitioners, after
initially developing a pool of high-p requests, may not continually assess compliance to
those requests over time. In these studies, initial compliance to high-p requests was not
related to compliance to high-p requests during intervention. These results suggest that
continual assessment of high-p request compliance is a key factor in maintaining the in-
tegrity of the intervention. Second, practitioners should provide positive consequences
for compliance with high-p requests. Positive consequences increase the level of rein-
forcement within the response class, building a momentum-like effect that helps main-
tain compliance even when low-p requests are delivered. Finally, practitioners should be
careful to maintain temporal contiguity between high-p and low-p requests. Long delays
(e.g., more than 10 sec) may disrupt the contiguity between the high-p sequence and sub-
sequent low-p request and decrease the effects of the intervention.

Implications for Researchers

Researchers may want to further examine the parameters under which the high-p se-
quence operates effectively. More specifically, an examination of history and schedule of
reinforcement may aid in the development of more effective interventions. For example,
what type of high-p sequences and how much reinforcement is needed to increase com-
pliance for targeted behaviors that have been reinforced for many years? A second area
of research may focus on high-p tasks that are embedded directly into materials. Accord-
ing to these results, embedding high-p sequences into academic tasks was only moder-
ately effective. Embedding high-p tasks into academic materials is advantageous in that
teachers can set up the intervention prior to class, thus leaving more time to teach. No di-
rect consequences (e.g., verbal praise) were delivered by the experimenters for comple-
tion of high-p or low-p tasks in studies that examined embedded high-p requests (e.g.,
Belfiore et al., 1997; Hutchinson & Belfiore, 1998). These researchers hypothesized that
high-p task completion may act as a conditioned reinforcer to help establish a momen-
tum of compliance. However, given that the students in these studies were selected on the
basis of their academic noncompliance, the academically based high-p sequence may not
have been a particularly powerful conditioned reinforcer, which would account for the
modest results. Researchers should examine new ways to determine high-p sequences
and establish conditioned reinforcers to enhance the effects of this promising practice
with academic materials.
 152 LEE

Killu (1999) suggested that high-p request sequences are advantageous in that they
can be used to prevent noncompliant behavior, do not require close physical proximity to
a potentially violent person, and can be implemented by a variety of intervenors across
settings. The results of this meta-analysis confirm those findings by Killu and expand the
knowledge base on high-p request sequences by outlining areas of potential interest for
both practitioners and researchers.
The results of this meta-analysis must be viewed within the context of the limitations
of this review. First, only studies published in peer-reviewed journals were used in the
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

analysis. However, unpublished works (e.g., doctoral dissertations) do not go through a

peer review process beyond an initial advisory committee, which may weaken the level
of confidence associated with that study (Kavale, Hirshoren, & Forness, 1998). In addi-
tion, there is some indication that unpublished studies are eventually published in refer-
eed journals (Mostert, 1996). Second, data for all of the variables examined were not
available for every study and may have biased the results. For example, compliance to
high-p requests during intervention phases and exact IPTs were not given for every study.
Researchers should try to include this potentially important information in future work.
Finally, there is controversy surrounding the adequacy of the PND metric as a method to
synthesize findings from single-subject studies. This controversy often has centered
around the use of one number (the PND) to summarize the complexities of single-subject
data (see Scruggs, Mastropieri, & Casto, 1987b). However, summarizing data across in-
dividuals does provide information about the breadth of studies conducted in a given area
of the literature. Although not a substitute for individual rigorous empirical studies of in-
tervention effectiveness, meta-analyses allow us to examine a variety of variables that
may help practitioners use techniques more effectively and provide heuristics for future
basic and applied research in a given area.

REFERENCES

References marked with an asterisk indicate studies included in the meta-analysis.

*Ardoin, S. P., Martens, B. K., & Wolfe, L. A. (1999). Using high-probability instruction sequences with fad-
ing to increase student compliance during transition. Journal of Applied Behavior Analysis, 32, 339–351.
*Belfiore, P. J., Lee, D. L., Scheeler, M. C., & Klein, D. (2002). Implications of behavioral momentum and aca-
demic achievement for students with behavior disorders: Theory, application, and practice. Psychology in
the Schools, 39, 1–9.
*Belfiore, P. J., Lee, D. L., Vargas, A. U., & Skinner, C. H. (1997). Effects of high-preference single-digit math-
ematics problem completion on multiple-digit mathematics problem performance. Journal of Applied Be-
havior Analysis, 30, 327–330.
Brandon, P. K., & Houlihan, D. (1997). Applying behavioral momentum to practice: An examination of the be-
havioral momentum metaphor. Behavioral Interventions, 12, 113–131.
Cohen, J. (1988). Statistical power analysis of the behavioral sciences (2nd ed.). New York: Academic.
Davis, C. A., & Brady, M. P. (1993). Expanding the utility of behavioral momentum with young children:
Where we’ve been, where we need to go. Journal of Early Intervention, 17, 211–233.
*Davis, C. A., Brady, M. P., Hamilton, R., McEvoy, M. A., & Williams, R. E. (1994). Effects of
high-probability requests on the social interactions of young children with severe disabilities. Journal of Ap-
plied Behavior Analysis, 27, 619–637.
 BEHAVIORAL MOMENTUM 153

*Davis, C. A., Brady, M. P., Williams, R. E., & Hamilton, R. (1992). Effects of high-probability requests on the
acquisition and generalization of responses to requests in young children with behavior disorders. Journal of
Applied Behavior Analysis, 25, 905–916.
*Davis, C. A., & Reichle, J. (1996). Variant and invariant high-probability requests: Increasing appropriate be-
haviors in children with emotional-behavioral disorders. Journal of Applied Behavior Analysis, 29, 471–482.
*Davis, C. A., & Reichle, J. E. (2000). High-probability requests and a preferred item as a distractor: Increasing
successful transitions in children with behavior problems. Education and Treatment of Children, 23,
423–440.
*Davis, C. A., Reichle, J., Southard, K., & Johnson, S. (1998). Teaching children with severe disabilities to uti-
lize nonobligatory conversational opportunities: An application of high-probability requests. Journal of the
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

Association for Persons with Severe Handicaps, 23, 57–68.

*Ducharme, J. M., & Worling, D. E. (1994). Behavioral momentum and stimulus fading in the acquisition and
maintenance of child compliance in the home. Journal of Applied Behavior Analysis, 27, 639–647.
Hagopian, L. P., & Thompson, R. H. (1999). Reinforcement of compliance with respiratory treatment in a child
with cystic fibrosis. Journal of Applied Behavior Analysis, 32, 233–236.
*Harchick, A. E., & Putzier, V. S. (1990). The use of high-probability requests to increase compliance with in-
structions to take medication. Journal for the Association for Persons with Severe Handicaps, 15, 40–43.
*Horner, R. H., Day, H. M., Sprague, J. R., O’Brian, M., & Heathfield, L. T. (1991). Interspersed requests: a
nonaversive procedure for reducing aggression and self-injury during instruction. Journal of Applied Behav-
ior Analysis, 24, 265–278.
Houlihan, D., & Brandon, P. K. (1996). Compliant in a moment: A commentary on Nevin. Journal of Applied
Behavior Analysis, 29, 549–555.
*Houlihan, D., Jacobson, L., & Brandon, P. K. (1994). Replication of a high-probability request sequence with
varied inter-prompt times in a preschool setting. Journal of Applied Behavior Analysis, 27, 737–738.
*Hutchinson, J. M., & Belfiore, P. J. (1998). Adding a sequence of high-preference mathematics problems to
increase low-preference mathematics problems performance. Proven Practice, 1, 12–16.
Kavale, K. A. (2001). Meta-analysis: A primer. Exceptionality, 9, 177–183.
Kavale, K. A., Hirshoren, A., & Forness, S. R. (1998). Meta-analytic validation of the Dunn and Dunn model of
learning-style preferences: A critique of what was Dunn. Learning Disabilities Research & Practice, 13,
75–80.
*Kennedy, C. H., Itkonen, T., & Lindquist, K. (1995). Comparing interspersed requests and social comments as
antecedents for increasing student compliance. Journal of Applied Behavior Analysis, 28, 97–98.
Kern, L., Childs, K. E., Dunlap, G., Clarke, S., & Falk, G. D. (1994). Using assessment-based curricular inter-
vention to improve the classroom behavior of a student with emotional and behavioral challenges. Journal of
Applied Behavior Analysis, 27, 7–19.
Killu, K. (1999). High-probability request research: Moving beyond compliance. Education and Treatment of
Children, 22, 470–494.
*Killu, K., Sainato, D. M., Davis, C. A., Ospelt, H., & Paul, J. N. (1998). Effects of high-probability request se-
quences on preschoolers’compliance and disruptive behavior. Journal of Behavioral Education, 8, 347–368.
*Mace, F. C., & Belfiore, P. J. (1990). Behavioral momentum in the treatment of escape-motivated stereotypy.
Journal of Applied Behavior Analysis, 23, 507–514.
*Mace, F. C., Hock, M. L., Lalli, J. S., West, B. J., Belfiore, P. J., & Brown, D. K. (1988). Behavioral momen-
tum in the treatment of noncompliance. Journal of Applied Behavior Analysis, 21, 123–141.
*Mace, F. C., Mauro, B. C., Boyajian, A. E., & Eckert, T. L. (1997). Effects of reinforcer quality on behavioral
momentum: Coordinated applied and basic research. Journal of Applied Behavior Analysis, 30, 1–20.
Mathur, S. R., Kavale, K. A., Quinn, M. M., Forness, S. R., & Rutherford, R. B. (1998). Social skills interven-
tions with students with emotional and behavioral problems: A quantitative synthesis of single-subject re-
search. Behavioral Disorders, 23, 193–201.
*McComas, J. J., Wacker, D. P., & Cooper, L. J. (1998). Increasing compliance with medical procedures: Ap-
plication of the high-probability request procedure to a toddler. Journal of Applied Behavior Analysis, 31,
287–290.
*McComas, J. J., Wacker, D. P., Cooper, L. J., Peck, S., Golonka, Z., Millard, T., et al. (2000). Effects of the
high-probability request procedure: Patterns of responding to low-probability requests. Journal of Develop-
mental and Physical Disabilities, 12, 157–171.
 154 LEE

Mostert, M. P. (1996). Reporting meta-analyses in learning disabilities. Learning Disabilities Research &
Practice, 11, 2–14.
Nevin, J. (1996). The momentum of compliance. Journal of Applied Behavior Analysis, 29, 535–547.
Nevin, J. A., Mandell, C., & Atak, J. R. (1983). The analysis of behavioral momentum. Journal of the Experi-
mental Analysis of Behavior, 39, 49–59.
Olejnik, S., & Algina, J. (2000). Measures of effect size for comparative studies: Applications, interpretations,
and limitations. Contemporary Educational Psychology, 25, 241–286.
Olmi, D. J., Sevier, R. C., & Nastasi, D. F. (1997). Time-in/time-out as a response to noncompliance and inap-
propriate behavior with children with developmental disabilities: Two case studies. Psychology in the
Schools, 34, 31–39.
Downloaded by [Washington State University Libraries ] at 21:33 18 November 2014

Piazza, C. C., Moes, D. R., & Fisher, W. W. (1996). Differential reinforcement of alternative behavior and de-
mand fading in the treatment of escape-maintained destructive behavior. Journal of Applied Behavior Analy-
sis, 29, 569–572.
*Romano, J. P., & Roll, D. (2000). Expanding the utility of behavioral momentum for youth with developmen-
tal disabilities. Behavioral Interventions, 15, 99–111.
*Rortvedt, A. K., & Miltenberger, R. G. (1994). Analysis of a high-probability instructional sequence and
time-out in the treatment of child noncompliance. Journal of Applied Behavior Analysis, 27, 327–330.
*Sanchez-Fort, M., Brady, M. P., & Davis, C. A. (1995). Using high-probability requests to increase
low-probability communication behavior in young children with severe disabilities. Education and Training
in Mental Retardation and Developmental Disabilities, 30, 151–165.
*Santos, R. M., & Lignugaris/Kraft, B. (1999). The effects of direct questions on preschool children’s re-
sponses to indirect requests. Journal of Behavioral Education, 9, 193–210.
Schlosser, R. W., & Lee, D. L. (2000). Promoting generalization and maintenance in augmentative and alterna-
tive communication: A meta-analysis of 20 years of effectiveness research. Augmentative and Alternative
Communication, 16, 208–226.
Scruggs, T. E., & Mastropieri, M. A. (2001). How to summarize single-participant research: Ideas and applica-
tions. Exceptionality, 9, 227–244.
Scruggs, T. E., Mastropieri, M. A., & Casto, G. (1987a). The quantitative synthesis of single-subject research:
Methodology and validation. Remedial and Special Education, 8(2), 24–33.
Scruggs, T. E., Mastropieri, M. A., & Casto, G. (1987b). Response to Salzberg, Strain, and Baer. Remedial and
Special Education, 8(2), 49–52.
*Singer, G. H. S., Singer, J., & Horner, R. H. (1987). Using pretask requests to increase probability of compli-
ance for students with severe disabilities. The Journal of the Association for Persons with Severe Handicaps,
12, 287–291.
Strain, P. S., Kohler, F. W., & Gresham, F. (1998). Problems in logic and interpretation with quantitative synthe-
ses of single-subject research: Mathur and Colleagues (1998) as a case in point. Behavioral Disorders, 24,
74–85.
Walker, H. M., & Walker, J. E. (1991). Coping with noncompliance in the classroom: A positive approach for
teachers. Austin, TX: PRO-ED.
*Wehby, J. H., & Hollahan, M. S. (2000). Effects of high-probability requests on the latency to initiate aca-
demic tasks. Journal of Applied Behavior Analysis, 33, 259–262.
Westling, D. L., & Fox, L. (2000). Teaching students with severe disabilities (2nd ed.). Upper Saddle River, NJ:
Prentice Hall.
*Zarcone, J. R., Iwata, B. A., Hughes, C. E., & Vollmer, T. R. (1993). Momentum versus extinction effects in
the treatment of self-injurious behavior. Journal of Applied Behavior Analysis, 26, 135–136.
*Zarcone, J. R., Iwata, B. A., Mazaleski, J. I., & Smith, R. G. (1994). Momentum and extinction effects on
self-injurious escape behavior and noncompliance. Journal of Applied Behavior Analysis, 27, 649–658.