Professional Documents
Culture Documents
2018 - Deshais - A Comparison of Group Contingencies On Academic Compliance
2018 - Deshais - A Comparison of Group Contingencies On Academic Compliance
SUNGWOO KAHNG
UNIVERSITY OF MISSOURI
We used a reversal design with an embedded multielement design to compare the effects of an
independent group contingency and a randomized dependent group contingency on compliance
with assigned literacy worksheets in a first-grade general education classroom. Nine participants
were selected based on low levels of compliance in baseline or by teacher identification. Results
indicated that both group contingencies increased compliance relative to baseline for the major-
ity of participants. The independent condition produced higher levels of compliance for four
students and the randomized dependent condition produced higher levels of compliance for one
student. For four students, the two group contingencies were equally effective. A preference
assessment indicated that the majority of target students preferred the randomized dependent
condition and the majority of nontarget students preferred the independent condition. A num-
ber of potential explanations for our preference findings, including the possible role of obtained
reinforcement, are discussed.
Key words: classroom, compliance, group contingency, obtained reinforcement, preference
Group contingencies are arrangements in research has been conducted in educational set-
which a consequence is delivered contingent on tings (Barrish, Saunders, & Wolf, 1969;
the behavior of one member, some members, Gresham & Gresham, 1982). Group contin-
or all members of a group (Litow & Pumroy, gencies are well suited to schools because they
1975). Group contingencies have been evalu- allow a single teacher to change the behavior of
ated in a number of settings, such as hospitals a large group of students (Gresham &
(Bowman et al., 2016), residential facilities Gresham, 1982). In addition to being efficient,
(Brown & Redmon, 1990; Frankosky & group contingencies possess other advantages in
Sulzer-Azaroff, 1978), open-pit mines (Fox, educational settings. They enable teachers to
Hopkins, & Anger, 1987), and community- avoid singling out students for individualized
based methadone clinics (Kirby, Kerwin, interventions (Elliot, Turco, & Gresham,
Carpenedo, Rosenwasser, & Gardner, 2008); 1987), utilize the effects of peer influence and
however, the majority of group-contingency attention (Skinner, Cashwell, & Dunn, 1996),
and facilitate the delivery of teacher attention
Meghan A. Deshais, Department of Behavioral Psy- contingent on appropriate behavior (Davis &
chology, Kennedy Krieger Institute; Alyssa B. Fisher,
Department of Behavioral Psychology, Kennedy Krieger Blankenship, 1996).
Institute; SungWoo Kahng, Department of Health Psy- Litow and Pumroy (1975) outlined three
chology, University of Missouri. types of group contingencies: independent,
Meghan A. Deshais is now at the Department of Psy-
chology, University of Florida. interdependent, and dependent. In an indepen-
Correspondence concerning this article should be dent group contingency, the same criteria are
addressed to SungWoo Kahng, University of Missouri, applied to every member of the group and indi-
205 Portland St., Columbia, MO 65211.
E-mail: KahngS@health.missouri.edu vidual members who meet criteria are granted
doi: 10.1002/jaba.505 access to reinforcement (e.g., Brantley &
© 2018 Society for the Experimental Analysis of Behavior
116
GROUP CONTINGENCIES 117
first-grade worksheets such as “circle,” “write,” response was not written within the provided
“underline,” etc. The number of trials (oppor- trial space, or if the response was incompatible
tunities to respond) in each section and the with the instructions at the top of the
total number of trials per worksheet varied section (i.e., the student drew a picture in the
based on the worksheet. The mean number of provided blank instead of writing a word or the
trials per worksheet during the baseline condi- student underlined words instead of circling
tion was 18 trials (range, 6-38) and 14.5 trials them). Our rationale for requiring students to
(range, 3-25) during the first and second base- provide a response that was compatible with
line phases, respectively. The mean number of the instructions was that responding in accor-
trials during the randomized dependent group dance with written or vocal directions is a nec-
contingency condition was 17.3 trials (range, essary skill for success on standardized
10-25). The mean number of trials per work- assessments.
sheet during the independent group contin- A second observer independently collected
gency condition was 14.8 trials (range, 7-26) compliance data for all students during 27% of
and 16.7 trials (range, 12-24) during the first baseline sessions, 22% of randomized depen-
and second phases, respectively. The number of dent sessions, and 22% of independent ses-
trials per worksheet and worksheet difficulty sions. Interobserver agreement was calculated
were not explicitly controlled or manipulated for each worksheet following the conclusion of
by the experimenters. the study using the exact trial-by-trial agree-
Compliance was expressed as a percentage ment method. An agreement was defined as a
and calculated by dividing the number of trials trial in which both observers scored “compli-
the student complied with on their worksheet ance” or both observers scored “noncompli-
by the total number of trials on the worksheet ance.” A disagreement was defined as a trial in
and multiplying the result by 100 (this yielded which the primary observer scored “compli-
a compliance score for a single session). We cal- ance” and the second observer scored “non-
culated percentage compliance after each liter- compliance” or vice versa. The total number of
acy period. Compliance was scored on a trial if agreements was divided by the total number of
the student provided a response within the pro- trials (agreements and disagreements) and mul-
vided trial space and the response was compati- tiplied by 100 to yield an agreement score for
ble with the instructions at the top of the each worksheet. Agreement scores for all work-
section. For example, if the instructions indi- sheets in a given condition were averaged.
cated the student should write a word in the Interobserver agreement (IOA) scores were
fill-in-the-blank to complete the sentence and 97% (range, 75%-100%) for baseline sessions,
the student wrote a word in the provided 99% (range, 88%-100%) for randomized
blank, we scored compliance for that trial. If dependent sessions and 99% (range, 78% to
the instructions indicated that the student 100%) for independent sessions. Only three
should circle all of the misspelled words in a individual worksheet IOA scores were below
sentence and the student circled at least one 80% across all conditions. Two of those scores
word, we scored compliance for that trial. For (77%, 78%) were obtained from a nontarget
compliance to be scored, student responses had student whose pencil marks were very light on
to be written in pencil unless the instructions two of his worksheets, making it difficult for us
for that trial contained the word “color,” in to score his responses. The remaining low score
which case a response written in crayon was (75%) was obtained from a nontarget student
accepted. Noncompliance was scored on a trial if on a worksheet with only four trials (the two
the student failed to provide a response, a observers disagreed on one trial).
120 MEGHAN A. DESHAIS et al.
Our second dependent variable was student response within the provided trial space, the
preference for the group contingency condi- response was compatible with the instructions
tions. During the preference assessment, we at the top of the section, and the response was
brought each student into the hallway and correct. For example, if the instructions indi-
asked the student which condition they pre- cated that the student should circle all of the
ferred (e.g., “Which days did you like best, the misspelled words in a sentence and the student
red days or the blue days?”). We recorded stu- circled all of the misspelled words, we scored
dent responses with paper and pencil. If the that trial as accurate. If the student underlined
student did not indicate a preference (“I liked all of the misspelled words, or circled only
them both the same”), we recorded that the some of the misspelled words, we scored that
student had no stated preference. trial as inaccurate.
Our third dependent variable was target and
nontarget students’ obtained reinforcement
Procedures
during the two group contingency conditions
conducted in the multielement phase of the General procedures. We conducted one ses-
experiment. Obtained reinforcement was calcu- sion per day during a morning literacy period
lated by dividing the number of sessions in in Ms. Lucy’s classroom, in the context of the
which a student earned the reinforcer in a given normal classroom routine. The class was
condition by the total number of sessions the divided into four groups of four to six students,
student was present for in a given condition based on ability level. See Table 1 for student
and multiplying the result by 100 to yield a groupings. Target students are listed by name;
percentage. nontarget students are listed by gender
Additional variables of interest were: (B1 = nontarget boy #1). During the session,
(a) mean compliance for target students and the groups rotated through four classroom sta-
nontarget students in the various conditions tions (a small-group instruction station, a work-
and (b) mean accuracy for the target students sheet station, a reading station, and a centers
and nontarget students in the various condi- station); each group went to each station once.
tions. Mean compliance for target students was Each rotation lasted 9-15 min depending on
calculated by adding all of the target students’ the day; however, on a given day all groups had
compliance scores in a given condition and the same duration of time at each station.
dividing the result by the total number of com- According to Ms. Lucy, 9 to 15 min was suffi-
pliance scores collected for target students in cient time for students to finish the assigned
that condition. The same calculation was com- worksheets. The worksheet station was the sta-
pleted for the nontarget students. We collected tion in which students were expected to com-
and analyzed accuracy data after the completion plete the assigned seatwork.
of the study. We randomly selected a sample of
Table 1
five worksheets per student from each of the Literacy Groupings in Order from Highest (Group 1) to
three conditions and scored each worksheet for Lowest Literacy Skills (Group 4)
accuracy. Accuracy was expressed as a percent-
age and calculated by dividing the number of Group 1 Group 2 Group 3 Group 4
trials to which the student responded correctly B1 G1 G4 G5
by the total number of trials on the worksheet B2 G2 Charlotte Amelia
B3 G3 Ramona Nancy
and multiplying the result by 100 (this yielded B4 B5 Wilbur Laura
an accuracy score for a single session). Accuracy B6 Madeline Peter
George
was scored on a trial if the student provided a
GROUP CONTINGENCIES 121
The worksheet station consisted of one hex- feedback was given to the students regarding
agonal table and chairs. A bowl of crayons and the worksheets and no programmed contingen-
a basket of books were located in the middle of cies were in place during baseline.
the table. Pencils were freely available to stu- Group contingency. The experimenters were
dents throughout the classroom. The teacher in the classroom during all group-contingency
and experimenters did not provide instructions sessions. The two group-contingency condi-
or assistance to students at the worksheet sta- tions were alternated daily, except for one
tion. Students were permitted to whisper qui- instance in which two randomized dependent
etly to one another provided they were sessions were conducted consecutively.
discussing the worksheet or helping each other. Ms. Lucy selected an activity that she called
If students finished early, they were expected to “Dance Breaks” to serve as the reinforcer dur-
read a book quietly. There were no pro- ing both group-contingency conditions. Dance
grammed consequences for problem behavior; Breaks were short (2-5 min) videos from “Just
Ms. Lucy usually delivered corrective verbal Dance Kids” that were projected onto the
feedback to students when they engaged in Smartboard™ in the front of the classroom.
off-task or other minor disruptive behavior. Com- Students stood in front of the Smartboard™
pliance data were collected via permanent product and followed the video or danced on their own.
when students were not in the classroom. Dance Breaks were not available to students at
Baseline. Experimenters were not in the class- any other time during the course of this study.
room during baseline sessions. We created Dance Break tickets by writing stu-
Presession. Ms. Lucy explained the assigned dents’ names on the back of store-bought raffle
worksheet to the class immediately prior to the tickets. During both group contingency condi-
session. She held up the worksheet and read tions, students who earned tickets that day par-
aloud the instructions at the top of each sec- ticipated in the Dance Break. Students without
tion. After explaining every section in the tickets had to remain “in the audience” (seated
worksheet, Ms. Lucy answered all student ques- on the rug) during the Dance Break.
tions. She asked students to confirm they Presession. Procedures were similar to baseline
understood the instructions (e.g., “Give me a with a few exceptions. After Ms. Lucy
thumbs up if you know how to do today’s explained the assigned worksheet, we read a
worksheet.”). She reminded students that she script explaining the group contingency in
could not help them with their worksheets effect that day and answered student questions.
because she would be busy at the small-group Both scripts stated that the students needed to
station and if they needed help they could “ask “finish their packet” and “do their best to
a friend” in their group. Ms. Lucy told the stu- answer each question” (see Supplemental Mate-
dents to put their worksheets under the pencil rial for the daily scripts that we read to the stu-
box located at the worksheet station when they dents). For purposes of discrimination, each
were finished or at the end of the rotation. condition was assigned a color and the two
Session. Ms. Lucy reported that baseline ses- conditions were referred to as “blue days” and
sions were identical to the daily classroom rou- “red days.” When the independent condition
tine as described above in general procedures. was in effect, a single blue folder labeled, “All
Postsession procedures. Following the literacy of Ms. Lucy’s friends” was placed at the work-
period, students left the classroom for lunch sheet station. When the randomized dependent
and Ms. Lucy collected worksheets. We scored condition was in effect, two red folders labeled,
the worksheets, collected compliance data, and “Boys” and “Girls” were placed at the work-
returned the worksheets to the teacher. No sheet station.
122 MEGHAN A. DESHAIS et al.
Session. Procedures were similar to baseline did not receive tickets and we did not reveal
except for the presence of the experimenters the identity of the secret student. The experi-
and the colored folders at the worksheet menter delivered the same feedback as for the
station. independent condition (general, nonspecific
Postsession procedures. Following group con- feedback on performance) to teams that did
tingency sessions, we collected the worksheets not earn tickets.
and students left the classroom for lunch. We
scored the worksheets, collected compliance
data, determined which students met criteria, RESULTS
and created tickets for those students. When Figure 1 displays graphs for the four target
the students returned from lunch, we delivered students for whom both group contingency
the tickets and the Dance Break was conditions were equally effective at increasing
conducted. compliance. Charlotte’s compliance was vari-
During the independent condition, we able in the first baseline phase. High and stable
issued a ticket to each student who met the cri- levels of compliance were observed in both
terion (100% compliance with assigned work- group contingency conditions. Compliance
sheet). The experimenter called each student to decreased and variability increased during the
the front of the classroom to receive his or her reversal to baseline. Charlotte’s compliance was
ticket. The student then walked to the dance high and stable during the best-treatment
floor, deposited the ticket into a jar, and waited phase. Laura’s compliance was low during the
for the Dance Break to start. Students who did first baseline phase. During all subsequent
not meet criteria did not receive tickets. After phases, including the second baseline phase,
all of the tickets were delivered the experi- Laura’s compliance was high and stable. Ramo-
menter gave general, nonspecific feedback to all na’s compliance was low and variable in the ini-
of the students who did not earn tickets tial baseline phase. Increased compliance was
(i.e., “many students forgot to circle the mis- observed in both group contingency conditions
spelled words today in the second section”). relative to baseline. Compliance decreased dur-
During the randomized-dependent condi- ing the reversal to baseline but did not reach
tion, we issued tickets to the girls’ team and levels as low as the first baseline phase. Ramo-
the boys’ team contingent on the performance na’s compliance was high and stable during the
of “secret students” (one girl and one boy). We best-treatment phase. Wilbur’s compliance was
selected secret students by randomly choosing low during the first baseline phase. Compliance
one worksheet from the girls’ folder and one increased slightly with the introduction of the
worksheet from the boys’ folder. Our selection group contingencies, although a high degree of
was truly random; any student’s worksheet variability was observed in both conditions.
could be selected, even if they had been chosen During the reversal to baseline, Wilbur’s com-
as the secret student in a previous session. If pliance decreased to levels lower than the first
the secret student met the criterion (100% baseline. Compliance increased during the best-
compliance with assigned worksheet), every treatment phase to levels similar to those in the
member of the team received a ticket. When multielement phase.
secret students met the criterion, the experi- Figure 2 displays graphs for the four target
menter called the secret student to the front of students for whom the independent group con-
the classroom and let the student pass out tingency produced higher levels of compliance
tickets to his or her teammates. If the secret than the randomized-dependent group contin-
student did not meet the criterion, the team gency. Amelia’s compliance was low and
GROUP CONTINGENCIES 123
100 100
80 80
60 60
PERCENTAGE COMPLIANCE
Baseline
40 40 Dependent
Independent
20 20
CHARLOTTE LAURA
0 0
5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45
100 100
80 80
60 60
40 40
20 20
RAMONA WILBUR
0 0
5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45
SESSIONS
Figure 1. Percentage compliance across daily sessions for the four target students for whom both group contingency
conditions were equally effective at increasing compliance.
100 100
80 80
60 60
PERCENTAGE COMPLIANCE
Baseline
40 40 Dependent
Independent
20 20
AMELIA GEORGE
0 0
5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45
100 100
80 80
60 60
40 40
20 20
MADELINE NANCY
0 0
5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45
SESSIONS
Figure 2. Percentage compliance across daily sessions for the four target students for whom the independent group
contingency condition was the most effective condition at increasing compliance.
variable during the first baseline phase. During Lower, variable levels of compliance were
the multielement phase, Amelia’s compliance recaptured following the reversal to baseline.
increased in both group contingency conditions Compliance was high and stable during the
compared to baseline, with higher, more stable best-treatment phase, replicating Amelia’s
levels observed in the independent condition. results from the independent condition of the
124 MEGHAN A. DESHAIS et al.
100
PERCENTAGE COMPLIANCE
80
60
40
20
PETER
0
5 10 15 20 25 30 35 40 45
SESSIONS
Figure 3. Percentage compliance across daily sessions for Peter, the only target student for whom the randomized
dependent group contingency condition was superior at increasing compliance.
multielement phase. George engaged in variable decreased during the reversal to baseline. The
levels of compliance during the initial baseline effects of the independent condition were repli-
phase, although an increasing trend was cated with Nancy in the best-treatment phase.
observed prior to the introduction of the multi- Figure 3 displays the results for Peter, the
element phase. George’s compliance was highly only student for whom the randomized-
variable in both group-contingency conditions, dependent group contingency produced higher
although slightly higher levels of compliance levels of compliance than the independent
were observed in the independent condition. group contingency. Peter’s compliance was low
Upon reversing to baseline, compliance during the initial baseline phase. During the
decreased to levels lower than those observed in multielement phase, Peter’s compliance
the first baseline. Finally, in the best-treatment increased relative to baseline, with slightly
phase, George’s compliance was high and sta- higher levels of compliance observed in the ran-
ble. Madeline’s compliance was low and highly domized dependent condition. During the
variable during the initial baseline phase. Com- reversal to baseline, compliance decreased
pliance increased slowly following the introduc- slightly. Finally, in the best-treatment condi-
tion of the group contingency conditions but tion, Peter’s compliance increased initially but
variability was observed in both conditions. a decreasing trend was observed towards the
Madeline’s compliance was somewhat higher in end of the study.
the independent condition but absences from Table 2 displays group means for target stu-
school resulted in fewer data points in that con- dents and nontarget students during each con-
dition. Upon reversing to baseline, compliance dition. Target students’ mean compliance was
decreased. During the best-treatment phase substantially lower than nontarget students dur-
Madeline’s compliance increased slightly com- ing the initial baseline phase. For both groups
pared to the second baseline phase but variabil- of students, compliance increased in both
ity persisted. Nancy’s compliance was low group-contingency conditions relative to base-
during the first baseline phase. Compliance line with slightly higher levels of compliance
increased in both conditions during the multie- observed in the independent condition.
lement phase, although higher and more stable Although target student compliance improved
levels of compliance were observed in the in both group-contingency conditions, they did
independent condition. Nancy’s compliance not attain levels of compliance as high as the
GROUP CONTINGENCIES 125
Table 2
Mean (and Ranges) Compliance in Each Condition
Group contingency
Best
Subjects Baseline I Indep. Random. Dependent Baseline II Indep.
Target students 51% 85% 79% 60% 89%
(n = 9) (0-100%) (0-100%) (5-100%) (0-100%) (0-100%)
Nontarget students 75% 96% 91% 85% 97%
(n = 11) (0-100%) (64-100%) (0-100%) (13-100%) (58-100%)
Table 3 Table 4
Mean Accuracy in Each Condition (Sample of Five Stated Preference for Group Contingency Conditions
Sessions)
Random. No stated
Subjects Baseline Indep. Random. Dependent Subjects Indep. Dependent preference
Table 5 Table 6
Obtained Reinforcement in Group Contingency Secret Students Selected for Each Team During
Conditions (%) Randomized Dependent Group Contingency Condition
increasing compliance (average compliance across assigned seatwork. A cost–benefit analysis might
the two conditions was within 5%). The results be beneficial to educators who seek to improve
for these four participants are consistent with the academic performance but are limited by time
results of Lynch et al. (2009) and Speltz constraints.
et al. (1982), who reported that independent Despite the importance of increasing on-task
and randomized-dependent group contingencies behavior or compliance with assigned tasks
were equally effective at increasing compliance (Williamson et al., 2009), improvements in
with assigned seatwork or homework. For five tar- these behaviors do not necessarily result in
get students, slightly higher levels of compliance increased academic productivity (Klein, 1979).
were observed in one of the group contingencies. Previous research on the effects of group contin-
For four students (Amelia, George, Madeline, and gences on both compliance and accuracy has
Nancy), higher levels of compliance were been limited (Lynch et al., 2009; Reinhardt,
observed in the independent condition and for Theodore, Bray, & Kehle, 2009). One possible
one student (Peter), higher levels were observed in reason for this gap in existing literature is the
the randomized-dependent condition. Although issue of motivational deficits and skill deficits.
we were primarily interested in the effects of two Axelrod and Greer (1994) note that the use of
group contingencies on the performance of target group contingencies to address motivational def-
students, the performance of nontarget students icits is acceptable but it is not appropriate to
also improved during the group contingencies. apply group contingencies to academic behaviors
This is especially noteworthy because the nontar- when some students lack the necessary skills to
get students had higher levels of compliance than meet the expectations outlined by the contin-
target students in the initial baseline phase and gencies. Our pre-experimental observations in
therefore, less opportunity for improvement. Ms. Lucy’s classroom suggested that many of
Although only one target student performed the students probably lacked the necessary skills
better in the randomized-dependent group con- to complete literacy worksheets at 100% accu-
tingency compared to the independent, the racy, which is why we focused on compliance
randomized-dependent contingency did produce rather than accuracy. When we explained the
higher levels of compliance relative to baseline group contingency conditions to the students
for the majority of students. Williamson prior to each session, we were careful to never
et al. (2009) noted that one distinct advantage imply that their worksheets needed to be 100%
of randomized-dependent group contingencies is correct; we stated that they needed to “finish”
that they take substantially less time and effort their worksheet and “try their best to answer
on the part of the implementer than an inde- each question.” One risk presented by this deci-
pendent group contingency. We can report that sion was the possibility that students would
the randomized-dependent group contingency detect the distinction and begin responding hap-
took less time to implement than the indepen- hazardly without regard for accuracy. However,
dent group contingency because two worksheets our post-hoc analysis of accuracy suggested that
can be scored faster than twenty worksheets. this did not occur. Both group contingencies
However, we did not explicitly measure the time produced improvements in accuracy relative to
required to implement the two group contingen- baseline with only a few exceptions.
cies in this study, thus, we cannot draw conclu- Our second objective was to evaluate student
sions about the relative efficiency of these two preference for the two group contingencies.
arrangements. Future researchers might want to Many research studies that have evaluated pref-
consider evaluating the efficiency of group con- erence for group contingencies have used treat-
tingencies aimed at improving compliance with ment acceptability ratings as an indicator of
128 MEGHAN A. DESHAIS et al.
preference or hypothetical vignettes without preference findings. In other words, the fre-
direct exposure to the contingencies (Elliot quency with which students contacted rein-
et al., 1987; Shapiro & Goldberg, 1986; forcement in the two conditions varied, and
Turco & Elliot, 1990). Our preference assess- generally, students preferred the condition in
ment was simple; we simply asked the students which they received reinforcement most fre-
which condition they liked best and recorded quently. If we conceptualize the students’ stated
their answers. There are a number of potential preference as a measure of choice between two
explanations for our preference findings. First, it available alternatives, these findings are consis-
is possible that more students preferred the inde- tent with matching law theory (Baum, 1974;
pendent group contingency condition due to a Herrnstein, 1961). The matching law predicts
recency effect. The preference assessment was that organisms will allocate responding across
conducted prior to the reversal to baseline and two available alternatives according to the rate
the last session of the multielement phase was of reinforcement for each alternative. Admit-
an independent session. However, the difference tedly, a single assessment of preference based
between the number of students who preferred on self-report is not the same as measuring
the independent contingency over the response distribution over an extended period
randomized-dependent group contingency was of time. Nonetheless, the majority of students
minimal (10 students and 8 students respec- preferred the condition in which they obtained
tively), so it seems unlikely that our results were a higher reinforcement rate. To date, no other
a function of the most recent exposure. A sec- studies have examined the role of obtained
ond potential explanation for our preference reinforcement as it relates to student preference
findings is that student literacy skills might have for different group contingencies. It might be
influenced preference. The majority of target interesting for future researchers to systemati-
students preferred the randomized-dependent cally manipulate rates of obtained reinforce-
condition and the majority of nontarget students ment across group contingency conditions and
preferred the independent condition. That is, measure the effects of those manipulations on
the low-performing students preferred that their preference.
access to the reinforcer be contingent on some- Finally, our preference findings might have
one else’s performance and the high-performing been influenced by an unanticipated variable. It
students preferred that their access to the rein- is possible that, for some students, being
forcer be contingent on their own performance. selected as the secret student and “winning”
These results are similar to the preference find- might have had an impact on their reported
ings reported by Speltz et al. (1982). When we preference. Seven students “won” during the
asked the students why they preferred a given course of this study (G1, G4, G5, B2, B5, B6,
condition, they responded by saying things such and Wilbur). Students G5 and B5 were the
as, “on red days you pick a special student” or two students whose preference did not align
“on blue days you don’t actually have to worry with the condition in which they obtained a
about someone else doing a good job, you only higher rate of reinforcement. They both pre-
worry about yourself.” ferred the randomized-dependent group contin-
Our third objective was to examine the rela- gency even though they contacted
tionship between obtained reinforcement and reinforcement more frequently in the indepen-
student preference for the group contingency dent group contingency. Additionally, one of
arrangements. Differential rates of obtained the students (B2) without a reported preference
reinforcement across the two group contingen- for either condition had “won” for his team.
cies is another potential explanation for our He also contacted reinforcement more
GROUP CONTINGENCIES 129
frequently in the independent group contin- forgot to circle the misspelled words today in
gency condition but reported that he liked both the second section”) during the group contin-
the red folder days and blue folder days equally. gencies. It is possible that this feedback
For these three students, the unprogrammed increased the likelihood that students paid
reinforcers (social rewards) that they received closer attention to their work during the group-
from their teammates when they “won” may contingency sessions or when Ms. Lucy
have influenced their preference more than did explained the worksheets before those sessions.
the rates of obtained reinforcement across the Although our feedback may have influenced
two conditions. Thus, it might be possible for responding during the group contingencies rel-
teachers to sway the preference of high- ative to baseline, it should not have affected the
performing students by strategically selecting multielement comparison.
them to “win” when a randomized-dependent A final limitation of this study is that student
group contingency is introduced. compliance in a given session could have been
This study contains a number of limitations influenced by a number of variables that we
that warrant discussion. The first limitation of did not explicitly control. These variables might
this study is that we did not collect treatment account for some of the within-subject variabil-
integrity data on the implementation of the ity we observed. As stated earlier, we did not
group contingencies or interobserver agreement control or manipulate the length and difficulty
data on student preferences. Future researchers of each worksheet. Ms. Lucy followed her pre-
should collect treatment integrity data when established literacy curriculum and selected the
evaluating the effects of group contingencies on worksheets independently, without input from
student’s academic performance. Additionally, the experimenters. That said, we have no evi-
it would be beneficial for future researchers to dence that the difficulty did not change system-
collect interobserver agreement data when mea- atically over time, which presents a threat to
suring student preference for different group internal validity. Additionally, the varying
contingency arrangements. length and difficulty of worksheets across ses-
A second limitation of this study is that the sions might have presented two disadvantages
experimenters were present in the classroom to target students. First, our criterion of 100%
during the group contingency conditions but compliance might have been an unrealistic goal
absent during baseline conditions. We chose to for target students on days in which the work-
be present during the group contingency condi- sheets were particularly long or difficult. Future
tions to ease the burden for the teacher. Our researchers interested in the effects of group
presence in the classroom might have influenced contingencies on academic performance might
students’ compliance during the group contin- consider using flexible criteria for reinforcement
gencies (i.e., observer reactivity). However, we based on students’ performance in baseline or
had a prior history with the students in this on standardized assessments rather than a fixed,
classroom. Additionally, the differential effects unchanging criterion for all students. Second,
observed across the two group contingencies for the class was divided into groups according to
many target and nontarget students suggest that literacy skills. Groups 1 and 2 consisted of stu-
the contingencies, rather than the mere presence dents with the strongest literacy skills and each
of the experimenters, exerted control over stu- contained only one target student. Groups
dent responding. 3 and 4 consisted of the students with weaker
A third limitation of the current evaluation literacy skills and both contained at least three
is that we provided general feedback to students target students. These groupings might have
who did not meet criteria (e.g., “many students impacted the students’ ability to help each
130 MEGHAN A. DESHAIS et al.
Behavior Analysis, 20, 215-224. https://doi.org/10. Master list of Dolch sight words. Retrieved from http://
1901/jaba.1987.20-215 bogglesworldesl.com/dolch/lists.htm
Frankosky, R. J., & Sulzer-Azaroff, B. (1978). Individual Packard, R. G. (1970). The control of “classroom atten-
and group contingencies and collateral social behav- tion”: A group contingency for complex behavior.
iors. Behavior Therapy, 9, 313-327. https://doi. Journal of Applied Behavior Analysis, 3, 13-28. https://
org/10.1016/S0005-7894(78)80075-6 doi.org/10.1901/jaba.1970.3-13
Gresham, F. M., & Gresham, G. N. (1982). Interdepen- Reinhardt, D., Theodore, L. A., Bray, M. A., &
dent, dependent, & independent group contingencies Kehle, T. J. (2009). Improving homework accuracy:
for controlling disruptive behavior. Journal of Special Interdependent group contingencies and randomized
Education, 16, 101-110. https://doi.org/10. components. Psychology in the Schools, 46, 471-488.
1177/002246698201600110 https://doi.org/10.1002/pits.20391
Hansen, S. D., & Lignugaris/Kraft, B. (2005). Effects of a Shapiro, E. S., & Goldberg, R. (1986). A comparison of
dependent group contingency on the verbal interac- group contingencies for increasing spelling perfor-
tions of middle school students with emotional distur- mance among sixth grade students. School Psychology
bance. Journal of Behavioral Disorders, 30, 170-184. Review, 15, 546-557. https://eric.ed.gov/?id=
https://doi.org/10.1177/019874290503000204 EJ346440
Heering, P. W., & Wilder, D. A. (2006). The use of Skinner, C. H., Cashwell, C. S., & Dunn, M. S. (1996).
dependent group contingencies to increase on-task Independent and interdependent group contingen-
behavior in two general education classrooms. Educa- cies: Smoothing the rough waters. Special Services in
tion and Treatment of Children, 29, 459-467. http:// the Schools, 12, 61-78. https://doi.org/10.1300/
www.jstor.org/stable/42899896 J008v12n01_04
Herrnstein, R. J. (1961). Relative and absolute strength of Speltz, M. L., Shimamura, J. W., & McReynolds, W. T.
response as a function of frequency of reinforcement. (1982). Procedural variations in group contingencies:
Journal of the Experimental Analysis of Behavior, 4, Effects on children’s academic and social behaviors.
267-272. Journal of Applied Behavior Analysis, 15, 533-544.
Jones, M., Boon, R. T., Fore III, C., & Bender, W. N. https://doi.org/10.1901/jaba.1982.15-533
(2008). “Our mystery hero!” a group contingency Turco, T. L., & Elliot, S. N. (1990). Acceptability and
intervention for reducing verbally disrespectful behav- effectiveness of group contingencies for improving
iors. Learning Disabilities: A Multidisciplinary Journal, spelling achievement. Journal of School Psychology, 28,
15, 61-69. https://eric.ed.gov/?id=EJ817874 27-37. https://doi.org/10.1016/0022-4405(90)
Kirby, K. C., Kerwin, M. E., Carpenedo, C. M., 90034-5
Rosenwasser, B. J., & Gardner, R. S. (2008). Interde- Vidoni, C., & Ward, P. (2009). Effects of fair play
pendent group contingency management for cocaine- instruction on student social skills during a middle
dependent methadone maintenance patients. Journal school sport education unit. Physical Education and
of Applied Behavior Analysis, 41, 579-595. https://doi. Sport Pedagogy, 14, 285-310. https://doi.org/10.
org/10.1901/jaba.2008.41-579 1080/17408980802225818
Klein, R. D. (1979). Modifying academic performance in Williamson, B. D., Campbell-Whatley, G. D., &
the grade school classroom. Progress in Behavior Lo, Y. Y. (2009). Using a random dependent group
Modification, 8, 293-321. https://doi.org/10.1016/ contingency to increase on-task behaviors of high
B978-0-12-535608-4.50012-9 school students with high incidence disabilities. Psy-
Litow, L., & Pumroy, D. K. (1975). A brief review of chology in the Schools, 46, 1074-1083. https://doi.
classroom group-oriented contingencies. Journal of org/10.1002/pits.20445
Applied Behavior Analysis, 8, 341-347. https://doi.
org/10.1901/jaba.1975.8-341 Received January 10, 2017
Little, S. G., Akin-Little, A., & O’Neill, K. (2015). Group Final acceptance November 5, 2017
contingency interventions with children—1980-2010: Action Editor, Claire St. Peter
A meta-analysis. Behavior Modification, 39, 322-341.
https://doi.org/10.1177/0145445514554393
Lynch, A., Theodore, L. A., Bray, M. A., & Kehle, T. J.
(2009). A comparison of group-oriented contingen- SUPPORTING INFORMATION
cies and randomized reinforcers to improve home- Additional Supporting Information may be
work completion and accuracy for students with
disabilities. School Psychology Review, 38, 307-324. found in the online version of this article at the
https://eric.ed.gov/?id=EJ856390 publisher’s website.