DOCUMENT PIMP
ID 100 4,7 PS 007 SS5
AUTHOR Buckholdt, David R.; And OthersTITLE The Effects of Different Peinforcement Systems on
Cooperative Behaviors Exhibited by Children inClassroom Contexts.
SPONS AGENCY National Inst. of Mental Health (OHM, Rockville,Md. Center for Studies of Crime and Delinquency.
PUB DATE Aug 74MOTE 4mp.; Paper presented at the Annual Meeting of the
American Psychological Association (82nd, NewOrleans, Louisiana, August 3C-September 3, 1974)
PPIC? ME-$0.75 HC-$1.85 PLUS POSTAGED?FCPIPTORS *Academic Achievement; Behavior Change; *Contingency
Management; *Elementary School Students; InteractionProcess Analysis; Language Development; LearningTheories; Literature Peviews; Mathematics Education;Peer Relationship; *Peer Teaching; *PreschoolChildren; Peading Development; *Reinforcement;Tutoring
TD*NTTFI?PS *Competition Cooperation
DASTPACTThis paper describes a series of experiments which
indicate how different reinforcement systems affect cooperative,competitive, and individualized learning structures. Following abrief literature review of prior research, the experimentsinvestigated (1) individual reinforcement for peer tutoring and (2)shared-group reinforcement for peer tutoring. They examined theeffects of reinforcement by demonstrating how reward structures caninteract with F,everal independent variables in controlling thecregu.4ncy of cooperative behavior. The studies support the idea thatchil'ren from various socioeconomic classes, 3-to 11-years-old, inclassrooms of 4-17 members, can work together effectively incoopervAve instructional situations involving curriculum materialsouch as mathematics, vocabulary development, and reading. Also, theresults indicate that when appropriate reinforcement is provided forcooperative behavior, helping behaviors as well as studentperformance are increased. Several research topics are suggested forsuture research or 'he levelopment ani evaluation of cooperative goalz;rlic*nreNs. (JH)
US O. Piligli40114/ Of MEAL TNPutati011till lett AM,
411411004M. 1416 tine 0.Mita/Kir_
ve . .tt t! it t st
01 .4 .4 tt
, ao. 41' tit.1 4 -.
THE EFFECTS OF DIFFERENT REINFORCEMENT SYSTEMS ON COOPERATIVEBEHAVIORS EXHIBITED BY CHILDREN IN CLASSROOM CONTEXTS
by
David R. BuckholdtDepartment of sociologyMarquette !JniversityMilwaukee, Wisconsin
John S. WodarskiAssociate ProfessorSchool of Social WorkUniversity of TennesseeKnoxville, Tennessee
and
ReseRrch DirectorIntei7ration Project
Jewish Community Centers AssociationSt. Louis, Missouri
PRrer rre7nnteil t the, Annual Meeting of the American Psychological AsIoclationlovult ';"w LA:ulniana
F'r0.7aratlnn of t:11n WA3 facilitted through fundim.: fromy the !.-ttlonal in,;titute of rental
!'or anA
ABSTRACT
The merits of cooperative, competitive, and individualized
learning structures are reviewed. A series of experiments con-
ducted over several years which indicate how different rein-
forcement systems affect these structures are elaborated. Also
discussed are issues for future research.
The notion of children working cooperatively for purptsee of
instruction is not a new idea. As early as the first century, Quintilian
argued that students could benefit from teaching one another (Gartner,
et al., 1971). Johann Amos Comenius (1592-1670) believed that students
would benefit both by teaching and being taught by other students (Comenius,
1921). In late eighteenth century England, Joseph Lancaster and Andrew
Bell made extensive use cf cooperating student gro...ps, and the idea was
apparently tried for the first time in the United States when a Lancastrian
School was opened in New York City in i806 (Cubberley, 1934).
In recent years, the idea of cooperating children working in in-
structional or problem-solving groups of two or more, has enjoyed somewhat
of a revival in this country. Some of this recent interest in cooperative
groups has undoubtedly been spurred by problems of tight school budgets
and overcrowded classrooms, since student teachers can provide instructional
assistance without extra cost. A more positive stimulus for the new
experimentation, however, has arisen from the belief that group instruction
and problem solving may offer significant advantages for students. A
complete list of potential advantages is beyond the scope and intent of this
paper but we can offer several possibilities (Gartner, et al., 1971; Johnson
and Johnson, 1974).
1. Problem solving and overall performance may be better in
group than in individual contexts.
2. ';coup Interaction nay improve interpersonal and cooperative
skIlin In group members.
3. rAern.2e,...tIve or role tAkIni4 and empathy may he
heiw.htened in arms, r.ontexta.
,ntu-tion fro, 'r noo;erattve 2nW:lem nolvinw with other
atudentn nay redu-e antlety e!aunel t.y ntatun, awe, and back,'round
tik.S1 COPY MIME
2
differences between students and adult teachers. That is
tutors may be able to communicate more readily with children
who have slow learning rates, or when an adult creates anYiety
in a studfynt another child may be a more effective teac:ar%
5. Other children can ',rovide more individualized instruction,
including direct and immediate feedback.
6. Teaching another student may increase understanding as well
as build self-confidence, ego strength, and lead the
acquisition of teachine: behavior, a general skill that can
be very useful in an adult society.
7. The opportunity to teach another student may provide motivation
that was absent when a student was responsible only for himself.
8. Peer tutors may be more patient than teachers with children
who have slow learning rates,
C. Tutoring. may reinforce prior learning, may lead to reformulation,
and to deeper understanding A' the basic structure of the
material.
Few of theme potential positive effects have been directly investigated
in experiTental situations, but there have been a number of evaluative
studies risli,nel to test the effectiveness of cooperative tutoring arranF$)-
mmntr, .:rder cw.ltrolled conditions (Ansrla and Leith, 1969; Cloward, 1967:
7raver and ';torn, 1f70; 7artner, Kohler and Tlesrman, 1971;
ire' 1-P2. 'arril and 'herman, 1972; Barris, - herrnln,
Ard /; -;einver and ',la, 19C,Q; (,ai'nor, And
='; Y,;ir!qr#,Pr ar' i.vy, 1-471; !..!orw..an and Toy, 1T/01 NielArnrvAr,
1i q r p Ind N7Jrro 7711, 1c; ,, Ire?, 1 -)71! :Aylor, IrtwrIgnt, :tryi
BEST C J9 44AILABLE
3
1970s Willis, Crowder, and Morris, 1972: Zach, Horner, and Kaufman, 1969).
In comparison with most other educational research, where different learn-
ing environments or different styles of education maku little or no
difference for the instructional effectiveness the results of cooperative
tutoring is promising (Stephens, 1967). However, it is obvious from the
in^onsistent results that some tutoring arrangements are more effective
than ethers. Investigators have sup ested several reasons for this in-
consistency in results, including differences in the training of group
members, the age and background of the group members, the amount of time
devoted to instruction, and the quality of the instructional materials.
Cne set of potentially powerful factors has generally been neglected,
howover, in the explanations of differential effects, and that is the goal
structure of the learning environment which can function either to facilitate
or hinder cooperative tutoring. Johnson and Johnson (1974) have identified
three goal structures: individualistic, competitive, and cooperative.
IndivlIuallItic structures are pre3ent when the behaviors and accomplishments
of chillren have no consev.nces for anyone but thetzelves. That is, their
performance or attainment (1,)es not affect the goal attainment of other
chillrIn. For example, in a completely individualized instructional program
or tn an indiviluAlly administered reinforcement progran, the success or
fall.ir," of ono has little or ro effect on the evalu,7.tion, gr'a'des,
rewa7-!,, or ' ;t1', 1 or otho :n a conneti tive itructurt the sucr!ons
or fa41,1re of one
II I
1oirly affe7t th chanf, for ;;w7cen9 of other
^_hiHren V*" '701r re.' y1 tt or arothr.r in It fynt-7., of limited
")^ r)r nee 1y 74,1)^,,, the n----)rt,Iritic. for r al attain-,'
r.fn rt nr. p 1:11 "O;41 ettrlirtir,vi
4
revolves around groups rather than individuals. Group members share in
the success or failure of the group's performance. A cooperative goal
m!ructure should theoret'Acally promote more helping behaviors ')etweurl
children and better tutor!.ng. In individualistic systems, the performance
o: others is irrelevant to a student's own goal attainment while in a
competitive system thl success of others threatens one's own achivement.
In both of these structures we would expect a low incidince of cooperative
behavior. In a cooperative structure, however, the success or failure of
indviduals has implications for other students in the cooperative group.
Thus, if rewards are str'ctured that axe contingent upon the group's per-
formance the incldence of cooperative behavior eaould increase with the
proportion of reinforcement that is contingent upon the group's performance.
For our present purposes, one way to conceptualize the goal structure
is to focus on the structure of rewards (Kelly and Thibaut, 1969; Coleman,
19;1). In a comnetitive structure where the rewards are limited and compari-
sons are mr.le between individual children, the success of one individual lg.
gaining maximum rewards reduces the probability of success of others. In an
ihdYilualiltie7. system, judgment is not made on comparative bases and rewards
are lintrIbuted acrording to pernoaal achlevemcnt, possibly judged against
one's 7,-;tt aehieyemont. In a ccoperAtive system, individual rewards are
earnei ;VI 1 f-lnnfluon of the quaLty of croup work. Student tutorin.; itnd
rellt^d renratlye behayirr should be partially dependent on the reward
0,tr'r7t!in of nll'Inrc-n. n or learn_nr 614untion, incroaninz an the proportion
'n-nn-,r.ne 1.,7Andflnt u;:on croup perfornance increa,;eu. Lata nef,n to
" '!On^rn.tIVr, ,t1h17!01-1 Inr^ 'e fr#!quontiy in a f.c-)n.,ratIv.
In! -1-11ffin.tntly innl frPTIont in a rr,np-tit1:n or
r 11
5
indlvidualistic reward structure (Hamblin, Buckholdt, Ferritor, Kozloff,
and Blackwell, 1971; Hamblin, Hathaway, and Wodarski, 1971; Wodarski,
Hamblin, Buckholdt, and Ferritor, 1772; Wodarski, Hamblin, Buckholdt, and
Ferritor, 1973; Wodarski, Hamblin, Buckholdt, and Ferritor, 1974; Johnson
and Johnson, 1974).
To date, there have been only a few experiment:3 that focus on the
reward structure for the group and the tutoring and other cooperative be-
haviors between students. Our purpose here is to review several of these
eyperiments conducted by a gr:vp at CENREL in 5t. Louis over the last few
years, and not only ".o summarize what they tell us about how reward structures
relate to effective cooperation, but also to acquaint us with some of the ways
that cooperative or tutoring programs may be structured. Moreover, the purpose
is to sw7est how these rtwari stvuctures may interact with other variables
such as .liabject e.:-sa, size of classroom, age, and so forth in controlling the
freluency of cooperaV ,e behavior and to review future relevant research
islues concerr.tn4 the use of different reinforcement structures to control
the frr.luency of cooperative behavior exhibited by children in classroom
contexts.
rcr 7.:tor
The ir-t, a pilot r.....udy oy 1:u(*ho1dt and (11Arzblin, et al.,
`r11 r7- t4e1vo firs c".11(Irrn who were from an inner
7=oc.)1 In An areA and beln tryJnir to leArn
thn th;tt of the alrhat.et over P. four week
4/i t:ne eir,r17-ent !)t..h,yr. chIllren ir. their clannen
1 in vr-).1:., n,,r0 of tn twelve had learnd
II 11
I - ,
6
period. In the experiment, the children were randomly divided into three
groups, and all were randomly assigned tutors from among the other first
graders who had been most successAil in learning their symbols. Each day
the child and his tutor were sent to a small room at the opposite end of the
building from the classrooms where they worked with the aid of a Language
Master. Alphabet symbols of the sounds (for example "a" or "th") were
placed in the upper lefthand corner of a Language Master card. The sound
itself plus a word beginning with that sound was recorded on one sound track
of the card. For example, if a card had an "m" on tie upper lefthand corner,
when a child ran the card through the Language Master, he would hear,
monkey, 'tic ".
Tutors worked with their tutee in several ways. Sometimes, they
would play a game were the tutee gave the sound for the first letter and
the tutor took the second letter. At other times, they would simply ask the
tutee to give the sound and then run the card through the Language Master
to check for accuracy. All tutors kept two piles of cards, one which included
the soures still to be learned and the other sounds not yet known. At the
Het
binning of the study, tutors were instructed to work with the children in
the same way regardless of the reinforcement condition. After each of the
scheduled twenty-minute periods, an adult tester would arrive to test the
pupil. The would hold up each of the thirty-three cards and ask the pupil
to ir1entify the r,ound symbolized on the card. If he i0entified the appropriate
mound within ten nsconds, he was riven credit for knowing the sound.
The r* wer c, three rolnforremont condition!;. In ':roue A, where both
tom' 711711 Ar,-1 tr.ie tutor W" rs. rninfnr,!ed aecordln to the tutee's progress in
:01-r!-; 1"0r--n*. ,re thr. retail rorrf,,!tly
il.ntf1,,1 Ill o' tnn nounin th,lt lelrrvid to critrion in earlIer
BEST C&W1 AVAILAUL
7
sessions, and two additional tokens were given for each new sound he had
learned to criterion. Tokens were used to purchase special activities,
such as free classroom time and extra recess time, crackers, sweet dry cereal,
toys, and tickets to play recreational games, such as pinball, target
shooting, and bowling. Also, each day that the student learned two new
sounds, both the pupil and his tutor earned a special prize--a cup of ice
cream, a candy bar, or a popsicle. In Croup B, the tutors and students were
each given eight tokens at the beginning of each of the twenty-minute
sessions which they received recardless of their progress. This group, as
well AS Group A, could trade their tokens for a variety of special privileges
or material items. Croup C received no tokens or other material reinforcers.
Rather, the tutor-student pair simply met every day for their twenty-minute
period.
The summary data for the three groups are given in Figure 1. The
greltest effect was obtained with Group A, who received the strongest rein-
forcers for effective learning and indirectly for effective tutoring. Their
mellan was thirty-three sounds learned to criterion (one hundred percent) by
the end of the twenty-day experiment. (..,rcups b and C, whose reinforcers for
lelrnIn tutorinp, were given non-contingently, did not do as well. In
the median -:oNndr. learned to criterion lc twenty-days W33 twenty-
r,reent); in 7froNp - 13.5 (or 41 percent).
, rt r
A -J-1!117 rro,Pvy with (7-).171rAbln reglilt1 by Pfolffer, :;h a,
'
4:!7'41Iro nt list 1r711 31!) in An er--erir,.nt with 16 white
nolPrl, 73_'; montt,7. -n nituLatIon !;6newnat
8
like that of the previous experiment except the children could work either
individually or cooperatively in pairs at the Language Master. In the A
periods, a male teacher sat as a nonparticipant observer in the room with
the children and with the children and the Language Masters. He did not
attend to or respond to the childi in any way. The purpose for this was
to reduce to zero the reinforcement given by the teacher for tutoring and
learning during the A periods, yet control for the presence of an adult.
During period Ala, however, reinforcement was given for individual vocabulary
gains demonstrated each day in a test after the fifteen-minute learning
period and curing experimental conditions Alb and A2 for mutual vocabulary
gains of cooperating pairs. In the 3 periods, the teacher walked around
the room from one pair of children to another, rraising them and handing
out tokens which were exchangeable for valued reinforcere if he found them
teaching one another but ignoring them if they were not tutoring. Wage
reinforcement was also given in the B periods for mutual vocabulary gains of
cooperating pairs as in condition Alb and A2. In condition C, tokens were
gradually phased-out and the teacher used only praise to reinforce coopera-
tive behavior.
Figures 2 and 3 about here
The data in Figures 2 and 3 indicate that the children tutored each
oth r more and they learned at a faster rate when the teacher intermittently
'RV, them tokens and praise for learning together at the Language Master
than they did lurinp; the A periods when they were simply reinforced for
vocA`Julary ,;:atnn. In other words, the data in Figure 2 suvq:est that these
47:,.111 r,)_ ly orp ftnotn-r unl IntArmittnt
reinfor,-Pnont for tutoring per ne was added to the reinforcement the pains
I
BEST cut MAILABLE
9
received for passing the test. Furthermore, the data indicate that while
the children did not learn from working individually at the Language
Masters, the rate of learning increased substantially during the B period
when they start$,A tutoring one another.
Hence, this experiment agai supports the geleral hypothesis that
effective tutoring and effective learning will increase as the strength of
the rein'orcers for effective cooperation r.r tutoring and effective learn-
inr: increases, However, it represents a nice twist in terms of the design
of tutorin pro rams. There was no stigmatization of the children in this
experimntl the children were simply paired and told to teach one another.
Under the prop]- reinfr:cement conditions they did: to fact, they generally
took turns at being the teacher and the pupil.
The third experiment in this series by Stoddard, Bushell and Hamblin
(Hamblin, et al., 19711 37-39) tested the same basic hypothesis, but the
nituation was somewhat different in that it did not involve Language Masters.
?tither, the 17 suburban preschoolers, ages 35-59 months, durin6 a standard
45-minute study period, could choose in all conditions of a BA B AB AC
exrweri:r.ent-11 series to work on iniividual lessons in one area of the class-
room with one of two te;chers or they would team up in pairs andwcrk in
ero'%:ler area of thn cla-Isroom with a third teacher. Careful records we -e
:opt on the pro,:ress of each individual child from the beginning and a, child
who ;(r.04 a nubj-ct wan asked to tutor another child who OA not. Throughout
the A -r]l 3 conditions, the children had a choice. They would receive
tca7he- :,rd tokens for workinil, either individually or in pairs.
kr; ,,hoy were completing a lesson they were given tokens in
tLe n.:rnor symbo13, words, numbers, etc., learned to t I, 4),1
10
criterion. However, in the B condition, the children who were working
together as tutor-pupil pairs were reinforced differentially. That is,
both received a special token when the tutored pupil answered questions
correctly on the assigned lesson material. The special token was larger and
was made out of more attractive plastic. More importantly, it was worth
four of the regular tokens. Condition C was similar to the B condition
except the attractiveness of the backup reinforcers, the things the children
could buy with their tokens, was increased. Field trips to the local fire
station, police station, pet shop, post office, etc., were added to the
usual purchasables.
Figure 4 about here
Note in Figure 4 that, in general, the children spent more time in
tutor-pupil teams during the B condition when the reinforcement for such
activity was increased over what it was in the A condition. Moreover,
when the children were reinforced still more in Condition C, by being able
to buy more with tokens, they spent still more time tutoring one anol.her.
Finally, test performance increased with tutoring; that is, the mean number
of test items correctly answered by the pupils increased with the time spent
tutoring.
In assigning lessons, the teachers in this experiment attempted to
give each child experience both as apupil and as a tutor. This was not
difficult "or the progress of any given child was somewhat uneven. Every
chill did some things better than others and he could generally tutor in
areas where he had strengths. For example, it was not uncommc-:;1 for one
chill to teach a second his numbers and then the second to teach the first
r-_k!1-7,:. 11-o, th,srr. rir- =ntiinAtr1 fro:n the 5y7,;tem.
T-'irst, the ot7.ervern were continually amazed at the way the children in-
nov,Jel effe-Alve t.e:!(7hin4 te,:hnlcues, 3ond, the childrim,once they got
the "hang of it", seemed to enjoy teaching one another and learning from4,1
BEST COPY AVAILABLE
11
one another. Thus, the situation was unstructured enough to allow the
children a great deal of freedom to develop their own approaches to teach-
ing and learning and that freedom to develop, it seems, was fun.
There are some educational experimentors who may be nervous about
reversal designs such as those used in the two previous experiments and
many believe that the data are confounded since several experimental
conditions were investigated through tine. In the following experiment
conducted durim: an eight week period, which tested the effects of rein-
forcement on tutoring and learning (Haeblin and Hamblin, 1972), an analysis
of variance design was used involving 32 inner city preschoolers, 26 white
and 6 b11,-:1% The experinent also tested the effects of adult versus peer
tutoring.
Figure 5 about here
The resultr; in iti.re 5 indicate that tokens for reading and peer
tutoring increased the number of books the children were able to read to
criterion and that the effects were additive (all interaction terms were
insignIficlnt). The high IQ children se_!med to benefit more since under
the peer tutorine condition they continued to be tutored by adults for 10
minutes ,,;:eh day and they tutored their slower peers for 10 minutes each
dal - - ^n tutorin time wrIn not constant. 3e that as it mrly, the low IQ
wen they receiv--: tokens for readinL; and were tutored by their
pe!rn, dll -0"1.7.1,!er-,_bly better than the hith and medium IQ children when the
latter wpr. 7,1ven token-, for attordin:r, end when they were ju3t tutored for
in min.:ta; hite theee d:.ta again suivp:A tnat reinforr!enent
f,-)r rirrA,-ti7S
rmnt7nnn 4h7.
le!irrlr.7 mar.,^11), inr.rv:t thn effP t'vempn!; of
r ;1-,^1r(-sri4dy on,. of the
1^n?Ilirm )17 1(...,xe nlidren 13 inhioitel when they
4
BEST COPY AVARABIT
12
are taught by what to them are giants. In addition to being smaller, leas
frightening and inhibiting, peer tutors generally use a language that
peers understand. The late Jules Henry, after observing inner city class-
rooms over an extended period, suggested that adult teachers tend Zo talk
over the heads of their children. Children, when tutoring their peers,
problhly do not have that problem. .;one may be clumsy teachers at first
but our experiencs, has been that given practice and reinforcement tor
effective tutor ilv, mont children can become rather good teachers. This
imnortant finding should be considered when educators decide whether or not
to institute peer tutoring. Some children may learn considerably better if
they have the opportunity to learn from peers.
Tuto-ihr. n.nd ,roup Contin- encies
Prnfenbrenner's (197C) Russian studies suggest that when children are
put under group instead of individual contingencies, peer tutoring occurs
natur-.1 y anal oecause of it, academic achievement is improved over what it
might bm were the children put under individual contirwsncies (presumably for
the same reinforcers). ihile L4ronfew:renner cites no experimental data to
demc,h:;trate the point, the ponsibility is intriguing because one of the
prlhlols of rettin,i up the conventional tutoring pro7ram is that it often
reriul a (7or.--.1'!ernle wosint of _1Pneroom mangvement by the teacher in terms
of ttTe ar entrvy involved in the orcani7ation of the academic tasks and the
rOmirli'ltrath of tre reInforrenent condit1')ns. If r,ronfonrenner were correct,
^ rAn-r"Inn -71wht r,y7.101 rv)it of troy e manaaement problems. Three
py7A-17ent7 hrt70 rfle'nt ter.n done w.i.th invent1,7ate thene matters in sore
14-At .
r)wever, 'ref ore reviewina, them, It 18 nenesAary to dinctmn precisely
I
13
To most people, group contingencies define situation where *embers
of a group are reinforced on the basis of the average performance of all
of the group members. Thus, if the members of group averaged 50 percent
on a test, they might receive five tokens each (if 10 were the maximum)
or If they averaged 90 percent on the test, they might receive nine tokens
each. When the performances of all group members are averaged and rein-
fcreed, it will be referred to as an averse mrformance group c_11L.n__Lcrerorlc.
Alternatively, the group may be reinforced on the basis of the high
performances in the group. Thus, if the top three scores on a test by
meabers of the group averaged 90 percent, all members of the group receive
8 tokens. This contingency is referred to as a hituerformance group
contingency. Additionally, a group may be reinforced on the basis of the
low performances in the group. Thus, if the low 3 scores on a test by members
of the group average 50 percent, all members of the group might each receive
5 tokens, or if the bottom 3 performances average 70 percent, hen each
member of the group might receive 7 tokens. This is referred to as low
virgencrfor5 .
These distinctions between average, high and low performance group
contingencies may seem unimportant, but theoretically they should lead, and
empirically they do lead, to quite different behavioral outcomes. In fact,
the purpore of the first experiment by Hathaway and Hamblin (Hamblin,
Hathaway, and .odarski, 1971) was to compare the effects on academic achieve-
ment of the average,hign, and low performance group contingencies with the
effect!: of two other types of individual continoncies. for performance and
for attenance. The experiment involved 5 experimental g,roupe of Inner city
firth ldern, 3 witn 7, 1 with 8, and 1 with 9 members. The groups etch
e7,mnt woe; linder 01.7n tie experilentAl (7onlitions whith wnre e.snii".ned
In ,-nur.01r tAlArv7ed or-ier drinw, their mAthemntirr, reading and
14
periods, which lasted for approximately 35 minutes each day. The teacher
used the regular curricula material provided by the school district and
progreso was Assessed by calculating standarized gain scores (average
progress for the entire class defined the unit) on weekly pre-post teats.
The d'Ata in Figures 6 and 7 indicate that the slow and the gifted
students had differential reactions. In particular, the gifted students
performed best on the high performance group contingency and the slow
students performed best on the low performance group contingency. In fact,
data suggest that the best performances turned in over the entire experiment
were by the slow students under the low performance group contingency. This
occurred in part because under such a contingency, the slow students evident-
ly felt considerable motivatini since their corZormance was so crucial to
the group outcome. The other reason for their rettorkabls progress was that
the sore gifted students spent much of their time tutoring the slower
students. Be that as it say, the second best performance turned in by the
gifted students was under the low performance group contingency and the
data (not shown here because of space limitations) indeed show that on
the average the groups, as a whole, did better under the low performance
group contin4ency than under any other.
Figures 6 and 7 a'potit here
The second experiment, by iodarski, Hamblin, 6uckholdt, and Ferritor
(Wodar5.,.1, et al., 1971), was designed to see if different mixes cf
individ.lal and low performance group contin7,encles night be more effective
in xr,.elratIrg acalenic achievement than was a straight low performance
("7--)." t ..Wnn^f. 7r1n the ware foArp exrerimental coniltion7 one In
which r.hilAren who were on 1 ^O T.serr!ent indlviduid rontini7en^le were paid
BEST COPY PARABLE
15
a play dollar for each problem they themselves got right, one in which they
were on 100 percent low performance group contingency. They were paid a
"dollar" for the average of the bottoms four performances by members of their
gToup, and there were two intermediate conditions where they were paid a
fraction of a "dollar" for their own performance and a fraction of a "dollar"
for the average of the bottom four performances of their group. This
experiment involved fifth grade inner city children, but only during a 25-
minute math period each day. Again the groups were rotated through the
experimental conditions in a different counter balanced order. On alternate
days in all conditions, the children were told that once they had finished
their own work, they could tutor other members of the group if they chose
to. Data ware gathered on the percent of the time the children spent tutor-
in,r, after they had completed their work and the rate of improvement on the
arithmetic testa as calculated for each of the groups for each of the experi-
mental 1 nditions. These data are presented in Figures 8 and 9, rote that
peer tutoring increased markedly as the proportion of group reinforcement
ine.!reased. Also note that in general the acquisition coefficients also in-
crfiAsed as the proportion of group reinforcenent incryased. 34th cf these
relltiohrhIn3 were highly significant statistically. However, further
analysis of the lata not shown here train suists that the mower students
herofited t)re than (lid the more lifted stadentn, that on the avprai;e the
gl,rtf0 st1,!..h4s ttor tho 10r) 1 orepnt low nerforrance
tran they on the Incilvi(luAl rontin7enies. Apt-irently
*J-.1 rIrlens And t t,Itoring -. lower hell'n,A tho 1Parn
thlr .!fly i,aarr,1rj hy j';n: rrr,
rr ,r ".
tPr,lod to do
t6
zure's P an 12 about here
The data are rather clear then. At least under low perfornanc( group
contingencies, children do organila themselves to teach one another, as
Bronfenbrenner suggested they might, and their academic achievement improves
an a resllt.
The third experiment employiry shared group reinforcement (Buckholdt,
et al., 1774) added a dimension of training in skills thought to be useful
for effective cooperation on instructional tasks. The 60 students were
fourth-wraders from schools which serve a poverty-stricken urban neighbor-
hood in A lArge mid -wes tern city. The experiment was a 2 x 2 randomized
desiwn, with the two factors being reinforcement con6ition (non-contingent
reinforcenent vr. shared group contingent reinfwcement) and tutoring (no
trainin7 for tutoring vs. training for tutoring). All students were
initially tented on a measure of reading comprehension and stratified into
low, midlle, And high levels. Ten small groups of six members each were
then cor er! within clartnrooms by randomly assigning two members of each
ability in7P1 to each group, for a total of six members to each group. Small
grou7:!; were tnen randomly assi-r.ed to experimental conditions. The group,
rather than the individuals, served as the unit of analysis.
All grou,h woeeed on a r.IAding ansi7nnlnt for twenty-five minutes a
dry, tree week for four weeks for a 2 minute period. The reading
we!-^ a:!17)t,.1 rolling :r1r1 exerciner wry-n
r,01,10- tn
177nf-""1
4-1tddie , levels 1r.a telr difficulty level reh-,lned
1-mm t.tu4'f. :tu(!entl wer,- -1vcr, Err:Ictico
Ar.--wer whlie -;;;r7-,of! of lien -''! q! ,7.ttenn to
17
to facilit; to either asking for Delp from other children or giving help.
ollowing the reading perio1, eaf,;h student took a 20-iter Nultiplo-choice
compreNension test. No tutoring or other assistance was permitted during
the test.
Dwrore the be5innin7 of the study, the experimenters rest with
telrhers and studentn to identify reinforcers which were desired by the
chil:!rrn and which were ,7,enerally available to schools without additional
conts. A free 1-eriod in the afternoon to sceialize, play cards, and dance
were the ncrA po:ular reinforcers selected. Others included extra tire at
myn and recen:; and opportunities to mpet popular local sport and entertain-
ment firurec.
Children in thn non-continr:ent reinforcement conditions received
reinfor,:enPnt merely for clrtici;at:n4 in the experiment and not for their
thP riadin,7 exernires. Children in the group-contingent
corlition rocetved nointn blned on the averve score of the group. One
ro!rt an -_-rr!(! for each Incrermnt above the standard of 200 so that a
gr,n with nr averai-P of 3 would earn 15 noints. The points were exchanged
on-A A f3r rr,irf.orcern. more points a grow) accunulatt the tore
In A rt.tr,forr..el tr.ey ^ouii pi:rchaf..,e or the longer they could remain
i r..- ar:tIvIty. Tho ^'-.11?-!r0n ir. the non-contin-ent group alpu
ro-rt./r,r! clrformanoe as well an
t- not ..,,;u r.' an nx,.hanwo of rItntn.
,-.11!r-r, 1r "tralrIr.g. In t.;:,o1-17-;,." cor!Ition r-oeivPri about
4W-) A' 1^.04 nir1-^/,'Jrre In tutorin.r,. rnurrr. Inclt,(1(.71
-7 I :yin: In how 4.o rarn-ni ttin t yro.i no ,,d Inr-P
le
and how to provide instruction, feedback, and reinforcement for other
students. All students also had supervised practice in both the role of
tutor and tutee. The children in the "no-training" condition did not,
receive this instruction.
Two dependent variables were used; one was a measure of "reading
comprehension" and the second was a measure of student "cooperative and
tutori - behavior." The mailing comprehension measure was calculated from
the performance of the several groups during the final three days of the
study. The measure of cooperative and tutoring behavior was obtained by
observers usin7 a time-sampling checklist used in a previous study (Wodarki,
et al., 1972).
Table 1 contains the analysis of variance for the reading comprehen-
sion measure. The main effect! for both factors are significant 0 .05
level), with group contingency groups outperforming non-contingency gzoups
and trained-tutor groups cutperforming groups without training. There was
no sirnificant Interaction between the two factors.
Table 2 contains the analysis of variance for the tutoring measure.
Thr rain ilffelt for both factors are eiznificAnt = .05 level). There
wig no r.1,!rifihitnt interfthtion retwnen the tvo factors.
in 2 1.)%t horo
nt0101 ru-..nar170d in tril ru!,7r7rt trot idra that chil(irlin
frhm 7Arl--r,' 'w-en the 047pn of 7 to 11, in c.1asn-
1' hP.n work t-4-rthrr effrct',vrly ir
4. e
19
mathematics, wocabulary development, reading, and so forth, and that they
can serve as teachers for one another. Moreover, the studies indicate that
when aperevriete rel.nforcement is provided for cooperative behavior, help-
my beheviorn as well as student performance can be increased. In more
general terms, we are arguinx that reinforcement strategies can be used tc
cre%te a coo^erative goal structure for dyads or larger groups of children
and that a coocerative goal structure will improve cooperative behavior
amor.1 chilies AA well as rroup academic performance.
The implications of these and similar research findings in the effects
of cooperative stWent groups for the design of instructional systems and
for the eeneral inrrovement of educational services are great. Current
educational research and innovation projects give major attention to the
id-nttflitton and teaching of so-called essential cognitive skills, to
the content and design of curriculum, and to the role and critical skills
of the te,_-!her. The social system of the children is generally either
ne:1.!tPl or viewed an a passive response to either teacher or curricular
cr:ntrol ( eomnetitLve classroom), as an active, but individualistic
r-)le, in cild ,7!-.00nes his indivtdual optionn from available alterna-
ti7,-1 dlaaAn:,n) or develops his own alternative (free schools). Polo
nowlvir, often do not follow prescribed models. 13ecause an
1(%^3 not exnilitly build-in role relationnhip3 between
rsh!l!r,n not rein !alit relationnhinl will develop. Or, becrAwle the
^.;r7- -1_1; 1 iriv11;a1:7 doea not ,.an that children will not compete
wi -)nm 1-, venr pr^ r)rmn to hold down performance. The
r.,+ts , 4.14.""-
p,..0 ^ON,.
;-ernon.1117-d -rft to beat
S 1 -
20
social system is created, and the norms of the system may not support the
formal goals of thl school. In fact, them is evidence to indicate that
when goal structures arm not systematically built into an instructional
enviornmont, vhillron will alort the interaction straterfies they know best,
that 11 arl connitItIve strate;.-les (Johnson and Johnson,
11"4). There is evidence to suport the arFument that individualistic,
co7ttItive nr,,te.71 -verve to limit student academic performance and aspira-
tion-. and to Ae7rs1 natilfaction with school (Coleman, 1968; Johnson and
J1nnri,
ducitional researchers and develorers then should attend to ways of
crestin zo:..Terative interaction between children for purposes of increasing
a^a.tellc rArfornance as well an for improvinz interpersonal communication
snot tinfle:tIon with school. The re-desicn of the structure of rewards
offer-, one route to Attack the problem. There are, of course, other promising
alteirAtive-. for re-focunirz, the goal structure of the student social system,
The Fve'.;tcr.3 of Thlenan (1"/.P) on tetw,..en-nchool competition and of
,tn1 drin (1c/73) on team cometition and academic games are, for
the rvitn-)rn, t-e nnt.,t excitin7 recent proposals on the problem,
In 7 tnin pa-er, the authors would like to suggest several
t7171 feel -!onerv-? hicn 7:-Iority in future rormarch on the
leie:q-,ent -n1 e-.1,!t+lor ,7-)A1 structuron. (urrent research
I .11_,4 1r, 1'i1nAry 1!-.; lt t :!:-Alo. However, art)rer3 to
;o1 Well as
our knowler!,701 and umier-
to bett,,r edlnAtionl syste7,11 for
BF,S1 fOrf
21
1. Do extrinsic reinforcers need to he used to develop cooperative goal
structures? A better question may be what types of rewards can be
used and which ones work best with various classes of students? We may
find, for example, that children with a history of failure in school
nay benefit from extrinsic reinforokirs while children who have had more
sqc.'!en3 or who have acquired more skills will acquire sufficient rein-
for,7,.nent from rarticipation in cooperative groups and from appropriate
lelrninA materials. Additionally, we need to inolate the effective
sechlnims for the delivery of reinforcement to increase cooperative
behaviors, such as tokens provided by adults or by children and praise
from children end/or adults.
2. lnl flwards (1073) have dcne sianificant research on important
prsco!.; varim!les which occur in group structures. However, further
nuly in this area is !ndicated. A children who need the most assitztance
fron othern receive less satinfaction and develop neaative images of
:n what ways do students pressure one another to improve
thetr And in peer influence constructive or destructive in
sart111A: ..)o children eve: attennt to sAbotAge group performance
And as 7a^. avoilel?r Zo cooperative rroups lead to greater satir-
flr:tt . 71tre---,r.al. 1rd intrat.ernonal relattonn?
3. 00%A'.1.4,..1. .,0., -0-:nonilvenesi tl reinforcPlent, a desire to help
a 1,el -;!* a%Ility, And no are needed to
tr!
it t:1,7
:Jo 7o:-It hale th.!,7,1 rA-.1113
r 7:r.*_11n^d ovmr loni
:1P. :
W-4()f tin'?
22
tutoring to indicate whether they continue to exhibit cooperative
behaviors in other contexts at later points in time or whether the
tutor and tutee maintain the academic gains.
5. Are group structures best for all subject areas or are there subjects
which should be handled in another way, with competitive or individual-
istic structures? In other words, we need to isolate the conditions in
which group contingencies work best with what type of curriculum
matevials. Likewise, (in some subjects more than in others) students
may be more effective tutors or may require extensive training in teach-
ing methods to be effective.
6. Is competition helpful between groups or should groups compete against
an abstract standard or their on past performance? Should there be no
competition of any kind? Can competition be eliminated or simply re-
directed?
7. What are the effective components of the tutoring process, the friend-
ship bond, interactional behavior, actual instruction in acquisition of
skills, reinforcement provided by peers and tutors, feedback, and so forth?
A. Very few studies exist which empirically compare the effects of coopera-
tive, competitive, and individualized structures created by various
rinforcenent systems. We need more investigations to provide hard data
to ascertain the effects of these various structur7s on essential depen-
dert variables in education.
Partial answer !A for come of these questions exist in the literature. The
tIT'e in rlre, how,ivor, for larce-scale and systematic research on the effects
and ImT,le.lentation of Fca,J....r;tructures which promote cooperative behaviors
- , nr, ,;:rolter n cial 7.rolr!m to t -,f2Ived tnan
23
invidtous competition, lack of communication wad empathy, and mutual
hostility among people. We can begin seeking the solution by teaching
our own children how to help one another.
BEST COPYAVAILABLE
REFERENCES
Maria, R., L. Birmt, and C. Leith. "Individual versus Cooperative Learning."Educational Research, 11, February 1969, 95-103
Bronfenbrenner, U. The Two Worlds of Childhood: U.S. and U.S.S.R. New YorktRussell Sage Foundation, 1970.
Buckholdt, D. R.. D. E. Ferritor, and S. Tucker. "Effects of Training inTutoring and Shared Group Consequences on Reading Performance andTutoring Behaviors." Paper presented at the American Educational ResearchAssociation Cc -ention, Chicago, April 1974.
Cloward, R. "Studies in Tutoring." The Journal of Experimental Education,36, Fall 1961', 14-25.
Coleman, J. S. "Academic Achievement and the Structure of Competition." InSocialization and Schools, Harvard Educational Review Reprint Series #1,1968, 1-23.
Comenius, J. A. The Great Didactic. London: A. and C. Black, 1921
Cubberly, E. P. Public Education in the United States. New York: Houghton-Mifflin Co.. 1934.
DeVries, D. L. and K. J. Edwards. "Learning Games and Student Teams: TheirEffect on Classroom Process." American Educational Research Journal,1971, 10, 307.318.
Etters, E. "Tutorial Assistance in College Core Courses." The Journal ofEducational Research, 60, May-June 1967, 406-407.
Fraser, C., A. L. Beaman, and R. T. Kelem. "Two Heads are Better than One:Modification o' College Performance by Peer Mm7it:ring." Paper read atWestern Psychological Association Convention, Par-%,J 4,nd, Oregon, 1972.
Frazer, S. and C. Stern. "Learning by Teaching." Reading Teacher, 23,February 1970, 403-405.
Gartner, A., M. Kohler, and F. Riensman. Children Teach Children. New York:Harper and 1,,ow, 1971.
Hamblin, R. L., L. Buckhol-i; D. Ferritor, M. Kozloff, and L. Blackwell. TheHumanization Processes. New York: John Wiley and Sons, 1971.
Hamblin, R. L., C. Hathaway and J. S. Wodarski. "Group Contingencies, PeerTutorirs,,, and Accelerating Academic Achievement." Pp. 41-53 in EugeneRamp and dill .71.otkins (:As.), A New LArection for Education: BehaviorAnalvnin. Lawrence, Kansas: The University of Kansas, department of!!uman J,,v0+10;..nolnt, 1771.
fir Si COPY NOILITLE
Hamblin, J. and R. Hamblin. "On Teaching Disadvantaged Preschoolers toReads A !!uccessful Experiment." American Educstional Research Journal,9, Spring 1972, 209-216.
Harris, V. W. and J. A. Sherman. "Effects of Peer Tutoring and Consequenceson the Math Performance of Elementary Classroom Students." Journal ofApplied 1,havior Analysis, 1972, in press.
Harris, V. W., J. ':;herman, D. Henderson, and M. Harris. "Effects of PeerTutoring on the Spelling Performance of Elementary Classroom Students."A New :;irection for Educations Behavior Analysis. Lawrence, KansassThe University of Kansas, Department of Hunan Development, 1972.
Hassini7pr, J. and M. Via. "How Much Does a Tutor Learn Through TeachingReading." Journal of Secondary Education, 44, January 1969, 42-44.
Johnson, D. d. and R. T. Johnson. "Instructional Goal Structures Cooperative,Competitive, or Individualistic." Review of Education Research, 44, 1974,213-241.
Kelley, H. H. and J. W. Thibaut. "Croup Problem 3olving." In G. Lindzey andE. Aronson (Eds.), The v.andbook of ;;octal Psychology. Reading, Massachu-settst Addison-Wesley Publishing Company, 1969, 4, 1-101.
Lucas, J. A., G. Gaither, and J. Montgomery. "Evaluating a Tutorial ProgramContaining Volunteer Subjects." The Journal of Experimental Education ,
36, Spring 196e, 78-81.
McWhorter, K. and J. Levy. "The Influence of a Tutorial Program UponTutors." Journal of Rending, 14, January 1971, 221-224.
Morgan, R. and T. Toy. "Learning by Teachings A Student-to-Student Compen-satory Tutoring Program in a Rural School System." Psychological Record,20, Spring 1970, 159-169.
Niederneyer, F. C. "Effects of Training on the Instructional behaviors ofStudent Tutors." Journal of Educational Research, 64, November 1970,119-123.
Pfeiffer, C. A. Teaching Preschoolers Complex Skills Using a ProgramasdExchange. Unpublished doctoral dissertation, Washington University, l9' .9.
Pierre, R. M. and 1. Sorrell. "White Tutors for Black Students." JournalCo11, P^rIonnel, 11, ',Lay 19?0, 169-172.
Ramirez, J. 7. "The' Effects of Tutorial Experience on the Problem-3olvingBehavior of 6th traders." California Journal of Educe-ttional :)enearch,22, ""rich 1171, &0-90.
'APIDhen, J. Y. Me Promn, of 'r!hooling. New Yorks Holt, tiirsehart, and
Stoddard, D. Commrative Analysis of Training Programs in behavior Mofi-fication. Unpublished doctoral dissertation, Washington University,1970.
Taylor, R. G., P. Cartwright, and C. Hanson. "Tutorial Program for Fresh-man Engineering Students: Effect on Grades and Attrition." TheJournal of Fxnerimental Education, 38, Spring 1970, 87-92.
Willis, J., J. Crowder, and B. Morris. "A Behavioral Approach to RemedialReading Using Students as Behavioral Engineers." In George Bomb (Ed.)A New iirection for '.Aucltiont Behavior Anal nis. Lawrence, Kansas:The University of Kansas, Jepartment of Human Development, 1972, inpreps.
Wodarski, J. S., R. Hamblin, D. Buckholdt, and D. Ferritor. "Effects ofIndividual and Croup Contingencies on Arithmetic Performance." Paperread at American Educational liesearch Association, New York, February1971.
Wodarnki, J. J., R. L. Hamblin, D. R. Buckholdt, and D. E. Ferritor. "TheEffects of Low Performance Croup and Individual Contingencies on Co-operative BehAviors Exhibited by Fifth Graders." The PsychologicalRecord, 22, 1972, 359-36e.
Wodarski, J. R. L. Hamblin, D. R. Buckholdt, and D. F. Ferritor. "In-dividual Consengences versus Different Shared Consequences Contingenton the Performance of Low-Achieving Croup Kembers." Journal of AppliedSocial Fn./0101°7, 3, 1973, 276-290.
Wooarski, J. R. L. Hamblin, D. R. 3uckholdt, and D. E. Feffitor. "Useof Croup reinforcement In .,chool Social York Practice." The Journal ofSchool :43,11a1 'dork, 1, spring 1974, 26-38.
Zach, L. Y. Frner, and J. Kaufman. "Tutoring in a Slum School." ElementarySchool Jo.ArnAl, 70, October 1969, 20-27.
30
20
1.1111010
So-
.4)- 0o- ;5 pi" , T4.4-7f1
Figure I
Cloup A
Group 11
group C
Bf SI COPY AVARAIILE
. Y 04(
tr'dR. A. r m11011.
10
Days15
Lffe7ts of peer tutorinr, for Iroup A with materialreinfor,7erl for leArnir., Iroun withmateria: nlinfor(7!_,r1 atten!ances anl Iroupwith r.o ratf!rial rt!infor7r1.
20
l 0
Iwo
figure 2
AL I Al t) I I I ;761
8c ri
20
11.1.1MI
dm.
I
I
I
I
I
I
II
I
A2 I 132 C
30 35 40 45 50 55School Days
a.n -:ercent of sc.s?..e ti:-.e spent cooperating at %an P,uage '.'tater eachAfter 1' l'fe Ala 'ware reinforcement wos
' ;en ; qr.-. tti , for 7...Itual . voerttul0.r7r.:! 7-11r7,. I reir.foreers ti in Alb
a.n Ft:. co-)7e atinc, et. :,anruagerein f'-,r-7e-r,; in
- .
10
Ala Aft
Figure 3
61
I
C
2
IMMomirl
111111
I0 25 30 35 40
School Days
45 55
n._1.-r.ner of wor-.13 learnt' to criterion per child each school dly.(Aster I re f r,
60
17-1
. II 0
0
roV
C F.-30
CI
on
I6,*
ror 20
r
VE.
10
BEST COPY AVA:LIGLE
e A B
D
/I'
B
ti9%
c, 'ti
\). e.
fiF
A
. d
.
C
tckli/7/
06
'
I
/
i
A
0
9
\ /%/
=1
...
60 /0 80Day (of Semester)
90
''enn 1.7e seventeen rres7hcolern spent stulyin7 in tutor-punil pnir3 andthe mean nu.-.ber cf tent cornIctiv cornietel by tutore! rum's 1)/d'ty throu,:n scrien exrerir-.ental rerio1.3.vorKel it token f-_)r witntutor eorre-!t,v%ems ni -.I 1 'Ix fat wc.rth our"ti r'; fc,r it.3;:.-r,e l le
trinn were
I . I
In the A conflitionnrerular token for
trokn, in :it themr,u1nr tokens were
7 Wt.: !;11.:11
(Artf...r :;t0.1-!nr.1,
j!
95
the te:vhersntu,lent-
token exch 1:117e
riven tonr to (!xcent
20
5
vt
I-
0 0L
.f)f-a
Cyr,
a.
5
o 0 1,
L
0 4-416- 3
Li.r)
I:
2
Cro V
0
ri5dre 5
titS1 Cal Mk
Preschoolers:
r---1 Low I.Q.
r---1 High and Medium I.Q.
1111.
Tokens for Tokens for Tokens for Token!., forReading Reading Attending Attending
Peer Adult Peer AdultTutoring Tutoring Tutoring Tutoring
The relationshi7 of books read (attendance controlled) to tokensani t..itcrinv, by levels. (After J. A. Hamblin, 1970.)
'1 ,:
'3 --
14 .1 .7
.:3 1 t r
L'
Performance
Group Contin.;ency
High Performance
Group Contingency
Averacle Performance
Group Contingency
Individual Perforriance
Contingency:
771 c
Individual Attendance
Contingency
Average Standardized Gain Scores
O
01
(9
.3
J :3:
.
...
:1.
:1 ...
Low Perforrance
Grou
Contingency
High Performance
Group Contingency
Average Performance
Group Contingency
Individual Performance
Contingency
Individual Attendan-e
Contingency
Aver:.je Standar.11-.ed :ai, .).ores
00
L0
euXsv
)00
fire
83
PO
72
64
56
/18
40
32
24
16
8
Figure 8
Co -T.,,:r:st.1 GroupF.Ypf.ri-a-it11 Group 1
A A Expfrir1-2ntu1 Group 11
a a Ex, cr Group IItx-)Lri;.ental Group IV
0'
° e°,00'
.0/;00 /
1,"°./ /
.00 .33 .6";
Proportion of Group Reinforcement1.00
Pf.rcenta,-,e of cooperative behaicr for each exnerimental groupai-,,ainst the proportion of group reinforcement composing
contingencri.
corrpariion group rrovi''.es a base inc for entire periol of thestu,!:1. .he fcar va:ue3 are for fletoh (onleclitive nev,nentof t.he experi7er.t. p:ettir.,7 of the n-pre_nr; of the! 14 af
*.(n ri sec7ents ir, ri!rr.enta.-?.s Or
. 48
. 140
. 36
. 32
. 28
_ .24.
.20
. 16
0
.12
. 0 Scr-
cz..014
. 00
-.014
-.08
-.12
Figure 9
00
0
O
00
0
0
0
00
.00 .33 .67 1.00
Proportion of Group Reinforcement
:Aefult-311,1r^7. rep,rensinn line for the r:oefficients for the 'whole
group plottei ri,:air-,zt the pron,ortion of r,roup reinforcement
:n ocntrni%, n,! m7o,,ffi,!ten*,1 for ',.he -orrtrisongrGur, for erth ner7rent o' the exTerir..ent
1..,!re... 07. -. . 4.. 0.
TABLE 1
UNIVARIATE ANALYSIS OF VARIANCEOF READING COMPREHENSION MEASURE
Source df MS F P...
Factor A (Reinforcement)1 276.14 14.85* p c: .001
Factor B (Tutoring)1 85.55 4.60* p ..:: .05
A X B1 28.75 1.55 p 2:0.05
Within cells36 18.59
*F-ratio sicylif;cant at least at .05 level
TABLE 2
UNIVARIATE ANALYSIS OF VARIANCEFOR MEASURE OF TUTORING BEHAVIOR
Source df MS F P
Factor A (Reinforcement)1 21,390.62 254.65* p .c: .001
Factor B (Tutoring)1 1,946.02 23.17* p ..001
A X B1 133.23 1.59 p >.05
Within cells 36 84.0
*F-ratio significant at least at .05 level