9
JOURNAL OF APPLIED BEHAVIOR ANALYSIS 1975, 8, 449-457 NUMBER 4 (WINTER 1975) AN ANALYSIS OF A PERFORMANCE FEEDBACK SYSTEM: THE EFFECTS OF TIMING AND FEEDBACK, PUBLIC POSTING, AND PRAISE UPON ACADEMIC PERFORMANCE AND PEER INTERACTION1 RONALD VAN HOUTEN, SHARON HILL, AND MADELINE PARSONS MOUNT SAINT VINCENT UNIVERSITY, ALEXANDER SCHOOL, HALIFAX, N.S., AND ATLANTIC INSTITUTE OF EDUCATION, HALIFAX, N.S. In the first experiment, after establishing baseline composition rates in each classroom, timing (announcing time limits) and feedback (student self-scoring) were introduced followed by the introduction, removal, and reintroduction of public posting of highest scores. Timing and feedback improved story writing performance and public posting of highest scores improved performance even further in both classrooms. Teacher praise produced further improvement in one classroom but had no effect on performance in the other. Changes in on-task behavior paralleled changes in writing rate. Comments made by children concerning their own work or work of their peers were recorded throughout the experiment. Although the baseline rate of performance comments was almost zero, the introduction of each variable markedly increased the rate of performance comments. In the second experiment, baseline rates on reading and language exercises were estab- lished in a fifth-grade classroom. The entire performance feedback system was introduced on a multiple baseline across the two behaviors and then removed during the final phase of the experiment. Introducing the system improved performance on both tasks. These results further increased the generality of some of the findings of the previous experi- ment and of previous research on the efficacy of the experimental package of timing, feedback, public posting, and praise. DESCRIPTORS: feedback, public posting, time limits, writing rates, academic per- formance, classroom research, self monitoring The development of techniques that acceler- ate progress toward academic objectives is an important concern of educators working with culturally deprived children. One technique that might prove effective is a performance feedback system consisting of timing, feedback, and pub- lic posting of student scores. Van Houten, Morrison, Jarvis, and McDon- ald (1974) improved children's composition 'This research was supported by Grant A 8352 from the National Research Council of Canada to the senior author. We would like to express apprecia- tion to the Halifax School Board and to Keith Sulli- van, Principal of Alexandra School, for their support and assistance. The authors would also like to thank Dr. John Wincze for his careful reading of the man- uscript, and Dr. James Sherman for comments that greatly determined the direction of this research. The second experiment was submitted by the third author as partial fulfillment for the M.Ed. at the Atlantic In- stitute of Education, Halifax, N.S. Reprints may be obtained from Ronald Van Houten, Department of Psychology, Mount Saint Vincent University, Halifax, Nova Scotia, Canada B3M 2J6. writing performance by introducing an experi- mental system consisting of timing, feedback (self-scoring), and public posting of student per- formance. One question raised by their results is how much each factor contributed to the overall effectiveness of the experimental system. Other research suggests that each of those elements can be effective when individually applied to non- academic behaviors. For example, feedback and self-scoring have proven effective in training teachers to use praise (Cooper, Thompson, and Baer, 1970), to train parents to use social con- sequences effectively (Herbert and Baer, 1972), to improve classroom behavior (Broden, Hall, and Mitts, 1971), and in the treatment of pho- bias (Leitenberg, Agras, and Thompson, 1968). Hence, it is reasonable to suspect that feedback might be effective in increasing academic pro- ductivity. Public posting of performance has also proven effective in a wide range of settings. Panyan, Boozer, and Morris (1970) found that publicly 449

how much each factor contributed to the overall be effective when

  • Upload
    ledang

  • View
    215

  • Download
    0

Embed Size (px)

Citation preview

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 1975, 8, 449-457 NUMBER 4 (WINTER 1975)AN ANALYSIS OF A PERFORMANCE FEEDBACK

SYSTEM: THE EFFECTS OF TIMING AND FEEDBACK,PUBLIC POSTING, AND PRAISE UPON ACADEMIC

PERFORMANCE AND PEER INTERACTION1RONALD VAN HOUTEN, SHARON HILL, AND MADELINE PARSONSMOUNT SAINT VINCENT UNIVERSITY, ALEXANDER SCHOOL, HALIFAX, N.S.,

AND ATLANTIC INSTITUTE OF EDUCATION, HALIFAX, N.S.

In the first experiment, after establishing baseline composition rates in each classroom,timing (announcing time limits) and feedback (student self-scoring) were introducedfollowed by the introduction, removal, and reintroduction of public posting of highestscores. Timing and feedback improved story writing performance and public posting ofhighest scores improved performance even further in both classrooms. Teacher praiseproduced further improvement in one classroom but had no effect on performance in theother. Changes in on-task behavior paralleled changes in writing rate. Comments madeby children concerning their own work or work of their peers were recorded throughoutthe experiment. Although the baseline rate of performance comments was almost zero,the introduction of each variable markedly increased the rate of performance comments.In the second experiment, baseline rates on reading and language exercises were estab-lished in a fifth-grade classroom. The entire performance feedback system was introducedon a multiple baseline across the two behaviors and then removed during the final phaseof the experiment. Introducing the system improved performance on both tasks. Theseresults further increased the generality of some of the findings of the previous experi-ment and of previous research on the efficacy of the experimental package of timing,feedback, public posting, and praise.DESCRIPTORS: feedback, public posting, time limits, writing rates, academic per-

formance, classroom research, self monitoring

The development of techniques that acceler-ate progress toward academic objectives is animportant concern of educators working withculturally deprived children. One technique thatmight prove effective is a performance feedbacksystem consisting of timing, feedback, and pub-lic posting of student scores.Van Houten, Morrison, Jarvis, and McDon-

ald (1974) improved children's composition

'This research was supported by Grant A 8352from the National Research Council of Canada tothe senior author. We would like to express apprecia-tion to the Halifax School Board and to Keith Sulli-van, Principal of Alexandra School, for their supportand assistance. The authors would also like to thankDr. John Wincze for his careful reading of the man-uscript, and Dr. James Sherman for comments thatgreatly determined the direction of this research. Thesecond experiment was submitted by the third authoras partial fulfillment for the M.Ed. at the Atlantic In-stitute of Education, Halifax, N.S. Reprints may beobtained from Ronald Van Houten, Department ofPsychology, Mount Saint Vincent University, Halifax,Nova Scotia, Canada B3M 2J6.

writing performance by introducing an experi-mental system consisting of timing, feedback(self-scoring), and public posting of student per-formance. One question raised by their results ishow much each factor contributed to the overalleffectiveness of the experimental system. Otherresearch suggests that each of those elements canbe effective when individually applied to non-academic behaviors. For example, feedback andself-scoring have proven effective in trainingteachers to use praise (Cooper, Thompson, andBaer, 1970), to train parents to use social con-sequences effectively (Herbert and Baer, 1972),to improve classroom behavior (Broden, Hall,and Mitts, 1971), and in the treatment of pho-bias (Leitenberg, Agras, and Thompson, 1968).Hence, it is reasonable to suspect that feedbackmight be effective in increasing academic pro-ductivity.

Public posting of performance has also proveneffective in a wide range of settings. Panyan,Boozer, and Morris (1970) found that publicly

449

RONALD VAN HOUTEN, SHARON HILL, and MADELINE PARSONS

posting the percentage of sessions conducted bythe staff of a training school for the retardedmarkedly improved staff performance. In an-other study, publicly posting the names of chil-dren who returned to class promptly after recessdecreased the number of students late for class(Hall, Cristler, Cranston, and Tucker, 1970).McKenzie and Rushall (1974) found that hav-ing swimmers publicly post their cumulative at-tendance and the number of training unitscompleted increased both attendance and workoutput. The results of these studies suggest thatpublic posting might have also been an effectivevariable in the Van Houten et al. (1974) experi-ment.

The present experiments were designed first,to determine how much each element contrib-uted to the effectiveness of the total package inthe Van Houten et al. (1974) experiment; sec-ond, to document experimentally the casual ob-servation that performance comments increasewith public posting of student performance. Inthe Van Houten et al. (1974) experiment, theexperimenter observed a large increase in socialinteractions involving comments on student per-formance after the experimental package wasintroduced similar to that observed by McKenzieand Rushall (1974) after they introduced publicposting. These casual observations may be ofimportance because peer comments affect class-room behavior (Soloman and Wahler, 1973).

The final purpose of the present research wasto determine whether the variables of timing,feedback, and public posting would be effectiveif applied to another task, such as reading.

EXPERIMENT I

METHOD

Subjects and SettingChildren in two fourth-grade classrooms, lo-

cated in a low socioeconomic multiracial area ofHalifax, Nova Scotia, served as subjects. Theclass sizes were 19 and 20 children for ClassesA and B respectively. The second author taughtlanguage arts to both fourth-grade classes at

separate times of the day. These students weredivided the previous year according to their read-ing ability, with the better readers placed inClass A and the poorer readers in Class B.

General Procedure

Before beginning the experiment, the teacherwas instructed to make a list of compositiontopics of approximately equal difficulty (in herjudgement). Topics were then randomly orderedand one was preassigned for each day of theprogram. Composition writing was then intro-duced into the class schedule every school dayexcept Wednesdays. Each child received a sheetof lined paper at the beginning of compositiontime. The topic was then written on the boardand read aloud by the teacher. The class wasthen told to write as much as they could on thetopic. While the class worked, the teacherhelped children with their spelling. If a child didnot know how to spell a word, the teacherwould write the word on a separate sheet ofpaper on the child's desk. Discipline was main-tained throughout with the use of instructionsand reprimands. At the end of 20 min for ClassA and 10 min for Class B, the teacher asked thestudents to stop writing and collected the papers.The teacher then moved on to the next topic.

MeasuresWriting rate. All papers were scored each day

by a research assistant who counted the numberof words in each composition. The number ofwords written was converted into a rate measureby dividing by the total available writing time.Nonsensical or repetitive sentences were not in-cluded. The definitions of these two classes werethe same as used in the Van Houten et al.(1974) study. The number of words excludedduring all phases was very low. However, repeti-tion did increase for three children in Class Aduring experimental conditions. One reason forthis may have been the length of the composi-tion period (20 min), possibly too long for thesechildren. A second scorer counted words on tworandomly sampled days during each experimen-

450

PERFORMANCE FEEDBACK SYSTEM

tal phase. Each scorer's counts were totalled andcompared (low total score/high total score).Mean rescoring agreement for Class A was97%, with a range of 94% to 100%. The meanrescoring agreement for Class B was 98%, witha range of 96% to 100%.

On-task recording. On-task behavior consistedof children sitting in their seats with their pen-cils in hand, either writing or looking at thematerial they had written. All other behaviors,such as looking around the room, talking to an-other student, or drawing on the paper wereconsidered off-task. Children with their handsraised were always considered on-task.

On-task behavior was observed for each childusing a time-sampling technique. The classroomhad five rows of desks. Each row was sampledfor 1 min. Every 10 sec the observer swept downthe row from front to back recording the num-ber of children in the row and the number ofchildren on-task in that row. After six record-ings, the observer moved on to the next row foranother minute until all five rows had been sam-pled. This procedure was then repeated untilthe end of the writing period.A second observer independently recorded

on-task behavior on two days during each exper-imental phase. Both observers began recordingat the start of the composition period andstopped at the end. Occurrence reliability wascalulated by dividing the total number of timesthat the two observers agreed by the total num-ber of observations. Mean reliability for Class Awas 98.4%, with a range of 93% to 100%.Mean reliability for Class B was 95.5 %, with arange of 93% to 98%.

Comments on performance. Remarks madeabout the composition were recorded by an ob-server seated at the back of the class, just before,during, and after the composition period, bymarking a separate box on the recording sheet.Only clearly audible comments such as: "Howdid you do?", "I beat my score!", "Did you beatyour score?", "How much did you write?", wererecorded. A second observer seated directly be-hind the first observer independently recorded

comments on two days during each phase. Inter-observer reliability was calculated by dividingthe larger count into the smaller count. Mean re-liability for Class A was 91%, with a range of83% to 100%. Mean reliability for Class B was93%, with a range of 79% to 100%.

Experimental DesignA reversal design was employed with both

classrooms in order to assess the relative contri-bution of several different variables on writingperformance. After stable baseline rates wereobtained in both classes, the timing and feed-back phase was introduced. This phase was fol-lowed by the introduction, removal, and reintro-duction of public posting, with timing and feed-back remaining in effect throughout. Finally,praise was added and removed to the previousphase (A-B-B+C-B-B+C-B+C+D-B+C).

Baseline. During this phase, the children werenot told that they were being timed and they re-ceived no feedback on their compositions. Theywere instructed not to write repetitious sentencesand were given several examples of this type ofsentence. The children in Class A were given 20min a day to write and the children in Class Bwere given 10 min. Class B wrote for a shortertime because the teacher felt that 20 min wouldbe too long for this group. At the end of the al-lotted time, the teacher said that it was time tomove on to other activities and the papers werehanded in. This baseline phase was in effect forfive sessions for both classes.

Timing + feedback 1. During the first day ofthis phase, children in both sections were toldthat they were being timed with a stopwatch for20 and 10 min respectively and that this pro-cedure would continue for all further composi-tions. At the end of the allotted time, they wereinstructed to stop writing and pick up a red pen.They were then to count all words and entertheir total at the top of the paper. These countswere not used for the experiment but served toprovide the children with immediate feedback.The children's count was compared for accuracy

RONALD VAN HOUTEN, SHARON HILL, and MADELINE PARSONS

with the experimenter's count by dividing thehigher of the two scores into the lower. Themean reliability of the children's counts through-out the experiment was 87.3% for Class A and81.1% for Class B. The timing and feedbackphase lasted for five sessions.

Timing + feedback + public posting 1. Dur-ing the first day of this phase, each child's namewas written on the board on a chart showing thegreatest number of words each child had writtenon any composition to date. They were then in-structed to see if they could beat their score. Onthe following day, and for each successive dayduring this phase, the scores on the chart werechanged each time a child exceeded his highestprevious score. These changes were always basedupon the experimenter's data, and scores werechanged just before the start of each day's com-position. In addition, a check was placed besidethe child's name whenever his score waschanged. Hence, it was possible for each childto determine how many times he had beaten hisscore by counting the number of checks besidehis name. Just before composition period, thechildren were instructed to look at the chart tosee whether or not they had beaten their score.In all other respects, this phase, which lasted fivesessions, was identical to the timing and feed-back phase that preceded it.

Timing + feedback 2. At the start of thisphase, the chart was erased from the board andthe teacher announced to the class that she wasno longer going to put their scores on the charton the board. In all other respects, this phase,which lasted five sessions, was identical to thefirst timing and feedback phase.

Timing + feedback + public posting 2. Onthe first day of this session, the chart again ap-peared on the board. The students were showntheir highest score to date and were asked to tryand beat this score. This phase was identical tothe first timing, feedback, and public postingphase.

Timing + feedback + public posting +praise. This phase was similar to the previousphase, with the addition of teacher praise. Chil-

dren were praised by the teacher whenever theywere on-task and whenever their score on thechart changed. Some examples of praise state-ments the teacher used were: "Goodyou beat your score.", "Good , yourscore came up.", "If you keep working like thatyou will likely beat your score .'. Thisphase lasted seven days.

Timing + feedback + public posting 3. Dur-ing this final phase, which lasted three sessions,praise was removed.

RESULTSChanges in the mean number of words writ-

ten per minute, the mean percentage of on-taskbehavior, and the total number of performancecomments made during each of the experimen-tal conditions are presented for both Classes Aand B in Figures 1 and 2 respectively.

Writing rate. During the baseline phase,Class A averaged approximately 4.5 words perminute and Class B 1.8. Introduction of thetiming and feedback condition increased themean response rate of Class A to 8.3 words per

PA150

01-10

z~ADZ

I-'au

u ' 70C

WO

Zx

O6 0

CLASS A TIMING l~ ~ +EEDBACK

LIC PUBLIC PKWLIC PUBLICBASELINE POSTING POSTING POSTING POSTING

- ~~~~~~~~~~PRAISE*. b. .

~ - -I -

I r.,V.--5 10 15 20 25

DAILY SESSIONS30 35

Fig. 1. The mean response rate in words per min-ute, the mean percentage of intervals on-task, and thenumber of performance comments in each daily ses-sion across all conditions for Class A. The broken ver-tical lines represent the introduction of the experi-mental conditions. The broken data line represents anunrecorded session.

452

)A)o)o

)o

Dd

PERFORMANCE FEEDBACK SYSTEM

CLASS S TIMINGFEEDBACKI PUBLC AULEC PUBLIC I>UUIC

SASELINE POSTING POSTING POSTING POSTINGPRAISE

1| ' S

esA_ _E~ ~ ~ ~ ~ -q

aI

to0

to

WI-

IU40

:12'0'

i

N,I'.P

I{Lc~~~~~~~~*S..*1I II: I4 1 I Ia-4tv.jvAe~l I

5 10 I5 20 23 -DAILY SESSIONS

Fig. 2. The mean response rate in words per min-ute, the mean percentage of intervals on-task, and thenumber of performance comments in each daily ses-sion across all conditions for Class B. The brokenvertical lines represent the introduction of the experi-mental conditions. The broken data line represents an

unrecorded session.

minute and Class B to 3.5, almost double thebaseline rates.

Introduction of public posting of scores fur-ther increased mean response rates to 10.8 and5.4 words per minute for Class A and B respec-tively. After removal of public posting, mean

rates decreased to 7.9 and 3.4 words per minutefor Class A and B. These rates are very close to

the mean rates during the previous timing andfeedback phase. Reintroduction of public post-

ing increased the mean rate of Class A to 11.0and the mean rate of Class B to 5.0.

Introduction of praise increased the rate ofClass A to 13.0 but had little effect on Class B(mean rate of 5.3 words per minute). Removalof praise decreased rate for Class A to 10.0words per minute and produced no change forClass B (6.1 words per minute).

On-task. The daily mean percentage of on-

task behavior for each class shows changes thatparallel changes in writing rate. This correspon-

dence is more striking for Class B than forClass A.

Performance comments. During the baselinephase, no performance comments were observed

in Class B and only one comment was observedin Class A. After timing and feedback were in-troduced, the number of performance commentsmade in both classes increased markedly. Themean number of comments per day during thisphase was 13.3 for Class A and 8.5 for Class B.A further increase to a mean of 19.8 per day forClass A and a mean of 23.0 per day for Class Bfollowed introduction of public posting ofscores and a decrease to a mean of 11.6 and 5.2respectively followed its removal. Reintroduc-tion of public posting increased the number ofcomments in Class A and B to 17.6 and 12.0respectively. The introduction of praise resultedin a very large increase in comments for Class Aand a smaller increase for Class B (Class A,47.1; Class B, 19.0). Removal of praise resultedin a large decline in comments for Class A anda smaller drop for Class B (Class A, 15.3; ClassB, 12.0). Throughout all phases, almost allcomments were made either before or just afterthe timed writing period. Comments were rarelymade during the timed writing period itself.

DISCUSSIONThe results demonstrated that timing + feed-

back, public posting of scores, and praise eachcontributed somewhat to the effectiveness of thetotal package. Further, increased writing ratewas correlated with increased on-task behaviorand an increased rate of performance commentsmade by the children. Although each of thethree experimental phases improved the per-formance of children in Class A, the praise con-dition did not seem to be effective with Class B.Why teacher praise was effective with one classand not with the other remains an unansweredquestion.

Although the highest rate of performancecomments followed introduction of the publicposting and praise condition, comments did in-crease with introduction of the timing and feed-back phase. This increase may have resultedfrom the emphasis that feedback places on per-formance and the emphasis timing places oncompetition.

453

I

RONALD VAN HOUTEN, SHARON HILL, and MADELINE PARSONS

Story quality was also rated by two indepen-dent observers who were unaware of the purposeof the experiment. All stories were rated on fiveaspects of story quality as described by VanHouten et al. (1974). The results of the primaryrater were in accord with the changes in writingrate, but the reliability rater tended to rate thestories written in all phases to be of similar qual-ity. This failure to find reliable differences instory quality may be related to the children'spoor reading and spelling skills.One question raised by the results of this ex-

periment concerns the effect of performancecomments on writing rate. The correlation be-tween these two variables suggests that a func-tional relationship may exist between commentsand writing performance. Further research isneeded to establish the existence of such a rela-tionship.

EXPERIMENT II

METHOD

Subjects and SettingNineteen children from a fifth-grade class of

31 children taught by the third author, and lo-cated in a low socioeconomic area of Halifax,Nova Scotia, served as subjects.

General ProcedureComprehension exercises. Before beginning

the experiment, the teacher prepared all of thecomprehension questions for each of the storiesleft in the Grade 4 reader, Adventure Awaits,Ginn and Company. A set of 10 questions wasprepared for approximately every eight pages ofthe reader. These assignments were then ran-domly ordered, and preassigned to different ses-sions. Items consisted of how, why, when, where,what, and who questions about the assignedreadings. At the beginning of each reading pe-riod, students were given a stencilled sheet withthe 10 questions about the story and a space towrite in the answer. The children were in-structed to answer each question in sentenceform on their stencilled sheets. Any new words

in the readings were put on the board and theirmeanings explained. If a student encounteredadditional unfamiliar words, the teacher helpedhim with them. Students were not permitted totalk or to help one another. At the end of 25min, all papers were collected and the studentsmoved on to reading workbook exercises. Theassignment was scored at the end of each day' bythe teacher and returned on the following day.Students then went over the questions and cor-rected their work before beginning a new exer-cise. An individual who was not aware of thepurpose of the experiment scored the assign-ments on three randomly sampled days duringeach phase of the experiment. Each scorer'scount was then totalled and compared (low totalscore/high total score). Mean rescoring reliabil-ity was 95.3%, with a range of 90.5% to98.7%.

Multiple word-meaning exercises. Before be-ginning the experiment, the teacher prepared allof the multiple word-meaning exercises for theduration of the experiment. Multiple word-meaning questions consisted of words takenfrom the Grade 4 vocabulary with a list of threedifferent meanings. Below these meanings werethree specific sentences, each representing a dif-ferent use of the word. A space was providedbelow each sentence for the student to write inthe correct definition, depending upon how theword was contextually used. Each definition wasused once in each exercise.

At the beginning of the language period, eachstudent was given a stencilled sheet with 10words requiring three responses each and in-structed to begin their work. Students were alsoinstructed to work independently and not totalk. At the end of 15 min, the work was col-lected by the teacher and the children were in-structed to begin their next assignment. Themarking procedure for multiple word-meaningexercises was the same as for comprehension ex-ercises. An individual who was not aware of thepurpose of the experiment rescored the assign-ments during the three randomly chosen sessionsduring each phase. Each scorer's counts were

454

PERFORMANCE FEEDBACK SYSTEM

then totalled and compared (low total score/high total score). Mean rescoring reliability was

99.2%, with a range of 98.7% to 99.6%.

Experimental Design

A multiple baseline across behaviors, with re-

versal design was employed. After stable base-lines were recorded in both comprehension andmultiple word-meaning exercises, timing, feed-back, praise, and public posting of scores were

introduced, first during the comprehension ex-

ercises and later with the multiple-meaning ex-

ercises. A return to baseline conditions for bothtypes of exercises followed.

Baseline 1. During this phase, the studentswere not told that they were being timed andreceived no immediate feedback on either com-

prehension or multiple word-meaning exercises.Performance feedback system. On the eighth

day of the program, explicit timing, self-scoring,public posting of scores, and praise for improve-ment was introduced during the comprehensionexercises. On the first day of this condition, stu-

dents were given the following instructions:"Starting today, we are going to do somethingdifferent. While you are doing your reading, Iam going to time you with this stopwatch andyou are going to have exactly 25 minutes to doyour work. At the end of this time, you'll have a

chance to score your work and to see how many

problems you have done correctly. You willthen put your pencils in your desks and pick up

your red pens and draw a line under the lastproblem you have completed. You will thengrade it as we go along and count the numberof problems worked correctly and put this score

at the top of your page."The chart was then introduced and its pur-

pose explained. The chart was a 71 by 56 cm

sheet of white bristleboard with the followinginformation printed on it: each student's name;

the days of the week; a space for each of the fol-lowing, each student's daily score, the totaldaily group performance scores, the total weeklygroup performance scores, the total weekly per-

formance scores for each student, and the cumu-

lative number of perfect papers for each stu-dent. This sheet was covered in plastic and alldata were recorded with water-based pens. Itwas placed above the blackboard at the front ofthe class in clear view.

It was explained that the purpose of the chartwas to see if the group as a whole could beattheir highest total daily and weekly score, andwhether individual students could beat theirown total weekly score. The number of perfectpapers was also cumulated, since the fixed num-ber of questions given each day placed a ceilingon how many problems a child could work cor-rectly each day. The scores on the chart werechanged by the teacher at the beginning of eachnew session. Students were praised if they ex-ceeded their highest weekly total and the groupwas praised whenever they exceeded their high-est daily or weekly group score. Perfect scoreswere also pointed out and praise was deliveredby the teacher.

The same basic procedure was followed formultiple word-meaning exercises starting onDay 13 of the program, except that the childrenwere timed for only 15 min.

Baseline 2. At the start of this phase, bothcharts were removed and the teacher announcedthat she was no longer going to put their scoreson the chart. In all other respects this phase,which lasted seven sessions, was identical to thefirst baseline phase.

RESULTSThe mean number of comprehension and

multiple word-meaning problems worked cor-rectly and incorrectly during all experimentalphases are presented in Figure 3. The numberof comprehension and multiple word-meaningproblems worked correctly increased substanti-ally after the performance feedback system wasintroduced; the number of comprehension prob-lems worked correctly nearly doubled and thenumber of multiple word-meaning problemsworked correctly increased by approximately50%. Comprehension errors showed a slightdecrease after the experimental package was in-

455

RONALD VAN HOUTEN, SHARON HILL, and MADELINE PARSONS

troduced, while multiple word-meaning errorsshowed no initial change. However, multipleword-meaning errors increased slightly duringthe last 10 days of the experimental phase. Dur-ing the second baseline phase, the comprehen-sion correct rate decreased to baseline levelswhile errors seemed to increase above baselinelevels. Word-meaning correct rate decreased dur-ing the reversal phase while the error rate re-mained about the same as during the previousphase.

DISCUSSION

The results of this experiment demonstratedthat the experimental package of timing, feed-back, public posting of scores, and praise can ef-fectively improve performances on reading andlanguage exercises. This result increases the gen-

erality of the findings obtained by Van Houtenet 4l. (1974), as well as the results of Experi-ment 1.

It is interesting to note that although the ex-

perimental contingencies were placed only upon

the number of problems worked correctly, com-

prehension errors decreased during the experi-mental phase. Why a similar decrease was not

observed for word-meaning errors remains an

unanswered question. One way in which errors

might be further suppressed would be to post

them publicly along with the correct scores.

Although it was not possible for trained ob-servers to record performance comments madeby students during this experiment because oflogistical problems, the experimenters did makesome interesting observations. First, studentsfrequently praised and applauded students whoexceeded their highest score and second, studentsoften praised poorer students who showed im-provement. These comments tended to be simi-lar to those observed in Experiment I.

GENERAL DISCUSSION

The introduction of timing, feedback, publicposting of scores, and praise increased the per-formance of elementary school children in story

z

us~-

zoO v

B.

Oz- <

BASEUNE I EXPERIMENTAL

5 10 15 20 25DAILY SESSIONS

BASELINE 2

30 35 40

Fig. 3. The mean number of reading comprehen-sion and word-meaning problems worked correctlyand incorrectly in each daily session across all condi-tions for Classes A and B. The broken vertical linesrepresent the introduction of the experimental con-ditions.

writing, reading comprehension, and word-meaning exercises with the effects of those vari-ables showing no sign of weakening over time.These contingencies were effective spanningwork times from 10 to 25 min. Further analysisrevealed that timing and feedback and publicposting of scores each individually contributedto the effectiveness of the experimental packageupon story writing rate while praise was effec-tive with only one of the two classes. These re-sults increase the generality of the previous find-ing that the experimental package of timing,feedback, and public posting of scores improvedstory writing performance (Van Houten et al.,1974). Although these results further demon-strate the utility of this approach, it still remainsto be established whether or not an entire aca-demic program could be based on use of a per-formance feedback system. Only further re-search will determine the limitations of thisapproach.

In both experiments, all children used thesame textual materials. However, there is no rea-son why this approach could not be employed inan "open concept" classroom, where childrenare working on individualized materials. Thismight be accomplished by dividing each indi-vidual assignment into roughly equivalent unitsor lessons that each child could self-score. The

456

PERFORMANCE FEEDBACK SYSTEM 457

number of lessons completed during timed ac-tivity periods could then be posted for each in-dividual and for the class as a team.

Another result of interest was the marked in-crease in the rate of performance comments fol-lowing introduction of the experimental vari-ables. In light of Soloman and Wahler's (1973)finding that peer comments can affect classroombehavior, it is possible that these comments mayhave mediated many of the changes in perform-ance observed. Further research should focusupon the direct manipulation of performancecomments in order to assess their role in theclassroom.

REFERENCES

Broden, M., Hall, R. V., and Mitts, B. The effect ofself-recording on the classroom behavior of twoeighth-grade students. Journal of Applied Behav-ior Analysis, 1971, 4, 191-199.

Cooper, M. L., Thomson, C. L., and Baer, D. M. Theexperimental modification of teacher attendingbehavior. Journal of Applied Behavior Analysis,1970, 3, 153-157.

Hall, R. V., Cristler, C., Cranston, S. S., and Tucker,B. Teachers and parents as researchers using

multiple baseline designs. Journal of AppliedBehavior Analysis, 1970, 3, 247-255.

Herbert, E. W. and Baer, D. M. Training parents asbehavior modifiers: self-recording of contingentattention. Journal of Applied Behavior Anaylsis,1972, 5, 139-149.

Leitenberg, H., Agras, W. S., Thompson, L. E., andWright, D. E. Feedback in Behavior Modifica-tion: an experimental analysis in two phobiccases. Journal of Applied Behavior Analysis,1968, 1, 131-137.

McKenzie, T. L. and Rushall, B. S. Effects of self-recording on attendance and performance in acompetitive swimming training environment.Journal of Applied Behavior Analysis, 1974, 7,199-206.

Panyon, M., Boozer, H., and Morris, N. Feedbackto attendants as a reinforcer for applying operanttechniques. Journal of Applied Behavior Analysis,1970, 3, 1-4.

Soloman, R. W. and Wahler, R. G. Peer reinforce-ment control of classroom problem behavior.Journal of Applied Behavior Analysis, 1973, 6,49-56.

Van Houten, R., Morrison, E., Jarvis, R., and Mc-Donald, M. The effects of explicit timing andfeedback on compositional response rate in ele-mentary school children. Journal of Applied Be-havior Analysis, 1974, 7, 547-555.

Received 18 November 1974.(Final acceptance 21 May 1975.)