118 Journalof Cwurulum and Supervsf tWinter 1991, Vol. 6, No. 2,118-129

TESTING A NEW SUPERVISORY PROCESSFOR IMPROVING INSTRUCTION

GERALD H. ELGARTEN, The City University of New York

The most frequently used technique to evaluate teachers and to improveinstruction involves a classroom observation, followed by a post-observationconference and sometimes preceded by a pre-observation conference' Al-though supervisors might engage in some or all of these components, thegeneral purpose is to increase teachers' effectiveness. 2 Some researchers haveproposed that supervisors can increase teacher effectiveness by identifyingthe techniques and behaviors that are not enhancing instruction and by givingthe teacher alternative approaches and strategies that might accomplish lessonobjectives. Research shows, however, that teachers believe they learn littlefrom this process. 3

Therefore, we need to overcome deficiencies in the traditional supervi-sory process by developing alternatives that will enable teachers to learn morefrom their interactions with supervisors.4 This exploratory study constructeda new supervisory process for improving instruction and tested the processwith secondary school supervisors and mathematics teachers.

The test of this new model of supervision consisted of comparing twosupervisory processes. One, the local traditional supervisory process, con-sisted of an observation and post-observation conference where the supervi-sor and teacher collegially, through verbal interaction, developed a list ofsuggestions for improving the teacher's performance. The other was the

'Ben M. Harris, Supernsory Bebavior In Education, 3rd ed (Englewood Cliffs, NJ Prentice-Hall, 1985); Patricia E. Holland, 'Implicit Assumptions about the Supervisory Conference AReview and Analysis of Literature," Journal of Curriculum and Supervisfon 4 (Summer 1989)-362-379.

'Carl D. Ghckman, Supervision oflnstructfon A DevelopmentalApproacb, 2nd ed (BostonAllyn & Bacon, 1990); David A. Squires, William G. Huitt, and John K. Segars, EffeCftve Scboolsand Classrooms. A Researcb-based Perspective (Alexandria, VA. Association for Supervision andCurriculum Development, 1984).

3Arthur Blumberg, Supervisors and Teacberm A Private Cold War, 2nd ed. (Berkeley, CA

McCutchan, 1980);, Milbrey W. McLaughlin, "Teacher Evaluation and School Improvement,"Teacbers ColegeRecord86 (Pall 1984): 193-207.

4JerryJ. Bellon and Elner C. Bellon, Clazroom Supervision and nstructionallmprovement

A $ynergetC Procese 2nd ed. (Dubuque, IA. Kendall-Hunt, 1982),James M Cooper, "Observationskills," in Developing Skll'sfor Instuctional Supervison, ed. James M Cooper (New YorkLongman, 1984), pp. 79-111.

Gerad H. Elga rten 119

experimental supervisory process, which was the same as the traditionalmodel, except that after observing the teacher's lesson the supervisor taughtthe same lesson to the teacher's next class. The only deviations in thesupervisor's lesson were the demonstrations of behavioral changes that thesupervisor believed would make the teacher more effective. While observingthe supervisor teach, the teacher recorded the differences in his own and thesupervisor's lessons. These differences in the two lessons became the focus ofthe post-observation conference and then, in turn, the suggested behavioralchanges for the teacher. In effect, the study determined whether supervisorsmore effectively transfer suggested behavioral changes to teachers verbally orthrough demonstration.

Research conducted in an industrial setting indicates that observationallearning (also referred to as modeling) effectively changes adult behavior.sObservational learning is defined as acquiring cognitive skills and behaviorpatterns by observing others' performance. 6 According to Bandura's theory,four subprocesses govern observational learning: (1) attentional, referring toobservers' ability to focus on the relevant aspects of the modeled behaviors,(2) retention, referring to observers' ability to remember the aspects of theintended modeled activities, (3) production, referring to observers' ability toconvert symbolic concepts into appropriate behavior, and (4) motivation,referring to observers' ability to use what they observed. Bandura believesthat effective modeling requires the demonstrator to attend to all foursubprocesses.

RESEARCH QUESTION

Is Bandura's observational learning theory applied to supervising instruc-tion effective in changing teacher behavior?

Hypothesis 1. Teachers who are supervised with the use of modeling willshow a significantly greater rate of implementation of a supervisor's suggestedchanges in classroom behavior than teachers who are supervised withoutmodeling.

Hypothesis 2. Teachers who leave the post-observation conference witha list of behavioral changes will show a significantly greater rate of implemen-tation of a supervisor's suggested changes in classroom behavior than teacherswho leave without a list.

'Phillip J. Decker, 'Effects of Symbolic Coding and Rehearsal In Behavior-modeling Behav-Ior,"JournalofApplledPsycbology65 (December 1980). 627-634, PhilippJ Decker, 'The Enhance-ment of Behavior ModelingTrainingof Supervisory Skills bythe Inclusion of RetentionProcesses,'Personnel Pscbology 35 (Summer 1982). 323-332, Gary P. Latham and ULise M. Saari, "Applicationof Social LearningTheory toTraining Supervisors through Behavior Modeling,'JournalofAppltedPsycbology64 (June 1979): 239-246.

'Albert Bandura, Socal Foundations of Tbought and Actionm A Social Cognitive bTheory(Englewood Cliffs, NJ: Prentice-liall,.1986).

GensldH. Flgaten 119

120 Testing a New Supervisory Process for Improving Instruction

Hypothesis 3. An interaction effect will occur: Teachers who leave thepost-observation conference that used modeling with a list of suggestions willcarry out a significantly greater number of the suggestions than teachers in anyother treatment group (modeling without a list, traditional with a list, tradi-tional without a list).

I used the following rationale for the study:

* evidence suggesting that teachers' classroom behavior can be changedto increase student learning7

* Bandura's observational learning theory8

* substantial empirical evidence showing that observational learningeffectively changes adult behavior9

A RESEARCH MODEL OF OBSERVATIONAL LEARNING THEORY

The experimental modeling supervisory process is a research model ofBandura's theory. (1) A one-to-one correspondence exists between the con-structs in the theory and the activities of the constructs in the modelingsupervisory process, (2) to preserve the structure of the theory, its constructsappear in the supervisory process in the same order as they appear in thetheory.' ° Table 1 outlines the modeling process.

Design of the Exploratory Field Experirnent

- In each of the six participating schools, the mathematics supervisoradministered both the experimental modeling and the local traditional super-visory processes, eliciting from each teacher four suggested changes in be-havior. The local traditional supervisory process is the same as the ex-perimental process without the supervisor's demonstration. The traditional

'John T. Lovell and Kimball Wiles, SupervisionforBetter Scbools, 5th ed (Englewood Cliffs,NJ: Prentice-Hall, 1983).

8Albert Bandura, Social Foundations of Thbought and Action A Social Cognitive Theory

(Englewood Cliffs, NJ: Prentice-Hall, 1986).9Phillip J. Decker, 'Effects of Symbolic Coding and Rehearsal in Behavior-modeling Be-

havior," Journal of Applied Psycbology 65 (December 1980) 627-634, Phillip J Decker, "TheEnhancement of Behavior Modeling Training of Supervisory Skills by the Inclusion of RetentionProcesses," Personnel Psycbology 35 (Summer 1982) 323-332, Karen King, 'Modeling vsLecture/Discussion m Training Undergraduates as Teachers," Perceptual and Motor Skills 51(October 1980). 527-531, Gary P. Latham and Uise M Saarl, "Application of Social-Learning Theoryto Traning Supervisors through Behavior Modeling," Journal of Applied Psycbology 64 (June1979)-. 239-246;, Herbert H. Meyer and Michael S. Raich, "An Objective Evaluation of a BehaviorModelingTtrainig Program," PersonnelPsychology 36 (Winter 1983): 755-761.

°Joseph E. Hill and AugustKerber, Models, Metbods, and AnalyticalProcedures in Educa-tionalResearcb (Detroit: Wayne State University Press, 1967).

121

A. Teach the lesson.

(Attenatonal)(Retentionm Symbolic coding)

B. Observation and analysis--the teacher responds to threequestions:

1. What do I Isupervisorl do thatyou don't do too often?

2. What do we both seem to do?3. What good things do you do

that I don't seem to. do?

(Retention: Symbolic rebearsal)

C. Post-observation conference1. The teacher responds to

questions 1, 2, and 3.

2. The teacher records changesin behavior to carry out infuture lessons.

3. The teacher records changesin behavior to carry outimmediately.

4. The teacher responds toquestions.

(Production Reinforcement)

D. The teacher practices behaviorchanges with the behavior listdeveloped at the post-observation conference inview when teaching.

E. The teacher remains in contactwith the supervisor, providingfeedback about progress inincorporating new teachingbehaviors into lessons.

Gerald H. Elgarten

Table 1. Experimental Supervisory Process Using Modeling

Supervisor Teacher

A. Observation and Analysis1. Observe lesson.2. Identify behaviors that

need to be introduced orchanged.

B. Teach the same lesson to theteacher's next class. Keepclose to the teacher's lessonand style.1. Model strengths of the

teacher's lesson.2. Model new behaviors that

would increase learning inthe classroom.

C. Post-observation conference1. Ask the teacher to

respond to questions 1, 2,and 3.

2. Ask the teacher to recordbehavior changes to carryout in future lessons.

3. Ask the teacher to recordchanges in behavior tocarry out immediately.

4. Ask the teacher

a. How will you carry outthe suggestions?

b. How will youremember to carry outthe suggestions?

(Motivation Feedback)D. The supervisor informally

observes the teacher onceduring the next week to givefeedback.

E. The supervisor informally asksthe teacher about progress incarrying out the changes everythree days.

122 Testing a New Supenvisory Process for Improving Instruction

process is the comparison treatment because it is the dominant approach allsupervisors in the school system use. I made no attempt to change the commonsupervisory practices being used.

Sample

In each school, 8 mathematics teachers were randomly assigned to oneof four treatments. experimental modeling supervisory process, with theteacher leaving the post-observation conference with a list of suggestedbehavioral changes, experimental modeling, with the teacher leaving withouta list; local traditional supervisory process, with the teacher leaving thepost-observation conference with a list of suggested behavioral changes; andlocal traditional, with the teacher leaving without a list (for all treatmentgroups, n - 12). For the 48 teachers, the following contextual data werecollected. 29 were male (60 percent), and 19 were female (40 percent); themedian years teaching, 6; the median age, 37, 41 were liberal arts majors (85percent) and 7 education majors (15 percent); the supervisors rated theteachers' general performance level as excellent for 3 teachers (6 percent),good for 17 teachers (35 percent), average for 23 teachers (48 percent), andweak for 5 teachers (11 percent).

Therefore, this sample was representative of all high school mathematicsteachers in New York City on teachers' gender but not the median yearsteaching (13). The mathematics chairs participating in the study had an averageof 6 years' experience as supervisors. The average experience of mathematicssupervisors in the school system is about 7 years.

One week after the post-observation conference, the respective supervi-sor revisited the teacher's class and compiled an implementation score rangingfrom 0, indicating that the teacher carried out no suggestions, to 4, indicatingthat the teacher carried out all suggestions.

Threats to Internal Validity

The supervisors' observation was the instrument used for gathering dataThis instrument presents potential problems, such as possible rater bias to aparticular treatment and variability in how the raters recorded the data In thisstudy, the raters (mathematics supervisors who administered all the treatmentsin their schools) determined, by a yes or no rating, whether teachers hadcarried out the suggestions. This design introduced another problem: Theraters might want to show that they were effective supervisors by being lenientin recording the teacher's use of their suggestions. The supervisors' trainingexperience and lack of involvement in designing the experimental supervisoryprocess minimized these potential threats to the accuracy of collecting data(instrumentation).

Cera H. Elgarten 125

To control the variability in how the raters recorded the data, the trainingprogram conducted before the experiment clearly delineated what to acceptas implementation of a behavior change. Simulations identified what consti-tuted an implemented behavior. Also, because all 6 raters observed all groups,any variability should occur equally among all groups and should thereforenot be a factor.

The supervisors in this experiment have successfully passed the NewYork City licensing examination for secondary school supervisors of mathe-matics. These experienced supervisors suggested changes in teacher behaviorand then acted as rating teachers in their own departments to determinewhether the teacher carried out the suggestions. These activities were normaljob responsibilities of these mathematics supervisors.

All supervisors were aware that their teacher ratings would be compiledwith all other schools and not analyzed separately. Also, because the supervi-sors were carrying out all treatments and were not involved in designing anyof them, they should have no bias toward a particular treatment. In addition,before the experiment, the supervisors indicated that they did not find a'highdegree of implementation, by teachers, of their suggestions. This admissionindicates that they would not be too lenient in evaluating the implementationof their suggestions as a way of concealing their own weaknesses. Therefore,the possible threat to instrumentation caused by rater bias to a treatment, orby raters trying to show effectiveness by giving high ratings, was minimal.

DATA ANALYSIS

Table 2 summarizes the mean implementation scores for teachers in thefour treatment groups. To determine whether there were any significantdifferences in the means reported in Table 2, I1 computed a two-way analysisof variance (ANOVA). Table 3 summarizes the results.

DISCUSSION AND CONCLUSIONS

Hypotbesis 1. The two-iway ANOVA indicated significant differencesbetween the groups on the number of suggestions carried out. The differenceswere due to the type of supervision alone. The mean for the modeling group(M - 3.17) was significantly different from that of the traditional group (M-2.13), F(1, 44) - 15.88 (P < .001). Based on the significant difference in theteachers' implementation rate of their supervisors' suggestions for the model-ing and traditional supervisory processes, we can conclude that teacherstended to carry out more suggestions when they saw the suggestions modeled(demonstrated) by their supervisors. This finding supports Hypothesis 1 andis consistent with the findings of empirical studies conducted in an industrial

Genzld H. Elganen 123

124 Testing a New Supenrvsoy Processfor Improving Insruction

Table 2. Implementation Scores for Teachersin Four Treatment Groups

Leaving post-observationconference

Type ofsupervision With list No list Total

ModelingM 3.42 2.92 3.17SD .67 .90 .82n 12 12 24

TraditionalM 2.25 2.00 2.13SD 1.14 .85 .99n 12 12 24

TotalM 2.83 2.46SD 1.09 .98n 24 24

NOTE. Scores ranged from 0, indicating the teacher carried out no suggestions, to 4, indicatingthe teacher carried out all the suggestions.

Table 3. Two-Way Analysis of Variance of Implementation Scoresfor Teachers in Four Treatment Groups

Sum of MeanSource squares df square F

Main EffectsType of

supervision (A) 13.021 1 13.021 15.878'With list/no list(B) 1.687 1 1.687 2.058

Interaction(A x B) .187 1 .187 .229

Residual 36.083 44 .820

Total 50.979 47 1.085

·p < .001

Gerald H. Egarten 125

environment," as well as in the training of student teachers.'2 Both of thesestudies reported that modeling improved the subjects' performances in thebehaviors being modeled.

For effective observational learning, Bandura identifies four subprocessesthat the model must attend to. The first subprocess is attent/onaland refers tothe model's ability to focus observers' attention on the behaviors beingmodeled. When the supervisors taught the same lesson to the teachers' nextclasses, they were incorporating the attentional subprocess into the experi-mental supervisory model. The supervisors asked the teachers, "What do yousee me doing that you don't do too often?" Their responses, developed whilethey observed their supervisors teaching their classes the same lesson they hadjust taught, indicate that the teachers were able to identify the modeledbehaviors. Follow-up interviews with the participating supervisors also con-firmed this finding.

The second subprocess is retention and relates to observers' ability toremember the important behaviors being rfiodeled. Gerst points out thatdeveloping symbols to represent observed behaviors aids retention." The rawdata from the modeling groups indicate that from the same process used foridentifying modeled behaviors (teachers observing their supervisors teachtheir classes the same lesson), the teachers produced symbolic representationsof the modeled behaviors--their own written description of the behaviors.The teachers wrote the behavioral descriptions in response to the supervisors'question, "What do you see me doing that you don't do too often?"

Bandura has found that when symbolic codes are rehearsed, observersincrease their retention of the behaviors.' 4 Follow-up interviews with theparticipating supervisors reveal that in the groups using modeling, the

"Robert F. Burraska, 'The Effects of Behavior Modeling Training upon Managers' Behaviorsand Employees' Perceptions,' Peisonnel Psycbology 29 (Autumn 1976). 329-335, William C.Byham, Diane Adams, and Ann Kiggins, 'Transfer of Modeling Training to the Job," PersonnelPsycbology 29 (Autumn 1976). 345-349, Phillip J. Decker, Effects of Symbolic Coding andRehearsal in Behavior modeling Behavior,' Journal of Applied Psycbology 65 (December 1980).627-634, PhilipJ. Decker, 'The Enhancement of Behavior Modeling Training of Supervisory Skillsby the Inclusion of Retention Processes,' Personnel Psycbdology35 (Summer 1982). 323-332, GaryP. Latham and Lise M. Saari, 'Application of Socai LearningTheory to TraimngSupervisorsthroughBehavior Modeling,' Journal of Applied Psychology 64 (une 1979). 239-2

46; Herbert H. Meyer

and Michael S. Raich, 'An Objective Evaluation of a Behavior Modeling Training Program,"Personnel Psychology 36 (Winter 1983); 755-761, Preston E. Smith, "Management ModelingTraining to Improve Morale and Customer Satisfaction," PersonnelPsychology 29 (Autumn 19"6).351-359.

"Karen King, "Modeling vs. Lecture/Discusslon in Training Undergraduates as Teachers,'Perceptual and MotorSkills 51 (October 1980): 527-531.

"Marvin D. Gerst, 'Symbolic Coding Processes in Observational Learnng," in PsycologicalModeling, ed. Albert Bandura (Chicago: Aldine-Atherton, 1971), pp. 1-62.

"Albert Bandura, Social Foundations of bThought and Actionr A Social Cogntrie 7Theory(Englewood Cliffs, NJ: Prentice-Hall, 1986).

Gerald H. Elgarten 125

126 Testing a New Supervisoty Processfor Improving Instruction

teachers' lists of identified good teaching behaviors became the focus of thepost-observation conferences. The supervisors' focus on the teachers' lists ofpotential behavior changes during the post-observation conferences showsthat the model attended to the retention process of Bandura's theory.

Theproduction process refers to observers' abilityto convert the symboliccodes into appropriate behaviors. Bandura points out that people will modelless than they observed if they are not actively encouraged and do not receivepositive feedback." Follow-up interviews reveal that 5 of the 6 participatingsupervisors did not actively encourage or give positive feedback to theirteachers. Therefore, 5 of the 6 models (supervisors) did not attend to theproduction process. The teachers' tendency to carry out a significantly greaternumber of suggestions when the supervisory process used modeling maycontradict Bandura's emphasis on positive feedback during the productionprocess. Alternative explanations exist for the teachers' high implementationrate of suggestions without positive feedback from their supervisors: The lackof positive feedback did not overcome modeling's powerful effect whenexamining the implementation one week after the post-observation confer-ence; a decrease in the teachers' implementation rate may occur several weeksafter the post-observation conference; the teachers may have carried out evenmore suggestions if their supervisor had used positive feedback; or thefeedback was not necessary.

The motivation process refers to observers' ability to put what theyobserved into practice. Bandura, Adams, Hardy, and Howells and Schunkbelieve that when observers see a successful behavior carried out, they tendto believe that they can also perform the behavior.' 6 All the supervisors saidthey believed a major strength of the experimental supervisory process wasthat the teachers were able to see the suggestions actually carried out withtheir own students. This process of demonstrating behavioral changes in thecontext of their own class may have increased the teachers' sense of efficacyand, in turn, may have motivated the teachers to carry out the supervisors'suggestions."

Hypothesis 2. The two-way ANOVA indicated no significant differencesbetween the groups leaving the post-observation conference with or withouta list of suggestions on the number of suggestions carried out. The mean for

"Ibid.'6Albert Bandura, N. E. Adams, A. B. Hardy, and G. N. Howells, "rests of Generality of

Self-efficacy Theory,' Cognitive Thberapy and Researcb 4 (No. 1, 1980)- 39-66; Dale H Schunk,'Self-effcacy Perspective on Achievement Behavior' (Rockville, MD, August 1982, ERIC Docu-ment No. 226 293).

"Dale H. Schunk, 'Self-efficacy Perspective on Achievement Behavior" (Rockville, MD,August 1982, ERIC Document No. 226 293).

Gerald H. Elgarten 127

the group leaving the post-observation conference with a list of suggestions(M - 2.83) did not significantly differ from that of the group leaving thepost-observation conference without a list of suggestions (M- 2.46), F(1, 44)- 2.058 (p < .05).

Therefore, giving teachers a list of suggestions at the end of a post-observation conference does not tend to affect how well they carried out thesuggestions. This finding does not support Hypothesis 2, which is based onmy prediction, from the results of a pilot study, that giving teachers a list ofsuggested behavior changes would increase their implementation rate of theirsupervisors' suggestions. In follow-up interviews, however, the supervisors allagreed that they found no evidence in their teacher contacts following thepost-observation of the teachers actually keeping the list of suggestions. Thesupervisors said they did not see any of the lists in view when they revisitedthe teachers' classrooms. Therefore, this study did not determine the effectson the teacher's implementation rate of their supervisor's suggestions thatwould stem from keeping a list of suggestions in view while teaching.

Hypothesis 3. The two-way ANOVA indicated no interaction effects be-tween the groups in the number of suggestions carried out, F(1, 44) - .229 (p< .05). This finding does not support Hypothesis 3, which was based on myprediction, from the results of a pilot study, that giving teachers a list ofsuggested behavior changes would significantly increase their implementionrate of the suggestions for both the traditional and the modeling supervisoryprocesses.

A possible reason for the lack of a significant interaction effect was thatthe teachers did not use the list of suggestions as a reminder to carry out thebehavioral changes. Perhaps expecting professionals to carry a list aroundwith them was unreasonable. This behavior appears not to fit with mostteachers' self-images. If teachers had the option of leaving the post-observation conference with or without a list, however, those selecting the listmay be more inclined to use it.

Limitations

1. Thbe qualifications oftbepersons using theproposedsupervisory modelAll supervisors in this study were subject-area specialists. Any attempt togeneralize these results to supervisors who are not specific subject specialistswould be questionable.

2. The classroom situation where the supervisory process was beingcarried out. All modeling behavior changes occured in mathematics dass-rooms with the specific teacher's class, teaching the same lesson. Any attemptat using the results of this study in a teaching situation that differs from thispattern would be questionable.

Gerald H. Elgarten 127

128 Testing a New Supersoy Process for Improving Instruction

Recommendations

1. When modeling is used in the supervisory process, teachers carry outsignificantly more suggestions than when the traditional supervisory processis used. Secondary school principals might encourage their mathematicssupervisors and other subject-area specialists to use modeling in the supervi-sory process. Principals should also consider using modeling in the supervi-sory process when observing classes in their own subject area of special-ization. Schools with no subject-area specialists or supervisors should considertraining master teachers to use modeling with their peers.

2. Two evaluation studies by the Rand Corporation have reported asignificant relationship between teachers' sense of efficacy and studentachievement.'8 Based on the findings of the Rand studies, researchers fromthe University of Florida conducted a teacher-efficacy study 9 The purpose ofthe research was to further investigate the relationship between the teachers'sense of efficacy and student achievement. In a final report in 1983, Ashton,Webb, and Doda cite the major findings as supporting the Rand results 20 Theyreport that the teachers' sense of efficacy was (1) significantly related tostudents' achievement in high school basic skills classes, (2) related to schoolorganizational structures (team organization and teacher involvement in deci-sion making); and (3) related to a warm, accepting classroom climate.

In their conclusions, Ashton and others have supported the findings ofBandura, Guskey, Hillman, and Schunk that perceived self-efficacy is not afixed construct.2 Instead, self-efficacy is negotiated daily with each teacherinteraction with students, parents, colleagues, and administrators These re-searchers suggest that future research explore the processes by which instruc-tional techniques and administrative strategies can "reduce the threats and

'8David Armor, Patricia Conry-Osequera, Millicent Cox, Nicelma Kin, Lorraine McDonnel,

Anthony Pascal, Edward Pauly, and Gail Zellman, 'An Analysis of the School Preferred ReadingProgram in Selected Los Angeles Minority Schools" (Santa Monica, CA, August 1976, ERICDocument No. 130 243); Paul Berman, Milbrey McLaughlin, Gall Bass, Edward Pauly, and GallZellman, "'Federal Programs Supporting Educational Change, Vol. VI] Factors Affecting Imple-mentation and Continuation" (Santa Monica, CA, April 1977, ERIC Document No 140 432)

'9Carolyn H. Denham and Joan J. Michael, "Teacher Sense of Efficacy A Definition of theConstruct and a Model for Further Research," Educational Researchb Quarterly 6 (Spring 1981)39-63, Sherri Gibson and Myron H. Dembo, "Teacher Efficacy A Construct Validation," JournalofEducatioralPsychology 76 (August 1984): 569-582.

a'Patricia T. Ashton, Rodman B. Webb, and Nancy Doda, 'A Study of Teachers' Sense ofEfficacy, Final Report" (Gainesville, FL, July 1983, ERIC Document No. 231 833)

nAlbert Bandura, Social Foundations of Thbought and Action. A Social Cognitive Theory(Englewood Cliffs, NJ. Prentice-Hall, 1986), Thomas R. Guskey, 'Teacher Efficacy, Self-concept,andAttitudes toward the Implementationof Mastery Learning" (paper presented at the71st annualmeeting of the American Educational Research Association, Washington, DC, April 1987, ERICDocument No. 281 838), SusanJ. Hillman, 'Contributions to Achievement The Role of Expecta-tions and Self-efficacy in Students, Teachers, and Principals," Dissertation Abstracts International,44/12-A.3588A (University Microfilms No. DA84-06810), Dale H Schunk, 'Self-efficacy Perspec-tive on Achievement Behavior' (Rockville, MD, August 1982, ERIC Document No 226 293)

---

Gerard H. Elgarten 129

increase the support of teachers' sense of efficacy."2 Based on the empiricalevidence of the strong relationship between teachers' sense of efficacy andstudent achievement and further evidence that our perceived self-efficacy canchange, I recommend that future studies using modeling in the supervisoryprocess examine the effects of modeling on teachers' sense of efficacy.2 3

3. The use of modeling in the supervisory process is effective in changingmathematics teachers' classroom behaviors. Modeling might have similareffects on principals. Therefore, I recommend applying the same supervisoryprocess outlined in this study to principals' inservice training. For example, anexperienced principal might be assigned to help a novice principal improvefaculty conference presentations. After observing the novice principal make apresentation at a morning faculty conference, the more experienced principalwould make the same presentation, modeling sonie suggested behaviorchanges, at an afternoon conference. During the afternoon confdience, thenovice principal would be responding to the same three questions asked inthe experimental supervisory process. (1) What do I [the supervisor] do thatyou don't seem to do? (2) What do we both seem to do? (3) What good thingsdo you do that I don't seem to do? The responses would then become thefocus of the post-observation conference.

GERALD H. ELGARTEN is Associate Professor of Mathematics Education, The CityCollege, The City University of New York, North Academic Center 6/207, ConventAvenue at 138th Street, New York, NY 10031.

'Patricia T. Ashton, Rodman B. Webb, and Nancy Doda, 'A Study of Teachers' Sense ofEfficacy, FinalReport" (Gainesville, FL,July 1983, ERIC Document No. 231 833), p. 38.

"Susan M. Tracz and Sherri Gibson, "Effects of Efficacy on Academic Achievement" (paperpresented at the annual meeting of the California Educational Research Associatfon, Marina delRay, CA, November 1986, ERIC Document No. 281 853).

Geald H. Elganen 1Z9

Copyright © 1991 by the Association for Supervision and Curriculum Development. All rights reserved.