You are on page 1of 8

Journal of Educational Psychology

1887. Vol. 58, No. 3,181-188

EFFECT OF FEEDBACK FROM TEACHERS


TO PRINCIPALS1
ROBERT W. DAW2 AND N. L. GAGE
Stanford University

Each member of an experimental group of 151 elementary school


principals was given feedback concerning his teachers' ratings of their
actual and ideal principals on 12 behaviors. These principals were sub-
sequently found to differ significantly, in the direction of teachers'
preferences, from 143 principals in a control group. Initial differences in
ratings were controlled by analysis of covariance. A 2nd, nonpretested
control group did not differ from the pretested control group; hence the
pretest itself did not produce the effect and difference between experi-
mental and control groups was attributable to the feedback itself. 2
intervals between feedback and 2nd rating, 2 forms of feedback, the
principal's age and experience, and the sequence and direction of the
rating-scale items were found to be nonsignificantly related to the effect
of the feedback. The results suggest that feedback of this kind im-
proves the behavior of elementary school principals.

It is highly plausible that feedback changes behavior has been developed


regarding how others feel about one's by Gage, Runkel, and Chatterjee
behavior will affect one's behavior. (1960). In brief, the rationale is that
Whether this maxim will hold under a the feedback will inform some princi-
given set of practical circumstances pals that their teachers evaluate their
must, however, be determined empiri- behavior less favorably than the prin-
cally. In the present experiment, ele- cipals might desire. If we assume that
mentary school principals were told principals respect their teachers' opin-
how their teachers rated them and an ions, we can expect this information
ideal principal; other principals, simi- to create in the principals a condition
larly rated, were not given this in- of imbalance (Heider, 1958), asym-
formation. metry (Newcomb, 1959), incongruity
One theoretical justification for hy- (Osgood & Tannenbaum, 1955), or
pothesizing that feedback of this kind dissonance (Festinger, 1957). To re-
move or reduce this condition, that is,
'A more detailed presentation of the to restore a condition of equilibrium or
data and instruments is available in the consistency, the principals are likely
first author's doctoral dissertation, "Chang- to change the behaviors concerned in
ing the Behavior of Elementary School
Principals Through the Use of Feedback," the directions desired by the teachers.
on file in the Stanford University Library. After enough time has elapsed to al-
The dissertation was written under the di- low such behavior changes to occur
rection of the second author. Support for and to be perceived by the teachers,
the research was provided by the Graduate a second description of the principals'
Division of Stanford University and by a
small grant (MHI544-01) from the United behaviors by their teachers will re-
States Public Health Service. The present flect such changes.8
report was written by the second author
during tenure as a Fellow at the Center for "William McGuire (personal communi-
Advanced Study in the Behavioral Sciences cation, August 15, 1966) has suggested that
and as a special Fellow of the National self-esteem theory—that is, that persons be-
Institute of Mental Health. have so as to maximize self-esteem, not to
8
Now at Santa Maria Joint Union High minimize inconsistencies, or discrepancies—
School District, Santa Maria, California. is more relevant to our experiment. This
181
182 ROBERT W. DAW AND N. L. GAGE

Positive results in experiments on TABLE 1


the effect of feedback of ratings have DISTRIBUTIONS, MEANS, AND RANGES ON
previously been obtained (Bryan, VARIOUS VARIABLES FOR EXPERIMENTAL
AND CONTROL GROUPS
1963; Gage, 1963; Gage, Runkel, &
Chatterjee, 1963). In those experi- Experi- Control Control
ments, teachers were rated by their Variable mental group 1 group 2
group (Ci) (C,)
pupils and then, on a subsequent (E)
rating by the same subjects, were Kind of district
found to differ significantly, and in the District with superintend-
ent other than principal 105 111 115
direction of the raters' ideals, from a District where the principal
also acts as the superin-
control group not given the ratings. tendent 46 32 46
The present experiment was aimed Sex
Male 137 130 146
at determining whether the same ef- Female
Education
14 13 15
fects would be found with feedback No B.A. degree
B.A. degree
1
49
0
40
1
44
from teachers to principals. It also M.A. degree 97 100 115
Ed.D. degree 3 3 1
incorporated some refinements in de- Ph.D. degree 1 0 0
Number of students in the
sign that permitted testing rival hy- school
potheses as to the cause of the change M 909 512 607
Range 185-1011 185-1187 200-1175
in rated behavior. In addition, data M 42 43 41
were gathered concerning different Range
Number of years of experi-
27-63 29-63 27-63
time intervals, forms of feedback, ence as a principal
M 8.3 8.6 7.9
and personal characteristics of the Range 1-36 1-32 1-35
principals. Number of teachers in the
school
M 14.8 15.3 15.5
Range 8-34 8-37 8-39
METHOD
Note.—For the Experimental Group, N = 151; for
Subjects Control Group 1, N = 143; for Control Group 2, If = 161.

The subjects were 455 elementary school


principals in all the counties in California a single principal who also acted as super-
in the fall of 1962. Because they were as- intendent, the superintendent-principal was
signed at random, the principals in the ex- directly invited to participate. Although
perimental and control groups had about initial attrition was about 25%, the subse-
the same number of pupils and teachers, quent rate of participation was never less
and were similar in sex, age, educational than 93% of the principals invited. That
level, and years of experience as a princi- is, of the 1007 original contacts, 752 yielded
pal, as shown in Table 1. Besides the ex- the name of the person who participated.
perimental (E) group, there were two con- To insure the anonymity of the teacher
trol groups: Control Group Ci, which was responses, all schools with less than eight
rated on both the first and second occasions, teachers were eliminated, leaving 500 prin-
and "posttest only" Control Group Ca, cipals. Of these, 455 completely met all
which was rated only on the second occa- other requirements for inclusion in the ex-
sion. periment. Obviously, this final group may
Each superintendent of an elementary have been biased toward containing super-
school district (including unified districts) intendent-principals (a) interested in what
in California with more than one principal their teachers thought of their actions or
was invited to send the name of the first (b) reluctant to refuse to participate.
principal in the alphabetical listing of prin-
cipals in his district, and then that principal Methods and Schedule of Data Col-
was invited to participate. In districts with lection
kind of alternative to consistency theory The study was made in the school year
has been outlined by Deutsch, Krauss, and 1962-63. Letters of invitation were sent on
Rosenau (1962), Steiner and Rogers (1963), November 30. To minimize the possibility
and McGuire and Millman (1965). that principals would discuss the project
EFFECT OF FEEDBACK FBOM TEACHERS TO PRINCIPALS 183

with one another, only one principal was principal was given, for each of the 12 items,
invited from each district. By January 5, histograms showing the percentages of his
752 principals' names had been received; teachers who described him and their ideal
of these, 252 were excluded because their principal with each of the six response-
schools had less than eight teachers. Be- alternatives. A randomly chosen half of the
tween January 2 and 9, booklets entitled experimental group (the "a + h" group)
"What Do They Expect?" (WDTE) were received both the histograms and the me-
mailed to 340 principals in Groups E and dians (indicated by arrows) of the ratings
Ci. These booklets, presenting the research of the actual and ideal principal; the re-
as the tryout of a "Principal's Information maining principals (the "a" group) were
Project," required the principal to rate him- given only the medians of the ratings by
self and his ideal principal on 12 items. their teachers. These RYTO booklets were
In addition, the principal was asked to pro- withheld from Groups Ci and Ca until after
vide the information summarized in Table the second round of ratings of principals
1 and to indicate how many Teacher Opin- by their teachers.
ion Booklets would be needed to collect A follow-up questionnaire, designed to
ratings of the principal from his teachers. encourage careful reading of the RYTO's,
(The WDTE booklet was mailed to Group was mailed on February 25 to all principals
Ca just prior to the mailing of the posttest in Group E. This questionnaire and a re-
materials.) Principals were randomly as- minder elicited responses concerning the
signed to Groups E, Ci, and Ca prior to the RYTO from 93% of the experimental group.
mailing of the WDTE booklet. The rate of The second round of Teacher Opinion
return of these booklets for the three groups Booklets was mailed to a randomly chosen
ranged from 94.0 to 98.7%. half of Groups E and Ci on March 25. A
Teacher Opinion Booklets were mailed letter had been mailed a week in advance
to the principals on January 21; only eight asking the principal to set aside a specific
of the 336 booklets mailed were not re- day for the administration of these instru-
turned. In these booklets, the teachers ments. They were returned with little delay.
rated their actual and ideal ("best imagina- Teacher Opinion Booklets were mailed to
ble") principal on the same 12 items. the second half of Groups E and Ci on
On February 11, booklets entitled "Re- May 6. The use of two posttest dates per-
port on Your Teachers' Opinions" (RYTO) mitted determining whether the effect of
were mailed to the principals in Group E. the feedback changed with the length of in-
In these booklets, as shown in Figure 1, the terval between feedback and posttest.

Item: Acts Promptly to Fulfill Teacher Requests

Percent giving indicated answer Percent giving indicated answer


0 10 20 30 40 50 60 10 20 30 40 50 60
i LIKE my principal

principal

LIKE the best principal

attest UNLIKE my principal A little bit UNLIKE the best principal

-jSomewhat UNLIKE my principal Somewhat UNLIKE the best principal

Very much UNLIKE my principal very much UNLIKE the best principal
0 10 20 30 40 50 60 0 10 20 30 40 50 60
FIG. 1. Form of feedback, arrow-plus-histogram, in "Report on Your Teachers' Opinions."
184 ROBERT W. DAW AND N. L. GAGE

Items Concerning Principal Behavior shown in Figure 1. Then he was asked to


rate his ideal principal on the same 12 items.
The teachers described their actual prin- The teachers read the directions silently
cipal and their ideal principal on the fol- while the principal read them aloud. The
lowing 12 items concerning principal be- teacher wrote his responses on a card, put
havior : the card in an envelope, and sealed the
1. Encourages teachers with a friendly envelope. On the cover of his booklet, the
remark or smile. teacher read: "Your answers will be sealed
2. Gives enough credit to teachers for in an envelope by you and sent directly to
their contributions. Stanford University. No one at your school
3. Does not force opinions on teachers. or in your district will know how you an-
4. Enforces rules consistently. swered these questions." Further to insure
5. Criticizes without disparaging the ef- the teachers' privacy, the principal was di-
forts of teachers. rected to stand far enough away from his
6. Informs teachers of decisions or ac- teachers to prevent him from seeing their
tions which affect their work. papers, to make certain that all teachers'
7. Gives concrete suggestions for improv- answer cards were sealed in the envelope,
ing classroom instruction. to require the teachers to put their sealed
8. Enlists sufficient participation by envelopes into the large mailing envelope,
teachers in making decisions. to permit his teachers to see him moisten
9. Demonstrates interest in pupil prog- and close the clasp of the large envelope,
ress. and finally to ask one of his teachers to
10. Interrupts the classroom infrequently. drop the envelope in a United States mail
11. Displays much interest in teachers' box. The administration of the question-
ideas. naires was standardized as fully as possible
12. Acts promptly in fulfilling teacher with printed directions to the principals
requests. and teachers so as to assure both groups
The items were based on ideas obtained that their anonymity would be preserved.
from Campbell and Gregg (1957), Gross, It is noteworthy that all answer cards came
Mason, and McEachern (1958), Guba and back in sealed envelopes.
Bidwell (1958), Medsker (1956), and vari-
ous elementary school teachers. Each item "Report on Your Teachers' Opinions"
was worded in both positive and negative
directions in Forms A and B, respectively. The "Report on Your Teachers' Opin-
Forms Ai and Bi placed the items in reverse ions" consisted of a booklet containing two
order from that in Forms As and B2.' charts for each of the 12 items, as shown in
The items were intended to deal with Figure 1. The discrepancy between the
behaviors that could be expected to occur teachers' descriptions was implicit in the
frequently, could be briefly described with- vertical distance between the two arrows
out qualifying phrases, and could be changed indicating medians. If the two arrows were
by the principal within the time span of at the same point on these scales, the prin-
the research in a way that could be recog- cipal could infer that his teachers saw no
nized by the teachers. difference between him and their ideal prin-
The 70 items originally written were re- cipal in that kind of behavior. In propor-
duced to the final 12 on the basis of ratings tion to the distance between the two arrows,
of their importance, improvability, and no- the principal could infer that he departed
ticeability; the ratings were made by psy- in the given direction from his teachers'
chologists, professors of educational ad- ideal for that kind of behavior.
ministration, teachers, and principals. To determine the principal's reaction to
the RYTO and to encourage him to review
Teacher Opinion Booklet these reports, he was asked to answer six
questions on a reaction sheet. The re-
In the Teacher Opinion Booklet, the sponses indicated that high percentages
teacher was asked to rate his principal on (86-96%) of Group E found the RYTO
each item using one of the six alternatives interesting, understandable, and informa-
tive.
4
The various forms—Ai, Aa, Bi, and Experimental Design
Ba—were randomly assigned across schools;
that is, every teacher in a given school re- The experimental design is shown in
ceived the same form. Figure 2. Here, X represents the expert-
EFFECT OF FEEDBACK FROM TEACHERS TO PRINCIPALS 185
Croup Tntttt Tre»tMnt 6 veeU 12 w»t» signified the desirable end, and 6 the
E
i-ihort undesirable end, of the continuum.
••long Pre-ACT
The random assignment of princi-
pals to Groups E and Ci should have
made them equivalent. But the pre-
i-long
ACT means of these groups differed
at the .05 level on five of the 12 items
and on the mean of the 12 items. Also,
the direction of the difference was the
' 1-long "11
same for all 12 items, namely, the
C
direction favoring Group E. Presuma-
2-ihort °13
bly, however, the analysis of covari-
ance, to be described below, elimi-
c
2-long °U nated this pretest bias to the extent
Fw. 2. The experimental design. that the pretest scores were reliable.
The difference between Groups E
mental treatment (feedback), O refers to and Ci in pre-ACT means might have
the process of observation (ratings), Xs and resulted from a greater tendency of
Os in a given row are applied to the same
subjects, the left-to-right dimension indi- subjects receiving unfavorable feed-
cates temporal order, and the rows represent back on pre-ACT to drop out of
equivalent samples of persons. X. repre- Group E. In that event, the remaining
sents "arrow-only" feedback; Xith repre- members of Group E would be those
sents "arrow-plus-histogram" feedback.
Group d (posttest only) was used to allow who had received more favorable pre-
comparisons free of any effect attributable ACT ratings. But when this possi-
to unintended feedback received by Group bility was investigated, it was found
Ci through participating in the pretest; that, in fact, the few drop-outs from
these comparisons would show whether the Groups E and Ci did not differ on pre-
pretest itself produced changes in behavior.
ACT in this way, and the difference
RESULTS remained unexplained.
In describing the results, we shall The two experimental subgroups
refer to the protocols obtained from differing in the type of feedback pro-
the teachers as follows: vided (arrow-only and arrow-plus-
Pre-ACT—the teacher's description histogram) did not differ significantly
of his actual principal on the pretest. on any of the 12 pre-ACT means. Nor
Post-ACT—the teacher's descrip- were there any significant differences
tion of his actual principal on the in pre-ACT means among the sub-
posftest. groups given forms differing in the
Pre-IDL—the teachers' description direction of wording or the sequence
of his ideal principal on the pretest. of the items.
For each item, the mean of the
ratings received by a group of princi- Adjusted Post-ACT
pals was computed over the medians The effect of the feedback was
of the ratings of each principal by his measured by the difference between
teachers. The ratings of all items, Groups E and Ci. To adjust for the
regardless of whether they were origi- pre-ACT differences between the
nally worded positively or negatively, groups, analysis of covariance was
were converted to a scale in which 1 used. The pre-ACT rating served as
186 ROBERT W. DAW AND N. L. GAGE

TABLE 2 to perceive in the period allowed.


MEANS OF TEACHERS' RATINGS OF ACTUAL. Item 10, "Interrupts the classroom
PBINCIPALS IN EXPERIMENTAL GROUP AND infrequently," may have been rated
CONTROL GROUPS Ci AND C 2
too favorably on the pretest to allow
Means sufficient room for improvement; the
Difference
pre-ACT means on Item 10 were the
Item pre-ACT
Adjusted between
adjusted P
most favorable (had the lowest nu-
post-ACT post-
ACT post-ACT merical value) of all 12 items.
Means
E Ci E Ci C, E Ci Posttest-Only Control Group
1 1.82 1.39 1.31 1.50 1.44 1.38 1.43 .05 11.6" The "posttest-only" control group
2 1.47 1.57 1.45 1,63 1.60 1,50 1.58 .08 5.0' (C2) was used to eliminate certain
3 1.54 1.61 1.50 1.77 1.77 1.61 1.66 .05 16.5"
4 1.68 1.75 1.59 1.84 1.95 1.69 1.74 .05 13.5" possible attenuating effects on the
5 1.34 1.51 1.36 1,56 1.56 1.42 1.49 .08 6.7*
a 1.42 l.« 1.38 1.61 1.67 1.48 1.51 .03 17.0" comparisons. Such effects might have
7 1.72 1.84 1.64 2,04 1.93 1,81 1.87 .06 25.1"
8 1.58 1.98 1.50 1.80 1.86 1.63 l.«7 .04 19.0" resulted from any feedback, or sensi-
9 1,28 1.37 1.27 1 36 1.38 1,30 1.34 .04 2.2
10 1.19 1.26 1.20 1.29 1.30 1.23 1.26 .03 3.3 tization to the items of behavior, that
11 1.40 1.55 1.40 1 59 1.57 1 45 1.54 .09 4.4"
12 1.51 1.61 1.44 1,66 1.71 1.52 1.57 .05 8.7" might have been received by the pre-
1-12 1.49 1.55 1.42 1.64 1.65 1.49 1.57 .08 19. 4»* tested control group (Ci) as a result
Note.—In this and all other tables, means refers to a of their participation in the pretest.
scale in which a score of 1 was assigned to "Very much
LIKE" rating scale alternative, 2 to "Somewhat LIKE," If such sensitization occurred, Group
etc., to 6 for "Very much UNLIKE." For the Experi- Ci would differ less from Group E
mental Group (E), N <= 151; for Control Group 1 (Ci),
N = 143; for Control Group 2 (C«), N = 161.
* p < .05.
than would a nonpretested control
" p < .01. group (C2) which received neither the
feedback nor the pretest, Here the
the control variable, the post-ACT comparisons must be made in terms
rating as the dependent variable, and of unadjusted posttest means, since
the feedback as the independent vari- there were no pretest means with
able. Although the 12 items are corre- which to adjust the posttest means of
lated, it is considered worthwhile to Group C 2 .
examine results for each item indi- Table 2 also shows the means for
vidually as well as for the mean over Group C 2 . The means for Group Ci
all 12 items. Analyses of covariance and C2 did not differ significantly.
for each of the 12 items and for the Both of these groups did differ, in the
mean over Items 1-12 yielded the re- same direction, from Group E. Hence,
sults shown in Table 2. The difference the pretest in itself did not affect
between the adjusted post-ACT means Group Ci, and the improvement in
was significant at the .05 level or Group E must be ascribed to the feed-
better for all but two of the 12 items; back alone, not to the feedback plus
for the mean over Items 1-12, it was the pretest.
significant at the .01 level. In all
cases, the difference between the ad- Adjusted Post-ACT Minus Pre-IDL
justed post-ACT means favored the The teachers' initial ratings of their
experimental group. Only on Items 9 ideal principal (pre-IDL) make possi-
and 10 did the adjusted post-ACT ble an interpretation of the direction
means not differ at even the .05 level of the difference between adjusted
of significance. But even on these post-ACT means. It should be re-
items, the difference favored the ex- called that the feedback informed
perimental group. Item 9, "Demon- Group E as to how their teachers rated
strates interest in pupil progress," may the ideal principal. The feedback
have been too difficult for the teachers should influence the principal to
EFFECT OF FEEDBACK FROM TEACHERS TO PRINCIPALS 187

TABLE 3 6-week or a 12-week interval group.


ADJUSTED MEAN Posi-ACT AND PRE-IDL Analysis of covariance revealed no
RATINGS FOB THE EXPERIMENTAL GROUP significant difference in the two groups'
AND CONTROL GROUP Ci
adjusted post-ACT means. Further,
Adjusted Adjusted there was no significant effect due to
post-ACT pre-IDL U post-ACT the Feedback X Interval interaction.
if
minus
Item pre-IDL In short, the effect of the feedback
was not a function of the interval over
E Ci E Ci E Ci
which it was measured.
1 1.38 1.43 1.09 1.08 .29 .35 One randomly chosen half of the
2 1.50 1.58 1.13 1.14 .37 .44 experimental groups received feed-
3 1.61 1.66 1.21 1.22 .40 .44 back in the form of median ratings
4 1.69 1.74 1.20 1.17 .49 .57 (arrows) only, and the other half
5 1.42 1.49 1.13 1.17 .29 .32
6 1.48 1.61 1.09 1.10 .39 .41 received median ratings plus the fre-
7 1.81 1.87 1.19 1.25 .62 .62 quency distributions (histograms) of
8 1.63 1.67 1.20 1.23 .43 .44 the ratings. Either kind might rea-
9 1.30 1.34 1.10 1.13 .20 .21 sonably by predicted to be the more
10 1.23 1.26 1.15 1.19 .08 .07 effective: The median-only might be
11 1.45 1.54 1.13 1.18 .32 .36
12 1.52 1.57 1.13 1.16 .39 .41 sharper, less ambiguous; on the other
1-12 1.49 1.57 1.14 1.17 .35 .40 hand, the median-plus-distribution
might be more convincing. But no
Note.—For the Experimental Group, significant difference was found for
N = 151; for Control Group Ci , N = 143.
any item between the adjusted post-
change in that direction. Hence, the ACT means of the subgroups receiv-
difference between adjusted post-ACT ing these two kinds of feedback.
and pre-IDL should be smaller for When the principals were divided
Group E than for Group Ci. As Ta- into subgroups on the basis of age
ble 3 shows, most of these differences (40 or younger versus 41 or older),
were indeed smaller for the experi- experience (5 years or less as ele-
mental group. At the time of the post- mentary school principal versus 6
test, the principals who received feed- years or more), and form (A t , A 2 ,
back came closer to their teachers' BI , B2) of the Teachers' Opinion
desires. Since the items are interde- Booklet, analyses of covariance
pendent, they do not, of course, pro- yielded no significant difference at
vide 12 separate tests of the overall the .05 level due to main effects or in-
hypothesis. But the means over all teractions. Thus, the significant re-
12 items also differed in the expected sults that occurred due to feedback
direction. did not seem to vary with age, experi-
ence, or form. The latter findings show
Interval, Form of Feedback, and that positional or directional sets did
Other Variables not significantly affect the teachers'
In the experiment by Gage, Runkel, responses.
and Chatterjee (1960), experimental
group subjects changed more if they DISCUSSION
had a longer interval between feedback All in all, the results indicate that
and posttest. But the time interval the feedback affected changes in the
was relatively short, ranging from 18 principals' behavior. Various questions
to 59 days. In the present experiment, remain, however, for subsequent re-
the experimental and control groups search. First, we must recall the un-
were assigned at random to either a explained pretest differences between
188 ROBERT W. DAW AND N. L. GAGE

the experimental and control groups CAMPBELL, R. F., & GBEQG, R. T. (Eds.)
despite their random assignment. Administrative behavior in education.
New York: Harper, 1957.
Second, we must ask whether methods DEUTSCH, M., KBAUSS, R. M., & ROSENAU,
of measuring behavior other than re- N. Dissonance or defensiveness? Journal
ratings by the same teachers would of Personality, 1962, 30, 16-28.
reveal the same kinds of behavior FESTINOER, L. A theory of cognitive dis-
change. Would observations by trained sonance. Evanston, 111.: Row, Peterson,
1957.
observers produce confirming evi- GAGE, N. L. A method for "improving"
dence? Would interviews of the princi- teacher behavior. Journal of Teacher
pals reveal the process by which feed- Education, 1963, 14, 261-266.
back operates, and show whether the GAGE, N. L., RUNKEL, P. J., & CHATTEHJEE,
B. B. Equilibrium theory and behavior
principals were consciously attempt- change: An experiment in feedback from
ing to change their behavior? Would pupils to teachers. (Report No. 6 in
disguised tests of teacher "morale" Studies in the generality and behavioral
reflect the desirable changes in rated correlates of social perception.) Urbana:
principal behavior? Bureau of Educational Research, College
of Education, University of Illinois, 1960.
Third, we should determine better GAGE, N. L., RUNKEL, P. J., & CHATTERJEE,
whether the improvements in the post- B. B. Changing teacher behavior through
ACT ratings reflect mere improved feedback from pupils: An application of
"halo effect" or actual changes in equilibrium theory. In W. W. Charters
specific behaviors. One way to proceed & N. L. Gage (Eds.), Readings in the
social psychology of education. Boston:
on this question would be to collect Allyn & Bacon, 1963. Pp. 173-181.
post-ACT ratings on uncorrelated GROSS, N., MASON, W., & MCEACHERN,
items not dealt with in the feedback; A. W. Explorations in role analysis: Stud-
if behavior also improves on such ies of the school superintendency role.
New York: Wiley, 1958.
items, the significance of the feedback GUBA, E. G., & BIDWELL, C. E. Adminis-
on specific items must be questioned. trative relationships. Chicago: Midwest
Similarly, items differing widely in Administration Center, University of Chi-
"changeability" could be compared cago, 1958.
as to the amount of change principals HEIDER, F. The psychology of interpersonal
relationships. New York: Wiley, 1958.
exhibit in them; presumably, if the McGuiRE, W. J., & MILLMAN, S. Anticipa-
ratings reflect more than general im- tory belief lowering following forewarn-
pressions, relatively unchangeable be- ing of a persuasive attack. Journal of Per-
haviors should be rated as changing sonality and Social Psychology, 1965, 2,
471-479.
If research allays the skepticism MEDSKER, L. L. The job of the elementary
school principal as viewed by teachers.
implicit in these suggestions, further Unpublished doctoral dissertation, Stan-
attention should be given to ways ford University, 1956.
of enhancing the effectiveness of feed- NEWCOMB, T. M. Individual systems of ori-
back. The behavior of teachers, princi- entation. In S. Koch (Ed.), Psychology:
pals, and persons in many similar A study of a science. Vol. 3. New York:
roles could be made more effective by McGraw-Hill, 1959. Pp. 384-422.
OSGOOD, C. E., & TANNENBAUM, P. H. The
applying the results of such a pro- principle of congruity in the prediction
gram of research. of attitude change. Psychological Review,
REFERENCES 1955, 62, 42-55.
STEINER, I. D., & ROGERS, E. D. Alternative
BRYAN, R. C. Reactions to teachers by stu- responses to dissonance. Journal of Ab-
dents, parents and administrators. (United
States Office of Education, Cooperative normal and Social Psychology, 1963, 66,
Research Project No. 668.) Kalamazoo: 128-136.
Western Michigan University, 1963. (Received July 11,1966)

You might also like