16 I The Urban Review

Pygmalion T h e basic question to be answered in this chapter is whether in a period

of one ),ear or less the children of whom greater intellectual growth is ex-
pected will show greater intellectual growth than the undesignated c o n t ~ l -

in group children. T h e r e are also four important subsidiary questions. If

there were some advantages to a child whose teacher had favorable extCec-
rations for his intellectual development, would these expectancy advantages
be greater for:

the 1.
2.
3.
Children
Children
Children
in
in
of
the lower grades or higher grades?
the fast track, or m e d i u m track, or slow track?
one sex rather than the other?

Classroom 4. Children

THE MAJOR VARIABLES

Age
of minority group or nonminority group status?

by R o b e r t R o s e n t h a l ~ L e n o r e Jacobson T h e folk knowledge of our culture, current theories of h u m a n developme,nt,

especially psychoanalytic theory, and the work of the developmental and
experimental psychologists and of the ethologists are in agreement on the
In 1965 the authors conducted an experiment in a public elemen- importance of age as a factor in determining the degree to which an~or-
ganism can be shaped, molded, or influenced (Scott, 1962). In general, "the
tary school, telling teachers that certain children could be expected younger the organism, the greater is thought to be the degree of suscepti-
to be "growth spurters," based on the students" results on the Har- bility to social influence. In his classic monograph, Coffin (I941) concluded
vard Test of lnflected Acquisition. In point of [act, the test was that infiuenceability increased from infancy to ages seven to nine but de:
nonexistent and those children designated as "'spurters'" were chosen creased after that. More recently in a summary of the evidence bearing~on •
at random. What Rosenthal and Jacobson hoped to determine by overt social influence on children, Stevenson (1965) reported the gre~,~eI
influenceability of five-year-olds than twelve-year-olds, a finding consisten|
this experiment was the degree (if any) to which changes in teacher
with Coffin's summary. Both Coffin and Stevenson were writing about more
expectation produce changes in student achievement. overt social influence than the subtle, u n i n t e n d e d influence of teachers'
prophecies. Still, it would be interesting to know whether influence processes
of a more subtle, unintended form would also show younger children te
be the more susceptible.

Ability
We are also interested in learning whether the children of the three tracks
dither in the degree to which they profit from the teachers' favorable ,ex;
pectations. In the case of ability, however, the literature is not so helpful
in telling us what we might find. Stevenson (1965) suggested that suscel~,ti.
bility to social influence may not be too contingent on the child's intet.
lectual status, and we know that the three tracks differ considerably in
average IQ. One of the most recent discussions of intellectual gains is b}
T h o r n d i k e (1966) who reports that there are only modest correlations]oe,
tween initial intellectual status and changes in intellectual status. In the
present research, in any case, we are not so much interested in gains pep se
but rather in the excess of gain that might be show'a by the "special" q_hiI.
dren over the "ordinary" undesignated children. In short, we are interested
in differences among the tracks in the degree of expectancy advantage thin
may be found, but we hardly know what to expect. T h e matter is furkhe~
complicated by the fact that the other two variables in which we are in.
terested, sex and minority group status, are not independent of tr~c}
placement. In the last chapter we saw that boys tend to overpopulate ~th~
slow track relative to girls who tend to overpopulate the fast track. Mexica1:
children, Oak School's minority group, tend to overpopulate the slow trac]~
and underpopulate the fast track.

Sex
Whether boys or girls are the more susceptible to social influence processe~
Robert Rosenthal is professor of social psychology at Harvard University. depends on whether the influencer is male or female (Stevenson, 1965)
Lenore Jacobson is an elementary school principal in the South San Francisco Since the overwhelming majority of Oak School's teachers are females, th(
Unified School District. findings from research with lady influencers interest us most. Those findings
* Copyright @ by Holt, Rinehart & Winston, lnc. All rights reserved. This excerpt,
summarized by Stevenson (1965), suggest that boys should be the more sus
from Chapter 7, is reprinted with permission of the publisher and author. ceptible to social influence. As in the case of the children's age, however
 September 1968 I 17

the social influence processes employed were neither u n i n t e n d e d nor very

subtle. Effects of teachers' expectations are likely to be both.

Minority Grou~ S t a t u s
T~e reasons for our interest in the variable of minority-group status need
" q e justification. So much of the literature on the disadvantaged child
,acuses on the minority-group child that "disadvantaged" almost means
"minority group." One of the best known publications dealing with the
disadvantaged is called Youth in the Ghetto. We shall be especially inter-
ested, then, if expectancy advantages occur at all, in whether they benefit
mlnority-group children more o'r less than nonminority-group children.
At Oak School the minority-group child is Mexican. T h e definition of
a minority-group child in this research, however, was more stringent than
simply whether the name was Mexican. T o qualify as a "minority-group
Table 1
ch~,ld," either the child himself or his parents had to come from Mexico, MEAN GAIN IN TOTAL 1Q AFTER ONE YEAR B Y EXPERIMENTAL- A N D
Spanish had to be spoken at home, and the child had to be present for the CONTROL-GROUP CHILDREN IN EACH OF SIX GRADES

administration of certain procedures. These procedures, in connection with Control Experimental Expectancy Advantage
another study (Jacobson, 1966), included administration of an IQ test in Grade N Gain N Gain 1Q Points One-Tail p < .05 a
Spanish, a test of reading ability, and the taking of photographs of the child
1 48 +12.0 7 +27.4 +15.4 ,002
himself. W i t h i n this sample of Mexican minority-group children there were 2 47 + 7.0 12 +16,5 + 9.5 .02
3 40 + 5.0 14 + 5.0 - 0,0
vaeiations in how "Mexican" each child looked. A group of ten teachers 4 49 + 2,2 12 + 5.6 + 3.4
5 26 +17.5 ( - ) 9 +17.4 (+) -- 0.0
with no connection to Oak School or its children rated each photograph 6 45 +10.7 11 +10.0 -- 0.7
Total 255 + 8.42 65 +12.22 + 3.80 .02
o n "how Mexican the child looked." T h e definition of how clearly Mexican
a L~hild "really" looked was the average rating of all ten teachers. These a Mean square within treatments within classrooms = 164.24.
ratings were highly reliable. T h e average rating of the same children by the
teachers of Oak School was correlated .97 with the ratings of the judges
who were not associated with Oak School.

INTELLECTUAL GROWTH
Exe'pectaneyAdvantageby Grades
T h e bottom row of T a b l e 1 gives the over-all results for Oak School. In
the year of the experiment, the undesignated control-group children gained
over eight IQ points while the experimental-group children, the special Figure 1
GAINS IN TOTAL IQ IN SIX GRADES
ch~idren, gained over twelve. T h e difference in gains could be ascribed to
chance about 2 in 100 times (F = 6.35).
Fhe rest of T a b l e 1 and Figure 1 show the gains by children of the two Control Group ~ Experimental Group

groups separately for each grade. W e find increasing expectancy advantage 30

as we go from the sixth to the first grade; the correlation between grade 27.4
level and magnitude of expectancy advantage (r = --.86) was significant at 27

'th~ .03 level. T h e interaction effect, or likelihood that at different grades

24
there were significantly greater expectancy advantages, was significant at
th~ .07 level (F : 2.13). (Interactions, however, are not sensitive to the 21

ordering of differences unless one makes them so with further statistical 18

efforts; that is, the p of .07 is conservative.)
In the first and second grades the effects of teachers' prophecies were 15

dve~matic. T a b l e 1 shows that, and so does T a b l e 2 and Figure 2. T h e r e we 12

find the percentage of experimental- and control-group children of the first
two grades who achieved various amounts of gain. In these grades about 9

every fifth control-group child gained twenty I Q points or more, but of the 6
~pecial children, nearly every second child gained that much.
So far we have told only of the effects of favorable expectancies on total 3

IQ, but Flanagan's T O G A yields separate IQs for the verbal and reasoning 0
spheres of intellectual functioning. These are sufficiently different from each First Second Third Fourth Fifth Sixth
other so it will not be r e d u n d a n t to give the results of each. In the case Grades
o P v e r b a l I Q the control-group children of the entire school gained just
less than eight points, and the special children gained just less than ten,
a difference that could easily have arisen by chance. T h e interaction term
was not very significant (p < .15) so that we can not conclude greater ex-
pectancy advantage at some grade levels than at others. But we do have
a special interest now in the first and second graders, and it will do no harm
18 I The Urban Review

Table 2 to see what happened there in particular. In those combined grades, the
PERCENTAGES OF FIRST A N D SECOND GRADERS G A I N I N G A T LEAST TEN,
T I V E N T Y , OR T H I R T Y TOTAL 1Q POINTS control-group children gained 4.5 verbal IQ points, and the special chil:
dren gained exactly 10 points more, or 14.5. If we may have the t test (ev~en
1Q Gain at Least Control N = 95 Experimental N = 19 One-Tail p of Difference
though the interaction effect was not significant) we would find t = 2.24,
10 points a 49% 79% .02
20 points b 19% 47% .01 p < .02, one-tail.
30 points 5% 21% .04
For grades three through six the control gained 1.6 points more tile
a Includes children gaining twenty and thirty points or more. the experimental group, a difference not nearly significant. T a b l e 3 sun.
Includes children gaining thirty points or more.
marizes these results.
T h e advantage of favorable expectations showed itself more clearly~in
reasoning I Q as shown in T a b l e 4. For the school as a whole, the advantage
of favorable expectations was a seven point net gain in reasoning I Q ( [ * =
6.98), and there were no significant differences in the six grades in degree
of expectancy advantage. Once again, the younger children benefited most.
While we are not especially interested in the magnitude of IQ gain of the
Figure 2 control group, it does seem remarkable that the younger children of e,~.n
PERCENTAGES OF FIRST A N D SECOND GRADERS G A I N I N G the control group should gain so heavily in reasoning IQ. T a b l e I shows
T E N , T W E N T Y , OR T H I R T Y TOTAL IQ P O I N T S
that control-group children gained substantially in total I Q and not o{dy
Control Group ~ Experimental Group at the younger ages where we might expect practice effects to be m~-st
dramatic. T h e r e is no way to be sure about the matter (we shall return
to it in a later chapter), but it may be that experiments are good for chil-
80
dren even when the children are in the untreated control group.
70
Expectancy Advantage by Tracks and Sex
60
None of the statistical tests showed any differences among the three tracks
50 in the extent to which they benefited from teachers' favorable prophecies.
T h a t was the case for total IQ, verbal IQ, and reasoning IQ. W h e n the
40
entire school benefited as in total IQ and reasoning IQ, all three tracks
30 benefited; and when the school as a whole did not benefit much, as~-~in
verbal IQ, none of the tracks showed much benefit. For all three ] ( ~
20
measures, the tendency was for the middle track, the more average childf, n,
10 to benefit most from being expected to grow intellectually, b u t the ,ii~-
ference could easily have occurred by chance.
0
10 IQ Points 20 IQ Points 30 IQ Points In total IQ, girls showed a slightly greater advantage than boys of having
been expected to show an intellectual spurt; but to see what really hap-
pened we must look at boys' and girls' expectancy advantages for the two
subtypes of IQ. T a b l e 5 shows the gains in all three types of I Q by b#ys
and girls of the experimental and control groups. In verbal I Q it was the
boys who showed the expectancy advantage (interaction F = 2.13, p = .16);
in reasoning IQ it was the girls who showed the advantage, and it was
dramatic in size (interaction F = 9.27, p = .003). Just why that should be
Table 3
is not at all clear. On the pretest, boys had shown a higher verbal IQ than
M E A N G A I N I N VERBAL IQ AFTER ONE Y E A R B Y EXPERIMENTAL. A N D girls (4.4 points), and girls had shown a higher reasoning I Q than b~ys
CONTROL-GROUP CHILDREN I N GRADES O N E - T W O A N D T H R E E - S I X
(8.5 points). Apparently each group profited more from teachers' prol~he-
Control Experimental Expectancy Advantage cies in the area of intellectual functioning in which they were already a
Grades N Gain N Gain IQ Points One-Tail p < ,05 ~ little advantaged.*
It was mentioned earlier that expectancy advantage was not dependent
1-2 95 +4.5 19 +14.5 +10.0 .02
3-6 174 +9.6 49 + 8.0 -- 1.6 on placement in any one of the thJ:ee tracks. T h a t conclusion is modified'
Total 269 +7.79 68 + 9.85 + 2.06
when we examine expectancy advantages in the three tracks separately ~'oi
Mean square within = 316.40. boys and girls. Only for reasoning IQ is there a statistically significan~
effect (triple interaction F = 3.47,/9 < .04). T a b l e 6 shows the excess of gain
in reasoning I Q by the experimental over the control boys and girls in each
of the three tracks. We already knew that girls showed the greater e x p ~ t -
Table 4
M E A N G A I N I N R E A S O N I N G IQ AFTER O N E Y E A R B Y EXPERIMENTAL- A N D ancy advantage in reasoning IQ, and from Table 6 we see that this wa~
CONTROL-GROUP CHILDREN I N GRADES O N E - T W O A N D T H R E E - S I X
significantly more true in tile medium track, the track with the more
Control Experimental Expectancy Advantage average children.
Grades N Gain N Gain IQ Points One-Tail p < .05 ~ W e knew also that girls are over-represented in the fast track. Thes
1-2 95 +27.0 ( - ) 19 +39.6 (+) +12.7 .03
are the brighter girls from whom a lot is already expected. T h e slow track
3--6 160 + 9.1 ( - ) 46 +15.9 (+) + 6.9 .06 girls tend to be relatively very slow at Oak School, and we know that girls
Total 255 +15.73 65 +22.86 + 7.13 ,005
only rarely are placed there, and that they represent a real challeng( to
a M e a n square w i t h i n = 6 6 6 . 5 8 . Oak School's teachers. Of the middle-track girls there is little to say--
 September 1968 I 19
III I

teachers tend to find them uninteresting; pre-existing expectations about Table 5

MEAN GAIN IN THREE 1Q SCORES AFTER ONE YEAR BY EXPERIMENTAL
thek" intellectual ability are neither favorable as in the fast track nor very AND CONTROL BOYS AND GIRLS
tmfavorable
,-y and challenging as in the slow track. Perhaps when teachers Control Experimental Expectancy Advantage
are given favorable expectations about these children a greater increment N Gain N Gain IQ Points One.Tailp < .06
o~interest results than when expectations are given of girls in the outer
trajzks. That is a possible explanation of the greater effect in the average Total tQ
Boys 127 + 9.6 32 +12.5 + 2.9
crack of teachers' favorable expectations for girls' intellectual growth. Why Girls 128 + 7.3 33 +12.0 + 4.7 .04

.he'growth should be in reasoning IQ in particular is not at all clear, but Verba! tQ

Boys 136 + 8.4 ( - ) 34 +13.9 (+) + 5.6 .06
~y;do know that for the girls in this experiment when there are advantages Girls 133 + 7.2 34 + 5.8 -- 1.4
of teacher prophecies they tend to occur in the reasoning sphere of intel- Reasoning IQ
lecZual functioning. Boys
Girls
127
128
+19.2
+12.3
32
33
+15.3
+30.2
- 3.9
+17.9 ,0002
A pupil's sex turned out to be a factor complicating the amount of
expectancy advantage found in the three tracks. Sex also complicated the
magnitude of expectancy advantage found in the younger children of the
fit~$ two grades compared to the older children of the upper four grades.
Fable 7 shows the number of IQ points by which the gains of the experi-
mdntal-group children exceeded the gains of the control-group children.
~'bese expectancy advantage scores are shown separately for each of the
Table 6
three IQ measures for boys and girls in the lower and upper grades. For EXCESS OF GAIN IN REASONING IQ BY EXPERIMENTAL OVER CONTROL
..Jta'-I IQ, although the "special" boys of the lower grades did profit from BOYS AND GIRLS IN THREE TRACKS AFTER ONE YEAR
being expected to grow intellectually, the girls of the lower grades gained Track Boys Girls
nearly three times as many IQ points as a function of favorable expectations
Fast -- 2.6 + 9.1
(triple interaction F = 2.96, p = .09). For verbal IQ there was no difference Medium -12.0 +42.0 a
Slow - 0.3 +12.5
between boys and girls at either grade leveI in the amount of profit from Total -- 3.9 +17.9
favorable expectations (triple interaction F < 1), although, as we learned
ear./l~ier, boys and girls of the lower grades were helped more than children p = .00003, one-tail,
of ,,upper grades three through six. For reasoning IQ, boys and girls at
digerent grade levels did show very different magnitudes of expectancy
ad~antage. Boys in higher grades performed better in contrast to girls in
o~.,er grades who performed better when they were expected to do better
Itriple interaction F = 8.14, p < .005). Most of that effect was due to the
extraordinary performance of the first- and second-grade girls of the ex-
Table 7
perimental group who gained over forty IQ points more than did the EXCESS OF GAIN IN THREE 1Q SCORES BY EXPERIMENTAL OVER CONTROL
control-grou p girls of the first and second grade. BOYS AND GIRLS IN TWO GRADE LEVELS AFTER ONE YEAR
To summarize our somewhat complex findings involving pupil's sex as a Boys Girls
factor, we may say most simply that girls bloomed more in the reasoning
Total IQ
Grades 1 - 2 + 6.1 +17.1~
Grades 3-6 + 2.3 -- 0.1
* This footnote will serve to illustrate the complexity of nature and the need for
non~omplacency in the behavioral researcher. Preliminary results of a study conducted Verbal IQ
Grades 1-2 +10.8 a + 9.5
with Judy Evans give just the opposite results and with an equally significant Grades3-6 + 2,8 -- 5.8
probability level. The same basic experiment conducted at Oak School was repeated in Reasoning IQ
two elementary schools located in a small Midwestern town. Unlike Oak School, which Grades 1-2 -10.7 +40.2 ~
Grades 3 - 6 + 3.6 +10.0 a
drew its pupils from a lower-class community, these schools drew their pupils from a
~ub-~tantial middle-class community. Oak School's student body included a large
p < ,05, one-tail (or .10 two-tail).
proportion of minority-group members; the two Midwestern schools did not. The mean p ~ .0002, one-tail.
pret,2st total I Q at Oak School was 98, compared to the pretest total I Q of 105 found p < .00002, one.tail.
in these Midwestern schools. Eight months after the teachers had been given the names
o/iheir "'special" children, retests were administered. The results of the studies at the
two schools were su~ciently similar that the results could reasonably be combined. No
~xpectancy advantage was found for either boys or girls as measured by total I Q or
verJoal IQ. For reasoning I(2, however, the results were opposite to those found at Oak
~chool. Now it was the boys who showed the benefits of favorable teacher expectations.
Th@'e who had been expected to bloom gained over sixteen I Q points compared to the
less than nine gained by control-group boys. Among the girls it was the control=group Table 8
:hildren who gained about fifteen I Q points while those of the experimental group CORRELATIONS BETWEEN MEXICAN FACIAL CHARACTERISTICS AND ADVANTAGES
OF FAVORABLE EXPECTATIONS AFTER ONE YEAR
ga;~ned just over five I Q points. ( T h e interaction F was 9.10, p ~ .003.) In these
schools, just as in Oak School, boys had shown higher pretest verbal 1Qs than girls while Boys Girls Total
gir~s had shown higher pretest reasoning 1Qs than boys. Therefore, in these middle-class
N r N r N r
schools it was not true that each sex benefited most from favorable teacher expectations
~n ~7vose areas in which they were already somewhat advantaged. At the time of this Totall0 7 +.70 a 9 -.14 16 +,27
writing'there appears to be no ready explanation for this dramatic and very highly VerbalIQ 7 +.54 10 --.11 17 +.21
.t . ReasoninglQ 7 +.75 ~ 9 --.01 16 +.14
statf~tzcally significant reversal (p = .00004) in the two studies. But now we know for
sure'gthat Oak School's results, like the results of all behavioral experiments, are a p = .08, two-tail.
not ~niversaL b p = ,05, two-tail.
20 I The Urban Review
II

sphere of intellectual functioning, and boys bloomed more in the verbal expectancies not because they necessarily are more malleable but rattier
sphere of intellectual functioning when some kind of unspecified blooming because they are believed by teachers to be more malleable.
was expected of them. Furthermore, these gains were more likely to occur A fourth interpretation suggests that younger children are more sensitive
to a dramatic degree in the lower grades. T h a t susceptibility to the un- to and more affected by the particular processes whereby teachers cdm-
intended influence of the prophesying teacher should be greater in the municate their expectations to children. Under this interpretation, it is
lower grades comes as no special surprise. All lines of evidence tend to possible that teachers react to children of all grade levels in the same way
suggest that it is younger children who are the more susceptible to various if they believe them to be capable of intellectual gain. But perhaps i(:is
forms of influence processes. T h e influence of a teacher holding favorable only the younger children whose performance is affected by the spe~al
expectations may not be so very different. W h y the boys gained more in things the teacher says to them, the special ways in which she says them,
verbal IQ when expected to gain intellectually, and why the girls gained the way she looks, postures, and touches the children from whom she ~x-
more in reasoning IQ is not so easily explained. Earlier we did mention pects greater intellectual growth.
the possibility that children profit more from vague teacher expectations A fifth interpretation suggests that the effects of teachers' expectations
• lk'

in those spheres of intellectual functioning in which they tend to be slightly were more effective in the lower grade levels not because of any difference
advantaged to begin with. In Oak School, the pretest verbal IQs were higher associated with the children's age but rather with some correlated sampling
for boys than for girls by over four points; the pretest reasoning IQs were "errors." Thus it is possible that the children of the lower grades are the
higher for girls than for boys by over eight points. children of families that differ systematically from the families of the chil-
dren of the higher grade levels.
A sixth interpretation also suggests that the greater IQ gain in younge?
Expectancy Advantageby Minority-GroupStatus
children attributable to teacher expectation is a result of sampling error,
In total IQ, verbal IQ, and especially reasoning IQ, children of the minor- not in the sampling of children this time but in the sampling of teach0rs.
ity group were more advantaged by favorable expectations than were the It may be that in a variety of demographic, intellectual, and personalit
other children though the differences were not statistically significant. variables, the teachers of the younger children differed from the teachers
For each of the Mexican children, the magnitude of expectancy advan- of the older children such that they may have (1) believed the com~mu
tage was computed by subtracting from his or her IQ gain the IQ gain nications about their "special" children more or (2) been more effective
made by the children of the control group in his or her classroom• T h e communicators to their children of their expectations for the children's
resulting magnitudes of expectancy advantage were then correlated with performance.
the "Mexican-ness" of the children's faces. Tables 7-8 show the correlations There is some evidence to suggest that teachers of the lower grades do
obtained among Mexican boys and girls when expectancy advantage was in fact differ from the teachers of the upper grades of Oak School. Two
defined by total, verbal, and reasoning IQs. For total IQ and reasoning administrators who were well acquainted with all the teachers rated them
IQ, those Mexican boys who looked more Mexican benefited more from on over-all effectiveness as teachers• The two administrators agreed well in
teachers' favorable expectations than did the Mexican boys who looked their ratings (r = +.88) and, although there were many exceptions, teachers
less Mexican. There is no clear explanation for these findings, but we can of the lower grades were judged to be more effective teachers by both ~d-
speculate that the teachers' pre-experimental expectancies of the more ministrators (average r between effectiveness and teaching grade = --.57.
Mexican-looking boys' intellectual performance was probably lowest of all. p < .02).
These children may have had the most to gain by the introduction of a T h e finding that only the younger children profited after one year frbm
more favorable expectation into the minds of their teachers. their teachers' favorable expectations helps us to understand better,~tbe
results of two other experimenters, Clifford Pitt (1956) and Charles Flowers
SOME DISCUSSION (1966). Pitt, it will be recalled, divided his sample of fifth-grade boys i~'to
The results of the experiment we have described in some detail provide three groups. For one group he reported the boys' IQ scores to the teachers
further evidence that one person's expectations of another's behavior may after having arbitrarily added ten points• For another group he reported
come to serve as a self-fulfilling prophecy. W h e n teachers expected that the boys IO scores after having deducted ten points. For the third gr6~up
certain children would show greater intellectual development, those chil- he reported the boys' actual IQ scores. Pitt found that there were no et~ects
dren did show greater intellectual development. For the basic year of the on school achievement at the end of the year of teachers having been given
experiment, the self-fulfilling prophecy was in evidence primarily at the false information about their pupils' IQ.
lower grade levels; it is difficult to be certain why that was the case. A T h e results of our own study suggest that after one year, fifth graderr
number of interpretations suggest themselves, and these are not mutually may not show the effects of teacher expectations though first and second
exclusive. ~aders do. Pitt's study differed in too many ways from our own to m~tke
First, younger children are generally regarded as more malleable, less direct comparisons possible, however. Pitt did not, for example, retest~the
fixed, more capable of change, more subject to the effects of critical periods children on IQ p e r se but only on school achievement. More important
(Scott, 1962). It may be, then, that the experimental conditions of our perhaps, is the fact that Pitt's teachers knew their pupils for nearly ~'wc
experiment were more effective with younger children simply because months before being given pupils' IQ scores. T h a t was long enough ~o~
younger children are easier to change than older ones. (It should be re- teachers to have developed realistic expectations of pupils' performane"
called that when we speak here of change we mean it as change relative more powerful than the expectations that could have been induced by
to control-group change. Table 1 showed that even fifth graders can change adding or deducting IQ points.
dramatically in IQ, but there the change of the experimental-group chil- T h e equivocal results of Flowers' experiment are also not directly corn,
dren was not greater than the change of the control-group children.) parable to our own data. Flowers' pupils were also older children (seventh
A second interpretation is that younger children within a given school graders) and each child had many different teachers rather than just c~e:
have less well-established reputations within the school. It then becomes Perhaps the effects of teachers' expectations were diluted by being distrib-
more credible to a teacher to be told that a younger child will show intel- uted over many teachers• In the case of Flowers' study, we must bear,in
lectual growth• A teacher may "know" an older child much better by mind, too, that the classes arbitrarily labeled as brighter had been assig~aed
reputation and be less inclined to believe him capable of intellectual growth different teachers than had been assigned to the control-group classes.
simply on someone else's say-so. Therefore, any differences between the experimental- and control-group
A third interpretation is a combination, in a sense, of the first two. It classes could have been due to differences in the quality of teachers as-
suggests that younger children show greater gains associated with teachers' signed to each. •