MOTHERS' PERCEPTIONS OF THEIR CHILDREN'S
INTELLECTUAL ABILITIES AND THEIR
RELATIONSHIP TO ACADEMIC ACHIEVEMENT
By
MARIA DELGADOHACHEY
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
To my husband, John, for all his love and support.
ACKNOWLEDGEMENTS
I would like to thank Dr. Scott A. Miller, the chairman
of my committee, for all his help in the design of this
study, for his excellent editorial assistance and for all
the guidance and support he has given me throughout the past
years. I would also like to thank the members of my com
mittee, Dr. Walter Cunningham, Dr. Patricia Miller,
Dr. Richard A. Griggs and Dr. Patricia Ashton,for all their
valuable suggestions and their help in the preparation of
this study.
My sincere appreciation is expressed to Dr. James
Algina for his assistance in the statistical analyses and
interpretation of the data. His expertise and patience were
invaluable.
I would also like to thank Mr. Larry Scott, principal
of Southside Estates Academy; Sister Eithne, principal of
San Jose Catholic School; and Ms. Maureen Thiec, principal
of The Chappell School, for their assistance in the recruit
ment of subjects and for providing the facilities at their
schools to conduct this research. Finally, I would like to
thank all the children who participated in the study for
their cooperation during the testing sessions and the
mothers of these children for giving of their time.
TABLE OF CONTENTS
PAGE
ACKNOWLEDGEMENTS........................................ iii
LIST OF TABLES........................................... v
ABSTRACT................................................ vii
CHAPTER
ONE INTRODUCTION...................................... 1
Parents' and Teachers' Expectations: Antecedents
and Effects..................................... 1
The Accuracy of Parents' Perceptions of their
Children's Intellectual Abilities ............... 17
Goal of Thesis................................... 21
TWO METHOD............................................ 29
Subjects ......................................... 29
Procedures....................................... 32
Variables........................................ 35
THREE RESULTS........................................... 53
Descriptive Results................. ............ 53
Intercorrelations among the Variables ............ 74
Results of the Multiple Regression Analyses...... 87
FOUR DISCUSSION....................................... 105
APPENDICES
A RECRUITMENT LETTER............................... 120
B MOTHERS' QUESTIONNAIRE........................... 122
C HUMAN SUBJECTS CONSENT FORM....................... 129
D INTERCORRELATIONS AMONG ALL VARIABLES............ 132
REFERENCES............................................... 134
BIOGRAPHICAL SKETCH ..................................... 136
LIST OF TABLES
TABLE PAGE
1 Age and Sex Distribution of the Children .......... 55
2 Percentage of the Children's Fathers at Each
Level of the Hollingshead's Educational Scale.... 56
3 Percentage of the Children's Fathers at Each
Level of the Hollingshead's Occupational Scale... 57
4 Percentage of Mothers at Five Different
Education Levels................................. 58
5 Percentage of the Children at Five Different
Levels of IQ..................................... 59
6 Percentage of Mothers Who Were Accurate and Who
Overestimated and Underestimated their
Children's IQ Scores.............................. 63
7 Average Frequency Ratings Given by the Mothers to
Each of the 14 Situations Listed in Question #9.. 65
8 Average Frequency Ratings Given by the Mothers to
Each of the 14 Situations Listed in Question #10. 67
9 Percentage of Mothers at Each Level of Minimum
Demands for Academic Achievement................. 68
10 Percentage of Mothers at Each Level of Pleasing
Demands.......................................... 70
11 intercorrelations among the Mothers' Estimates
of their Children's Intellectual Abilities and
their Children's Real IQ Scores................... 77
12 Intercorrelations among the Children's IQs, GPAs,
SATs, the Mothers' Level of Education and the
Families' SES..................................... 79
13 Intercorrelations among the Mothers' Demands for
Academic Achievement and the Mothers' Estimates
of their Children's Abilities .................... 81
14 Correlations between the Global Scores of the
Frequency Measures and Each of their Respective
Items............................................ 84
TABLE PAGE
15 Percentage of Mothers at Three Education Levels
Who Overestimated, Underestimated, or Were
Accurate in Predicting their Children's IQ
Scores........................................... 93
16 Intercorrelations among All Variables............. 133
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy
MOTHERS' PERCEPTIONS OF THEIR CHILDREN'S
INTELLECTUAL ABILITIES AND THEIR
RELATIONSHIP TO ACADEMIC ACHIEVEMENT
By
MARIA DELGADOHACHEY
April 1984
Chairman: Dr. Scott A. Miller
Major Department: Psychology
The purpose of this study was to determine (1) whether
the accuracy of mothers' perceptions of their children's in
tellectual abilities could predict their children's academic
achievement, (2) whether the demands made by mothers for
their children's academic achievement varied as a function
of their perceptions of their children's abilities, and
(3) whether these demands could predict their children's
academic achievement.
The participants were 70 elementary school children and
their mothers. The mothers were asked to estimate their
children's IQs and to indicate at what level of their chil
dren's academic achievement they would let them know they
were pleased with their performance and at what level they
would be dissatisfied. The children were administered a
standardized IQ test (WISCR) and data were gathered on
their school grades and Stanford Achievement Test scores.
The accuracy of the mothers' perceptions was determined by
taking the difference between the mothers' IQ estimates and
the children's real IQs. These accuracy scores and the moth
ers' level of demands were then used in several multiple
regression analyses to answer the main questions of the
study. The following variables were included as controls:
the children's age, sex, IQs, the mothers' education
levels and the families' SES. It was predicted that
the children with the most accurate mothers would have the
highest level of academic achievement. It was also pre
dicted that the mothers' demands would vary as a function of
their perceptions of their children's abilities and that
these demands in interaction with the children's IQs would
predict the children's actual school performance.
The results were the following: (1) the mothers' per
ceptions were found to be relatively accurate but the ac
curacy did not predict their children's academic achieve
ment; (2) a positive relationship was found between the
mothers' demands and their beliefs about their children's
abilities; (3) a positive relationship was found between the
absolute level of the mothers' demands and their children's
school performance. Since the study was correlational in
nature, no causal interpretations could be made about the
above relationships. It was concluded that the results
supported the model of the cycle of influences between
expectations and behaviors.
VI i
CHAPTER ONE
INTRODUCTION
Parents' and Teachers' Expectations:
Antecedent and Effects
There are wide individual differences in academic
achievement among children. Numerous studies have been
conducted trying to determine what variables are respon
sible for these individual differences. Several variables
have already been identified as good or moderate predictors
of school achievement. Among them are certain demographic
characteristics of students such as socioeconomic status
(SES), race, ethnic background, ordinal position, family
size, etc. (Henderson, 1981). Performance on standardized
tests of mental abilities or intelligence quotient (IQ)
scores have also been well documented as good predictors of
school achievement (Stanley & Hopkins, 1972). In addition,
certain specific personality characteristics of students
such as achievement need (as measured by projective tests or
personality inventories), self concept, locus of control and
others have also shown to be predictive of school perform
ance (Naylor, 1972; Purkey, 1970).
Among the many variables that have been studied as pos
sible predictors of school achievement is the expectancy
variable. The effect that expectations have on achievement
behavior has been studied from many different perspectives.
Some researchers have focused on the expectations held by
I
teachers of their students' achievement behavior (Braun,
1976); others have focused on the expectations held by
parents of their children's school performance (Callard,
1968; Entwisle & Hayduk, 1978; Mahan, 1975; Seginer, 1983);
and still others have focused on the expectations held by
the students of their own achievements (Entwisle & Hayduk,
1978; Rappaport & Rappaport, 1975; Stipek & Hoffman, 1980).
Perhaps one of the most publicized studies of the in
fluence of expectations on intellectual behavior is Rosen
thal and Jacobson's 1968 study. These investigators tried
to determine whether manipulating the teachers' beliefs or
expectations regarding the abilities of their students would
produce changes in the children's intellectual behavior.
The manipulation of the teachers' beliefs regarding the
abilities of their students was done by giving the teachers
a list of names of children who supposedly had been identi
fied as potential academic "spurters." More specifically,
the teachers were told that the children had been adminis
tered the "Harvard Test of Inflected Acquisition." This
test, they were told, could identify those children who were
likely to experience a spurt in academic and intellectual
performance during the coming school year. In actuality,
Rosenthal and Jacobson had administered Flanagan's Test of
General Ability which is a standardized IQ test. This test
was readministered again during the middle and at the end of
the school year. In addition, Rosenthal and Jacobson gath
ered data on the children's general academic achievement
3
test scores, the children's school grades and the teachers'
ratings of the children's behavior in the classroom. At the
end of the school year, Rosenthal and Jacobson compared the
intellectual performance of the children who had been la
belled potential "spurters" to that of a control group of
children. The results of the study showed that at the end
of the school year greater intellectual gains were obtained
by the experimental children than by the control children.
The authors also reported that the teachers rated the behav
ior of the experimental children in more positive ways than
that of the control children. Rosenthal and Jacobson con
cluded that the expectations held by the teachers with re
gards to the experimental children were probably responsible
for the intellectual gains observed among these children.
Rosenthal and Jacobson's (1968) study was later criti
cized on many accounts and the "Pygmalion effect" they
claimed to have demonstrated failed to replicate in numerous
subsequent studies. Elashoff and Snow (1971) have reviewed
in detail many of the design, sampling and measurement prob
lems that plagued the original Rosenthal and Jacobson study.
Among the many criticisms discussed in this review were the
following: poor and illdefined procedures for assigning
the children to the experimental and control groups, sub
stantial and differential subject attrition from the experi
mental and the control groups, the use of a standardized IQ
test which had not been fully normed for use with younger
children, the use of untrained teachers for the administra
tion of the IQ test, lower than normal pretest scores among
4
the younger children as well as extremely large pre to
posttest gains in IQ scores among these children. The most
serious of the criticisms, however, was with regards to the
validity of the experimental procedure itself. Rosenthal
and Jacobson (1968) included a teacher interview and memory
test at the end of their experiment. These procedures were
included to validate the effects of the experimental manipu
lation technique used in the study, that is, to make sure
the deception of the teachers had worked and their expecta
tions of the experimental children had, in fact, been
changed. The results of the interviews and the memory tests
showed, according to Rosenthal and Jacobson (1968), that the
teachers could not remember the names of the experimental
children and that many of them had only casually looked at
the lists of names given to them by the experimenters.
Elashoff and Snow (1971) criticized the authors for failing
to see the important implications of the teachers' reports
and concluded: "Evidently the Pygmalion effect, if any, is
an extremely subtle and elusive phenomenon that acts through
teachers without conscious awareness on their part" (p. 42).
Numerous studies on teachers' expectations followed the
initial Rosenthal and Jacobson's (1968) study. Some of these
follow up studies were direct attempts at replicating the
Pygmalion effect. Others were simply related studies attempt
ing to further explore teacher expectancies. Baker and Crist
(1971) reviewed several of these follow up studies, the major
ity of which failed to replicate Rosenthal and Jacobson's find
ings. According to these authors, the studies which were the
least likely to replicate their findings were those which fol
lowed Rosenthal and Jacobson's (1968) procedures more closely.
They argued that the manipulation technique used by Rosenthal
and Jacobson to increase the teachers' expectations of the ex
perimental children was too weak to produce the type of effects
expected. The authors also pointed out that significant ef
fects of expectations were more likely to be found in studies
that did not try to manipulate teachers' expectations but
rather assessed the effects of the teachers' expectations which
already existed naturally.
Dusek and O'Connell (1973) showed that naturally formed
teacher expectancies are, in fact, more likely to affect
students' achievement behavior than experimentally induced
expectations. In their study, these authors asked the
teachers of a group of second and fourth grade students to
rank their students in terms of how well they thought they
would perform at the end of the school year in language and
arithmetic skills. The experimenters then randomly divided
the students into a control and an experimental group. The
teachers were given the names of the experimental children
and were told that these children had been administered a
test which had shown that they would show large improvements
in language and arithmetic skills throughout the school
year. In actuality, the children (both the experimental and
the control group) had been administered the Stanford
Achievement Test. This test was again administered at the
middle and at the end of the school year. The results of
6
the study showed that the students who were initially ranked
high by their teachers had higher SATs at all three adminis
trations of the test than the children who were ranked
lower. The results also showed that the experimental manip
ulation of teachers' expectations had no effect on the chil
dren's SAT scores. Most researchers today would agree that
the formation of expectations is a complex phenomenon in
volving many variables and that experimentally induced ex
pectations may not have the same kind of influence on
achievement behavior as naturally formed expectations
(Braun, 1976).
Despite all the problems with the Rosenthal and Jacob
son (1968) study and the failure of the follow up studies to
replicate their findings, research on expectancy effects has
continued to date although it has taken many new and different
directions. The theoretical background or the logical basis
underlying research on expectancy effects is a sensible one
and perhaps this is one reason why research in this area has
continued (Braun, 1976). Expectations are believed to affect
intellectual behavior in the following way: The expectations
held by teachers or parents about a given child affect their
own behavior towards that child. The behavior of the teacher
or parent in turn affects the intellectual behavior of the
child which then serves to confirm and reinforce the initial
expectations of the teacher or parent. This sequence of in
fluences creates a cycle which is selfperpetuating. In addi
tion, the expectations of teachers and parents are believed to
7
affect the child's own selfexpectations which also influence
his intellectual or achievement behavior. Figure 1 illus
trates this hypothesized cycle of influences between expec
tations and behavior. It should be noted that this illus
tration is a simplified version of the models presented by
other authors (Braun, 1976; Seginer, 1983). Braun's (1976)
model, for example, is more detailed but it is useful only
to illustrate the effects of teachers' expectations of their
students and not the effects of parents' expectations.
Braun's model includes a number of variables that are like
ly to influence the natural formation of teachers' expecta
tions of their students. Among the variables included are
the following: the sex of the students, their IQ score,
their physical appearance, their previous achievement
scores, their cumulative folders, their ethnic background,
the students' names, the teachers' knowledge of the stu
dents' siblings, the SES of the students' families, and the
students' present achievement behaviors. In addition,
Braun's model also includes a number of teachers' behaviors
that are likely to vary as a function of the expectations
they hold of their students. These behaviors are believed
to influence the students' selfexpectations and achievement
behaviors. Among the teachers' behaviors listed are the
following: quantity of interaction with the students, differ
ential grouping of the students within the classroom, differen
tial activities and questions provided for the students, and
Figure 1. Hypothesized cycle of influences between
expectations and behavior.
9
qualitative differences in the questioning of students
(i.e., in prompting and waiting for the students' answers).
Seginer's (1983) model is also more detailed than the
one presented in Figure 1. Her model, however, is only use
ful to illustrate the effects of parents' expectations of
their children. Like Braun (1976), Seginer includes in her
model some of the antecedent variables which may influence
the formation of expectations except her variables apply to
the expectations held by parents not teachers. Among the
variables she considers influential she lists the parents'
own educational aspirations, the feedback provided by the
schools of their children's achievement behaviors, and a
variable she terms "parental knowledge." This last variable
refers to the general knowledge parents may have about the
development of children, to their knowledge of the perform
ance of children on intellectual tasks and to their assess
ments of their own children's development and performance.
In her model, Seginer also includes some of the parental be
haviors which may vary as a function of the expectations
they hold of their children. In particular, she lists a
category of behaviors she calls "achievement supporting be
haviors" and another she calls "differential reinforcements."
Although there are many differences between Braun's
(1976) and Seginer's (1983) models, the basic hypothesized
sequence of influences between expectations and behaviors is
the same in both models. It is this basic sequence of
influences underlying both models that is illustrated in
Figure 1. It is interesting to note that both Braun and
Seginer have expanded and provided details in their models
in similar areas. More specifically, they have both tried
to delineate some of the antecedent variables that may in
fluence the natural formation of expectations and some of
the specific behaviors which are influenced by these expec
tations. This is probably a reflection of some of the new
directions that research on expectancy effects has taken
since the initial Rosenthal and Jacobson (1968) study.
The search for variables that influence the natural
formation of expectations has, in fact, been one of the
areas where research on expectancy effects has been ex
panded. This research has been most productive with re
gards to teachers' expectations of their students and
students' expectations of themselves. Braun (1976) has re
viewed many of these studies and his model includes most of
the variables that have been found to influence teachers'
expectations of their students. Unfortunately the search
for variables that affect the natural formation of parents'
expectations of their children has not been as productive.
Seginer (1983) has reviewed several studies of parental ex
pectations. Her review shows that although there are some
studies which have focused on these important antecedent
variables, the bulk of the search remains to be done.
As mentioned before, Seginer (1983) delineated three
important variables which may influence the formation of
parental expectations: the parents' own educational aspira
tions, school feedback, and parental knowledge. She cites
11
studies in support of all three of these variables but this
evidence is scarce. Among the more interesting studies she
reviewed is one by Entwisle and Hayduk (1978) which shows,
among other things, that school feedback may, in fact, in
fluence parents' expectations of their children. In their
study, Entwisle and Hayduk asked the parents of a group of
children entering first grade to predict the school grades
they thought their children would get. Their sample was
drawn from two schools, one with predominantly middle class
children and the other with mostly working class children.
They followed the children's school performance longitudi
nally until the end of their second year in school. They
also asked the parents to predict their children's grades
three more times: towards the end of first grade, at the
beginning of second grade, and again towards the end of
second grade. In addition, the children themselves were
asked to predict their own grades at the beginning and end
of each school year. The results showed that the feedback
parents received from the school about their children's
grades appeared to have an influence on their future predic
tions of the grades their children would get. That is,
parents apparently adjusted their expectations so that they
were more in line with their children's actual school per
formance. Entwisle and Hayduk also reported that when these
adjustments in expectations occurred they were more likely
to be upward adjustments. That is, expectations were more
likely to rise than to fall. They referred to this phenom
enon as the "buoyancy effect." It should be mentioned that
Entwisle and Hayduk also found that the initial predictions
parents made about their children's grades were very closely
related to the children's actual IQ scores. This suggests
that parental knowledge about their children also probably
plays a role in the formation of their expectations.
Again, this supports Seginer's (1983) model. More will be
said about other results of the Entwisle and Hayduk study
later in this chapter.
Seginer's (1983) suggestion that parental knowledge
about children in general and their own children in
particular may be an important variable influencing their
expectations is an interesting one. Unfortunately, there is
very little research on parental knowledge and beliefs about
children. In a recent 1980 article McGillicuddydeLisi has
called for research on variables that may affect this type
of parental knowledge. In this article, McGillicuddyde
Lisi argues that parental belief systems may be an important
variable affecting their parental practices towards their
children. She emphasizes the importance of studying these
belief systems and uncovering the variables that may have an
influence in their formation. Among the variables she sug
gests for study are the following: the parents' amount of
experience with children, the number of children they have,
the sex distribution and spacing of the children, the
parents' SES and others. These variables are supposedly
important because according to McGillicuddydeLisi parental
beliefs about children undergo progressive changes whenever
13
parents encounter new and discrepant information and try to
assimilate it into their current belief systems. The belief
systems of parents who have more children or have children
of different sexes are likely to be different from those of
parents with only one child or children of only one sex.
This is so because events such as the birth of a second
child or a child of a different sex expose parents to new
information about children and give them a chance to vali
date and modify their beliefs. McGillicuddydeLisi (1980)
reports some evidence that supports her claim that the above
variables may have some influence on parental belief sys
tems. More specifically, she reports having found that
parental beliefs about how children come to understand cer
tain concepts vary as a function of family configuration,
SES, the sex of the parent and the sex of the child.
Further research is needed to determine how parental beliefs
are influenced by the above mentioned variables.
It is interesting to note that although McGillicuddy
deLisi does not use the word "expectations," the kind of
relationship she proposes between parental belief systems
and parental behaviors towards their children is very simi
lar to that proposed for expectancy effects. McGillicuddy
deLisi (1980) conceptualizes the family as a "system of
mutual influences." She argues that not only are parental
beliefs and behaviors shaped by the variables she suggested
above, they are also shaped by the children's reactions to
the parents' beliefs and behaviors which in turn produce
further changes in the parents' belief systems and behaviors.
Research is needed to determine how parental beliefs about
children are formed and to determine what roles their
beliefs play in the regulation of their parental practices.
Another popular area of expansion for research on ex
pectancy effects has been the search for the different kinds
of behaviors that are influenced by expectations. Ini
tially, research on expectancy effects focused only on try
ing to determine whether teachers' expectations of their
students had an effect on the students' achievement behav
iors. Later, researchers began to focus their attention on
how the teachers' expectations of their students affected
the teachers' own behaviors towards those students. The
teachers' behaviors towards their students and the students'
own selfexpectations were believed to be the two mediating
processes by which expectations could affect the children's
academic achievement. Therefore, researchers made an effort
to delineate the specific teacher behaviors which may vary
as a function of the expectations they hold of their stu
dents. Braun (1976) has reviewed many of these studies.
The studies he has reviewed suggest that teachers do, in
fact, treat and interact with their students differently
depending on whether they believe the children to be high or
low achievers. Behaviors such as the amount of praise they
give to their students, the way in which they physically
structure the classrooms and assign children into different
learning groups, and all the other behaviors listed by Braun
15
(1976) in his model have been found to vary as a function of
teachers' expectations.
Unfortunately, research on how parents' beliefs and ex
pectations affect their own behavior towards their children
is very scarce. Seginer (1983) suggested in her model that
parents' achievement supporting behaviors may be one poten
tial category of behaviors which may vary as a function of
their expectations of their children. She also suggests
that parents may use some kind of reinforcement procedures
to make their children conform to their expectations, that
is, differentially reinforcing achievement behaviors that
are consistent with their expectations and ignoring or pun
ishing those which are not. This is a very important area
of research that deserves further study.
A study by Crandall, Dewey, Katkovsky and Preston
(1964) explored the relationship between parents' attitudes
about their children's achievement behaviors and their chil
dren's actual academic performances. They also examined how
certain selfreported parental behaviors related to the
children's actual academic performances. In this study, the
experimenters interviewed the parents of a group of second,
third and fourth grade children. In the interviews, the ex
perimenters gathered data on the following variables:
(1) the degree of importance or value the parents attached
to their children's intellectual achievements, (2) the par
ents' beliefs about their children's level of intellectual
competence, (3) the amount of dissatisfaction or satisfac
tion the parents felt about their children's intellectual
achievements, (4) the parents' minimal standards for their
children's achievement performances, (5) the frequency and
intensity with which the parents attempted to increase their
children's participation and competence in intellectual
activities, (6) the frequency and the extent of the parents'
participation with their children in intellectual activi
ties, (7) the frequency and intensity of the parents' posi
tive reactions to their children's intellectual accomplish
ments, (8) the frequency and intensity of the parents' nega
tive reactions to their children's lack of intellectual in
terests and accomplishments. These investigators also gath
ered data on other parental selfreported behaviors which
were nonspecific to the children's intellectual perform
ances. In addition, the investigators administered the
StanfordBinet Intelligence Test to the children and gath
ered data on the children's performances on the California
Achievement Test. The results of this study showed that
only a few of the parents' attitudes and selfreported
behaviors were related to the children's actual academic
performances. Also, many of the relationships found were
specific only to the mothers' attitudes and behaviors but
not the fathers'. For example, the mothers' beliefs about
their children's level of intellectual competence and their
degree of expressed satisfaction/dissatisfaction with their
children's performances were found to be positively related
to the children's actual academic achievements. These re
lationships, however, were not significant for the fathers'
beliefs. Another interesting finding was that only the
fathers' reactions (positive or negative) to their daughters'
achievements were found to be related to their daughters'
actual school performance. Finally, the parents' frequency
of participation with their children in intellectual activi
ties and the degree to which they attempted to increase
their children's involvement in these activities related
negatively to the children's actual school performance. It
should be pointed out that the children sampled for this
study were well above average in intellectual performance.
The average IQ for the sample was 124 with a standard
deviation of 16 and approximately 40% of the sample had IQs
above 130! Thus, it should be kept in mind that the results
reported by Crandall et al. (1964) may not generalize to
other samples of children with more normal levels of intel
lectual abilities. Despite this problem, however, this
study is an interesting attempt at trying to determine what
kind of parental attitudes and behaviors may affect chil
dren's academic achievement. It is also unfortunate that
these investigators did not try to determine whether the
parents' selfreported behaviors varied as a function of
their beliefs about their children's level of intellectual
competence. This is exactly the type of question that
Seginer and McGillicuddydeLisi would like to see answered.
The Accuracy of Parents' Perceptions of
Their Children's Intellectual Abilities
The accuracy of parents' perceptions of their chil
dren's intellectual abilities is an important variable
which unfortunately has been mostly ignored in research
studies on expectancy effects. These studies usually have
been done with the underlying assumption that expectancy
effects on achievement behavior can be produced by simply
increasing expectations. The level of ability of the child
is usually not taken into account. That is, it is usually
assumed that higher expectations will produce higher academic
achievement regardless of the level of ability of the child.
Mahan (1975), for example, attempted to raise student
achievement behavior by manipulating parental expectations.
Her sample consisted of a group of low SES elementary school
children who had scored in the bottom twothirds of the
Stanford Achievement Test. The parents of the experimental
children were contacted by their children's teachers who had
been instructed to tell the parents that their children were
capable of doing better in school. The teachers also met
with the parents of the control children but did not attempt
to raise parental expectations among them. The results of
the study showed that the scores of the Stanford Achievement
Test at the end of the school year were no different for the
experimental children than for the control children. The
study did not clearly show whether the parents' expectations
of the experimental children had been successfully manipu
lated although the author did report that more of the
experimental parents reported being dissatisfied with their
children's school work after talking to their children's
teachers.
19
Mahan's study had many serious flaws most of which were
addressed by the author in discussing her results. However,
one of the main problems with the study which was not
addressed was the fact that it did not take the students'
level of ability into account. The author simply assumed,
as is often done in research on expectancy effects, that ex
pectations could affect achievement behavior regardless of
the level of ability of the children involved. This assump
tion needs to be examined in future research.
Hunt and Paraskevopoulos (1980) have recently argued
that the accuracy of parents' perceptions about their chil
dren's intellectual abilities may play an important role
in their children's cognitive development and actual intel
lectual performance. These authors start out with the
premise that children benefit most from cognitive experi
ences which are moderately discrepant from their current
level of cognitive development. They then argue that trying
to achieve this moderate level of discrepancy when present
ing a task to a child requires accurate perception of the
child's current level of cognitive development. They refer
to this as the problem of the "match." According to Hunt
and Paraskevopoulos, parents with accurate perceptions of
their children's level of cognitive abilities should be able
to produce better "matching" experiences for their children.
Thus, this should lead to better cognitive development among
children with accurate parents.
Hunt and Paraskevopoulos (1980) attempted to find evi
dence in support of their hypothesis. In their study they
asked a group of 50 mothers to predict how their children
(39 to 54 years of age) would respond to a set of 96 test
items taken from three different standardized tests. They
also administered these same test items to the children to
determine the accuracy of the mothers' perceptions. The
authors expected to find a negative correlation between the
inaccuracy of the mothers' predictions and the children's
level of cognitive development. The results of the study
supported their hypothesis. A correlation of r = .80 was
found between the number of incorrect predictions given by
the mothers and the number of test items passed by the chil
dren. A closer examination of the results, however, showed
that this correlation could have resulted from a methodo
logical artifact. Apparently, the accuracy of mothers' pre
dictions was correlated to the children's level of cognitive
performance because the more items of the test the children
passed, the fewer overestimations or false predictions their
mothers could make. This possible methodological artifact
made the results of Hunt and Paraskevopoulos' study incon
clusive. Further research is needed to test their hypothe
sis and to determine whether in fact mothers' accuracy of
perception of their children's cognitive abilities is pre
dictive of their children's level of cognitive development.
Entwisle and Hayduk's (1978) study, which was described
earlier, also found some evidence which is congruent with
Hunt and Paraskevopoulos'claims. In their study, they found
that there was a racial difference in the level of parents'
initial expectations of their children's first grades in
school. White parents' initial expectations tended to be
very conservative (averaging slightly under a "B" grade) and
were highly correlated with their children's actual IQ
scores. Black parents' expectations were found to be un
realistically high and failed to correlate with their
children's IQ scores. The authors reported that when the
parents' expectations were too discrepant from the chil
dren's actual level of performance they failed to show any
effect on future grades. However, a slight discrepancy
between the parents' expectations and their children's
grades tended to predict a change in the children's future
grades. Whenever the children's grades showed change they
tended to change towards achieving greater consistency with
the parents' expectations. Interestingly, the children's
own expectations were also found to be very high and un
realistic regardless of the children's race. Entwisle and
Hayduk's results emphasize the need to take children's level
of ability into account when assessing the effects of expec
tations on children's achievement behavior. Their results
also point to the fact that parental accuracy in perceiving
their children's abilities may play an important role in the
formation of parents' expectations and in the effects these
expectations have on their children's achievements.
Goal of Thesis
There are several issues on expectancy effects that
seem to deserve further study. One of these issues is the
role that children's level of intellectual abilities may
play in determining the effects that expectations have on
their academic achievement. Research on expectancy effects
has been characterized by a failure to take this important
variable into account when assessing the effects of expecta
tions on academic achievement. Dusek and O'Connell (1973),
for example, tried to determine whether naturally formed
teachers' expectations had different effects on children's
academic achievement than experimentally induced expecta
tions. They found that naturally formed expectations pre
dicted the children's level of achievement while experimen
tally induced expectations did not. These investigators,
however, failed to control for differences in the children's
level of ability when assessing the effects of the two types
of expectations on the children's academic achievement. It
is possible that the reason why the teachers' naturally
formed expectations related to the children's achievement
measure was that these expectations may have been based on
their assessments of the children's actual intellectual
abilities. These assessments were probably fairly accurate
and thus related to the children's actual achievement per
formance.
Another way in which researchers have neglected to con
sider children's intellectual abilities as an important
variable influencing expectancy effects is by assuming that
children's academic achievement can be improved by simply
raising parents' and/or teachers' expectations of these
children (Mahan, 1975). Entwisle and Hayduk's (1978)
23
research, however, has indicated that the absolute level of
expectations may not be as important in predicting changes
in children's academic achievement as the fit or the degree
of discrepancy between these expectations and the children's
actual academic performance. They have also shown that when
expectations are unrealistically high they fail to have any
effect on children's academic achievement. It seems that if
expectations are to have any effect on achievement behavior
they must be based on relatively accurate assessments of
children's abilities. This issue of accuracy in the percep
tion of children's intellectual abilities is an important
one because the accuracy of perception helps determine the
degree to which the expectations held by teachers or parents
will be realistic given the level of ability of the children.
Hunt and Paraskevopoulos (1980) have also suggested the
important role that the accuracy of parents' perceptions of
their children's intellectual abilities may be playing in
their children's overall level of cognitive development.
Although the results of their study were inconclusive due to
a methodological artifact, their hypothesis is an interest
ing one which needs to be reexamined.
The goal of the present study was to determine whether
a specific aspect of parental beliefs about their children's
abilities, namely, the accuracy of their perceptions, pre
dicted their children's academic achievement. The study
also examined some of the variables that may affect the
accuracy of parents' perceptions of their children's abili
ties. This study differed from others in that it included
important controls of variables which are known to affect
children's academic achievement and which other studies have
typically ignored. It also differed in the way in which
mothers' perceptions of their children's intellectual abili
ties were assessed. Hunt and Paraskevopoulos (1980), for
example, asked mothers to predict how their children would
respond to a set of specific test questions. In the present
study the mothers' perceptions of their children's abilities
were assessed in a more global manner. A group of mothers
of elementary school children were asked to rate their chil
dren's overall level of intellectual abilities and to give
estimates of their children's IQ scores. It seems reason
able to assume that most parents have an overall impression
of their children's level of ability. It is the accuracy of
this overall impression that this study attempted to assess.
The accuracy of the mothers' perceptions of their children's
abilities was determined by computing a deviation score
which indicated how far or how close the mothers' estimates
of their children's IQ scores were to the children's real
IQs. These accuracy scores were then used in several
analyses to try to answer a number of questions about the
variables that may influence the accuracy of mothers'
perceptions and to answer the question of whether the
accuracy of mothers' perceptions could predict the chil
dren's academic achievement. All of the analyses performed
included the following variables as controls: the age of
25
the children, their sex, their IQ scores, the mothers' level
of education and the families' SES.
The more specific questions asked and the predictions
made were the following:
1. Does the accuracy of the mothers' perceptions of
their children's abilities vary as a function of the fol
lowing variables: the children's age, sex, IQ scores, the
mothers' level of education and the families' SES? It was
expected that the accuracy of the mothers' perceptions would
vary as a function of the mothers' level of education and
the families' SES. That is, mothers with more years of
formal schooling were expected to be more accurate than
mothers with fewer years of formal schooling. Likewise, the
mothers of children from higher SES families were expected
to be more accurate than those from lower SES families. In
addition, the mothers' accuracy was expected to vary as a
function of the children's ages. It was expected that moth
ers with older children would be more accurate than mothers
with younger children. The rationale behind this prediction
was that the mothers of the older children have had a chance
to receive more feedback from the schools about their chil
dren's intellectual performance than the mothers with
younger children. These mothers have also had more chances
to observe their children and to adjust their overall im
pressions of their children's abilities. Therefore, they
should be more accurate than the mothers of younger
children.
26
2. Does the accuracy of mothers' perceptions predict
the children's level of academic achievement? Data
were gathered on the children's grades in school and their
Stanford Achievement Test scores. These measures were
then used to determine whether academic achievement varied
as a function of the mothers' accuracy scores. It was ex
pected that the children with relatively accurate mothers
would have the highest level of academic achievement. The
children with inaccurate mothers (both those who overesti
mated and those who underestimated their children's abili
ties) were expected to have lower levels of academic
achievement.
3. Is there a relationship between the accuracy
of mothers' perceptions of their children's intellectual
abilities and the frequency of opportunities mothers
have to observe and compare their children's intellectual
abilities? The mothers were asked to report how frequently
they had the opportunity to observe their children's intel
lectual abilities under a variety of circumstances. They
were also asked to report how frequently they had the oppor
tunity to compare their children's abilities to those of
other children of their child's age. A relationship was
expected between the mothers' reports of the frequency of
opportunities they have to observe and compare their chil
dren and their accuracy scores. More specifically, mothers
who reported having more frequent opportunities to observe
and compare their children's abilities were expected to be
27
more accurate than those who reported having less frequent
opportunities.
Another aspect of expectancy effects research that
seems to deserve further study is the relationship between
parents' beliefs about their children's abilities and their
own behavior towards their children. Previous research has
shown that the behavior of teachers towards their students
varies as a function of the teachers' expectations of those
students (Braun, 1976). There is, however, very little re
search on how parents' behaviors towards their children vary
as a function of the parents' beliefs about their children.
The second goal of the present study was to examine a spe
cific type of parental behavior, namely the demands mothers
make for their children's academic achievement, and to deter
mine whether they vary as a function of the mothers' beliefs
about their children's abilities. It should be pointed out
that in the present study there was no direct measure of the
mothers' behaviors or the demands they made of their chil
dren. Instead, a selfreport measure was used in which the
mothers were asked to indicate the level of academic achieve
ment at which they would let their children know that they
were very pleased with their school work and the level at
which they would let them know they were dissatisfied. This
measure was similar to the Crandall et al. (1964) measure of
mothers' minimal standards of academic achievement which was
found to be related to their daughters' actual academic
achievement. The following specific questions were asked in
28
this study about the mothers' selfreported demands for
their children's academic achievement:
1. Do the mothers' demands vary as a function of
their perceptions of their children's intellectual abili
ties and the following variables: the children's age, sex,
IQ scores, the mothers' level of education and the families'
SES? Mothers who perceived their children as having higher
levels of intellectual ability were expected to make higher
demands of their children than mothers who perceived their
children as having lower levels of intellectual ability.
Also, the mothers of older children, children with higher
IQs, and children from higher SES families were expected to
make higher demands than the mothers of younger children,
children with lower IQs and children from lower SES fam
ilies. Finally, the more educated mothers were also ex
pected to make higher demands of their children than the
mothers with lower education levels.
2. Is there an interaction effect of the mothers'
level of demands and the children's IQ scores which
serves to predict the children's level of academic achieve
ment? Higher demands for academic achievement were expected
to predict higher levels of actual academic achievement only
for children with certain levels of IQ. The rationale be
hind this prediction is that the appropriateness of mothers'
demands, given their children's actual level of ability, may
be more important than the absolute level of the demands.
CHAPTER TWO
METHOD
Subjects
The participants of this study consisted of 70 children
and their mothers. The subjects were recruited through the
cooperation of three elementary schools in the Jacksonville,
Florida, area: San Jose Catholic School, The Chappell School
and Southside Estates Academy. Permission to conduct the
research at these schools was obtained by contacting the
principals of 15 private schools in the area. Only private
institutions were contacted because the Duval County School
Board had already denied permission to conduct the research
at any of their public schools. The above three schools
were the only private schools interested in participating in
the study.
The schools from which the sample was drawn were com
parable in many respects. The student enrollment at all
three of these schools is relatively small: San Jose Cath
olic has approximately 250 children enrolled in grades 1st
through 6th; The Chappell School has 185; and Southside
Estates Academy has 183. The slightly larger number of
students at the Catholic school is due to the fact that this
school has two sections of 5th and 6th graders. The average
class size, however, is about the same in all three schools,
that is, approximately 30 students per class. The schools
30
are also similar with respect to their tuition charges which
range from 135 to 150 dollars per month. Also, none of
these schools has any special selection criteria for the
admission of its students. The academic curriculum of these
schools is very similar to that of most public schools ex
cept for the addition of some specialized classes to the
private schools' curriculums. San Jose Catholic School, for
example, includes a Catholic catechism class in addition to
their basic academic curriculum; The Chappell School in
cludes a Spanish course; and Southside Estates Academy in
cludes a Bible class.
In order to recruit the subjects, 618 letters were sent
to the mothers of all the children of elementary school age
attending the above schools (grades 1st through 6th). The
recruitment letter explained the purpose of the research
study and asked the mothers to volunteer as participants.
The letter also let them know that their participation would
consist of a 45minute session in which they would be asked
to answer a questionnaire about their children's intellec
tual abilities and school work. In addition, this letter
asked the mothers for their permission to test their chil
dren and to have access to their children's school records.
A copy of the recruitment letter sent to the mothers is in
cluded in Appendix A.
The overall response rate to the recruitment letter was
approximately 12%. Among the mothers who responded, five
had to be excluded from the study due to the fact that they
already knew their children's intelligence quotient scores
31
(IQ scores) from previous administrations of intelligence
tests to their children. In addition, 15 of the mothers who
responded had more than one child in the age range required
for participation in the study. In these cases, only one of
the children was included in the study. The choice of which
child to include was made at random by the experimenter.
Seventeen children were excluded from the study for this
reason. The final participation rate by school was approxi
mately 14% (34 children) from San Jose Catholic, 11% (20
children) from The Chappell School, and 9% (16 children)
from Southside Estates Academy.
The children who participated in the study ranged in age
from 6 years 3 months to 12 years 6 months. The average age
was 9 years 5 months and the standard deviation was 1 year 10
months. Approximately half of the children were males (34)
and half were females (36). Most of them were white (90%)
with the exception of three black and four oriental children.
In addition, five of the white children were Spanish surnamed.
The majority of the children came from families in the
middle to uppermiddle socioeconomic status (SES) as indi
cated by their fathers' occupations and level of education.
Approximately 74% of the sample had fathers with educations
beyond the high school level and at least 80% of the chil
dren's fathers were engaged in white collar and/or profes
sional occupations. Since SES is one of the variables of
interest in this study, more will be said about the fami
lies' SES levels later in the Results chapter. It should be
noted that two children in the sample came from homes where
32
there was no father present. In one of these cases it was
due to the father's death and in the other it was due to
divorce. The majority of the children, however, came from
intact families (77%) and 19% came from families in which
the parents had been divorced but the mother was remarried.
Most of the mothers who participated in the study had
educations beyond the high school level and only one mother
had failed to complete high school. The number of years of
formal schooling for the total sample of mothers ranged from
9 to 20 years with a mean of 13.87 years and a standard
deviation of 1.81 years. More will be said about the moth
ers' education in the Results chapter.
The majority of the mothers interviewed for the study
had occupations outside their homes or were fulltime college
students (77%). The remainder were homemakers. For the
most part, the mothers who worked outside their homes were
engaged in white collar traditional female occupations such
as teaching or clerical and secretarial positions.
Procedure
The mothers who responded to the recruitment letter
were contacted by phone and an appointment was made to meet
with them at a time of their convenience. All the mothers
were seen individually at their own homes by a 26year old
female investigator. The meetings lasted approximately 30
to 45 minutes. At these meetings the mothers were asked to
complete a questionnaire about their children's intellectual
abilities and school work. A copy of this questionnaire is
included in Appendix B.
33
Before giving the mothers the questionnaire the experi
menter explained the purpose of the research and reminded
them of what would be required of them and their children as
participants of the study. At this point the mothers were
asked to read and sign a human subjects informed consent
form. A copy of this consent form is included in Appen
dix C. The initial explanations given by the experimenter
were usually brief since the same explanations were restated
in the consent form in lengthier form. After the subject
had read and signed this consent form and all her questions
had been answered, the experimenter introduced the question
naire by saying the following: "This is the questionnaire
that I would like you to fill out. I would like you to fill
it out with (child's name) in mind. Since we are inter
ested in finding out what you think about your child's
abilities, there are no right or wrong answers to these
questions. It is very important, however, that you try to
be as honest as you can when answering these questions so
that we know exactly how you feel about your child's intel
lectual abilities. Please feel free to ask any questions
you may have while filling out the questionnaire."
After the mothers had completed the questionnaire the
experimenter checked it to make sure no questions had been
left unanswered. The mothers were then told the date and
time at which their children would be tested. The mothers
were also asked to send a note to their child's teacher on
the day the child would be tested to let the teacher know
the time at which the child would be taken out of class. In
34
addition, the mothers were asked to remind their children on
the date of their test that they would be taken out of class
for about an hour to be tested. They were also instructed
to tell the children that the test was not a "school" test
and that the results of the test would not affect their
grades in any way. Since the experimenter who administered
the questionnaire to the mothers was the same who tested all
the children, the experimenter had a chance to meet the
children at their own homes before the date of their test.
The experimenter tried to establish rapport with the chil
dren at the time of this first meeting and also reassured
the children that the test would not be hard. If the
children asked why they had to take this test, they were
told that the experimenter had to test 70 children to be
able to graduate from college and that they were doing this
as a favor to the experimenter.
The test administered to the children was the Wechsler
Intelligence Scale for ChildrenRevised (WISCR). All the
children were tested individually at offices provided by
their schools. The administration of the test lasted ap
proximately 45 minutes to an hour depending on the child.
At the beginning of the test the children were told to try
to do their best but not to worry if they could not answer
all the questions because some of the questions were made
for older children. They were also reminded that the re
sults of the test would not affect their school grades and
that neither their teachers nor their mothers would be given
the results.
35
Variables
The following is a list of the variables used in the
study:
1. Age of the children. This variable was used as a
continuous variable. A large age range of children was
sampled in order to insure reasonable variability in the
data. This was important because the analysis of the data
was correlational in nature. As mentioned before, the
children ranged in age from 63 to 126 years (mean = 95,
s.d. = 110).
2. Sex of the children. Approximately equal numbers
of male and female children were sampled for the study.
This variable was the only categorical variable included in
the analysis of the data and was used mainly as a control
variable.
3. Socioeconomic status (SES) of the children's fami
lies. This variable was quantified by using Hollingshead's
Two Factor Index of Social Position (Bonjean, Hill &
McLemore, 1967). This measure utilizes the fathers' occupa
tions and education levels to arrive at a numerical SES
score. The procedure followed to derive the SES numerical
scores was the following: First, the fathers' occupations
and education levels were rated on two separate 7point rat
ing scales designed by Hollingshead (Bonjean et al., 1967).
The seven positions of these occupational and educational'
scales are listed in Tables 3 and 4 in the Results chapter.
The scale scores assigned were then used in the following
formula to compute the numerical SES score or the "index of
36
social position score" (Bonjean et al., 1967, p. 385):
occupational scale score x 7 (factor weight) = partial score;
education scale score x 4 (factor weight) = partial score.
The index of social position score is the sum of the two
partial scores.
The possible SES scores that can be obtained by using
this index range from 11 to 77. It should be noted that a
low index score on this measure indicates a high social
status position and a high index score indicates a low
social status position.
4. Mothers' level of education. This variable was
used as a continuous variable. It was recorded in terms of
the number of years of formal schooling the mothers had
achieved.
5. Children's Intelligence Quotients (IQs). The
children's scores on the Wechsler Intelligence Scale for
ChildrenRevised were also used as a variable in the study.
Only the full scale IQ scores (combined verbal and perform
ance IQ scores) of the scale were used.
6. Children's Academic Achievement. Two parallel
measures of the children's academic achievement were used:
a. The children's national percentile scores on the
1983 administration of the Stanford Achievement Test
(SAT).
b. The children's overall grade point averages (GPAs)
for the academic year 19821983. An overall grade
point average score was computed for each child by
using the following scale: A = 4.0 points; A = 3.75
37
points; B+ = 3.25 points; B = 3.0 points; B = 2.75
points; C+ = 2.25 points; C = 2.0 points; C = 1.75
points; D+ = 1.25 points; D = 1.0 points; D = 0.5
points; F = 0 points.
In computing these GPA scores only the grades received
in the following courses were used: Reading, Mathematics,
English, Spelling, Science & Health, and Social Studies.
The children's grades on courses such as Physical Education,
Music, Art, Spanish or Religious Education were excluded
when computing the GPA scores. This was done because many
of these courses were not assigned letter grades. Also,
some of these courses were not taught at all three schools.
It should be noted that the grading scales of the three
schools from which the children were sample differed
slightly. The following grading scales were printed on the
report cards of each of the three schools:
San Jose Catholic The Chappell School
A = 93100 A = 95100
B = 8792 B = 8594
C = 7686 C = 7584
0 = 6675 D = 7074
F = 65 and below F = 69 and below
Southside Estates Academy
A = Superior
B = Above Average
C = Average
D = Below Average
F = Failure
38
No attempt was made to adjust the children's grades to one
comparable scale. This would have been an impossible task
since one of the schools did not even provide numerical
equivalents for the letter grades they assigned.
In addition to the above variables, a number of meas
ures were derived from the questionnaire administered to the
mothers. The measures that follow were all derived from
this questionnaire.
7. Mothers' perceptions of their children's intellec
tual abilities. Three parallel measures of the mothers'
perceptions of their children's abilities were derived from
questions number 5, 7, and 8 of the questionnaire. These
questions asked the mothers to estimate their children's
intellectual abilities in three different ways:
a. Question #5 asked for a rating of their children's
intellectual abilities.
b. Question #7 asked for a numerical estimate of their
children's IQ scores and a lower and upper bound number
of an interval within which they thought their chil
dren's IQ scores would fall.
c. Question #8 asked for a percentile estimate of
their children's IQ scores.
Question #5 of the questionnaire provided a 9point
rating scale for the mothers to rate their children's intel
lectual abilities. This rating scale had the following
scale positions:
1. exceptionalhe is an extremely bright child,
gifted for his age.
39
2. well above averagehe is a very bright child,
brighter than most children his age.
3. above averagehe is slightly brighter than the
average child his age.
4. slightly above averagehe is slightly brighter
than the average child his age.
5. averagehe is as capable as the average child of
his age.
6. slightly below averagehe is slightly less
capable than the average child of his age.
7. below averagehe is less capable than the average
child his age.
8. well below averagehe has difficulty keeping up
with most children of his age.
9. extremely below averagehe is not capable of
keeping up with children of his age.
After the mothers had rated their children's intellec
tual abilities on the above scale and before they were asked
to give a numerical estimate of their children's IQ scores,
they were provided with the following information in ques
tion #7:
The results from tests measuring children' intelli
gence quotients (IQs) show that out of every one thou
sand children tested approximately
1 child will have an IQ above 145
22 children will have IQs between 130 and 145
136 children will have IQs between 115 and 130
341 children will have IQs between 100 and 115
40
341 children will have IQs between 85 and 100
136 children will have IQs between 70 and 85
23 children will have IQs below 70
Based on this information what would be your best esti
mate of your child's IQ? Please keep in mind that an
IQ score is a relative measure. That is, it reflects
how well a child performs on the test as compared to
other children of his same age. Also keep in mind that
the average IQ score is 100. The majority of children
score within 15 points plus or minus 100 (between 85
and 115). Scores within this range are considered
normal.
This information was provided in order to maximize the
chances that the mothers would give accurate and sensible
estimates of their children's IQ scores. The mothers were
also provided with a definition of a percentile score before
they were asked to give the percentile estimates of their
children's IQ scores. This definition was provided in ques
tion #8 and read as follows:
A percentile indicates where your child's IQ score
ranks in comparison to other children who have taken
the same IQ tesL. A percentile score of 50 would mean
your child's score is in the middle. Half of the other
children who took the test would have scored above him
and half would have scored below him. A percentile
score of 80 would mean your child did better than 00%
of the other children who took the test and worse than
20%. A percentile score of 25 would mean your child
41
did better than 25% of the other children who took the
test and worse than 75%.
Again, the above definition was provided in order to
maximize the chances that the mothers would give accurate
and sensible percentile estimates of their children's IQ
scores.
In this study it was essential to obtain a valid meas
ure of the mothers' perceptions of their children's intel
lectual abilities since the purpose of the study was to de
termine how accurate these perceptions were and whether or
not they predicted the children's academic achievement. The
three types of estimates that the mothers were asked to give
were included in the questionnaire for the above reason.
Theoretically, all three of these measures should correlate
highly with each other since they are basically asking the
same thing in three different ways. It was decided before
hand that the numerical estimate would be used to determine
the mothers' accuracy scores as long as it correlated highly
with the mothers' ratings of their children's abilities,
that is, as long as it appeared to be a sensible and valid
measure of the mothers' perceptions. The numerical estimate
was preferred because the mothers' accuracy scores could be
easily computed from it by taking the difference between
each mother's numerical estimate and her child's real IQ
score. The percentile estimate would be used to compute the
accuracy scores only in the event that they appeared to be
better measures of the mothers' perceptions than the numeri
cal estimates.
42
8. Accuracy of the mothers' perceptions of their chil
dren's abilities. The accuracy scores of the mothers were
computed by taking the difference between the mothers' nu
merical estimates of their children's IQ scores and their
children's real IQ scores. The accuracy scores obtained by
using this procedure were negative if the mothers had under
estimated their children's IQ scores and positive if they
had overestimated their children's IQ scores. Also, the
closer these accuracy scores were to zero, the more accurate
the mothers had been in estimating their children's IQ
scores.
The mothers' numerical estimates of their children's IQ
scores were used to compute the accuracy scores because the
preliminary analysis of the data suggested that these esti
mates were relatively sensible measures of the mothers' per
ceptions. The correlation between the numerical estimates
and the mothers' ratings of their children's abilities was
fairly high (r = .75, p = .0001). Table 11 in the Results
chapter shows the intercorrelations obtained for the three
different type of estimates and the children's real IQ
scores. More will be said about these intercorrelations in
the Results chapter.
Perhaps it should be mentioned here that although the
mothers' numerical estimates appear to be fairly valid meas
ures of their perceptions, the reliability of the accuracy
scores derived from these estimates remains in question. As
mentioned before, these accuracy scores were obtained by
taking the difference between the mothers' numerical
43
estimates of their children's IQ scores and their children's
real IQ scores. Difference scores of this type, and differ
ence scores in general, tend to be less reliable measures
than single scores. The unreliability of difference scores
is mainly due to the fact that the errors of measurement
associated with each measure used to compute the difference
score contribute to the overall error variance of the dif
ference score. Usually, if the measures used to obtain the
difference score have high initial reliabilities, then the
reliability of the difference score will be considerably
higher than if the measures had poor initial reliabilities
(Mehrens & Lehmann, 1975). In the case of the accuracy
measure computed for this study, only the reliability of one
of the measures used to derive this score is known (the
average standard error of measurement for the WISCR full
scale IQ score is 3.19 points)(Wechsler, 1974); therefore,
it will not be possible to estimate the reliability of the
accuracy scores. Not knowing the reliability of the moth
ers' accuracy scores may be a problem in this study since
one of the main hypotheses has to do with the relationship
between these accuracy scores and the children's academic
achievement. It is possible, for example, that no relation
ship between the accuracy of the mothers' perceptions and
their children's level of academic achievement is found
simply because the accuracy scores were too unreliable
rather than because no relationship actually exists. In
order to be able to discern which of these interpretations
is more likely if such negative results were found, it would
44
be helpful if it could be shown that the mothers' accuracy
scores correlate with another measure which theoretically
should be related to them. The two measures that follow
were included in the study for this reason.
9. Frequency of the mothers' opportunities to observe
their children's intellectual abilities. This measure was
included in the study to determine whether the accuracy of
mothers' perceptions varied as a function of how frequently
the mothers had the opportunity to observe their children.
As mentioned before, this measure was also included hoping
it would help clarify the results in the event no relation
ship was found between the accuracy of mothers' perceptions
and their children's academic achievement. Theoretically,
the amount of time mothers spend observing their children's
intellectual abilities should correlate positively with how
accurately they predict their children's IQ scores. That
is, mothers who spend more time with their children should
be more accurate than mothers who spend less time with them.
The measure of the frequency of the mothers' opportunities
to observe their children's intellectual abilities was
derived from question #9 of the questionnaire. This ques
tion read as follows:
The following is a list of instances in which
parents have had the opportunity to observe their
child's intellectual performance. Please indicate
whether or not you have had the opportunity to ob
serve your child's intellectual abilities under
these circumstances. Also indicate how frequently
45
you have had this opportunity by putting a number
from 0 to 5 by the activity to reflect the follow
ing frequencies:
0 = never
1 = very infrequently, less than once a month.
2 = not very often, at least once every two weeks.
3 = regularly, at least once a week.
4 = often, at least three times a week.
5 = very frequently, almost every day.
listening to your child name letters of the
alphabet or read.
explaining to your child the meaning of a
word.
listening to your child count or solve
arithmetic problems.
observing your child work on a jigsaw
puzzle.
helping your child with his school work or
looking over his school work.
teaching your child the words to a song,
poem or prayer.
discussing with your child the plot of a
television program, movie or book.
playing reasoningtype games with your child
or observing him play these games with other
children.
playing games that require remembering a set
of rules or observing your child play these
46
sort of games (e.g., table games, card
games, sports).
teaching your child how to do a specific
task
observing your child put something together
or working on a craft.
observing your child talking and interacting
with other children.
looking over your child's art work.
playing video games with your child or ob
serving him play these games with other
children.
A global score reflecting the total frequency of oppor
tunities mothers have had to observe their children's abili
ties was computed for each mother. This global score was
computed by adding up all the frequency ratings for all 14
situations. The possible range of scores was from 0 to 70.
A score of zero would be given to a mother who indicated
never having the opportunity to observe her child's intel
lectual abilities under any of the 14 situations listed. A
score of 70 would be given to a mother who indicated having
had the opportunity to observe her child's abilities under
all 14 situations "very frequently."
10. Frequency of mothers' opportunities to compare
their children's intellectual abilities with those of other
children. This measure was also included in the study to
help clarify possible negative results. Again, it was ex
pected that this measure would correlate positively with the
47
mothers' accuracy scores. This measure was derived from
question #10 of the questionnaire which read as follows:
The following is a list of instances in which you
may have had the opportunity to compare your
child's intellectual abilities to the abilities of
other children of your child's same age. Please
indicate whether or not you have had the opportu
nity to compare your child's performance with that
of other children under the following circum
stances. Also indicate how frequently you have
had these opportunities by putting a number from 0
to 5 to indicate the following frequencies:
0 = never
1 = very infrequently, less than once a month.
2 = not very often, at least once every two weeks.
3 = regularly, at least once a week.
4 = often, at least three times a week.
5 = very frequently, almost every day.
listening to other children of your child's
age name letters of the alphabet or read.
explaining the meaning of a word to other
children of your child's age.
listening to other children of your child's
age count or solve arithmetic problems.
observing other children of your child's
age working on a jigsaw puzzle.
helping other children of your child's age
with their school work or looking over their
school work.
teaching other children of your child's age
the words to a song, poem or prayer.
discussing the plot of a television program,
movie or book with other children of your
child's age.
playing reasoningtype games with children of
your child's age or observing them play
these games.
observing children of your child's age
playing games that require remembering a set
of rules.
teaching other children of your child's age
how to do a specific task.
observing other children of your child's age
putting something together or working on a
craft.
observing other children of your child's age
talking and interacting with each other.
looking over the art work of other children
of your child's age.
playing with or observing other children of
your child's age playing video games.
A global score was also computed for this measure by
adding up all the frequency ratings given by the mothers for
49
all 14 situations listed. As before, the possible range of
scores on this measure was from 0 to 70.
11. Mothers' demands for their children's academic
achievement. Two measures of the mothers' demands for their
children's academic achievement were included in the study.
These measures were derived from questions #2 and #3 of the
questionnaire. These questions read as follows:
How low would your child's grades in school have
to get before you let him know that your are not
satisfied with his school work?
I would be dissatisfied
A's.
I would be dissatisfied
A's and B's.
I would be dissatisfied
B's.
I would be dissatisfied
than C's.
I would be dissatisfied
C's.
I would be dissatisfied
than D's.
with more B's than
with grades lower than
with more C's than
with grades lower
with more D's than
with grades lower
I would never let him know that I am dis
satisfied.
How high would your child's grades in school have
to get before you let them know that your are very
pleased with his school work?
50
very pleased
with mostly A's and a
I would be
few B's.
I would be
few A's.
I would be
few C's.
I would be
few B's.
I would be
few D's.
I would be
I would be
very
very
pleased
pleased
with
with
no F's.
whatever grades
he brought.
The responses given by the mothers to the first ques
tion represent the minimum level of demands mothers make of
their children. The responses given to the second question
represent the pleasing level of demands, that is, the mini
mum level of grades the children have to get in order to
please their mothers. Numbers from one to seven were as
signed to each of the responses to both of these questions.
A score of one on the minimum demands measure was given to a
mother who answered that she would be dissatisfied if her
child got more B's than A's. A score of seven was given to
a mother who answered that she would never let her child
know that she was dissatisfied with her child's school per
formance. Likewise, a score of one on the pleasing level of
demands measure was given to a mother who answered that she
would be very pleased if her child got mostly A's and B's.
very pleased with mostly B's and a
very pleased with mostly B's and a
very pleased with mostly C's and a
very pleased with mostly C's and a
51
A score of seven was given to a mother who answered that she
would be very pleased with whatever grades her child brought
home.
In addition to all of the above measures derived from
the questionnaire, a few other questions were included in
the questionnaire for other purposes. Question #1 was
included mainly as a warmup question in order to get the
mothers thinking about their children's school performance.
This question read as follows:
What are your child's grades in school?
he gets mostly A's with very few B's.
he gets mostly A's and B's only.
he gets mostly B's with some A's and C's.
he gets mostly C's with some B's and D's.
he gets mostly D's with some C's and F's.
he gets mostly F's with some D's and C's.
Numbers from one to six were assigned to each of the
responses given to this question. A score of one was given
to the mothers who answered that their children were getting
the highest grades and a score of six to those who answered
they were getting the lowest grades.
A couple of openended questions (#4 and #6) were also
included in the questionnaire. The answers to these ques
tions were not included in the main data analyses. These
questions were included to generate hypotheses for future
studies dealing with parents' perceptions of their chil
dren's intellectual abilities. These questions read as fol
lows:
52
#4. Please comment on your child's school work.
If he is doing well in school, why do you
think he is doing well? If he is not doing
well in school, why do you think he is not
doing well?
#5. Can you describe any of the things that your
child does or has done in the past which have
led you to believe that his intellectual
abilities are at the level that you have
indicated in question #5?
Perhaps it should be mentioned that the majority of the
mothers filling out this questionnaire had no problems
understanding the questions. When they did have a problem,
it usually had to do with questions 9 and 10 of the ques
tionnaire. These were the two measures of the frequency of
opportunities mothers have to observe and compare their
children's intellectual abilities. Usually the mothers men
tioned that when their children were younger they had had
the opportunity to observe their abilities more frequently
than now that they were older. The mothers usually wanted
to know whether their answers should reflect how frequently
they observe their child now at his present age, or how
frequently they had observed their child in the past. When
this question was asked the experimenter told the mothers
to answer how frequently they had the opportunity to observe
their children now, at the child's present age.
CHAPTER THREE
RESULTS
The results of this study will be presented in three
parts. The first part will include the descriptive statis
tics of each of the variables used. The second part will
include a description of how these variables intercorrelate
with each other. Finally, the third section will cover the
results of the regression analyses performed to answer the
major questions of the study.
Descriptive Results
1. Age of the children. The children ranged in age
from 63 to 126 years. The mean age was 95 and the
standard deviation was 110. The age distribution of the
children was very similar in all three of the schools sam
pled. A oneway analysis of variance was conducted with
"school" as the independent variable to make sure the age of
the children did not differ by school. The results of this
analysis indicated that there were no significant differ
ences among the schools in terms of the age of the children.
It should be mentioned that five additional separate
analyses of variance were conducted to determine whether the
following variables differed by school: the IQ scores of
the children, the children's SAT and GPA scores, the
mothers' level of education, and the families' SES levels.
Since none of these variables were found to differ
54
significantly among the schools the data for all three
schools were pooled for the remaining analyses.
2. Sex of the children. There were 34 males and 36
female children in the study. Table 1 shows the sex distri
bution of the children by age. As can be seen, there were
comparable number of male and female children throughout the
different age groups in the sample.
3. Socioeconomic status of the children's families.
The Hollingshead's index of social position scores computed
for the children's families ranged from 11 to 73. The aver
age score was 32.17 and the standard deviation was 15.16.
Tables 2 and 3 show the percentage of the children's fathers
falling into the different levels of the Hollingshead's edu
cational and occupational scales. As mentioned in the
Method chapter, the ratings on these two scales were used to
compute each family's index of social position score.
4. Mothers' level of education. The number of years
of formal schooling the mothers had achieved ranged from 9
to 20 years. The mean was 13.87 years and the standard
deviation was 1.81 years. Table 4 shows the percentage of
mothers at several different education levels. As can be
seen, the majority of the mothers had educations beyond the
high school level (over 68%).
5. Children's intelligence quotients (IQs). The chil
dren's full scale IQ scores on the WISCR ranged from 81 to
139. The mean was 109.08 and the standard deviation was
11.92. Table 5 shows the percentage of children at five
different levels of IQ. As can be seen from this table, the
Table 1
Age and Sex Distribution of the Children
Age
Sex 67 89 1012 Total
Males 11 10 13 34
Females 10 10 16 36
Total 21 20 29 70
Table 2
Percentage of the Children's Fathers at Each Level of the
Hollingshead's Educational Scale
1. Professional
(M.A., M.S., M.E.,
M.D., Ph.D., LL.B.)
2. Fouryear college
graduate
3. 13 years of college
4. High school graduate
5. 1011 years of school
6. 79 years of school
7. Under 7 years of school
Total
Percentage
of the
total sample
20.00
18.60
35.70
22.86
0.00
2.86
0.00
100.00
Number
of
fathers
14
25
16
0
2
0
70
Note. The seven scale positions for the Hollingshead's
Educational Scale were listed in Bonjean et al.,
1967, p. 383.
Table 3
Percentage of the Children's Fathers at Each Level of the
Hollingshead's Occupational Scale
Percentage Number
of the of
total sample fathers
1. Higher executives of larger 22.86 16
concerns, proprietors and
major professionals
2. Business managers, proprietors 18.60 13
of medium sized businesses
and lesser professionals
3. Administrative personnel, 20.00 14
owners of small businesses
and minor professionals
4. Clerical and sales workers, 18.60 13
technicians and owners of
little businesses
5. Skilled manual employees 11.43 8
6. Machine operators and semi 7.14 5
skilled employees.
7. Unskilled employees. 1.43 1
Total 100.00 70
Note. The seven scale positions for the Hollingshead's
Occupational Scale were listed in the Bonjean et al.,
1967, p. 383.
Table 4
Percentage of Mothers at Five Different Education Levels
Percentage Number
of the of
total sample mothers
1. Professional 5.71 4
(17 years of school or more)
2. College graduate 14.29 10
(16 years of school)
3. Some college 48.57 34
(1315 years of school)
4. High school graduate 30.00 21
(12 years of school)
5. No high school 1.43 1
(under 12 years of school)
Total 100.00 70
Table 5
Percentage of the Children at Five Different Levels of IQ
Classification
Very Superior
Superior
High Average
Average
Low Average
Percentage
of the
total sample
5.71
11.43
22.86
54.29
5.71
100.00
Note. The IQ classifications were taken from Wechsler, D.,
1974, p. 26.
130+
120129
110119
90109
8089
Total
Number
of
children
4
8
16
38
4
70
majority of the children had average to above average IQs.
At least 80% of the children had IQ scores of 100 or more.
6. Children's academic achievement. The children's
scores on the Stanford Achievement Test ranged from the 17th
to the 99th percentile. The mean SAT percentile score was
71.23 and the standard deviation was 19.85. Over 84% of the
children had SAT scores above the 50th percentile. Thus,
the sample as a whole performed well above national averages
on the 1983 administration of the SAT. The children's grade
point averages also showed above average performance in aca
demic achievement. They ranged from 0.69 to 4.0 with a mean
of 2.94 and a standard deviation of 0.69. According to the
scale used to compute the children's GPA scores, a GPA of
2.94 would fall between a B and a B letter grade. At least
90% of the children had GPAs above a 2.0 or a C level and
approximately 50% of them had GPAs above a 3.0 or above a B
level.
7. Mothers' perceptions of their children's intellec
tual abilities. The mothers' ratings of their children's
intellectual abilities ranged from 1 (exceptional) to 6
(slightly below average). The average rating was 3.49 which
would fall between a 3 (above average) and a 4 (slightly
above average) rating. The standard deviation was 1.21.
Approximately 4% of the mothers rated their children as
exceptional; 19% rated them as well above average; 29% rated
them as above average; 23% as slightly above average; 24% as
average and only one mother (1.43%) rated her child as
61
slightly below average. These results indicate that when
using this rating scale the mothers were very reluctant to
rate their children's abilities as anything but average or
better. A slightly different picture emerges from the re
sults of the mothers' numerical estimates of their chil
dren's IQs. These estimates ranged from 75 to 143. The
average IQ estimate given by the mothers was 114.86 and the
standard deviation was 14.01. Although the majority of the
mothers still answered that their children's abilities or IQ
scores were above average (over 91% estimated their chil
dren's IQs to be at or above 100), a greater percentage of
them gave below average estimates than when using the rating
scale. Over 8% gave IQ estimates below 100. The mothers'
percentile range estimates of their children's IQ scores
ranged from "4049th" percentile to "over 95th" percentile.
The average percentile range estimated by the mothers was
80th89th and the standard deviation was over 10 percentile
points. As with the rating scale, when giving percentile
estimates of their children's IQs the mothers were very re
luctant to say their children's percentile IQs were below
average or below the 50th percentile. Only one mother said
her child's percentile IQ would fall below this level.
8. Accuracy of the mothers' perceptions of their chil
dren's abilities. As mentioned in the Method chapter, the
mothers' accuracy scores were computed by taking the differ
ence between their numerical estimates of their children's
IQ scores and their children's real IQ scores. The mothers'
accuracy scores indicated that a great number of them were
fairly accurate in predicting their children's IQ scores.
Approximately 37% predicted their children's IQs within six
points (plus or minus) of their children's real IQs. The
majority of the mothers, however, were wrong by more than
six points. When the mothers were wrong, they usually erred
in the direction of overestimating their children's IQs.
Over 47% of the mothers overestimated their children's
scores by more than six points. These overestimations
ranged from 7 points to 39 points. The average overestima
tion was 15.88 points and the standard deviation was 7.77.
Although the majority of the mothers overestimated their
children's IQs, at least 16% of them underestimated their
children's IQs by more than six points. The average under
estimation was 13.18 points and the standard deviation was
9.80. One mother underestimated her child's IQ score by as
much as 41 points! Table 6 shows the percentage of mothers
who were accurate within six points and the percentage of
those who overestimated and underestimated their children's
IQ scores.
Perhaps it should be mentioned that although many of
the mothers were fairly accurate in predicting their chil
dren's IQ scores,the majority of them expressed a great deal
of uncertainty when giving their estimates. Many of them
told the experimenter when filling out the questionnaire
that they had no idea what their child's IQ was and that
their estimate was simply a "wild guess." When asked to
give the numerical estimates the mothers were also asked to
Table 6
Percentage of Mothers Who Were Accurate and Who
and Underestimated their Children's IQ Scores
Overestimated
Percentage
of the
total sample
Overestimated by 15 points
or more
Overestimated within 714
points
Accurate within 6 points
Underestimated with 714
points
Underestimated by 15 points
or more
18.57 13
28.57 20
37.14 26
12.86 9
2.86 2
Total
Number
of
mothers
100.00
64
give a range of IQ scores within which they thought their
child's IQ would fall. The majority of the mothers gave
very wide ranges which probably reflect the uncertainty they
felt about their estimates. The size of the ranges given by
the mothers ranged from 5 to 45 IQ points. The average
range size was 23.67 IQ points and the standard deviation
was 11.14.
9. Frequency of the mothers' opportunity to observe
their children's intellectual abilities. The global fre
quency scores computed for each mother ranged from 17 to 62.
The average score was 40.24 and the standard deviation was
11.27. An average global score of 40.24 when divided by the
14 different situations listed in question #9 of the ques
tionnaire yields an average frequency score of 2.87 per sit
uation. That is, mothers reported having the opportunity to
observe their children with some degree of frequency falling
between "not very often" (equivalent to a rating of 2) and
"regularly" (equivalent to a rating of 3). Table 7 shows
the average frequency rating given by the mothers for each
of the 14 situations asked about in question #9. As can be
seen from this table, the mothers reported having the oppor
tunity to observe their children "talking and interacting
with other children" more often than any other type of situ
ation. They also reported having very few opportunities to
observe their children "working on a jigsaw puzzle" or
"playing video games."
10. Frequency of the mothers' opportunities to compare
their children's intellectual abilities with those of other
Table 7
Average Frequency Ratings Given by the Mothers to Each of the
14 Situations Listed in Question #9
Listening to your child name letters
of the alphabet or read
Explaining to your child the meaning
of a word
Listening to your child count or solve
arithmetic problems
Observing your child working on a
jigsaw puzzle
Helping your child with his school work
Teaching your child the words to a
song, poem or prayer
Discussing with your child the plot of
a TV program, movie or book
Playing reasoningtype games with your
child or observing him play these games
Playing games that require remembering
a set of rules
Teaching your child how to do a specific
task
Observing your child put something
together or working on a craft
Observing your child talking and inter
acting with other children
Looking over your child's art work
Playing video games with your child or
observing him play these games
Mean
3.46 (1.55)a
3.17 (1.35)
2.97 (1.31)
1.61 (1.20)
3.96 (1.32)
2.24 (1.42)
3.19 (1.26)
2.43 (1.24)
2.71 (1.32)
2.91 (1.13)
2.68 (1.14)
4.01 (1.10)
3.17 (1.19)
1.81 (1.63)
Note. A frequency rating of 1 = very infrequently, 2 = not
very often, 3 = regularly, 4 = often, 5 = very
a frequently.
Numbers in parentheses indicate the standard deviations.
children of the same age. Overall, mothers reported having
less frequent opportunities to compare their children's
intellectual abilities than to observe them. The average
global frequency score for this measure was 18.99 with a
standard deviation of 14.20 as compared to an average fre
quency score of 40.24 for the previous measure. An average
global score of 18.99 is equivalent to an average frequency
score of 1.36 for each of the 14 situations asked about in
question #10. That is, mothers reported having the oppor
tunity to compare their children's abilities with some
degree of frequency falling between "very infrequently"
(equivalent to a rating of 1) and "not very often" (equiva
lent to a rating of 2). Table 8 shows the average frequency
rating given by the mothers for each of the 14 situations
asked about in question #10 of the questionnaire.
11. Mothers' demands for their children's academic
achievement. The minimum level of demands mothers reported
making of their children ranged from "I would be dissatis
fied with more Bs than As" (equivalent to a rating of 1) to
"I would never let him know that I am dissatisfied" (equiva
lent to rating of 7). The average minimum level of demands
for the sample as a whole was 2.88 and the standard devia
tion was 1.17. An average demand level of 2.88 would fall
somewhere between "I would be dissatisfied with more Cs than
Bs" (equivalent to a rating of 3) to "I would be dissatis
fied with grades lower than As and Bs" (equivalent to a
rating of 2). Table 9 shows the percentage of mothers at
Table 8
Average Frequency Ratings Given by the Mothers to Each of the
14 Situations Listed in Question #10
1. Listening to other children of your child's
age name letters of the aphabet or read
2. Explaining the meaning of a word to other
children of your child's age
3. Listening to other children of your child's
age count or solve arithmetic problems
4. Observing other children of your child's
age working on a jigsaw puzzle
5. Helping other children of your child's
age with their school work
6. Teaching other children of your child's
age the words to a poem, song or prayer
7. Discussing the plot of a TV program, movie
or book with other children of your child's
age
8. Playing reasoningtype games with other
children of your child's age
9. Observing children of your child's age play
ing games that require remembering rules
10. Teaching other children of your child's
age how to do a specific task
Mean
1.51 (1.43)a
1.52 (1.28)
1.40 (1.40)
1.00 (1.20)
1.00 (1.42)
0.82 (1.07)
1.27 (1.29)
1.25 (1.31)
1.71 (1.41)
1.24 (1.26)
11. Observing other children of your child's 1.12 (1.19)
age putting something together
12. Observing other children of your child's age 2.75 (1.75)
talking and interacting with each other
13. Looking over the art work of other 1.27 (1.11)
children of your child's age
14. Playing with or observing other children 1.04 (1.05)
of your child's age playing video games
Note. A frequency rating of 1 = very infrequently, 2 = not
very often, 3 = regularly, 4 = often, 5 = very
frequently.
Numbers in parentheses indicate the standard deviations.
Table 9
Percentage of Mothers at Each Level of Minimum Demands for
Academic Achievement
Dissatisfied with more Bs
than As
Dissatisfied with grades
lower than As and Bs
Dissatisfied with more Cs
than Bs
Dissatisfied with grades
lower than Cs
Dissatisfied with more Ds
than Cs
Dissatisfied with grades
lower than Ds
Never dissatisfied
Total
Percentage
of the
total sample
12.86
27.14
22.86
35.71
0.00
0.00
1.43
100.00
Number
of
mothers
9
19
16
25
0
0
1
70
69
each level of minimum demands. It should be noted from this
table that almost all the mothers (except for one who
answered she would never let her child know that she was
dissatisfied) demanded grades of at least Cs or better from
their children.
The pleasing level of demands mothers reported making
of their children were, on the average, higher than the min
imum level of demands. The pleasing level of demands ranged
from "I would be pleased with mostly As and a few Bs" to "I
would be very pleased with whatever grades he brought home."
The average pleasing level of demands was 2.2 and the stand
ard deviation was 1.31. An average demand level of 2.2
would fall between "I would be very pleased with mostly Bs
and a few As" to "I would be very pleased with mostly Bs and
a few Cs" (equivalent to ratings of 2 and 3, respectively).
Table 10 shows the percentage of mothers at each pleasing
level of demands. As can be seen, over 71% of the mothers
reported they would be pleased with As and Bs only (equiva
lent to ratings of 1 and 2). Only two mothers answered they
would be pleased with whatever grades their child brought.
12. Responses to the warmup and openended questions.
As was mentioned in the Method chapter, a warmup question
asking the mothers about their children's grades in school
was also included in the questionnaire as well as a couple of
openended questions. The mothers' responses to the warmup
question indicated that the mothers knew and remembered
accurately what kind of grades their children were getting
in school. The responses to the warmup question ranged
Table 10
Percentage of Mothers at Each Level of Pleasing Demands
Percentage Number
of the of
total sample mothers
1. Pleased with mostly As and 32.86 23
a few Bs
2. Pleased with mostly Bs and 38.57 27
a few As
3. Pleased with mostly Bs and 15.71 11
a few Cs
4. Pleased with mostly Cs and 7.14 5
a few Bs
5. Pleased with mostly Cs and 2.86 2
a few Ds
6. Pleased with no Fs 0.00 0
7. Pleased with anything 2.86 2
Total 100.00 70
from "he gets mostly As with very few Bs" to "he gets mostly
Ds with 'some Cs and Fs." The average response on this
question was a rating of 2.53 with a standard deviation of
1.0. A rating of 2.53 in this scale translates to grades
mostly above a C letter grade which is exactly what the
majority of the children were getting (90% had GPAs above a
2.0 or a C level). The correlation between what the moth
ers said their children's grades were and the children's
actual grades also indicates that the mothers were very
aware of their children's performance in school. More will
be said about the correlation between these two variables
later in this chapter.
The mothers' responses to the first openended question
were rather interesting. This question asked them why they
thought their children were doing well or poorly in school.
Approximately half of the mothers (52%) said their children
were doing well in school. The remainder said their chil
dren were not doing well or not doing as well as they could.
The mothers who responded that their children were doing well
usually attributed their children's good school performance
to some specific personality characteristic of their chil
dren. Over 58% of the mothers' responses made mention to
personality characteristics such as maturity level, ability
to concentrate, selfconfidence, possession of good study
habits, eagerness to learn and eagerness to please parents
and/or teachers. The most common response was eagerness to
learn. Interestingly enough, the children's natural
abilities or intelligence was mentioned only by two of the
mothers. Over 24% of the responses given by the mothers
attributed the children's good school performance to some
particular behavior of the mother or the teacher toward the
child. The mothers usually mentioned that they gave their
children extra attention or described certain discipline
rules they had established at home (e.g., the children have
to do their homework before they are allowed to play).
Other mothers mentioned some characteristics of the teachers
or the school environment as the reason that their children
were performing well in school. Finally, over 12% of the
responses given by the mothers attributed their children's
good performance to the amount of "effort" the children were
putting into their school work.
It should be mentioned here that some of the mothers
gave more than one reason why they felt their children were
doing well or poorly in school. The percentages given above
reflect the percentages in terms of the total number of dif
ferent type of responses given rather than the percentage of
mothers giving the different type of responses.
The mothers who responded that their children were not
doing well in school also attributed their children's poor
performance to specific personality characteristics of the
children. The personality characteristics more commonly
mentioned were the lack of ability to concentrate, immatur
ity or a lack of a sense of responsibility, poor study
skills, introversion and boredom. At least 44% of the
mothers' responses made mention to these characteristics as
the main reason why their children were not doing well in
school. Approximately 37% of the mothers' responses attri
buted the children's poor performance to a lack of effort or
a lack of motivation to achieve on the part of the children.
The remainder of the mothers' responses (16%) blamed the
teachers or the schools for their children's poor grades.
They usually said the teachers were not spending enough time
with the children, not giving them enough positive rein
forcement or were putting too much pressure on the children
to do well. None of the mothers mentioned their child's
lack of natural ability as a possible reason why their child
may not be doing well in school. One mother did say that
she did not know why her child was doing so poorly in school.
The second openended question asked the mothers to
describe the things their children do or have done in the
past which have led them to believe their children's intel
lectual abilities are at the level they indicated in the
rating scale in question #5 of the questionnaire. The
majority of the mothers answered this question by giving
positive examples of their children's intellectual abili
ties. At least 19% of the responses given by the mothers
alluded to the children's grades as an indicator of their
children's intellectual abilities. It should be pointed
out, however, that most of the mothers mentioned grades in
combination with some other ability they had observed in
their children as an indication of their overall intellec
tual ability. The specific abilities the mothers said they
74
had observed in their children included
1. The speed with which their children learned new concepts
and ideas (15% of the responses made mention to this speci
fic ability).
2. The children's reading skills and early interest in
reading (14% of the responses).
3. The children's inquisitive nature and the sophisticated
level of the questions they asked (9% of the responses).
4. The children's memory skills and retention abilities (8%
of the responses).
5. The children's imaginations and ability to come up with
innovative ideas on their own (6% of the responses).
6. The children's vocabularies and communication skills (5%
of the responses).
7. The children's reasoning abilities, logic, and analytic
skills (4% of the responses). The remainder of the re
sponses made mention of other more specific characteristics
of the children such as a special interest in solving
puzzles or certain hobbies, social maturity, a competitive
nature, the ability to learn a specific skill on their own,
etc. Only four parents described negative characteristics
of their children. These parents said their children were
slow in learning and/or did not try hard enough to learn.
Intercorrelations among the Variables
Intercorrelations were computed for all 16 variables
included in the study. The matrix of Pearson product moment
correlation coefficients is included in Appendix D. Out of
a total of 120 correlations 57 were found to be significant
at the p 4 .05 oc level of significance or better. This
is not surprising given the nature of some of the variables
included in the study. Many of the variables were different
measures of the same construct and were expected to corre
late highly (e.g., GPAs and SATs were both measures of
academic achievement). Other variables were known to be
highly related to each other from previous research but
were included to be used as control variables in subsequent
analyses (e.g., IQ is a well known predictor of SATs and
GPAs). Finally, some measures were theoretically expected
to correlate with each other (e.g., the mothers' estimates
of their children's abilities were expected to correlate
with the actual measure of the children's abilities). For
the purpose of simplicity, rather than comment on all 57
significant correlations separately, comments will be made
on certain clusters of variables which were expected to cor
relate with each other for any of the above reasons.
The following variables were expected to correlate be
cause they were measures of the same or similar constructs:
1. The children's GPAs and SAT scores. These two measures
were expected to correlate because they were both measures
of the children's academic achievement. The correlation
coefficient for these two variables was r = .74, p .0001.
2. The three different type of estimates of the children's
intellectual abilities given by the mothers. These were ex
pected to correlate because they were all measures of the
mothers' perceptions of their children's abilities. The
76
intercorrelations among the three types of estimates (rating,
numerical IQ, and percentile IQ) are given in Table 11. As
can be seen, these correlations were all highly significant.
It should be mentioned that the negative signs in these cor
relations are an artifact of the way the questions in the
questionnaire were set up. The rating scale, for example,
was set up so that the lowest numerical rating (#1) was
equivalent to the highest ability level ("exceptional").
These correlations, although carrying a negative sign, are
really positive in nature. That is, each two related
variables covary in the same direction (they increase and
decrease together). This is also true of many of the
correlations listed in Appendix D. For simplicity purposes,
the reader should assume that the correlations listed are
positive in nature (regardless of their sign) unless
otherwise stated in the text.
3. The measure of the frequency of opportunities mothers
have to observe their children's abilities and the measure
of the frequency of opportunities mothers have to compare
their children's abilities to those of other children. Both
of these variables can be thought of as measures of how
often mothers have a chance to form an impression of the
level of their children's intellectual abilities. The cor
relation between these two variables was expected to be sig
nificant. The actual correlation found was r = .48,
p .0001.
4. The minimum and the pleasing level of demands. Both of
these were measures of the mothers' demands for their
Table 11
Intercorrelations among
Children's Intellectual
Scores
the Mothers' Estimates of their
Abilities and their Children's Real IQ
Children's
real IQs
Rated
intellectual
abilities
Percentile
IQ
estimate
Children's
real IQs
Rated
intellectual
abilities
Numerical
IQ
estimate
Percentile
IQ
estimate
1 .00
0.52
(.0001)
0.53
(.0001)
1 .00
0.75
(.0001)
0.41 0.68
(.0005) (.0001)
Note. Numbers in parent
significance.
eses indicate the C. level of
Numerical
IQ
estimate
1 .00
0.57
(.0001)
1.00
children's academic achievement and were expected to be re
lated. The correlation between these two variables was also
highly significant (r = .64, p .0001).
Among the variables which were expected to correlate
because they are known to be highly related to each other
from previous research were the following:
1. The children's IQ scores and the academic achievement
measures.
2. The families' SES and the children's academic achieve
ment measures.
3. The families' SES and the children's IQ scores.
4. The families' SES and the mothers' level of education.
5. The mothers' level of education and the children's aca
demic achievement measures.
6. The mothers' level of education and the children's IQ
scores.
The intercorrelations among these variables are shown
in Table 12. As can be seen, most of the above expected
correlations were replicated in this study. The only excep
tions were the relationship between the families' SES and
the children's academic achievement and the relationship
between the mothers' level of education and the children's
academic achievement. It is possible that since the sample
used in the study was highly selective, the variability in
the mothers' levels of education and the families' SES
scores may not have been enough to facilitate the finding of
significant correlations.
Table 12
Intercorrelations among the Children's IQs, GPAs, SATs, the
Mothers' Level of Education and the Families' SES
IQ GPA SAT SES Mothers'
education
IQ 1.00
GPA 0.50 1.00
(.0001)
SAT 0.61 0.74 1.00
(.0001) (.0001)
SES 0.30 0.01 0.15 1.00
(.0119) (.9325) (.2240)
Mothers' 0.25 0.08 0.20 0.47 1.00
Education (.0369) (.4979) (.0979) (.0001)
Note. Numbers in parentheses indicate the . level of
significance.
80
The following variables were expected to correlate with
each other for theoretical reasons:
1. The mothers' estimates of their children's intellectual
abilities and the children's actual IQ scores. As can be
seen in Table 11 all three types of estimates given by the
mothers were highly related to the children's actual IQ
scores. This supports the finding that the mothers had
somewhat accurate perceptions of their children's abilities.
It should be mentioned here that since the mothers' three
types of estimates correlated highly with the children's IQ
scores, these estimates also correlated with some of the
other measures which are known to correlate with IQ.
Examples of these measures are the children's GPAs, SATs,
the families' SES and the mothers' level of education. It
makes sense that since all these measures correlate highly
with the children's IQ scores they should also correlate
with the mothers' estimates of these scores. The above
reason explained at least 11 of the 57 significant correla
tions found. The specific correlations for these variables
can be found in Appendix D.
2. The mothers' estimates of their children's intellectual
abilities and the demands for academic achievement they make
of their children. Again, it makes sense that the mothers'
levels of demands would correlate with their perceptions of
their children's abilities. Table 13 shows the intercorre
lations among these variables. As can be seen, all the cor
relations are significant. The mothers' demands also corre
lated with the children's actual IQ scores. The correlations
Table 13
Intercorrelations among the Mothers' Demands for Academic
Achievement and the Mothers' Estimates of their Children's
Abilities
Rated intellectual
abilities
Numerical IQ
estimates
Percentile IQ
estimates
Minimum
demands
0.31
(.0080)
0.26
(.0269)
0.40
(.0007)
Pleasing
demands
0.27
(.0214)
0.38
(.0014)
0.34
(.0040)
Note. Numbers in parentheses
significance.
indicate the 0C level of
between the minimum and the pleasing level of demands and
the children's real IQ scores were r = .50, p .0001 and
r = .29, p .0153, respectively.
3. The mothers' demands for academic achievement and the
children's actual level of academic achievement. These
variables were also expected to correlate with each other
and they did. The correlations between the children's GPAs
and the mothers' minimum and pleasing level of demands were
r = .45, p .0001 and r = .29, p .0157, respectively.
Likewise, the correlations between the children's SATs and
the mothers' minimum and pleasing level of demands were
r = .42, p .0003 and r = .32, p .0075.
4. The mothers' accuracy scores and the measures of the
frequency of opportunities they have to observe and compare
their children's abilities. The mothers' accuracy scores
unfortunately were not found to correlate with either of the
frequency measures. This was contrary to what was expected.
It is not clear why these variables were not found to be
related. An item analysis was performed on each of the fre
quency measures to determine whether the global frequency
scores computed were accurate reflections of how the mothers
had answered each of the items of the measures. The item
analyses were performed by computing the correlations be
tween the global frequency scores and the frequency ratings
given to each of the items of the measures. The item
analyses showed that both frequency measures had high inter
nal consistencies. That is, each and every one of the items
83
in each of the measures correlated significantly with its
respective global frequency score. Table 14 shows the cor
relations obtained for the item analyses of both frequency
measures. It is important to point out that although both
measures were shown to have high internal consistencies, it
is still not clear whether they are valid indicators of the
frequency with which the mothers observe and compare their
children's abilities.
These frequency measures showed several unexpected cor
relations with other variables. The measure of the frequen
cy of opportunities mothers have to compare their children's
abilities, for example, was found to correlate significantly
with the children's IQ scores, the children's GPAs, the
mothers' percentile IQ estimates, and the mothers' ratings
of their children's abilities. These correlations indicated
that mothers whose children had high IQs and high GPAs re
ported having more frequent opportunities to compare their
children's abilities than mothers whose children had low IQs
and GPAs. Also,mothers who reported having more frequent
opportunities to compare their children's abilities rated
their children's abilities higher and gave higher percentile
IQ estimates than mothers who reported having less frequent
opportunities. These relationships may have emerged for
many sensible reasons. For example, the relationship be
tween the frequency of opportunities mothers have to compare
their children and their children's IQs may have emerged be
cause the children with the higher IQs may have mothers who
are more intelligent and spend more time comparing their
Table 14
Correlations between the Global Scores of the Frequency
Measures and Each of their Respective Items
Frequency measures
Item Frequency of
number opportunities to observe
the children's abilities
0.62
0.64
0.71
0.52
0.75
0.69
0.65
0.66
0.57
0.53
0.49
0.69
0.36
0.70
Frequency of
opportunities to compare
the children's abilities
0.82
0.77
0.89
0.67
0.82
0.82
0.86
0.85
0.82
0.83
0.68
0.78
0.38
0.79
Note. All the correlations
.0001.
were significant at p < than
children to other children. It should be pointed out,
however, that many mothers were not as careful when answer
ing this frequency question as they were when answering
other questions. Some mothers, for example, indicated that
they had very few opportunities to compare their children to
other children. They then proceeded to assign frequency
ratings of one across all situations asked about in the
question without taking care to read each particular situa
tion and adjust their ratings accordingly. Thus, the valid
ity of these frequency scores is questionable and it is not
clear whether the relationships found between this measure
and the other variables are reliable. The measure of the
frequency of opportunities mothers have to observe their
children did not show this problem. The mothers did answer
this question carefully. This measure was found to be nega
tively related to the age of the children (r = .30,
p .0124). That is, the mothers reported having more fre
quent opportunities to observe the younger children than the
older children. This relationship, although unexpected, is
a sensible one.
It could also be argued that the frequency measures
failed to correlate with the accuracy scores because the
accuracy scores themselves may not be valid or reliable
measures. This, of course, remains a valid possibility.
The accuracy scores were found to be significantly related
to all three types of estimates of the children's abilities
given by the mothers (rated abilities, r = .33, p .0050; IQ
86
numerical estimate, r = .61, p .0001; and IQ percentile esti
mate, r = .24, p .0426). All three of these relationships
were positive in nature. Mothers whose estimates were high
tended to overestimate their children's abilities while
mothers whose estimates were low tended to underestimate
their children's abilities. The accuracy scores were also
found to be related to the children's IQ scores (r = .36,
p .0024). This relationship indicated that mothers with
high ability children tended to underestimate their chil
dren's abilities and mothers with low ability children
tended to overestimate them. More will be said about the
mothers' accuracy scores and their relationship with the
other variables later in this chapter.
5. The children's academic achievement and the warmup
question which asked the mothers what their children's
grades in school were. These two measures theoretically
should correlate with each other. The correlation found
between the mothers' reports of their children's grades and
their children's actual grades or GPAs was r = .79, p .0001.
This shows that the mothers were very aware of their chil
dren's performance in school. The mothers' reports of their
children's grades also correlated with several variables
which had already shown to be related to the children's
actual grades. The following are examples of these vari
ables: the children's SATs (r = .73, p .0001), the
children's IQ scores (r = .59, p .0001), the mothers'
ratings of their children's abilities (r = .54, p .0001),
the mothers' numerical IQ estimates (r = .38, p .0011), the
mothers' percentile IQ estimates (r = .53, p .0001), the
mothers' minimum demands (r = .58, p .0001) and the mothers'
pleasing demands (r = .33, p .0060).
Finally, it should be mentioned that a negative rela
tionship was found between the children's age and the chil
dren's GPAs (r = .25, p .0370). It appears that at the
younger ages the children's GPAs are higher than at the
older ages. This may be a reflection of the grading poli
cies of the teachers who may be more lenient when grading
the work of the younger children. Since age will be in
cluded as a control variable in all subsequent analyses,
this relationship should not be a problem.
Results of the Multiple Regression Analyses
In order to answer the main questions raised in this
study it was necessary to conduct several multiple regres
sion analyses. A multiple regression analysis allows for
the evaluation of the relationship between several indepen
dent variables (Ys) and a dependent variable (X). Since so
many of the variables used in this study were interrelated,
this type of analysis was necessary in order to be able to
determine the relative contribution made by each of the in
dependent variables to the variance observed in the depen
dent variable. In this section of the chapter the results
of the regression analyses will be presented. These results
will be organized around each of the main questions raised in
the study.
88
The first question of interest in this study has to do
with the accuracy of the mothers' perceptions of their chil
dren's intellectual abilities. More specifically, the
question asked whether the mothers' accuracy scores varied
as a function of the children's age, sex, IQs, the mothers'
level of education, and/or the families' SES. It was pre
dicted that the mothers' accuracy scores would vary as a
function of the children's age. That is, the mothers were
expected to be more accurate in perceiving the abilities of
the older chidlren than the younger children. Mothers with
higher levels of education were also expected to be more
accurate than mothers with lower levels of education.
Finally, the mothers of the children in higher SES families
were expected to be more accurate than the mothers of the
children in lower SES families. A multiple regression
analysis was conducted to answer the above questions.
The multiple regression analysis was conducted with the
accuracy variable coded in such a manner that positive accuracy
scores indicate errors of overestimation, negative scores indi
cate errors of underestimation and scores of zero indicate per
fectly accurate predictions. It should be noted, that the
accuracy variable could have been coded such that no distinc
tion is made between the two different types of estimation
errors. That is, rather than use the difference scores with
their respective signs, the absolute value of the difference
scores could have been used such that zero scores would still
represent perfectly accurate predictions but anything greater
than zero would represent increasingly greater estimation
errors. In the analysis performed in which accuracy is not
coded in a truly linear manner, the possible linear relation
ships between accuracy and any other variables were examined
by including a quadratic term in the regression model. This
quadratic term was really testing for linear relationships of
the type that would emerge if the accuracy scores had been
coded in a truly linear manner. The linear term included in
the regression model was testing whether the two different
types of estimation errors (i.e., overestimations and under
estimations) differed in the way they related to the other
variables included in the analysis. It should be noted that
linear and quadratic terms were included in all the regression
analyses which included the accuracy variable.
The results of this analysis indicated that the
mothers' accuracy scores did not vary as a function of the
children's sex. Contrary to what was predicted, the
mothers' scores also did not vary as a function of the
children's age or the families' SES scores. A relationship
was found, however, between the accuracy scores and the
children's IQ scores. It appears that the mothers were less
accurate in predicting the IQ scores of children with either
very high or very low IQs and were more accurate in predict
ing the IQs of the children with average scores. The par
tial correlation between the mothers' accuracy scores and
the children's IQ scores after controlling for the effects
of all the other independent variables in the analysis was
r = .39, p .0013. This relationship was negative in nature.
90
That is, the mothers of children with low IQs had a tendency
to overestimate their children's IQs and the mothers with
children with high IQs had a tendency to underestimate their
children's scores.
It should be mentioned here that the correlation obtained
between the mothers' accuracy scores and the children's IQs may
be slightly inflated. This is so because whenever a correla
lation is computed between a variable x and a difference score
which has been derived using that same variable, the error of
measurement associated with the variable x will inflate the
correlation in the negative direction. It should be noted also
that it is possible that this relationship between the mothers'
accuracy scores and the children's IQ scores is artifactual in
nature. That is, if, for example, there was a tendency for
the majority of the mothers to predict their children's IQs
to be average or even slightly above average, then mothers
with children with very high or very low IQs would appear
less accurate since their children's actual IQ scores would
be further away from the average predicted scores. The re
lationship would arise not necessarily because the mothers
were less accurate in perceiving the abilities of the chil
dren with either very high or very low IQ scores but rather
because of a response bias on the part of the mothers who
may have preferred to give average or slightly above average
IQ predictions. A closer examination of the mothers'
predicted IQ scores, however, indicated that the mothers did
not exhibit such a response bias. The mothers gave a wide
range of IQ predictions and the distribution of their
predicted IQ scores closely parallels that of the actual IQ
scores of the children. Approximately 23% of the mothers
predicted their children's IQs to be in the 85 to 100 range
(24% of the children actually had IQ scores in this range),
30% predicted IQs between 101 and 115 (44% actual IQs fell
in this range), 25% predicted IQs between 116 and 129 (25%
actual IQs fell in this range). In addition, 19% predicted
IQs of 130 or higher (only 6% of the actual IQs were that
high) and only one mother predicted her child's IQ score to
be under 85 (only one child in the sample had an IQ score
that low). A response bias that would result in the type of
relationship found between the accuracy scores and the chil
dren's IQs is not clearly apparent.
A 'relationship was also found between the mother's
level of education and their degree of accuracy in esti
mating their children's IQ scores. The partial correlation
between these two variables after controlling for the
effects of all the other independent variables in the
analysis was r = .28, p .0245. This relationship indicated
that the higher the education level of the mother the more
likely she was to overestimate her child's IQ and the lower
the education level the more likely the mother was to under
estimate her child's IQ. In other words, contrary to what
was predicted, mothers with higher education levels did not
estimate their children's IQs more accurately than those
with lower education levels. They were, in general, more
inaccurate and more likely to overestimate their children's
92
IQs. Table 15 shows the percentage of mothers at three edu
cation levels who were accurate in predicting their chil
dren's IQ scores within 6 points and the percentage of those
who overestimated and underestimated their children's IQs.
As can be seen, a greater percentage of mothers with high
school educations were accurate than those with some college
or completed college. Also, the higher the education level
of the mother, the greater the percentage of overestimations
found.
The second major question raised in this study dealt
with the possible relationship between the mothers' accuracy
scores and the children's academic achievement. It was
hypothesized that children whose mothers had accurate
perceptions of their abilities would perform better in
school than those whose mothers had inaccurate perceptions
of them. Both overestimations and underestimations of the
children's abilities were expected to predict lower academic
performance than accurate perceptions. In order to test the
above hypothesized relationship, two parallel regression
analyses were conducted. One used the children's SATs and
the other used the children's GPAs as dependent variables.
The following independent variables were included in both
analyses: the mothers' accuracy scores, the children's age,
sex, IQs, the families' SES and the mothers' level of educa
tion. The last five of these variables were included for
control purposes. The linear models used in both analyses
included a linear and a quadratic term for the accuracy
variable. The results of these analyses showed that there
