SOCIOMETRIC CORRELATES OF LOCUS OF CONTROL IN
M. MICHAEL FAGAN
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
I would like to thank my chairman, Dr. Patricia Ashton. I
needed and appreciated her patient, thorough guidance. She helped
me to be more exact.
I would also like to thank the other members of my committee,
Dr. Mary Budd Rowe, Dr. Robert C. Ziller, and Dr. Walter A. Busby.
Dr. Rowe has helped to sharpen my thinking through her challenging
questions, Dr. Ziller has encouraged my creativity, and Dr. Busby
has expanded my viewpoint, showing me things that I have never been
aware of. I would also like to thank William Purkey for getting me
started on this project.
I am indebted to the students and teachers of Palatka Middle
School for participating in this project. I would also like to thank
my parents, Mathew and Rita, for giving me the confidence needed to
undertake and complete this project.
I would like to give special thanks to my wife, Patricia. She
worked very hard to give me the freedom to complete this study. With-
out her love and support it would have never been possible.
TABLE OF CONTENTS
ACKNOWLEDGMENTS . . ii
LIST OF TABLES . . v
ABSTRACT. . . ... ....... vi
I INTRODUCTION. . . ... 1
Background and Purpose. . ... 1
Significance of the Study . . 3
Theoretical significance . 3
Educational significance . 6
Definition of Terms . . 9
Assumptions and Limitations . 11
II REVIEW OF THE LITERATURE . .... 15
First Hypothesis: Internals Have More Friends than
Externals % . . 15
Second Hypothesis: Internals are More Popular on a Task-
Oriented Sociometric Item than on a Purely Social One 29
Third Hypothesis: People Prefer as Friends, Those Who
Are Similar to Them in Locus of Control ... 33
III METHODOLOGY . . .. 39
Subjects . . ... .. .39
Testing Procedure . .... 40
Instrumentation . . ... 40
The sociometric test ...... .......... 40
The Nowicki-Strickland Locus of Control Scale for
Children (N-S). . . .. 47
TABLE OF CONTENTS--continued
Statistical Analysis . . 50
First and second hypothesis . .. 50
Third hypothesis . . 52
Chapter Summary . . .. 53
IV RESULTS AND DISCUSSION . .. ....... 55
Results . . .. ... 55
Descriptive statistics . . .. 55
First hypothesis . . 58
Second hypothesis . . 60
Third hypothesis . . 62
Discussion of the Results . . 66
First hypothesis . . .. 66
Second hypothesis . . 69
Third hypothesis . . .. 70
V SUMMARY AND CONCLUSIONS . . .. 73
Summary . . 73
Conclusions . . .. 76
A The Nowicki-Strickland Locus of Control Scale for Children 81
B The Sociometric Questionnaire . .. 86
REFERENCES . . . 89
BIOGRAPHICAL SKETCH . . 100
LIST OF TABLES
4-1 Sex/race differences in mean locus of control scores 56
4-2 The magnitude and significance of variables used to predict
sociometric status on the best friends item. .. 59
4-3 The magnitude and significance of variables used to predict
sociometric status on the work-project item. .. 61
4-4 The magnitude and significance of variables used to predict
friends' locus of control. . ... 63
4-5 Chi square matrix indicating the locus of control of
friends chosen by internals, moderates, and externals. 64
4-6 Simple correlation and unique contribution of LOC in
predicting F-LOC for total sample, internals, and
externals ... .. . 65
5-1 Locus of control as a predictor of popularity as a friend
and work partner and friends' locus of control .... 76
Abstract of Thesis Presented to the Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
Sociometric Correlates of Locus of Control in
M. Michael Fagan
Chairman: Patricia T. Ashton
Major Department: Foundations of Education
This study investigated the relationship between adoles-
cents' locus of control and their peer preferences. Locus of
control was defined as the degree to which an individual expects
that his reinforcement is contingent upon his behavior; indi-
viduals with an internal.locus of control generally believe
that they control what happens to them, whereas those with an
external locus of control believe that their destiny is beyond
their control. and is determined by fate, luck or powerful others.
There has been a great deal of research indicating that in-
ternality is associated with academic success and cognitive super-
iority, but there has been relatively little investigation of
the social correlates of locus of control. This study related
locus of control to popularity as a friend, popularity as a
working partner, and interpersonal attraction. More specifically,
the following hypotheses were tested:
1. Internal adolescents are more popular (have more friends)
than their external peers.
2. Internals are more popular as working partners on an
academic project than they are as friends.
3. Adolescents tend to choose as friends those who are
similar to themselves in locus of control.
To investigate these hypotheses, 200 eighth graders were
administered the Nowicki-Strickland Locus of Control Scale for
Children and two sociometric items. The first sociometric item
was "Name the students who are your best friends," and the second
item was "If you had to work on a group project in English with
three other students who would you select to work with?"
Multiple regression was used to analyze each hypothesis.
For the first hypothesis, the dependent variable was sociometric
status on the best friends item (SMSbf); that is, the number of
times the subject was chosen as a best friend. The principal
independent variable was the subject's locus of control (LOC);
thus, locus of control was used to predict popularity. Sex and
race were controlled by forcing these two variables into the
multiple regression equation before locus of control. Also,
interaction between locus of control and sex (LOC x sex) and
locus of control and race (LOC x race) and curvilinear locus of
control term (LOC2) were checked for significance. Thus, the
complete multiple regression equation was as follows:
SMSbf = sex + race + LOC + (LOC x sex) + (LOC x race) + LOC2
The predictors for the other two hypotheses were the same
as above; the dependent variable for the second hypothesis was
sociometric status on the work-project item; and for the third
hypothesis,it was the subjects' friends' locus of control Thus,
the subjects' locus of control was used to predict their social
popularity, popularity as a working partner, and their friends'
locus of control.
In each of these multiple regression equations, with sex and
race partialed out, locus of control was a statistically signifi-
cant (p <.01) predictor of the dependent variable. However, locus
of control, with sex and race partialed out, accounted for less
than three percent of the variance in social popularity and less
than two percent of the variance in friends' locus of control.
Locus of control accounted for about eight percent of the variance
in sociometric status on the work-project item.
The data indicate that (1) the internal subjects were slightly
more popular, as friends, than the externals; (2) the internals
were considerably more popular as working partners; and (3) the
subjects, especially the internals, had a tendency to choose friends
of a similar locus of control, but most of this tendency was due
to racial and sexual preferences and racial and sexual differences
in locus of control.
Background and Purpose
Locus of control is a personality variable that has generated
a great deal of research in the past ten years (Phares, 1976). Rotter
(1966) developed the concept of locus of control and described it as
When a reinforcement is perceived by the subject as following
some action of his own but not being entirely contingent upon
his action, then,in our culture, it is typically perceived as
the result of luck, chance, fate, as under the control of
powerful others, or as unpredictable because of the great
complexity of the forces surrounding him. When the event is
interpreted in this way by an individual, we have labeled
this a belief in external control. If the person perceives
that the event is contingent upon his behavior or his own
relatively permanent characteristics, we have termed this a
belief in internal control. (p. 1)
Since its conceptualization, locus of control has been related to
academic achievement (Crandall, Katkovsky, & Crandall, 1965), informa-
tion seeking (Davis & Phares, 1967), birth control practice (MacDonald,
1970), socioeconomic status (Lefcourt, 1966) and a host of other vari-
ables. As Phares (1976) indicated:
The most basic characteristic of internal individuals
appears to be their greater efforts at coping with or
attaining mastery over their environments. This is
the most elemental deduction that could be made from
the nature of the I-E variable. Fortunately this
deduction has received widespread support from experi-
ments with many different populations in a variety of
situations, (p. 78)
Internals are generally more successful than externals at
achievement-oriented tasks, especially cognitive ones. For example,
as compared to externals, internals obtain better grades (Crandall
et al., 1965), are better problem solvers (DuCette & Wolk, 1973),
and utilize information more effectively (Seeman, 1963; Phares, 1968).
However, relatively little is known about the social correlates of
locus of control. It would be helpful to know the relationship, if
any, between locus of control and peer acceptance. Perhaps internals
are not only successful in school but also well-liked by their peers;
or perhaps their superior achievement is compensation for the lack of
social acceptance. Perhaps internals are more popular as working
partners than they are as friends. It would also be helpful to under-
stand the dynamics of interpersonal attraction with respect to locus
of control. Perhaps internals and externals complement each other and
become friends; or perhaps internals tend to associate with each other,
and externals, with fellow externals.
It was the purpose of this study to investigate the possibilities
mentioned in the preceding paragraph. More specifically the following
questions were asked:
1. Are internal adolescents more popular than their external
peers? (Are they selected more frequently as friends?)
2. Are internals more popular as working partners than they
are as friends? (Do they receive more selections as working partners
on an academic project than as friends?)
3. Do adolescents of a similar locus of control prefer each other
as friends? (Do internals tend to select other internals, and externals,
It was hypothesized that all of these questions would be answered
in the affirmative. In the next section the significance of these
hypotheses is discussed.
Significance of the Study
The theoretical and practical significance of this study are
discussed separately in the subsections that follow.
Theoretical significance. The concept of locus of control (Rotter,
1966) emerged from Rotter's (1954) social learning theory. The cor-
nerstone of Rotter's social learning theory is the principle that
behavior is a function of expectancy (Phares, 1976). According to
Rotter, one of the important factors that determines whether or not
a behavior will occur is the expectancy that the behavior will help
to produce a valued goal or outcome; the higher the expectancy the more
likely the occurrence of the behavior. By definition (Rotter, 1966),
internals generally expect that their behavior will produce desired
results, and, by definition, externals have a low expectancy that their
behavior will make a difference. Therefore, internals should behave
more effectively in goal-directed tasks. As previously mentioned, this
postulate has received strong empirical support in situations that
call for cognitive skill and mastery. However, the proposition has
rarely been tested in a social situation.
If Rotter's principle (behavior is a function of expectancy)
can be generalized from cognitive tasks to social ones, internals
should be more effective friend-seekers; and their effective friend-
seeking behavior should result in more friends. Thus, internals should
have higher sociometric status than externals on a sociometric item
that asks for "best friends."
The theoretical rationale for the first hypothesis of this study
is Rotter's principle that behavior is a function of expectancy. In
this case behavior is not measured directly, but rather, the inferred
result of behavior (friends) is measured by sociometric testing. The
expectancy variable is locus of control.
The rationale for the second hypothesis is an expansion of the
rationale of the first hypothesis. Not only should internals be more
effective friend-seekers, but they should also be perceived by their
peers as being more competent at academic tasks. Thus, internals should
receive some support (sociometric selections) on the basis of their
hypothesized social superiority (first hypothesis), plus additional
support for their cognitive superiority. It is expected that the
subjects will recognize the cognitive superiority of internals and
choose them as working partners. Therefore, the second hypothesis
is that internals have higher sociometric status on the task-oriented
sociometric item than on the purely social item ("best friends").
The rationale for the third hypothesis stems from the research
and theory in social psychology on interpersonal attraction. Newcomb
(1956, 1961, 1968) and Byrne (1971) have shown that similarity in
personality or attitude is a significant factor in interpersonal
attraction. Although there are many exceptions, people tend to be
attracted to those who are similar to themselves, especially if their
similarity is reinforcing (Byrne, 1971) and creates interpersonal
balance (Newcomb, 1968). According to Newcomb, a balanced relation-
ship occurs when a person favorably evaluates another person and both
agree in their evaluation of a third object or person. The simi-
larity hypothesis (the proposition that similar people are attracted
to each other) has been supported by a great deal of research (Byrne,
1971; Byrne & Griffitt, 1973).
The application of Newcomb's theory to locus of control suggests
that internals and externals would not have a balanced relationship,
because they would disagree on many important issues (for example,
the existence of luck or fate and the value of planning). Based on
Newcomb's theory, it was hypothesized that there would be a positive
correlation between a person's locus of control and the locus of
control of his friends.
Thus, this study has theoretical significance in that it tests
an extension of a social learning principle (behavior is a function
of expectancy) to a social situation. The study also tests the
similarity hypothesis with respect to locus of control. The studies
that support the theoretical rationale are discussed in the next
chapter, the review of the literature. In the next subsection the
educational significance of this study is discussed.
Educational significance. In the past 10 years educators have become
very interested in locus of control. McGee and Crandall (1968), Brown
and Strickland (1972), and Bartel (1971) found a positive relationship
between internality and academic success. However, the well-known
"Coleman Report" (Coleman, Campbell, Hobson, McPartland, Mood, Weinfield
& York, 1966) has been the most significant factor in alerting educators
to the significance of the locus of control variable. The authors of
this report concluded that
A pupil attitude factor, which appears to have a stronger
relationship to achievement than do all of the "school"
factors together, is the extent to which an individual has
some control over his destiny. (p. 23)
The evidence that internals tend to out-perform externals in
school, coupled with the evidence that internals are better adjusted--
psychologically--(Phares, 1976), has led many social scientists and
educators to advocate changing externals to internals; for example,
MacDonald (1973) wrote:
All of the research points to the same conclusion:
people are handicapped by an external locus of
control orientation. The prevailing belief is that
it is desirable to change people, especially those
who are not doing well in our society, in the direc-
tion of internality. (p. 170)
With this intention in mind, deCharms (1972), Stephens (1972),
Reimanis and Schaefer (1970), Nowicki and Barnes (1973), Hawes (1971),
and Ashton (1974) conducted experiments that were designed to help
children become more internal. The results of their work indicate
that certain methods help some people to become more internal and
that this change coincides with academic improvement or improvement
in skills related to academic success.
So far, the programs for changing locus of control have been
of an experimental nature. However, a group of educators and social
scientists (National Committee on Locus of Control and Self-Esteem,
1974) have recommended developing national norms for a locus of control
test, so that it can be administered to all public school students.
This standardized test would be used to measure growth in locus of
control (becoming more internal); teachers would be expected to improve
locus of control scores, as they are expected to improve their
students' academic achievement scores. The committee also recommended
several strategies for increasing internality.
Considering the movement toward internality training, an investi-
gation of the social correlates becomes crucial. If educators plan
to help students become more internal, they should know all that entails.
In this respect, this study should help in the following ways:
1. If the results of this study indicate that internals have
very few friends, educators need to systematically and thoroughly
investigate the social adjustment of internals, before initiating
any large-scale attempts to help students become more internal.
Perhaps gains in achievement are not worth possible social maladjust-
2. If the results of this study do not suggest that internals
are inadequate socially, educators can proceed with a little more
confidence in planning internality change programs.
3. If this study indicates that internality is positively
related to sociometric status and if subsequent experimental studies
indicate a cause-effect relationship between these two variables, pro-
grams designed to increase internality can be used to increase socio-
metric status, and/or programs designed to increase sociometric status
(Northway, 1944; Gronlund, 1959; Drabman & Lahey, 1974) can be used
to increase internality. Or, better yet, intervention programs can be
designed to work on both locus of control and peer acceptance.
4. If the third hypothesis is confirmed, indicating that externals
tend to associate with each other; it is likely that they perpetuate
their own externality through mutual reinforcement (Byrne, 1971) and
modeling. Thus,locus of control intervention programs could help
externals become more internal by structuring experiences where they
would interact with internals. Since externals are more persuadable
(Ritchie & Phares, 1969; Ryckman, Rodda & Sherman, 1972) and more
conforming (Crowne & Liverant, 1963) than internals, it is more
likely that internals would influence externals than the converse.
Thus, the study has practical significance in that it provides
educators with data about the social relationships of internals and
externals. These data are especially important now, because many
educators are thinking about increasing internal control in students.
Before educators begin to enact these plans, they need to know all
that it entails. Assuming that this study and subsequent studies
indicate that internals are not socially inadequate, this study may
suggest grouping techniques for helping externals to become more
internal. Thus, this study has both theoretical and practical signifi-
There are some limitations to the rationale of this study, and
they are discussed in the concluding section of this chapter. However,
before the assumptions and limitations are discussed, key terms are
defined in the section that follows.
Definition of Terms
The key terms used in this study are defined below. When appro-
priate, operational definitions are stated following the theoretical
1. Locus of control is a psychological construct that can be
viewed as a continuum reflecting the degree to which an individual
expects that his reinforcement is contingent upon his behavior.
At one end of the continuum is the expectancy that reinforcement
is contingent entirely upon one's behavior or aptitude internalityy);
at the other end is the expectancy that reinforcement is not at
all contingent upon one's behavior, but rather, reinforcement is
attributed to luck, fate or powerful others externalityy). Rotter's
(1966) definition of locus of control is cited on page 1 of this
report. Operationally, locus of control is defined as the score the
subjects receive on the Nowicki-Strickland Locus of Control Scale
for Children (N-S).
2. An internal is an individual who generally expects that his
reinforcement is contingent upon his behavior or aptitude. In this
study an internal is any subject who scores in the lowest third of
this sample on the N-S.
3. An external is an individual who generally expects that his
reinforcement is not contingent upon his behavior and aptitude,
attributing it to luck, fate or powerful others. For the purpose
of this study, an external will be operationally defined as a subject
who scores in the top third of this sample on the N-S.
4. A moderate is an individual who is not clearly internal or
external. In this study a moderate is a subject who scores in the
middle third on the N-S.
5. Sociometry is the method of determining social acceptance
by asking subjects to list associates with whom they would like to
participate in a certain activity. This technique, developed by
Moreno (1934), is discussed in more detail in the third chapter.
6. Sociometric status is one's relative position on a sociometric
test; that is, it is the number of selections one receives in compari-
son to his peers.
7. High sociometric status is the state of receiving many selec-
tions on a sociometric test, relative to one's peers. In this study
high sociometric status is considered the upper quartile of all sub-
jects ranked on the basis of the number of selections that they have
received. Researchers have used the term "high sociometric status"
interchangeably with "peer acceptance" (Adinolfi, 1970) and "popularity"
(Horowitz, 1967). This author will also use these terms interchangeably,
especially in the review of the literature.
8. Low sociometric status is the state of receiving few or no
selections from one's peers on a sociometric test. In this study low
sociometric status is operationally defined as the lowest quartile of
subjects ranked according to the numbers of selections they received.
Assumptions and Limitations
Much of the theoretical relevance of this study is based on the
assumption that both internal and external adolescents value friend-
ship. When Rotter (1954) postulated that behavior is a function of
expectancy, he included the value of the reinforcement in his formula.
Thus, in situations where it is expected that behavior will produce
reinforcement and the reinforcement is valued, behavior is likely to
occur. For this study it is assumed that the need for peer accep-
tance in early adolescence is so pervasive in our culture that (1)
internals and externals value friendship equally (although there may
be some externals who do not value friendship highly, there should
be about the same number of internals who do not value friendship
highly), or that (2) if there are differences between internals and
externals in the amount they value friendship, these differences are
not nearly as great as their difference in locus of control.
Also, the theoretical rationale is based on the assumption that
friendships are, in part, a product of behavior. This postulate
has been supported by sociometric research in that individuals of
different sociometric status behave differently toward peers (Gronlund,
1959; Cole & Hall, 1970) and that training an individual in social
skills can increase his sociometric status (Gronlund, 1959).
It is also assumed that a generalized measure of locus of control
(the N-S) includes social relationships and that the total locus of
control score reflects one's sense of control over social relation-
ships. The multidimensionality of locus of control is a concern;
a number of investigators (Mirels, 1970; Abrahamson, Schulderman &
Schulderman, 1973; Gurin, Gurin, Lao & Beattie, 1969), by using factor
analysis, have found that locus of control contains several dimensions.
For example, Mirels (1970) found two factors, personal control (feelings
of personal efficacy) and political control (ability to affect the
political system). According to social learning theory (Phares,
1976), a situation-specific measure of locus of control would be
more effective than a generalized measure in predicting behavior
relevant to that situation. Therefore, for this study the instrument
that measured the subjects' feelings of control about their peer
relationships would be ideal. Unfortunately, no such instrument exists.
There are several items on the N-S that relate to peer acceptance,
for example, item # 25: "Do you believe whether or not people like
you depends on how you act?" (see Appendix A). However, Nowicki (1973)
has factor analyzed the N-S and found no substantial factor that repre-
sents a sense of control over social situations. Therefore, out of
necessity, a global locus of control score is used to predict socio-
metric status, and it is assumed that this global score reflects the
subjects' sense of control over their interpersonal relationships. In
other words, it is assumed that there is a high correlation between
adolescents' global locus of control and their locus of control about
Concerning the practical significance of this study, it is assumed
that at least a minimum of friends is healthy and desirable for eighth
graders, and conversely that the absence of friends is not a desirable
condition for adolescents.
This study is limited, by intention, in that it is an exploratory
investigation of the relationship between two broad and general traits.
This study is not intended to tease out precise relationships between
locus of control, peer interaction and other variables; but rather,
it represents a necessary forerunner to such studies; it is expected
that this study will generate hypotheses about more.precise rela-
tionships between locus of control and adolescent peer interaction.
Considering the broad scope of this study, some of its limitations
are discussed below:
1. Due to the correlational nature of this study, a cause-
effect relationship between locus of control and peer acceptance can-
not be inferred.
2. Sociometric procedures do not indicate the quality of a
friendship. Perhaps internals have more friends, while externals have
fewer, but qualitatively superior, friendships which could be more
intimate, more loyal, or more enjoyable.
3. Both the N-S and sociometric testing are self-report inventories
and are subject to the criticisms inherent in that technique; the
main criticism being that subjects, consciously or unconsciously,
report inaccurately. *
In this chapter the purpose, rationale, and limitations of this
study have been discussed. The next chapter reviews the literature
relevant to the three hypotheses of this study.
REVIEW OF THE LITERATURE
Locus of control is a relatively new concept; most of the
research on this variable has been conducted in the past 15 years.
On the other hand, sociometric research was popular from about 1935
to 1950; since then, interest in sociometry has waned. As a result,
there have been very few studies (Nowicki & Roundtree, 1971; Adinolfi,
1970) that have directly related locus of control to a sociometric
Considering the paucity of evidence that directly relates socio-
metric variables to locus of control, most of the literature reviewed
for this study consists of indirect support of the hypotheses. For
example, locus of control has been related to some variables that
resemble or are related to sociometric status, and sociometric status
has been related to personal variables that are similar to locus of
control or related to it. These studies are discussed in this chapter, to-
gether with the few studies that provide more direct evidence. The
literature is reviewed as it relates to each hypothesis.
First Hypothesis: Internals Have More'Friends than Externals
Internality has been related to several social variables that
suggest internals should be more popular. For example, Midlarsky (1971)
found that internals were more helpful to strangers than were externals.
In Midlarsky's experiment each subject performed a task in the presence
of an accomplice of the experimenter. Each person was assigned a task
and was told that he could help the other person if he finished first.
The experiment was designed so that the true subject always finished
first. The internal subjects helped the accomplice significantly more
than the external subjects did. If internals are also more helpful
in their daily interactions, they should be more popular than externals,
because helpfulness has been associated with high sociometric status
(Reader & English, 1947; Cole & Hall, 1970).
Not only are externals less helpful than internals, but there
is also some evidence that externals desire more social distance than
internals, especially with strangers (Duke, 1973; Duke & Nowicki, 1972;
Duke & Fenhagen, 1975). In the Duke and Nowicki study subjects used
the Comfortable Interpersonal Distance Scale (CID) to indicate the
distance at which hypothetical people should stop when approaching
them; externals required more distance for strangers than did internals.
Duke and Fenhager (1975) administered the N-S and the CID to a group
of adolescent females; they found that externality was associated with
a greater preference for social distance. This suggests that externals
may be uncomfortable in social relationships, especially new ones,
and be less likely to interact with people and, therefore, less likely
to have friends.
Another reason for the hypothesized popularity of internals is
the finding that internality is more socially desirable than externality
(Bernhardson, 1968; Cone, 1971; Hjelle, 1971). Although earlier
studies have found no relationship between social desirability and
locus of control (Strickland, 1965; Tolor, 1967; Tolor & Jalowiec,
1968), the most recent evidence indicates that there is a moderate
relationship between internality and social desirability (Cone,
1971; Hjelle, 1971). The social desirability of a trait or the
instrument that measures the trait has been assessed in two ways.
One method is to have judges rate the social desirability of the
items of an instrument. For example, a judge rating the fourth
item of the N-S (item # 4. "Most of the time do you feel that
getting good grades means a great deal to you?") might conclude
that a "yes" response is more socially desirable than a "no" response.
In other words, it is desirable in this society to care a great
deal about one's grades. Using this method, Hjelle (1971) and
Bernhardson (1968) found that an internal view is more socially
desirable than an external view. Hjelle's finding suggests that,
since the internal point of view is valued more in this culture,
people who have this point of view should be more accepted and
popular than people who have the opposite point of view.
The second, and more frequently used, method of assessing social
desirability has been to use a standardized instrument. The
Edwards Social Desirability Scale (Edwards, 1957) and the Marlow-
Crowne Social Desirability Scale (Crowne & Marlow, 1964) are self-
report inventories that consist of items that mention socially
desirable or undesirable traits. Subjects who answer many items
in the socially desirable direction are describing themselves as
an ideal person. Cone (1971) correlated the locus of control scores
on the Edwards Social Desirability Scale; in all five samples inter-
nality was associated with high social desirability. Cone concluded
that internals may behave in a socially acceptable manner so that
they can influence people favorably. This conclusion supports the
rationale of the first hypothesis of this study, in that it suggests
that internals believe they can affect their social environment and
act accordingly. If this is true, internals should be more socially
competent than externals and should be more popular.
There have been two studies (Phares & Wilson, 1971; Nowicki &
Blumberg, 1975) that have related locus of control to interpersonal
attraction. Of course, these are directly related to the third
hypothesis (and are discussed in that context later in this chapter),
but the unexpected results of these two studies lend support to the
In the first of these two studies, Phares and Wilson (1971)
asked internal and external college students to evaluate an alleged
stranger based on his responses to Rotter's Internal-External Locus
of Control Scale (Rotter, 1966). Actually there were no strangers,
only two Internal-External scales--one completed in a totally internal
direction and the other in a totally external direction. The internal
subjects favored the internal "stranger," and the externals, as a
group, did not favor either.
As a follow-up to the Phares and Wilson study, Nowicki and
Blumberg (1975) carefully prepared four tapes, two of a male and
female internal discussing their views and two tapes of their external
counterparts. All four tapes were judged by the subjects and independent
raters to express accurately the intended locus of control orientation.
Forty college students were asked to judge the alleged strangers on
three criteria: (1) their general likability, (2) their desirability
as a working partner, and (3) their desirability as a roommate. The
internal "stranger" was significantly more attractive to both the inter-
nal and the external subjects on three criteria, but especially on the
third (desirability as a roommate). The Phares and Wilson study and
the Nowicki and Blumberg study suggested that, since internals are
preferred in an experimental situation, they should also be more popular
in a real-life situation, like a classroom.
Before discussing the two studies that directly relate locus of
control to sociometric status, the indirect evidence is summarized below
in support of the hypothesis that internals are more popular than exter-
nals. This research indicates that
1. Internals are more helpful.
2. Internals require less personal space or social distance.
3. Internality is a more desirable trait in this society.
4. In experiments where subjects were asked to express their pre-
ference for either an internal or an external "stranger" most subjects
preferred the internal.
The fourth point seems to offer the strongest support for the
hypothesis. However, both of these studies (Phares & Wilson, 1971;
Nowicki & Roundtree, 1975) were conducted under simulated conditions
with alleged strangers. The present study is concerned with on-going
relationships. Unfortunately, there is very little investigation
relating locus of control to real, on-going friendships. The extensive
literature review of this author has yielded only two such studies.
In the first of these studies, Nowicki and Roundtree (1971)
administered the N-S and two sociometric questions to a sample of
12th graders (N = 87). The first sociometric item asked the students
to list five classmates that would be good candidates for classroom
president; the second item asked the subjects to list five of their
best friends. Among males there was a significant relationship
between internality and nominations for class president, but there
was no relationship between locus of control and friendship. In
other words, internal males received more selections as class presi-
dent than did external males, but internals of either sex were no
more popular than externals as friends.
In the second of these studies, Adinolfi (1970) administered
a sociometric test to 600 college dormitory residents. From this
subject pool he selected four groups of 30 subjects each (15 males
and 15 females)--a highly accepted group, an unaccepted group, a
rejected group, and a control group. Then these subjects (N = 120)
were administered the Rotter Internal-External Locus of Control Scale
and several other personality tests. Among these four groups there
were no differences in locus of control scores.
The results of these two studies are contrary to the first
hypothesis of this study. However, it is expected that this study
will produce results different from those of Nowicki and Roundtree
and Adinolfi for the following reasons:
1. Their subjects were older than the subjects of this study.
According to studies by Costanzo and Shaw (1966) and Brownstone and
Willis (1971), younger adolescents are more influenced by their peers
than are older adolescents. This suggests that peer acceptance is
probably more important for a 13-year-old than for an 18-year-old.
Since peer acceptance is a crucial, ego-involved need for a 13-year-
old, it is likely that peer acceptance is related to his sense of
potency and control.
2. Adinolfi's subjects were volunteers from the original subject
pool; they were not randomly selected from their group. Campbell and
Stanley (1966) cited the use of volunteers as a possible source of
3. The previously mentioned locus of control studies suggest
that internals are more likable.
4. Many sociometric studies have related sociometric status to
personal variables that either resemble locus of control or have been
related to it. Many of these studies have used subjects similar in
age to the subject of this study in the same social environment,
the classroom. These studies constitute the remainder of the review
of the literature for the first hypothesis and are reviewed in the
paragraphs that follow.
A review of sociometric literature reveals two groups of studies
that support the hypothesis that sociometric status is related to
locus of control. Firstly, there are two clinically based studies
(Northway, 1944; Northway & Wigdor, 1947) that produced descriptions
of sociometrically high and low students that, in part, closely resemble
theoretical and empirical descriptions of internals and externals,
respectively. Secondly, sociometric status has been related to many
variables (e.g., self-esteem) that have also been related to locus
of control, such that internals parallel sociometrically high subjects
(e.g., high in self-esteem) and externals parallel the sociometrically
low (e.g., low in self-esteem).
Using the Rorschach test, Northway and Wigdor (1947) studied
eighth graders of high, average, and low sociometric status. The well-
liked subjects were described as having
S. a greater sensitivity to their environment--
almost an active, conscious striving in using the "feeling
tone" and social contacts of a situation to further their
ends. (p. 194)
It is logical to expect that an individual who believes that he
is in control of a situation will be more sensitive to and manipulative
of the environment. There is abundant evidence that internals are
more alert to environmental cues and better able to use these cues
than their external counterparts (Seeman & Evans, 1962; Seeman, 1963,
1966, 1967; Phares, 1968; Lefcourt & Wine, 1969; Williams & Stack,
1972). For example, Seeman (1963) found that internal prison mates
were more knowledgeable about parole policy than external inmates of
similar intelligence. Seeman concluded that internals had more
actively sought and more efficiently processed information that might
After reviewing the studies mentioned in the preceding paragraph,
Phares (1976) concluded that internalss seem to be eager to seek out
clues and to manipulate the situation so as to be better able to achieve
certain outcomes" (p. 63). Thus, Northway and Wigdor's description of
the well-liked eighth graders suggests that they are social internals;
that is, they actively seek out clues in the social environment and
use these clues rather effectively, as evidenced by their social success.
In the second of the clinically based studies, Northway (1944)
conducted an in-depth study of sociometrically low fifth and sixth
graders. Her results provided a holistic view of the personality
types of students who are not well accepted by their peers; these
personality profiles suggest relationships between low sociometric
status and locus of control. For two years Northway collected data
on 20 unpopular subjects through school records, interviews with
parents, intelligence tests, psychological interviews, classroom and
playground observation, and sociometric tests. Her investigation
yielded three types of unaccepted children. Of these 20 subjects
six were "recessive children" who were described as "listless, lacking
vitality, unhealthy, below normal in intelligence or underachieving,
unkempt, and seemingly uninterested in people, activity, or events of
the outside world."
There were nine "socially uninterested children" who Northway
described in the following manner:
These children superficially appear to be similar to
the truly recessive children. They are not liked by
others nor do they appear to make an effort in either
formal class activities or social affairs of the school.
They are often quiet and retiring. However, on closer
examination it is found that they are much better
developed in their care of person and possessions and
that they have interests. Those interests are personal
rather than social. A child's energy may be directed
towards art, music, science, hobbies, reading or to
affairs of the home. Some of these children are merely
quiet with and uninterested in other children; some
are shy and uncomfortable with them, some are bored
and critical of them, and some are rather objectively
and impersonally interested in observing what other
children do without in any way attempting to participate
with them. (p. 458)
The third group consisted of five "socially ineffective children"
who were described in the following manner:
Children of this group differ completely in their super-
ficial behavior from the former group; they are often
noisy, rebellious, delinquent in classroom affairs,
boastful and arrogant. They are a nuisance to the teacher
and the life of the classroom. They are diametrically
opposed to the recessive children. However, this is not
true, for they have in common the lack of acceptance by
classmates and these manifested forms of behavior seem
to have arisen as rather ineffective, naive attempts
to overcome the basic social insecurity and isolation
from group life that they experience. They have vitality
and are keenly interested in social affairs, but because
of failure in the establishment of social relations, they
make effortful, conspicuous,.and often foolish and futile
attempts to be recognized and accepted by the social group.
(Northway, 1944, p. 458)
Northway supplemented these psychological profiles with two case
studies from each group. The recessive state described by Northway
closely parallels'Seligman's (1976) concept of learned helplessness.
Learned helplessness is a state that in human subjects greatly resembles
reactive depression (Seligman, 1973) and is caused by an individual's
learning that there is no relationship between his behavior and what
happens to him. Seligman (1973) and Miller and Seligman (1973) have
related learned helplessness to locus of control. It seems that learned
helplessness is an extreme case of externality.
Thus,it is likely that the recessive children are inactive and
listless because they have come to believe that they have little or
no control over what happens to them. Recessive subjects should have
a very external generalized locus of control and should score highly
on the Nowicki-Strickland Scale.
Northway's description of the "socially uninterested" subjects
suggests that they have an external locus of control in social matters,
but an internal locus of control in personal matters. That is, they
probably expect that they can control most things that they do alone,
but they doubt their ability to affect other people, especially
their peers. The sample of this study will probably include some
socially uninterested adolescents; it is expected that they will
obtain low to moderate scores on the N-S, because they should answer
the social items internally and the nonsocial items externally.
Seven of the 40 items are clearly concerned with peer involvement.
"Socially ineffective children" seem to have a strong desire
to affect their peers, but lack the skill and self-confidence to
affect them positively. Northway's description and case studies sug-
gest that these subjects have deeply rooted, but disguised, feelings
of inferiority; and their verbal reports and behavior are often attempts
to compensate for their inadequacy. These subjects have a social
and probably generalized external locus of control; but their verbal
reports are likely not to indicate this, so their N-S scores could
Northway's study suggests that although there is a general trend
for almost all socially unaccepted children to have an external locus
of control in social matters, this trend will be obscured by the non-
social internality of the "socially uninterested" and the defensiveness
of the "socially ineffective." Therefore,only a moderate correlation
should be expected between sociometric status and the N-S scores of
There is some less direct, but statistically based, evidence
supporting the hypothesis that socially accepted adolescents tend to
be more internal. This evidence is the list of traits that have
been significantly related to both sociometric status and locus of
control. Of course, just because two variables are related to a third
variable is no indication that they are necessarily related to each
other; but if two psychological variables are related to several vari-
ables, it is likely that these two variables are related to each other.
As previously mentioned, sociometric status has been related to
academic achievement (Bonney, 1943; Grossman & Wrighter, 1948; Brown,
1954; Feinberg, 1953) and socioeconomic status (Brown, 1954; Cook,
1945; Neugarten, 1946; Grossman & Wrighter, 1948). Locus of control
is also related to academic achievement (Coleman et al., 1966; Crandall
et al., 1965) and socioeconomic status (Lefcourt, 1966).
Sociometric status is clearly related to self-concept (Baron, 1951;
Guardo, 1969; Horowitz, 1962; Reese, 1961; Videback, 1960), as is locus
on control (Fitch, 1970; Fish & Karabenick, 1971); and sociometric status
is negatively correlated with anxiety-(Baron, 1951; Mill, 1953), as
is locus of control (Butterfield, 1964; Watson, 1967; Feather, 1967;
Ray & Katahn, 1968). Subjects of low sociometric status have been found
to be depressed (Baron, 1951; Northway, 1944), as have externals
(Abramowitz, 1969; Miller & Seligman, 1973). Low sociometric status
has been associated with schizophrenic tendencies (Northway, 1944;
Northway & Widgor, 1947) and so has externality (Cromwell, Rosenthal,
Shakow & Zahn, 1961; Harrow & Ferante, 1969). With all of these traits
in common it is likely that most adolescents of high sociometric status
are internal, and most adolescents of low sociometric status are
Before presenting the literature relevant to the second hypoth-
esis, this section is summarized below:
1. Internals should be more popular than externals, because
they are more helpful, require less social distance, and possess a
viewpoint that is valued more by this culture.
2. In two experimental studies (Phares & Wilson, 1971; Nowicki
& Blumberg, 1975) subjects preferred an internal "stranger" to an
3. Two studies (Nowicki & Roundtree, 1971; Adinolfi, 1970) have
found no relationship between locus of control and sociometric status
for older adolescents.
4. Two studies (Northway, 1944; Northway & Wigdor, 1947) have
produced clinical descriptions of students of high and low sociometric
status; these descriptions resemble internals and externals, respectively.
5. Internals have many traits in common with subjects of high
sociometric status. As compared to externals and unpopular subjects,
internals and popular subjects tend to do better in school, come from
families of higher socioeconomic status, have positive self-concepts,
less anxiety, less depression, and less schizophrenic tendencies.
In the next section the literature relevant to the second hypoth-
esis is reviewed.
Second Hypothesis: Internals are More Popular on a Task-Oriented
Sociometric Item than on a Purely Social One
The second hypothesis compares the sociometric status of internals
on two different items: (1) "Name the students who are your best
friends" and (2) "If you had a group project in English (class) with
three other students, and half of your grade depended on this project,
which three students would you select to work with?" (These two items
are referred to as the "best friends" item and the "work-project" item,
The second hypothesis is that internals have higher sociometric
status on the work-project item than on the best friends item. The
rationale for this hypothesis is based on two well-documented findings:
(1) the nature of the sociometric item affects who is popular and who
is not, and (2) internals are academically superior to externals.
The difference in the nature of the sociometric items, relative to
this hypothesis has been discussed by Jennings (1947). She suggested
that there are two types of sociometric items, those related to an in-
formal, personal situation (like the best friends item) and those related
to more formal situations that usually involve a group project (such as
the work-project item); Jennings suggested that these two types of
items tap two underlying group structures, an informal system and a
formal system, respectively. According to Jennings, choices on items
that tap the formal system are based more on ability and less on social
popularity than are choices on items related to the formal system.
Jennings conclusion has been supported by Gronlund. After reviewing
several studies that compared sociometric status on different type
items, Gronlund (1959) concluded that there is a general social
acceptability factor for all items, but that
Where specific criteria indicate the need for knowledge
or skill, the social acceptability factor is surpressed
somewhat, in the choosing, in factor of success in the
activity. (p. 138)
Thus, in this study it was expected that the academically competent
students would have higher sociometric status on the working-project
item than on the best friends item. Since internals are better
represented among the academically competent, it was expected that
they would have higher sociometric status on the work-project item
than on the best friends item.
The studies which indicated that internals are academically
superior to externals were cited in the first chapter (Crandall
et al., 1965; McGee & Crandall, 1968; Brown & Strickland, 1972;
Bartel, 1971; Coleman et al., 1966). These studies were correlational.
For example, Crandall and her associates studied the relationship
between locus of control and school grades for students from the
third through 12th grades. Internal males averaged 2.54 (on the 4.0
system), and external males averaged 2.26; and internal males averaged
2.95, and external females 2.70. The grades of the internal students
were significantly (p < .01) higher than the externals' grades.
Several experimental studies (deCharms, 1972; Hawes, 1971; Stephens,
1972; Ashton, 1974) have also found a relationship between sense of con-
trol and academic performance. For example, deCharms (1972) investigated
the relationship between academic performance and "personal causation"
which he defined as "the initiation by an individual of behavior intended
to produce a change in his environment." Personal causation is similar
to locus of control in that they are concerned with an individual taking
control of his environment. Theoretically, personal causation is concerned
with intrinsic motivation, and locus of control is concerned with expec-
tancies about external reinforcement; but Phares (1976) and Lefcourt
(1966) have noted that there is a great deal of overlap between these
two concepts. Thus, deCharms' experiment has relevance to locus of con-
trol. In this study,he and his associates randomly assigned half of the
sixth grade teachers (N = 16) in an urban school district to an experi-
mental group; the other half constituted the control group. The experi-
mental group underwent "personal causation training",which was a program
designed to facilitate "origin" behavior among the teachers and the stu-
dents they taught. ("Origin" is a term that is similar to "internal.")
After the training sessions,the teachers conducted similar sessions in
their classrooms for the next two academic years. The results of the
study indicate that the experimental group, as compared to the control
group, had students who (1) rated their classrooms as more conducive to
independent, self-motivated study, (2) scored higher--more origin (inter-
nal)--on deCharm's Origin-Pawn measure, and (3) scored higher on stan-
dardized achievement tests.
Thus, the relationship between internality and high academic
achievement has been clearly established. Although there are excep-
tions, internals generally perform better in school than their
In this section it has been argued that the best friends socio-
metric item favors students who are academically competent. Since
internals are generally academically superior to externals, they
should have higher sociometric status on the work-project item. Also,
internals should have higher status on the work-project item than on
the best friends item, because of their academic superiority.
This researcher has been able to find only one study that has
related locus of control to two different sociometric items. In this
study, which was mentioned in the previous section, Nowicki and
Roundtree (1971) related the locus of control of 12th graders to
their sociometric status on a best friends item and on an item that
asked for good candidates for class president. Nowicki and Roundtree's
study is comparable to this one, since the first item is identical
to the first item of this study, and the second item is related to
the second item of this study in that they both have to do with
achievement oriented activities. The results of the Nowicki and
Roundtree study supported the second hypothesis of this study in that
internals, at least the internal males, had higher sociometric status
on the class president item than on the best friends item.
Thus,the hypothesis that internals should do better on a task
related item than on a purely social sociometric item has received
partial support. In the next section the literature relevant to
the last hypothesis is discussed
Third Hypothesis: People Prefer, as Friends, Those Who Are Similar
to Them in Locus of Control
Interpersonal attraction is one of the most studied phenomena
in social psychology. Laymen and psychologists alike have been
curious about why people are attracted to some people and not others.
One of the simplest explanations is that people prefer those who are
similar to themselves. There is abundant empirical evidence that this
explanation has some validity (Byrne, 1961; Byrne & Griffitt, 1973;
Newcomb, 1961), but only under certain conditions and only for cer-
The proposition that people are attracted to those who are
similar to themselves in attitude and personality has been called the
"similarity hypothesis." Byrne and Griffitt (1973) have described
three methods that have been used to test the similarity hypothesis,
Three basic research designs have been used to study
the influence of personality similarity on attraction.
In one approach, existing attraction pairs such as
friends, or spouses are selected and then assessed
with respect to one or more personality variables;
the scores of the series of pairs are correlated. In
a second approach, the personality measure or measures
are obtained, and then previously unacquainted subjects
are selected on the basis of their test scores and
placed in an interactive situation, followed by an
assessment of their attraction. .. In research
utilizing a third design. the subject's per-
sonality relevant behavior consists of his response
to the instrument used to assess personality char-
acteristics, and he is subsequently exposed to the
responses of the target on the same instrument with
other stimulus and elements controlled experimentally.
With this design it has been shown that attraction
is positively related to similarity. (pp. 320-321)
By using the third design it has been shown that people are
attracted to those who are similar to themselves in self-concept
(Griffitt, 1966, 1969), the need for approval (Goldstein & Rosenfeld,
1969), and dominance-submissiveness (Palmer & Byrne, 1970).
The third design has been used to test the similarity hypothesis
for locus of control in two studies, mentioned previously in this
chapter. In the first of these studies, Phares and Wilson (1971)
tested the hypothesis that internals are attracted to other internals
and that externals are attracted to fellow externals. As previously
mentioned their hypothesis was only partially confirmed; although
the internals subjects did prefer the internal "stranger," or the
externals showed no preference for either the internal or the
In the second of these studies, Nowicki and Blumberg (1975)
hypothesized that Phares and Wilson did not confirm the similarity
hypothesis, because the technique they used was "less than realistic."
They tested the same hypothesis using audio tapes, instead of com-
pleted Rotter scales. The hypothesis was still not confirmed; both
the internal and the external subjects preferred the internal "stranger."
In these two experiments the failure to confirm the hypothesis
may be due to the highly contrived conditions and the social desir-
ability of locus of control. Under these conditions, the subjects
may simply choose the person that seems more socially desirable
rather than choose the person that would really be a good friend.
Most of the studies that have supported the similarity hypoth-
esis and the studies by Phares and Wilson and Nowicki and Blumberg
belong to the third category described by Byrne and Griffitt
(1973). These studies have two factors in common: (1) their
method of eliciting interpersonal attraction is highly contrived,
and (2) they measure only initial attraction (first impression).
There have been some studies without these shortcomings, the
most noted being Newcomb's (1961) study. Newcomb invited 34 male
students transferring to the University of Michigan to live together
in a house near campus. In exchange for this rent-free opportunity,
the students agreed to participate in a research project which
entailed regular testing and interviewing. As mentioned previously,
Newcomb found that initial attraction was highly related to pro-
pinquity; generally the closer the rooms, the greater the chance of
interpersonal attraction. However, later attraction was more
strongly related to perceived similarity in attitudes. Newcomb's
study suggests that individuals in a group who have had the opportunity
to become acquainted with each other will associate with those group
members whom they perceive as similar to themselves.
Sociometric testing has also been used to test the similarity
hypothesis; this method is the first type described by Byrne
and Griffitt (1973). The results of these studies have been incon-
sistent. For example, Pinter et al., 1937, studied the friendships
of fifth through eighth graders and found that friends were not similar
in ascendance-submission, emotional stability, and extroversion-
introversion; and Thorpe found that classroom friends were.not sim-
ilar in neuroticism. However, Bonney (1945), Austin and Thompson
(1948), and Van Dyne (1940) found that adolescents and young adults
were more similar in personality to their friends than to nonfriends.
In these studies the personality trait that the friends had in com-
mon was a social nature, namely social and emotional adjustment
(Bonney, 1945), similarity of needs (Austin & Thompson, 1948), and
degree of dominance and sociability (Van Dyne, 1940).
The study of interpersonal attraction is further complicated
by the research of Winch (1958) which indicate that men and women
of complementary traits are often attracted to each other and main-
tain good marital relationships. Although married couples tend to
be similar in race, sociometric status, and religion, they often
complement each other on certain personality traits, especially those
related to dominance and submission, that is, people with dominant
needs tend to marry partners with submissive needs.
Hollander (1967) has evaluated the research on interpersonal
attraction and has concluded that
similarity is important for some kinds of relationships
and complementary for others. In most studies where
similarity is found to hold as a factor yielding a mutual
bond, attitudes and values are those elements which are
being measured. Complementary, on the other hand, may
be relevant to need satisfaction in an enduring inter-
action. (p. 193)
The peer relationships of 13-year-olds, although powerful in
an almost impersonal way, are still not so intimate and enduring
(Horrocks, 1976) that complementary need satisfaction would likely
be an overriding factor. It is only with mature, intimate rela-
tionships that complementary traits seem to be significant, as
evidenced by the fact that the complementary hypothesis ("opposites
attract" has been supported with research using married couples
(Winch, 1958; Hollander, 1967). Therefore, similarity in locus of
control, rather than complementarity in locus of control, should be
a significant factor in the peer relationships of young adolescents.
In the present study it is hypothesized that the social pre-
ferences of these subjects are related to perceived attitude similarity
and that the constellation of attitudes significant to these 13-year-
olds is related to their locus of control orientation. Therefore,
it is predicted that there is a moderate, but significant, relation-
ship between locus of control similarity and attraction; that is,
the subjects should choose classmates who are similar to themselves
in locus of control.
In this section there.was a brief review of the literature
investigating the similarity hypothesis. Studies were mentioned
which indicated that people tend to be attracted to those who are
similar to themselves in self-concept, need for approval, and
dominance-submissiveness. Two studies (Phares & Wilson, 1971;
Nowicki & Blumberg, 1975) were discussed which tested and did not
completely support the similarity hypothesis of locus of control.
It was hypothesized that this study would support the similarity
hypothesis for locus of control, because a more realistic method
(sociometric testing) was used to elicit interpersonal attraction.
In the next chapter the methodology of this study is discussed.
Two hundred eighth graders were administered a locus of control
instrument and a sociometric questionnaire. The results were analyzed
by multiple regression to investigate the following hypotheses:
1. With sex and race partialed out,
relationship between locus of control and
best friends item such that internals are
2. With sex and race partialed out,
relationship between locus of control and
work-project item such that internals are
than they are on the best friends item.
3. Also with sex and race partialed
there is a significant
sociometric status on the
more popular than externals.
there is a significant
sociometric status on the
more popular on this item
out, there is a significant
relationship between the subjects' locus of control and the locus of
control of their friends such that the subjects choose friends who
are similar to themselves in locus of control.
The subjects were 200 eighth graders from a middle school in a
city in North Central Florida with a population of 10,000. According
to the school administration, most of the school's students (80-85%) were
from families of lower middle and lower socioeconomic status. The
sample included 112 females and 88 males, 161 whites and 39 blacks.
The modal age was 13.
The sociometric test and the Nowicki-Strickland Locus of Control
Scale for Children (N-S) were administered by two English teachers
at the school in December of 1975. The teachers administered the
forms to all of their classes which met during different class periods
throughout the school day. The sociometric questionnaire was administered
first, and the N-S was administered one week later with a battery
of other tests. [The other data collected at this time were used
in a study by Damico and Purkey (1976.)] The students were assured
that their responses would remain confidential.
The sociometric test. The first sociometric item used for this study
was "Name the students who are your best friends." Three blank spaces
were provided below the items. The subjects were not directed to list
the friends in any specific order, nor were they explicitly told that
three names were expected. The subjects could list fewer or more than
three names, but most did list three and no more than three were
analyzed in this study.
The other item was "If you had to work on a group project in
English with three other students, and half of your grade depended
on this project, which three students would you select to work with?"
Three blank spaces followed this item. There were several other socio-
metric questions on this form (see Appendix B), but they were not be
analyzed in this study.
Although both the items used in this study are referred to as
sociometric, this is not exactly correct. Lindzey and Byrne (1968)
summarized Moreno's (1934) requirements of a sociometric test, as
1. The limits of the group should be indicated to the subjects.
For example, choices could be limited to members of a specific class-
2. The subjects should be permitted an unlimited number of
3. The subjects should be asked to indicate the individuals
whom they choose or reject in terms of a specific criterion. Each socio-
metric choice should be made with a meaningful activity in mind, for
example, "name the students you would like to sit next to in class."
4. Results of the sociometric question shou-ld be used to
restructure the group. Subjects should be told that their input will
be used in making such decisions.
5. The subjects should be permitted to make their choices privately.
6. The questions used should be gauged to the level of under-
standing of the sample.
Although the work-project item fulfilled most of these require-
ments, the best friends item violated the second, third, and fourth
requirements. The second requirement, unlimited choices, was violated
on both items, because unlimited choices would be more difficult to
manage in terms of statistical analysis. The third requirement, the
use of a specific activity, was not used for the best friends item;
because the first and third hypothesis of this study were intended to
investigate long-term relationships. It was assumed that when adoles-
cents were asked to list best friends, rather than to name peers
with whom they would like to interact in some activity, their choices
would be more likely to include actual relatively long-term friendships.
Lindzey and Byrne (1968) commented on the above-mentioned require-
ments as follows:
The requirements outlined above identify the sociometric
measure in a more or less pure form, and are generally
in agreement with Moreno's definition. However, rela-
tively few studies in this area meet all the requirements.
For example, the technique as used today seldom involves
the promise of restructuring the group .. One of the
more frequent modifications involves specifying the number
of choices the individual is required to make. (p. 455)
Thus, sociometric studies are not uniform in their methodology.
This poses a problem in assessing the reliability and validity of the
technique. However, since pure applications of sociometric testing
are the exception, rather than the rule, and since the variations of
the pure form seem to yield similar results; they will all be considered
together, and the reliability and validity of sociometric testing
(in its broad sense) will be said to apply to the procedure used in
this study. In fact, most of the studies that have assessed the
reliability and validity of sociometric tests have been variations
of the pure form.
Concerning the reliability of sociometric testing, a point made
by Gronlund (1959) needs to be considered:
perfect consistency from one test to another is neither
expected nor desirable, owing to the dynamic nature of
social relations. Revealing actual changes in social
relations is as important a requirement of the socio-
metric test as providing results that are constant
enough to have predictive value .. Thus, when
applied to sociometric testing, the various coefficients
of reliability refer to the consistency of choice behavior,
rather than to the characteristics of the test
itself. (p. 119)
Thus, the changes in sociometric results do not necessarily
reflect testing error; to a great extent they reflect actual changes
in the social perferences of the subjects. With this point in mind,
the internal consistency and the test-retest reliability of sociometric
tests will be discussed.
Grossman and Wrighter (1948) determined the internal consistency,
of their sociometric testing with four classrooms of sixth graders.
They randomly divided each group in half and then correlated the
sociometric status of each subject, as rated by one half of the class,
with his status in the other half of the class. They reported coef-
ficients of internal consistency from .93 to .97. Bass and White (1950)
and Ricciuti and French (1951), using college students as subjects,
reported internal reliability coefficients of .90. Ausubel, Schiff,
and Gasser (1952) reported coefficients from .54 to .86 for third,
fifth, and seventh graders and coefficients of .89 and .90 for 11th
and 12th graders.
Test-retest reliability has been determined by correlating the
sociometric status of subjects on one sociometric test with their
status on another test at a later date. Witryol and Thompson (1953)
and Thompson and Powell (1951) studied the consistency of sociometric
choices made by sixth graders at intervals of one week, four weeks,
and six weeks. Witryol and Thompson (1953) reported reliability coef-
ficients ranging from .60 to .90, and Thompson and Powell (1951)
reported correlations ranging from .89 to .92. In both studies, the
coefficients tended to decrease over time.
Using longer time intervals, Byrd (1951) found a reliability
coefficient of .89 with fourth graders selecting partners for a play
after a two-month interval, and Gronlund (1955) reported an average
reliability coefficient of .75 for fourth, fifth, and sixth graders
over a four-month interval. Bonney (1943) administered a sociometric
test, an IQ test, and an achievement test to a group of second graders
for four consecutive years. The correlations of the students socio-
metric status from one year to the next ranged from .67 to .84. The
sociometric status of the children in this study was as consistent as
their IQ and scores on the achievement tests.
Using high school students Northway (1947) reported coefficients
of .90 for a one-week interval and .60 for a one-year interval.
Jennings (1950) investigated the sociometric choices of adolescent
girls (12 to 16 years old) over time and found a correlation of .96
after four days and a correlation of .65 after eight months.
Thus,the results of sociometric testing are relatively consistent.
They are almost as reliable as typical intelligence and achievement
tests, as Bonney (1943) has demonstrated, and more reliable than most
As with reliability, the concept of validity, as it is typically
applied to testing and measurement, needs to be qualified in its
application to sociometric tests. If sociometric tests are supposed
to measure merely social choice, then they are by definition valid
(Pepinsky, 1949). However, if they are supposed to measure actual
social relationships, they will certainly fall short of this expecta-
tion, because, as Gronlund (1959) has mentioned:
An individual's actual associations are influenced by
environmental limitations, personal inhibitions, lack
of reciprocal feelings on the part of the desired
associates, and other related factors, as much as they
are by his preferences. Thus actual associations can
be expected to show some variation from the desired
associations indicated in sociometric choices. (p. 159)
Although sociometric choices do not correspond perfectly with
actual associations, there is a great deal of overlap. This overlap,
sociometric choices that are also actual associations, is considered
evidence of the validity of sociometric testing for this study;
because the first and third hypotheses of this study purport to
study friendships. Studies that have investigated the validity of
sociometric testing have related sociometric status and sociometric
choices with other measures of popularity and friendship choices.
Concerning sociometric status, the observations of teachers
and independent investigators correspond rather closely to the
results of sociometric tests. Bonney and Powell (1953) found that
sociometrically high first and second graders participated more
frequently in cooperative group activities and associated with more
children than did sociometrically low children. Newsletter, Feldstein,
and Newcomb (1938) found a correlation of .76 between the sociometric
status and the camp counselor ratings of popularity for 30 adolescent
boys. Gronlund (1951) had 40 sixth grade teachers rank their students
according to popularity; the average correlation between the teacher's
ranking and sociometric results was .60. Gronlund (1955, 1956) ob-
tained similar results in two other studies.
The individual choices of a subject are more complex and variable
(Gronlund, 1959) than sociometric status. Therefore, it is more diffi-
cult to observe and less likely to correspond closely with sociometric
results. Biehler (1954) compared the first sociometric choice of kinder-
garten children with their observed play companions. About 74 percent
of the chosen companions actually played with the children who chose
them. Gage, Leavitt, and Stone (1955) asked 103 fourth, fifth, and
sixth grade teachers to predict how each of their students would
respond to a sociometric item that asked for the students to list
five children in their room whom they would most prefer as classmates
if the class were divided into two groups. The average correlation
between the teacher's prediction and the sociometric results was .48.
Considering that observations are subject to error and that some
aspects of friendship can not be observed, the above-mentioned evidence
suggests that sociometric results correspond rather closely to actual
The Nowicki-Strickland Locus of Control Scale for Children (N-S)
Nowicki and Strickland (1973) developed the N-S (see Appendix A)
as the children's counterpart to Rotter's (1966) Internal-External
Locus of Control Scale; both scales were intended to be measures of
global locus of control. Nowicki and Strickland (1973) describe the
scale in the following manner:
The Nowicki-Strickland Locus of Control scale is a
paper-and-pencil measure consisting of 40 questions
that are answered either yes or no by placing a mark
next to the question. This form of the measure
derives from work which began with a large number
of items (N = 120), constructed on the basis of
Rotter's definition of the internal-external control
of reinforcement dimension. The items describe
reinforcement situations across interpersonal and
motivational areas such as affiliation, achievement,
and dependency. School teachers were consulted in
the construction of items. The goal was to make
the items readable at the fifth-grade level, yet
appropriate for older students. These items along
with Rotter's description of the locus of control
dimension were then given to a group of clinical
psychology staff members (N = 9), who were asked
to answer the items in an external direction. Items
on which there was not complete agreement among the
judges were dropped. This left 59 items, which made
up the preliminary form of the test. The 59-item
form of the test was then given to a sample of students
(N = 152) ranging from the third through eleventh
grades .. The results of further analysis as well
as comments from teachers and subjects in the sample
led to the present form of the test consisting of
40 items. (p. 151)
Split-half correlations correlated by the Spearman-Brown Formula
were .63 for third to fifth grades, .68 for sixth to eighth grades,
and .74 for ninth to eleventh grades; the sample size for each of these
three groups was about 300. Test-retest reliabilities with six-week
intervals was .63 for third graders, .66 for seventh graders, and .71
for tenth graders.
Nowicki and Strickland assessed the convergent validity of the
N-S by correlating it with two measures of locus of control--the
Intellectual Achievement Responsibility Questionnaire (Crandall et al.,
1965) and the Bialer-Cromwell Scale (Bialer, 1961). With 182 third
grade and 171 seventh grade blacks, the correlations between the Intel-
lectual Achievement Responsibility Questionnaire and the N-S were .31
and .51, respectively. With a sample of 29 children, ages 9-11, the
correlation between the Bialer-Cromwell Scale and the N-S was .41.
As evidence of the construct validity of the N-S, Nowicki and
Strickland (1973) cited significant correlations between academic
achievement and the N-S at varying age levels. They also demonstrated
that internality, as measured by the N-S, increases with age and offered
this trend as evidence of the validity of the instrument. Nowicki
(1973) has also demonstrated the construct validity of his instrument
by factor analyzing the test results of three different populations--
elementary school children, junior high school children, and high
school children. For the sample most relevant to this study (junior
high school subjects), there were three different factors. There
was one general factor that accounted for 38 percent of the variance
and was labeled "helplessness" (this factor was also present in the
other two samples). The other two factors accounted for 16ss variance,
8-19 percent; the second factor had to do with achievement, especially
through persistence and planning, and the third factor contained items
having to do with luck, persistence, and success in social areas.
Thus, it appears that part of the third factor (success in social
areas) is most relevant in this study. However, since it is only part
of one factor, contains only a few items, and accounts for so little
variance, it has not been used to predict sociometric status in this
Although the N-S is a rather new instrument and lacks the valida-
tion research of a more established instrument, the sparce amount of
evidence that is available suggests that it is about as reliable and
valid as most attitude/personality measures used in research.
MacDonald (1973) has evaluated most of the locus of control instru-
ments and has noted:
This test has been developed carefully by researchers
of solid reputation. Though of recent construction,
it has been used in many studies. Results presently
available indicate the scale to have adequate internal
consistency and temporal consistency. Data relevant
to divergent and convergent validity are encouraging.
In short, it appears to be the best measure of locus
of control as a generalized expectancy presently
available for children. (p. 185)
First and second hypotheses. In both of these hypotheses the dependent
variable was sociometric status (SMS) on the respective sociometric
item. Sociometric status was measured by counting the number of
selections that each subject received. Previous sociometric research
indicates that this distribution should be skewed to the right--fewer
popular children than unpopular ones (Groulund, 1959).
The principal independent variable was locus of control (LOC),
measured by the N-S. It was expected that this distribution would
be close to normal or skewed slightly to the right; there may be more
extreme internals than extreme externals (Nowicki & Strickland, 1973).
The variables of sex and race were forced into the equation before
LOC, as covariates, since sex and race may be confounding variables.
Thus the basic multiple regression equation was as follows:
SMS = sex + race + LOC
In addition to these basic variables, locus of control as a
nonlinear variable (LOC2) and interactions between locus of control
and sex and race were also checked for significance; so the complete
equation was as follows:
SMS = sex + race + LOC + LOC + (LOC x sex) + (LOC x race)
All of the independent variables were forced into the equation
in the order indicated above. The rationale for this approach has
been discussed under the rubrics of the "a priori ordering approach"
by Kerlinger & Pedhazur (1973).
Dummy coding (Kerlinger & Pedhazur, 1973) was used to code sex
and race. The data were submitted on cards to the University of
Florida's IBM 360-75 computer, assessing the multiple regression
program of the Statistical Package for the Social Sciences (SPPS)
(Nie, Hull, Steinbrenner & Bent, 1973). (In fact, all of the data
analysis of this study used programs from the SPSS library). If
the F-value was significant at the .05 level for the LOC variable,
the hypothesis'was accepted.
One of the assumptions of multiple regression is that the depen-
dent variable should be normal. However, Kerlinger and Pedhazur,
(1973) has noted that this is not always a problem:
It has convincingly been shown that the F and t tests
are "strong" or "robust" statistics, which means that
they resist violation of the assumptions. In general,
it is safe to say that we can ordinarily go ahead
with analysis of variance and multiple regression
analysis without worrying too much about assumptions.
Nevertheless, researchers must be aware that serious
violations of the assumptions, and especially of
combinations of them, can destroy results. We advise
students to examine data, especially by plotting,
and, if the assumptions appear to be violated, to
treat obtained results with even more caution than
usual. The student should also bear in mind the
possibility of transforming recalcitrant data,
using one or more of the transformations that are
available and that may make the data more amendable
to analysis and inference.
If the dependent variables differ markedly from the normal
distributions, they will be transformed, as Kerlinger has advised.
Third hypothesis. Multiple regression was also used in the analysis
of the third hypothesis. For this equation the locus of control of
the friends chosen by the subject (F-LOC) was the dependent variable.
It is expected that the distribution of F-LOC scores should be close
to normal, so none of the assumptions of multiple regression should be
violated in the analysis of the third hypothesis.
With F-LOC scores as the dependent variable, the principal inde-
pendent variable was the subject's own locus of control score. If
there is a significant positive correlation between the two, it would
indicate that these subjects tend to choose friends of a similar locus
of control orientation. A negative correlation would indicate a
tendency to choose friends of the opposite locus of control.
As with the first hypothesis, sex and race were forced into
the equation before the principal independent variable (the subject's
own LOC score). Also, interactions and a nonlinear relationship be-
tween LOC and F-LOC were investigated. Thus, the complete multiple
regression equation was as follows:
F-LOC = sex + race + LOC + LOC2 + (LOC x sex) + (LOC x race)
Two hundred eighth graders answered two sociometric questions:
one that asked the students to list their best friends, and another
that asked the student to name three classmates with whom they would
like to work an an academic project. One week later the subjects
completed the Nowicki-Strickland Locus of Control Scale for Children
The reliability and validity of sociometric testing and the N-S
were discussed; both testing procedures were as reliable and valid
as most psychological tests used in research. The subject's locus of
control (LOC) was the principal independent variable for all three
hypotheses. LOC and other variables were used in the multiple regres-
sion equations to predict sociometric status on the best friends item
(SMSbf), sociometric status on the work-project item (SMS ), and
the subject's friends locus of control (F-LOC). The complete multiple
regression equations were as follows:
1. First hypothesis: SMSbf = sex + race + LOC + LOC2
+ (LOC x sex) + (LOC x race)
2. Second hypothesis: SMS = sex + race + LOC + LOC2
+ (LOC x sex) + (LOC x race)
3. Third hypothesis: F-LOC = sex + race + LOC + LOC2
+ (LOC x sex) + (LOC x race)
The results of this analysis are discussed in the next chapter.
RESULTS AND DISCUSSION
In this chapter the results of this study are stated and discussed.
The first section presents the results, and in the next section the
significance of these results is discussed.
The results of this study are presented in sub-sections. The
first sub-section is a general presentation of the data which includes
the mean, standard deviation, and other aspects of the variables used
in the study. The remaining three sub-sections present the results
of the analysis of each hypothesis. These three sub-sections focus
on the inferential statistics of each hypothesis, and the first sub-
section focuses on the descriptive statistics of all of the variables.
Descriptive statistics. Two hundred subjects completed the Nowicki-
Strickland Locus of Control Scale for Children (N-S). Scores on the
N-S can range from 0 to 40--the higher the score, the more external.
The mean score of this sample was 15.99; the standard deviation was
5.54. The scores ranged from 3 to 30. Measures of the skewness and
kurtosis were taken and found to be -.22 and .20, respectively,
indicating that the distribution of N-S scores was close to normal.
The sample included 112 females and 88 males, 161 whites and
39 blacks. The female subjects had a mean N-S score of 15.21, and
the males, 16.96. (The lower the score, the more internal.) The
mean score for whites was 15.32, and for blacks it was 18.72. The
sex/race differences are presented in Table 4-1.
Table 4-1. Sex/race differences in locus of control scores
Subgroup N Mean Standard Deviation
White males 69 16.18 5.47
Black males 19 19.79 4.93
All males 88 16.96 5.31
White females 92 14.67 5.39
Black females 20 17.70 5.15
All females 112 15.21 5.32
All whites 161 15.32 5.42
All blacks 39 18.72 5.04
Sociometric status was measured by counting the number of
selections received. For the first hypothesis sociometric status on
the best friends item (SMSbf) was the dependent variable. The mean
number of selections received was 2.38; the standard deviation was
1.92. The most popular subjects (N = 2) received nine selections;
that is, nine of their classmates chose them as best friends.
Thirty-three subjects received no selections. Thus, SMSbf ranged
from zero to nine. The measures of skewness and kurtosis were
.97 and 1.03, respectively.
For the second hypothesis sociometric status on the work-project
item (SMS ) was the dependent variable. The mean number of selec-
tions received was 2.49; the standard deviation was 2.41. On this
item the most popular subject received 13 selections, and 41 subjects
received no choices. The measures of skewness and kurtosis were
1.53 and 3.14 for this variable. Like SMSbf, SMSwp was not signifi-
cantly related to either sex or race.
The indices of skewness and kurtosis indicated that SMSbf and
SMS p deviated considerably from a normal distribution (Nie et al.,
1975). Therefore, both distributions were transformed to T-scores,
a procedure commonly used to normalize recalcitrant data (Roscoe,
1969). The transformed variables which were used in the multiple
regression equations were labeled SMbf and SMwp. The correlation
between SMbf and SMp was .675.
Relevant to the third hypothesis, the 200 subjects of this study
made 413 usable choices on the best friends item (some subjects
chose friends who did not take the N-S, and these choices could not
be used). Of the 413 selections, 386 were friends of the same sex
and race of the chooser; 25 friends of the opposite sex were chosen,
and only two friends of another race were chosen.
The dependent variable in the third hypothesis was the locus
of control of the friends chosen by the subjects (F-LOC). Since
each sociometric choice, not each subject, was the individual unit
of analysis for this hypothesis, a subject's locus of control was
used as many times as he was chosen as a friend. The mean F-LOC
was 15.19 with a standard deviation of 6.08. Friends'locus of con-
trol ranged from 3 to 30.
As previously mentioned, the N-S test results of this study
were factor analyzed in the hope of finding a social factor. How-
ever, the factor analysis indicated that there was no substantial
The descriptive data have been presented above; the next three
sections present the results of the hypothesis testing.
First hypothesis. The first hypothesis postulated a significant
relationship between the subject's locus of control (LOC) and their
sociometric status on the best friends item (SMbf), such that inter-
nals would be more popular than externals. (As mentioned in the pre-
vious section, SMS was transformed to T-scores and labeled SM.) It
was hypothesized that the relationship between LOC and SMbf would be
significant with sex and race partialed out. Also, interactions
between locus of control and sex (LOC x sex) and locus of control
and race (LOC x race) and a curvilinear locus of control term (LOC2)
were checked for significance. The results of the multiple regression
equation used to predict SMbf are presented in Table 4-2.
00 LO "c
) 5- cO
U.O N._ N.
'-j N. C'.j
(a C S- 4-) 4-) 4-3
' M -;I- Lo ..
l r O o
X 0 -
The zero-order correlation between LOC and SMbf was -.147 which
is in the predicted direction and statistically significant (p <.01).
Sex and race were not significantly related to SMbf. The F-ratio
of LOC, as a predictor of SMbf, with sex and race partialed out was
5.41 (df = 3, 196) which is significant at the .01 level. However,
LOC accounted for less than three percent of the variance in SMbf.
With sex, race, and LOC entered into the multiple regression equation,
none of the other variables [(LOC x sex), (LOC x race), (LOC2)] was
a significant predictor of SMbf.
Second hypothesis. The second hypothesis predicted a significant
relationship between locus of control (LOC) and sociometric status
on the work-project item (SM ), when sex and race were partialed
out. It was also predicted that the relationship between LOC and
SMwp would be greater than the relationship between LOC and SMbf.
As in the first hypothesis, (LOC x sex), (LOC x race), and LOC2
were tested for significance. The results of the multiple regression
are presented in Table 4-3.
The zero-order correlation between LOC and SMwp was -.302 which
is in the predicted direction and significant at the .0001 level.
A formula recommended by Roscoe (1969) was used to determine the
statistical significance, if any, of the difference between this
correlation (r = .-302) and the correlation between LOC and SMbf
(-.147). These two correlations were not significantly
4-' .,- 4-)- -t
4-) 3 4-> co CMj Cr Y) LO a)
,o 0 ( C C\J C\j L0 -I C\J
ul (\II C. )\ C\_ C\J
4-) L) L- C.
"' O r- C) C) C)O
O "-)r- ) 0"
S X U
4- 3) m aI S
.0 S- 0 to
o ri ro(0.- c E ( V) s s S-
I > ,, 4-' a)
CL X X
-- a a) a)
4- a) 0
0 -" .r-
u ) C) m C)
a a o o) o o CD
1o- o- o
o I u
S4- .0 00 N
S- 0 0 I-
-' w-0) (n S- 4-3 4- 4-)
CD M Z O
.1 a) a) to
(U rc x x
S x u C-
Sa) ro C C) C)
F- II) -1 J 1J -J _
different from each other at the .05 level, but the probability that
they were not different was less than .10.
Sex and race were the first variables in the multiple regression
equation, but neither produced a significant F-ratio. With sex
and race partialed out, LOC produced a F-ratio of 18.23 (df = 3, 196)
which was significant at the .01 level. LOC, with sex and race con-
trolled for, accounted for about eight percent of the variance in SM .
With sex, race, and locus of control entered into the multiple regres-
sion equation, (LOC x sex), (LOC x race), and LOC2 were not statisti-
Third hypothesis. The third hypothesis postulated a significant rela-
tionship between a subject's locus of control (LOC) and the locus of
control of his friends (F-LOC), even when sex and race were accounted
for. Interactions between sex and locus of control (LOC x sex) and
race and locus of control (LOC x race) and a curvilinear locus of con-
trol term (LOC2) were also'checked for significance. The results of
the multiple regression equation used to predict F-LOC are presented
in Table 4-4.
The zero-order correlation between the subject's locus of control
(LOC) and the locus of control of their friends (F-LOC) was .228 which
was significant (p <.00001). With sex and race partialed out, LOC
produced an F-ratio of 7.15 (df = 3, 409) which was a significant (p <
.01) predictor of F-LOC, but it accounted for less than two percent
of the variance in F-LOC.
a0 C\M LO 00 I CM.
- ) \j r c) CMi CD
L3- r-- N i
() )D s-
0 'a a)
C S- X U
Cv') (\j C)
C L O) "-
Ln i- o <^
4-' -o -o _cI c 4-
CA C S.- 4-) 4-J J
u X 0 0 0 0
S/ __ _j _|
M 0 J0 J
With sex, race, and LOC partialed out, both interactions (LOC
x sex and LOC x race) added small, but statistically significant
(p < .01, p < .05, respectively), contributions to the equation.
An ex post facto analysis indicated that the interactions were such
that LOC was significantly related to F-LOC for females (r = .31,
p <.000001), but not for males (r = .03), and that LOC was signifi-
cantly related to F-LOC for whites (r = .20, p < .00011), but not
for blacks (r = .06).
With sex, race, and LOC partialed out, the curvilinear term
(LOC2) added a small, but statistically significant (p < .05),
contribution to the equation. The chi square matrix illustrated in
Table 4-5 suggests that the nature of the curvilinear relationship
between LOC and F-LOC is such that the similarity hypothesis is
valid for internals, but not for externals. Therefore, the multiple
regression equation was run separately for internals and externals.
Table 4-5.--Chi square matrix indicating the locus of control of
friends chosen by internals, moderates, and externals
Friends' locus of control
Internals Moderates Externals
(N = 154) 83 27 44
(N = 128) 47 40 41
(N = 131) 42 42 47
Chi square = 17.33, df = 4, p <.002.
Table 4-6 compares the ability of LOC to predict F-LOC in three
samples--the total sample, internals, and externals. For externals,
neither their sex nor their locus of control was significantly
related to the locus of control of their friends. For internals,
LOC was significantly related to F-LOC (r = .40, p < .00001); but,
with sex and race partialed out while LOC was statistically signifi-
cant (p < .01), it accounted for only 2.2 percent of the variance
Table 4-6.--Simple correlation and unique contribution of LOC in
predicting F-LOC for total sample, internals, and
Simple r Unique contribution F-ratio
Sample (LOC/F-LOC) of LOC with sex and race of unique
accounted for contribution
(N = 413) .23 .015 7.47**
(N = 154) .26 .022 3.89*
.(N = 131) .05 .000 .032
p < .05.
**p < .01.
Thus,locus of control was a statistically significant predictor
of SMbf, SM p, and F-LOC. The theoretical and practical significance
of these results are presented in the next section.
Discussion of the Results
The theoretical and educational significance of each hypothesis
is discussed in the following section. Also, some recommendations
for future research and theory development are discussed.
First hypothesis. Although the relationship between locus of control
and sociometric status on the best friends item was statistically
significant, even with sex and race controlled, the relationship was
so slight that it lacked practical significance. In fact, for all
practical purposes, this study indicated that internals and externals
are about equal in social popularity; this supports Nowicki and Round-
tree (1971) and Adinolfi (1970) who found no relationship between
locus of control and sociometric status on purely social items.
This researcher expected that locus of control would be a more
potent predictor of sociometric status. Its weakness as a predictor
may be due to any one or a combination of the following explanations:
1. Externals may value friendship more than internals.
According to social learning theory (Rotter, 1954), behavior is a
function of both an individual's expectancy about his behavior
obtaining reinforcement and the value the individual places on the
reinforcement in question. Therefore, a low expectancy could be
balanced out by a highly valued reinforcement. For example, a starving
person might strive desperately for food, even when he has a low
expectancy of obtaining food.
2. There may be a specific sense of control related to social
situations that is not highly correlated with generalized locus of
3. Locus of control may be related to goal-seeking behavior,
but not to the more spontaneous behavior of making and maintaining
The first explanation (externals may value friendship more than
internals) may have some merit. Therefore, future studies that in-
vestigate the relationship between locus of control and peer acceptance
should include a measure of the need for affiliation.
The second explanation (there is a specific sense of control
related to social situations) may have some merit. Even though studies
that have factor analyzed locus of control scales have not yielded a
purely social factor, it may be that the items of these scales do not
tap such a dimension.
The third explanation' (locus of control is related only to goal-
seeking behavior) is closely related to the previous explanation. Locus
of control has been more clearly related to goal-directed activities,
like academic work and problem solving. These activities are usually
step-by-step procedures that are aided by planning and directed at
specific, known goals. These activities also rely heavily on cognitive
ability. On the other hand, an adolescent's social world is not a
step-by-step process. Specific goals are usually not formulated; and
if they are formulated and adhered to in a rigid manner, the goal-seeker
is almost guaranteed social failure. Also, cognitive skill is less
important in social situations than it is in school and similar
activities. It seems that locus of control is most clearly associated
with methodical, cognitive effort; and this type of effort may not
be as helpful in formal social situations as it is in formal academic-
The processes of adolescent friendship need further explanation.
It would be helpful to know more about how and why adolescents associate
with each other. It would also be helpful to understand how adolescents
(and others) perceive their relationships with their friends. For
example, do people believe they can control their friends? If they do,
what do they mean by "control?" What effects, if any, do different
notions of control have on social relationships? It seems as though
present locus of control instrumentation is too restrictive for such
an exploration. Psychologists should begin with in-depth interviews,
field observation, and projective testing. Open-ended procedures may
produce data that can be used to formulate a concept of man's percep-
tion of his interaction with his environment that is more sophisticated
and more complex than Rotter's concept of locus of control.
Concerning the educational significance of this hypothesis, it
appears that internality is not clearly associated with popularity
or the lack of popularity. Therefore, educators (probably) need not
be concerned with possible social disadvantages of internality, at
least not on an individual basis. (Increasing internality may have
some disadvantages to the group; for example, is a team integrated
with internals and externals more productive and harmonious than a
team that is exclusively internal or external? In other words, this
study produced no evidence suggesting a child may lose peer acceptance
in becoming more internal. Therefore, educators can proceed with
more confidence in advocating internality change programs.
In summary, future studies that investigate the relationship
between locus of control and peer acceptance should use a measure of
the need for friendship, since social learning theory predicts that
the value of peer acceptance should interact with locus of control
and peer acceptance. Also, as described above, a more complex,
socially sensitive concept of sense of control should be used. Although
there may be no social disadvantages associated with internality on
an individual basis, the effects of locus of control on group harmony
and productivity should be investigated.
Second hypothesis. Locus of control was an effective predictor of
sociometric status on the academic project item. This finding suggests
that the subjects chose peers who did well in school, and those students
tended to be internal. This evidence adds support to the notion that
internality is associated with academic success.
This finding is closely related to the results of the Nowicki and
Roundtree (1971) study; in their study internals were not more popular
than externals on the best friends item, but internal males were more
popular on a sociometric item that asked for nominations for class
president. Thus,internals were given more consideration for tasks
requiring intelligence and responsibility.
Future studies that relate locus of control to popularity on a
task-oriented item should use appitude in that task as a covariate.
Third hypothesis. Like the first hypothesis, this hypothesis was
statistically significant but lacked practical significance due to the
small size of the relationship between the dependent and independent
variables. The data suggest that, given a biracial population of
eighth grade boys and girls, there is a tendency for these students,
especially the internals, to choose friends of a similar locus of
control. However, most of this tendency is due to racial and sexual
divisions within the classroom and racial and sexual differences in
locus of control. In this sample the students chose, almost exclu-
sively, friends of the same race and sex. Since black males, black
females, white males, and white females all differed in locus of con-
trol, there was a substantial correlation between a subject's locus of
control and his friends' locus of control on the basis of sex and race.
When sex and race were partialed out, there was only a slight tendency
for subjects to choose friends of a similar locus of control. This
tendency was especially evident for three overlapping subgroups within
the total sample--internals, females, and whites. The results of this
hypothesis parallel the results obtained by Nowicki and Blumberg (1975)
and Phares and Wilson (1971) in that internals showed a preference
for fellow internals and externals did not prefer other externals.
It appears that in real-life situations similarity in locus of
control is not an overriding factor in interpersonal attraction.
Perhaps a more sophisticated concept of control, one more applicable
to social situations than Rotter's (1966), would be a more significant
factor in interpersonal attraction. However, even a more sophisti-
cated model of sense of control would be limited by the similarity
hypothesis itself. This researcher does not recommend any further
research on similarity in locus of control as a factor in interper-
sonal attraction, because Phares and Wilson (1971), Nowicki and
Blumberg (1975), and-this study all produced the same finding--inter-
nals have a slight tendency to prefer fellow internals, and externals
do not prefer other externals.
There is a serendipitous finding of this study that deserves
mention. Only two of the 200 subjects chose a classmate of another
race as one of their "best friends"; a white girl and a black girl
chose each other. Considering that this study was conducted in a
school that has been desegregated, this finding suggests an important
question. How is desegregation to help black children assimilate
work-oriented attitudes, like an internal locus of control orientation,
if blacks and whites remain socially separate? Of course, this ques-
tion is based on the assumption that internality can be learned through
modeling in informal, social situations; the investigation of this
assumption would be an interesting study in itself.
In summary, the data indicated that (1) the internal subjects
were slightly more popular, as friends, than the externals, (2) the
internals were considerably more popular as working partners, and
(3) the subjects, especially the internals, had a tendency to choose
friends of a similar locus of control, but most of this tendency was
due to racial and sexual preferences and racial and sexual differences
in locus of control.
In this chapter the results have been presented and discussed.
In the next chapter the study is summarized and the conclusions are
SUMMARY AND CONCLUSIONS
The final chapter consists of two sections. The first section
is a summary of the purpose, rationale, methodology, and results
of the study. The second section states the conclusions, including
recommendations for future research.
This study investigated the relationship between adolescents'
locus of control and their peer preferences. Locus of control was
defined as the degree to which an individual expects that his rein-
forcement is contingent upon his behavior; individuals with an
internal locus of control generally believe that they control what
happens to them, whereas those with an external locus of control
believe that their destiny is beyond their control and is determined
by fate, luck, or powerful others.
There has been a great deal of research indicating that inter-
nality is associated with academic success and cognitive superiority,
but there has been relatively little investigation of the social
correlates of locus of control. This study related locus of control
to popularity as a friend, popularity as a working partner, and
interpersonal attraction. More specifically, the following
hypotheses were tested:
1. Internal adolescents are more popular (have more friends)
than their external peers.
2. Internals are more popular as working partners on an
academic project than they are as friends.
3. Adolescents tend to choose as friends those who are
similar to themselves in locus of control.
To investigate these hypotheses, 200 eighth graders were
administered the Nowicki-Strickland Locus of Control Scale for
Children and two sociometric items. The first sociometric item
was "Name the students who are your best friends"; and the second
item was "If you had to work on a group project in English with
three other students who would you select to work with?"
Multiple regression was used to analyze each hypothesis. For
the first hypothesis the dependent variable was sociometric status on
the best friends item (SMSbf); that is, the number of times the subject
was chosen as a best friend. The principal independent variable was
the subject's locus of control (LOC); thus locus of control was used
to predict popularity. Sex and race were controlled by forcing these
two variables into the multiple regression equation before locus of
control. Also, interactions between locus of control and sex (LOC x
sex) and locus of control and race (LOC x race) and curvilinear locus
of control term (LOC2) were checked for significance. Thus, the
complete multiple regression equation was as follows:
SMSbf = sex + race + LOC + (LOC x sex) + (LOC x race) + LOC2
The predictors for the other two hypotheses were the same as
above; the dependent variable for the second hypothesis was socio-
metric status on the work-project item, and for the third hypothesis
it was the subjects' friends' locus of control. Thus,the subjects'
locus of control was used to predict their social popularity,
popularity as a working partner, and their friends' locus of control.
In each of these multiple regression equations, with sex and
race partialed out, locus of control was a statistically significant
(p < .01) predictor of the dependent variable. However, locus of
control, with sex and race partialed out, accounted for less than
three percent of the variance in social popularity and less than
two percent of the variance in friends' locus of control. Locus
of control accounted for about eight percent of the variance in socio-
metric status on the work-project item. The following table sum-
marizes the efficacy of locus of control in predicting sociometric
status on the best friends item (SMSbf), sociometric status on the
academic work-project item (SMS ), friends' locus of control (F-LOC).
The conclusions of this study are stated in the next section.
Table 5-1.--Locus of control as a predictor of popularity as a
friend and work partner and friends' locus of control
With variables partialed out
Dependent Hpo i Zero-order
variable Hypothesis correlation Variance F-ratio
SMSbp 1st -.147 2.7% 5.41*
SMSwp 2nd -.302 8.4% 18.23*
F-LOC 3rd .228 1.5% 7.15*
*p < .01
The results of this study indicate the following:
1. Internal adolescents are slightly more popular than exter-
nals as best friends.
2. Internals are more popular as partners on an academic project
than they are as best friends.
3. In a biracial group of eighth grade boys and girls almost
all students choose best friends of the same sex and race.
4. Using the Nowicki-Strickland Locus of Control Scale to
measure the locus of control of eighth graders, white females are
more internal than white males who are more internal than black
females, and black males are the most external.
5. In a biracial population of eighth grade boys and girls
there is a significant tendency for these adolescents to choose
best friends of a similar locus of control, but most of this tendency
is due to sexual and racial preferences and sexual and racial dif-
ferences in locus of control.
6. The tendency to choose friends of a similar locus of control
was most evident for whites, females, and internals.
Based on the results of this study, the following suggestions
were recommended for future research in the social aspects of locus
1. Covariates like sex, race, socioeconomic status, and IQ
should be used in locus of control research. Perhaps some of the
variables that have been related to locus of control were overesti-
mated because of a third variable that was related to both locus of
control and the variable being investigated.
2. Interactions between locus of control and variables like
sex, race, socioeconomic status, and IQ should be examined. Perhaps
some significant relationships with locus of control have been over-
looked, because the investigator did not check for interactions.
3. When relating locus of control to peer acceptance, the need
for approval should be considered in that social learning theory
predicts that the need for approval should interact with locus of
control and peer acceptance.
4. Also, curvilinear relationships between locus of control
and relevant variables should be explored. Perhaps, as it was in
this study, the variable has a different relationship to internal
locus of control than to external locus of control.
5. There is a possibility that informal, social segregation
within desegregated schools is preventing blacks from acquiring
desirable middle class attitudes; this possibility needs to be
6. The effects of locus of control on group harmony and pro-
ductivity should be investigated. For example, a study could compare
the productivity and harmony of a group integrated with internals and
externals to groups that were exclusively internal or external.
7. The possibility that locus of control has special signifi-
cance for goal-directed behavior (like academic work), rather than
spontaneous behavior (like making friends), deserves investigation
and consideration in theory construction.
8. A theory more sophisticated than Rotter's concept of locus
of control needs to be developed to explain man's perception of his
interaction with his environment (especially his social environment).
This theory development should begin with a series of open-ended
These last two recommendations, especially the last one, suggest
major changes in locus of control theory. Although it will be
difficult to abandon the standardized instruments, the linear model,
and traditional psychological methodology; it will be necessary if
locus of control is to help psychologists make insightful discoveries
about the nature of man.
THE NOWICKI-STRICKLAND LOCUS OF CONTROL SCALE FOR CHILDREN
INSTRUCTIONS: There are 40 questions to this survey. Please answer
them by checking (/) the YES or NO column for each question. There
are no right or wrong answers--your honest opinion is what matters.
1. Do you believe that most problems will solve them-
selves if you just don't fool with them?
2. Do you believe that you can stop yourself from
catching a cold?
3. Are some kids just born lucky?
4. Most of the time do you feel that getting good grades
means a great deal to you?
5. Are you often blamed for things that just aren't your
6. Do you believe that if somebody studies hard enough
he or she can pass any subject?
7. Do you feel that most of the time it doesn't pay to
try hard because things never turn out right anyway?
8. Do you feel that if things start out well in the
morning that it's going to be a good day no matter
what you do?
9. Do you feel that most of the time parents listen to
what their children have to say?
Do you believe that wishing can make good things
When you get punished does it usually seem it's for
no good reason at all?
Most ,of the time do you find it hard to change a
friend's (mind) opinion?
Do you think that cheering more than luck helps a
team to win?
Do you feel that it's nearly impossible to change
your parent's mind about anything?
Do you believe that your parents should allow you to
make most of your own decisions?
Do you feel that when you do something wrong there's
very little you can do to make it right?
Do you believe that most kids are just born good
Are most of the other kids your age stronger than
Do you feel that one of the best ways to handle most
problems is just not think about them?
Do you feel that you have a lot of choice in deciding
who your friends are?
If you find a four leaf clover, do you believe that
it might bring you good luck?
Do you often feel that whether you do your homework
has much to do with what kind of grades you get?
Do you feel that when a kid your age decides to hit
you, there's little you can do to stop him or her?
Have you ever had a good luck charm?
25. Do you believe that whether or not people like you
depends on how you act?
26. Will your parents usually help you if you ask them
27. Have you felt that when people were mean to you it
was uaually for no reason at all?
28. Most of the time, do you feel that you can change
what might happen tomorrow by what you do today?
29. Do you believe that when bad things are going to
happen they just are going to happen no matter what
you try to do to stop them?
30. Do you think that kids can get their own way if they
just keep trying?
31. Most of the time do you find it useless to try to
get your own way at home?
32. Do you feel that when good things happen they happen
because of hard work?
33. Do you feel that when somebody your age wants to be
your enemy there's little you can do to change mat-
34. Do you feel that it's easy to get friends to do what
you want them to?
35. Do you usually feel that you have little to say about
what you get to eat at home?
36. Do you feel that when someone doesn't like you there's
little you can do about it?
37. Do you usually feel that it's almost useless to try
in school because most other children are just plain
smarter than you are?
38. Are you the kind of person who believes that planning
ahead makes things turn out better?
39. Most of the time, do you feel that you have little
to say about what your family decides to do?
40. Do you think it's better to be smart than to be
THE SOCIOMETRIC QUESTIONNAIRE*
A. Most classrooms have a few students who joke a lot and who make
others in the room laugh. These are the "Class Clowns." Please
list below the names (first and last) of the students you know
who clown around a lot. Students should be in yourlgrade.
Name the students who are your best friends.
C. Name the students who usually come up with the best ideas for
class projects or activities.
D. If you had to work on a group project in English with three other
students, and half of your grade depended upon this project, which
three students would you select to work with?
E. List the students that you think are liked by most everyone
in the class.
Damico and Purkey, 1976.
Questions used in this study.
Abrahamson, D., Schulderman, S., & Schulderman, E. Replication
of dimensions of locus of control. Journal of Consulting
and Clinical Psychology, 1973, 41, 320.
Abramowitz, S. I. Locus of control and self-reported depression
among college students. Psychological Reports, 1969, 25,
Adinolfi, A. A. Characteristics of highly accepted, highly rejected,
and relatively unknown university freshmen. Journal of Counseling
Psychology, 1970, 17, 456-464.
Ashton, P. T. The effect of internal control training on selected
personality and ability factors. Unpublished doctoral disser-
tation, University of Georgia, 1974.
Austin, M. C., & Thompson, G. G. Children's friendships: a study of
the bases on which children select and reject their best friends.
Journal of Educational Psychology, 1948, 37, 101-116.
Ausubel, D. P.s Schiff, H. M. & Gasser, E. B. A preliminary study
of developmental trends in sociempathy: accuracy of perception
of own and other's sociometric status. Child Development, 1952,
Baron, D. Personal-social characteristics and classroom social status:
a sociometric study of fifth- and sixth-grade girls. Sociometry,
1951, 14, 32-42.
Bartel, N. R. Locus of control and achievement in middle- and lower-
class children. Child Development, 1971, 42, 1099-1107.
Bass, B. M. & White, 0. L. Validity of leaderless group discussion
observer's descriptive and evaluative ratings for the assessment
of personality and leadership status. American Psychologist,
1950, 5, 311-312.
Bernhardson, C. S. Relationship between individual social desirability
ratings and hand acquiescense and I-E scores. Psychological
Reports, 1968, 22, 773-776.
Bialer, I. Conceptualization of success and failure in mentally
retarded and normal children. Journal of Personality, 1961,
Biehler, R. F. Companion choice behavior in the kindergarten.
Child Development, 1954, 25, 45-50.
Bonney, M. E. Personality traits of socially successful and
socially unsuccessful children. Journal of Educational
Psychology, 1943, 34, 449-472.
Bonney, M. E. A sociometric study of the relationship of some
factors to mutual friendships of the elementary, secondary,
and college levels. Sociometry, 1945, 9, 21-47.
Bonney, M. E., & Powell, J. Difference in social behavior
between sociometrically high and sociometrically low children.
Journal of Educational Research, 1953, 46, 481-495.
Brown, D. Factors affecting social acceptance of high school
students. School Review, 1954, 42, 151-155.
Brown, J. C., & Strickland, B. R. 'Belief in internal-external con-
trol of reinforcement and participation in college activities.
Journal of Counseling and Clinical Psychology, 1972, 38, 148.
Brownstone, J. E., & Willis, R. H. Conformity in early and late
adolescence. Developmental Psychology, 1971, 4, 334-337.
Butterfield, E. C. Locus of Control, test anxiety, reaction to
frustration, and achievement attitudes. Journal of Personality,
1964, 32, 298-311.
Byrd, E. A study of validity and constancy of choices in a socio-
metric test. Sociometry, 1951, 14, 175-181.
Byrne, D. Interpersonal attraction and attitude similarity.
Journal of abnormal and social psychology, 1961, 62, 713-715.
Byrne, D. The attraction paradigm. 1971, New York: Academic.
Byrne, D., & Griffitt, W. Similarity versus liking: a clarifica-
tion. Psychonomic Science, 1966, 6, 295-296.
Byrne, D., & Griffitt, W. Interpersonal attraction. Annual Review
of Psychology, 1973, 24, 317-376.
Campbell, D. T., & Stanley, J. C. Experimental and Quasi-experimental
Designs for Research. Chicago: Rand McNally, 1966.
Cole, H., & Hall, I. N. Psychology of Adolescence (7th ed.). New
York: Holt, Rinehart & Winston, 1970.
Coleman, J. S., Campbell, E. Q., Hobson, C. J., McPartland, J., Mood,
A. M., Weinfeld, F. D., & York, R. L. Equality of Educational
Opportunity. Superintendent of Documents Catalog No. FS 5.238:
38001, U. S. Government Printing Office, Washington, D. C., 1966.
Cone, J. D. Locus of control and social desirability. Journal of
Consulting and Clinical Psychology, 1971, 36, 499.
Cook, L. A. An experimental study of a stratified tenth grade
class. American Sociological Review, 1945, 10, 250-261.
Costanzo, P. R., & Shaw, M. E. Conformity as a function of age level.
Child Development, 1966, 37, 967-975.
Crandall, V. C., Katkovsky, W., & Crandall, V. J. Children's beliefs
in their own control of reinforcement in intellectual-academic
situations. Child Development, 1965, 36, 91-109.
Cromwell, R. L., Rosenthal, D., Shakow, D., & Zahn, T. P. Reaction
time, locus of control, choice behavior, and descriptions of
parental behavior in schizophrenic and normal subjects. Journal
of Personality, 1961, 29, 363-379.
Crowne, D. P., & Liverant, S. Conformity under varying conditions
of personal commitment. Journal of Abnormal and Social Psychol-
ogy, 1963, 66, 547-555.
Crowne, D. P., & Marlow, D. The approval motive. New York: Wiley,
Damico, S., & Purkey, W. W. The class clown phenomenon among middle
school children. A paper presented at the annual meeting of
the American Educational Research Association, San Francisco,
Davis, W. L., & Phares, E. J. Internal-external control as a deter-
minant of information-seeking in a social influence situation.
Journal of Personality, 1967, 35, 547-561.
deCharms, R. Personal causation training in the schools. Journal
of Applied Social Psychology, 1972, 2, 95-113.