PERSONALITY DETERMINANTS IN
ATTITUDES TOWARD DISABILITY
JOHN ROBERT NOONAN, JR.
A DIssERTATION I'RESENTED TO THE CRF.,rLiA lT COUNCIL 01
THE UNIVERSITY OF I LORDD,
IN PA.JRTL4L FULFLLLNMEN'T 01 THE BEQi
DEGREE 01 DOCTOR 01 PHILOOPHV
UNIVERSITY OF FLORIDA
The author wishes to express gratitude to Dr. Hugh C. Davis,
the chairman of his supervisory committee, who provided reliable
clarification and direction for the writer throughout the study.
To Dr. John R. Barry, original chairman of the committee, gratitude
is expressed for aiding the writer in his initial formulation of
the problem and for his constructive criticism in the planning of
the study. Dr. Henry S. Pennypacker, Dr. George H. Dunteman, Dr.
John T. Stone, and Dr. Louis D. Cohen are also thanked for their
comments and criticisms of various aspects of the dissertation.
Finally, the writer would like to express gratitude to his wife,
Doxie, who was an invaluable aid in every phase of the study, par-
ticularly so in the collection and scoring of the data.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS . . . . .
LIST OF TABLES . . . . .
I INTRODUCTION--STATEMENT OF
II LITERATURE REVIEW . .
III INITIAL HYPOTHESES . .
IV INSTRUMENTS . . .
V METHOD . . . . .
VI RESULTS OF INITIAL SAMPLE
VII RESULTS OF CROSS-VALIDATIOI
VIII DISCUSSION . . . .
IX SUMMARY . . . .
LIST OF REFERENCES . . . .
APPENDICES . . . . . .
A LETTERS TO THE STUDENTS .
3 INSTRUCTIONS TO THE RATERS
C INTER-RATER RELIABILITY OF
GRANOFSKY PICTURES . .
D MODERATOR CORRELATIONS FOR THE CROSS-
VALIDATION SAMPLE . . . . .
: : : : : :
LIST OF TABLES
1. Canonical Vectors of the
Initial Sample . . . . . ... .63
2. Nine-Variable Correlation Matrix
of the Initial Sample . . . .... .66
3. Correlations Between Field-Independence
and Attitudes Toward Disability . . .. .73
4. Correlations Between Social Conformity
and Attitudes Toward Disability . . . 74
5. Correlations Between Authoritarianism
and Attitudes Toward Disability . . 75
6. Correlations Between Ego-Strength
and Attitudes Toward Disability . . 76
7. Correlations Between Body-Satisfaction
and Attitudes Toward Disability . . 77
8. Correlations Between Body-Concern and
Attitudes Toward Disability . . ... .78
9. Correlations Between Social Desirability
and Attitudes Toward Disability . . 79
10. A Comparison of Relationships Found to Be
Significant in the Initial Sample with Those
Relationships Under the Same Moderator
Conditions in the Cross-Validation Sample 87
11. Nine-Variable Correlation Matrix of the
Cross-Validation Sample . . . ... .91
12. Canonical Vectors of the Cross-
Validation Sample . . . . . ... .97
INTRODUCTION STATEMENT OF PROBLEM
in the last two decades, the field of vocational rehabilitation
has made tremendous strides. With the passing of Public Law 565 in
1954, the Vocational Rehabilitation Administration was put on a firm
financial and administrative foundation. Since that time, there has
been a major Increase in research In rehabilitation, in the number of
professional specialists who work with the disabled, and in the facili-
ties needed to care for and train disabled people.
Despite the many improvements which have occurred, several basic
problems remain which interfere with the goals of rehabilitation.
One which continues to exist, and which appears to be particularly
resistant to change, is the negative, aversive way in which many people
respond to visibly physically disabled people (those Individuals whose
disabilities are readily apparent, e.g., amputees, disfigured people,
It is obvious that these negative attitudes are injurious to the
disabled person who is attempting to rise above or compensate for his
handicap. They have been shown to be a major factor in hiring prac-
tices, as it is generally accepted that unwarranted discrimination
exist in the employment of the disabled (Federation Employment and
Guidance Service, 1959; Jennings, 1951; Noland and Bakke, 1949;
Schletzer, et al., 1961). Rickard, et al. (1963), found that
although all disabled groups were subject to expressed prejudice from
employers, some were discriminated against much more than others.
The groups could be ranked according to the amount of discrimination
expressed toward them.
Attitudes of non-disabled people toward the physically disabled
are also of utmost importance in the day-to-day experiences which are
so crucial in influencing and forming the disabled person's feelings
about himself. There are countless descriptions of the frustrations
which disabled people encounter, and of the self-depreciation and
social withdrawal that inevitably follow (Wright, 1960; Barker, et al.,
1953; Dembo, et al., 1956).
Thus, the evil of discriminatory attitudes is twofold. The dis-
abled Individual is externally hindered in his efforts to "rise above"
his handicap by the vocational limits which his environment imposes
and which restrict the alternatives available to him in his quest for
fulfillment. He is hindered internally by the way he has learned to
view himself as the result of being the object of discrimination and
avoidance on the part of non-disabled people. Since he often sees
himself in an unnecessarily self-depreciating and limited way, the
realization of whatever true potential he might have is an extremely
Although much has been written about the negative attitudes which
are expressed toward the disabled, very few conclusive empirical studies
have been found to explain their etiology. Siller (1963 a, 1963 b) has
stressed the need for research into the personality factors involved in
how individuals relate to disabled people, but has bemoaned the lack of
a theoretical basis from which to attack the problem. Although it is
questionable whether theoretical bases are actually lacking (a thorough
review of the literature would suggest the contrary), Siller's studies
in this area seem to make that assumption. Granofsky (1956) also recog-
nized that empirical studies investigating personality variables in this
area were virtually non-existent. Although his major interest was in
whether negative attitudes toward the physically disabled could be modi-
fied by social contact, he did make an attempt to deal with personality
variables in the non-disabled. His study, which also could be charac-
terized as lacking in theoretical foundations, uncovered no relation-
ship between any certain personality variables and specific attitudes.
This study is an attempt to investigate and evaluate the various
theories which have been put forth as explanations of discriminatory
attitudes toward individuals possessing a visible physical disability.
An attempt will be made not only to appraise the validity of the various
theoretical approaches, but to give some indication of their relative
importance and degree of interrelation. Many studies have been done
attempting to establish a relationship between one variable, such as
prejudice, and attitudes toward the disabled. Since it is extremely
unlikely that any single variable can be cited as the determiner of
negative attitudes, these studies appear to have barely scratched the
surface. The present study will attempt to establish and describe inter-
actions and interrelations among those variables which have been con-
sidered important in the dynamics of negative attitudes. This study,
then, will not only serve to evaluate the theories which have been put
forth, but will hopefully result in a dynamic, rather than static,
appraisal of those personality variables which play a role in attitudes
toward physically disabled people.
The first step will be to review the studies which demonstrate the
presence of negative attitudes toward the disabled. When this has been
completed, the various theories purporting to account for this behavior
will be examined.
A. Studies Demonstrating the Existence of Negative Attitudes Toward
the Physically Disabled.
As early as 1921, Perrin had noted that the physical character-
istics and appearance of an individual "constitute a large group of
the total series of effects produced by that individual upon others."
In one of his initial studies in the area of interests, Strong
(1931) found that while a mildly favorable verbalized attitude was
expressed toward crippled and blind people, a more unfavorable attitude
was expressed toward deaf-mutes. Although the overall results indi-
cted generally positive verbalizations toward the disabled, there
was evidence of underlying biases and hostility.
Orgel and Tuckman (1935) investigated the kinds of nicknames given
to one another by a group of children living in an orphanage. They
found that the nicknames given by boys referred to physical defects
in 31.9% of the cases, and the nicknames of girls referred to physical
defects in 30.9% of the cases. Also, practically all of the nicknames
were derogatory, producing resentment.
Using college students in a study of verbally expressed attitudes
toward the disabled, Mussen and Barker (1944) directed them to rate
disabled people according to 24 personality traits. The data indicated
that these verbalized attitudes were generally favorable. However,
once again there was evidence of generalized attitudes or biases toward
the disabled, suggesting that they were being perceived in a stereotyped,
often degrading, manner.
Ray (1946) supported this finding of generalized public attitudes
in a study In which high school students served as subjects. The sub-
jects were asked to rank photographs of six college men according to a
list of behavioral personality traits. Half of the group was presented
with a photograph of an individual in a wheelchair; the other half was
presented with the same photograph with the wheelchair blocked out.
The results indicated that disabled people are perceived differently
(in a stereotyped manner) than non-disabled persons, i.e., they are
judged to be more conscientious, to feel more inferior, to be more un-
In a study involving social distance, Rusk and Taylor (1956) found
that 65% of 50 college students would not marry a person with an ampu-
tated leg, and 50% would not date such a person. Also, 85% would not
marry, and 72% would not date a deaf person.
Himes (1960) supported Rusk and Taylor's findings in a study in
which the Bogardus Social Distance Scale (Bogardus, 1927) was adapted
for use with blind students and collegiate normals. He was able to
show that the more intimate the proposed relationship with a blind per-
son, the more clearly and intensely he is rejected by his normal counter-
Using sociometric choices and lists of positive and negative traits,
Force (1956) found that physically handicapped children were not as well
accepted as normal children. The degree of disability was less Import-
ant than how the various disabilities were perceived by the normal sub-
In summarizing this group of studies, Barker and Wright (1954)
state that although public, verbalized attitudes toward the disabled
may be mildly favorable, an appreciable minority will openly express
negative attitudes, and a majority of individuals will reveal them in-
These studies have indicated the presence of underlying negative
attitudes toward disabled people. The various authors have not at-
tempted to explain their findings, only to show that these feelings
exist. The next group of studies will focus on attempts to explain
why these attitudes exist. Also, studies which support or contradict
the various hypotheses will be presented.
B. Theoretical Approaches Attempting to Explain Negative Attitudes
I. Theories and studies attributing negative attitudes to incon-
sistencies In the non-disabled's perceptual field. The first attempt
to understand why non-disabled people react in a negative manner toward
disabled people was made by Winkler (1931). He proposed that, because
of the unusual postures and movements of the crippled, such a person
serves as a strange, disturbing emotional stimulus. The behavior
aroused is that of suspicion and aversion, since there is little possi-
bility that the physically normal individual can establish an empathic
relationship with him. Winkler tested his hypothesis by displaying
actions pictures of healthy and disabled persons to 200 physically
normal subjects who were to judge the pictures with respect to charac-
ter and personality traits. Significantly more unfavorable judgments
were made of the pictures of disabled children, even though some of
these were not consciously recognized by the subjects as being disabled.
Schauer (1951), on the basis of psychiatric observations, felt
that it was the novel sight of a blind person which aroused fear in
normal people, thus lending some support to Winkler's views. He stated
that vague feelings of something "not being right" were the source of
In his theoretical framework involving neurophysiological concepts,
Hebb (1946) has attempted to explain fear responses to strange objects
and persons. His explanations remind one of Winkler's previously de-
scribed hypothesis. He postulates that a fear response occurs when on
object is seen which is like familiar objects in enough respects to
arouse habitual processes of perception, but in other respects arouses
incompatible processes. As in Winkler's theory, the inconsistency be-
tween expectancy and experience arouses fear, avoidance, and rejection.
The emphasis is on the effect of the individual's perceptual field on
his behavior. Personal responsibility for behavior is minimized, as
the individual is seen as a somewhat passive victim of the field.
Neither Hebb nor Winkler, however, is able to explain the individual
differences in reactions to the disabled, i.e., why some non-disabled
people are negatively affected by the visibly disabled, whereas others
are not, or at least, are affected in different degrees.
Another theoretical approach which is consistent with those of
Winkler and Hebb is that of Heider (1958), who contends that the nega-
tive impact of visibly disabled people is the result of dissimilarity
or unfamiliarity. He argues that the novel or different is experienced
as not fitting the structure of the matrix of the life space, as not
fitting one's expectations.
2. Studies and theories attributing negative attitudes to cultural
conformity. Several theorists have contended that attitudes toward the
disabled are culturally determined. Hanks and Hanks (1948), holding
this view, have outlined five status categories in which certain dis-
abled have been observed in various societies. These are: pariah,
economic liability, tolerant utilization, limited participation, and
Investigating 27 cultures, Jaques (1960) found that socioeconomic
factors play a major role in treatment of the disabled. Disabled
persons received protection in societies characterized by more pre-
dictable sources of livelihood, such as herding and agriculture, as
opposed to societies dependent on fishing, hunting, or gathering.
Mobility was another factor in the treatment of the handicapped, with
disabled persons receiving protection in the more stable societies.
Kolb (1959) recognized the importance of cultural attitudes in a
review of body-image disturbances by stating that a satisfactory social
adaptation among those with bodily defects depends more on cultural
dictates than upon the presence of the impairment.
MacGregor and Schaffner (1950) attribute much of the motivation of
patients requesting plastic surgery to cultural factors. They state
that most patients either wish to conform to cultural standards of physi-
cal beauty, to correct a deformity which might handicap social and eco-
nomic achievement, or to eliminate a socially perceptible trait which
may produce negative pre-judgments of their character or personality.
The ability of physical traits to influence people's judgments of per-
sonality has been described by Wright (1960), and labeled "spread."
Richardson, et al (1961), found a consistent preferential order
when children were asked to rank pictures of other children with physi-
cal disabilities. This order was stable across various subcultures,
seemingly determined by the broader cultural context.
This finding (that the pervasive cultural context is the important
determiner of reactions to disability), was indirectly supported by
Dow (1965), who found that social class was not a significant factor in
the conditioning of negative attitudes. He had hypothesized that the
reaction to disability was conditioned by the relative emphasis attached
to physique, and that this varied inversely with social class level.
Thus, the lower class would be expected to react more severely to physi-
cal impairment than would the middle or upper classes. His hypotheses
were not supported by the data, i.e., there was no significant class bias.
Trippe (1959) suggests that present cultural attitudes toward the
handicapped stem from earlier periods when such persons were a threat
to society. They could not produce goods or services, and reduced the
availability of already scarce necessities of life without themselves
contributing any material help to society.
Greenmum (1958) and Handel (1960) have attributed negative atti-
tudes toward the deaf and blind, respectively, to current cultural
beliefs, standards, and values. These cultural norms are particularly
important in the development and encouragement of stereotyping. Cowen
and Cowen (1964) found cultural differences in the attitudes of college
students to blindness and deafness, American students being more favor-
able to the blind while French students were more favorable toward deaf-
3. Theories and studies attributing negative attitudes to authori-
tarianism. Another group of theorists and researchers holds that dis-
abled people constitute a minority group. Like most minority groups,
they are subject to authoritarian attitudes (primarily manifested as
prejudice), are related to in a negative manner, and arouse hostility
and threat in certain members of the population. Dembo, et al., explain
that if a non-disabled's normal physical state is to be of value to him,
it is necessary for the abnormal physical state of the disabled to be
of lesser value. This process Is termed "devaluation." Through the
devaluation process, the disabled person is seen as being of less value,
and consequently is rejected by himself and others.
Gellman (1959, 1960) stresses prejudice as the source of rejection
of the disabled. He contends that this prejudice is a result of social
norms and customs, child rearing practices stressing health and normalcy,
and the non-disabled's own fear of being handicapped. Discriminatory
actions on the part of the non-disabled are classified into three groups:
(1) acquiescence to group standards, (2) displaced reaction to frustration,
(3) alleviation of personal fears and insecurities.
Cowen,et al. (1960), in developing an Attitudes Toward Blindness
Scale, found that negative attitudes toward blindness correlated sig-
nificantly with anti-minority, anti-Negro, pro-authoritarian attitudes.
Expanding Cowen, et al.'s study to Include disability in general,
rather than limiting it to blindness, Christie, et al. (1958), were
unable to provide support for the previous finding. No significant
correlations were found between the California F. Scale (Adorno, et
al, 1950) and the Attitudes Toward Disabled Persons (ATDP) Scale
(Yuker, et al., 1960).
On the basis of these studies, it would appear that blindness is
subject to authoritarian attitudes, but that disability in general is
not. One must question, however, whether something as diverse and
vague as "disability," which can range from the incapacitating to the
minute, and from the visible to the invisible, can be adequately
measured In an objective, uni-dimensional scale. Bell (i962), and
Siller and Chipman (1964) have previously raised this question. This
problem will be dealt with in greater detail in Chapter IV (INSTRUMENTS),
since the ATDP will serve as one of the criterion instruments in this
Granofsky (1956) tested the hypothesis that negative attitudes
toward the disabled were a form of prejudice, but that they could be
modified by social contact. He found that when attitudes were negative
prior to contact, eight hours of contact with disabled people did not
significantly modify them. In this study, the author developed a
Pictures Test to measure attitudes toward the disabled. Since It will
be utilized in the present study for that purpose, it will also be
described in detail in Chapter IV.
Somewhat contradictory to Granofsky's findings but consistent with
his hypothesis, Bateman (1962) found that contact and interaction with
blind children raised sighted children's ratings of them. Also, Wolman
(1958) found that blindness did not prevent a child from being accepted
by his sighted peers.
Rickard, et al. (1963), developed a social-distance scale to
measure prejudice toward disabled applicants for employment. All dis-
abled groups used in the study were found to be subject to expressed
prejudice. The disabilities could be ranked in terms of amount of preju-
dice expressed toward them.
Whiteman and Lukoff (1962) found evidence for the stereotyping of
blind people, i.e., seeing them as possessing certain traits, abilities,
and gifts. In a previous study Lukoff and Whiteman (1961) had found
that the term "blind people" encouraged negative stereotyping, but
"blindness" did not. There was a strong readiness to perceive blind
people as unhappy and lacking in Independence.
Jordan (1963) prefers "disadvantaged group" to "minority group" as
a description of the physically handicapped. Although he does present
a valid argument for his point of view, he admits that prejudice is
expressed and discrimination practiced toward the physically disabled.
Very probably the most in-depth explanation of the dynamics In-
volved in prejudice toward visibly disabled people is given by Adorno,
et al. (1950), who states:
The fact that the authoritarian person's helplessness as
a child was exploited by the parents and that he was forced
into submission must have reinforced any existing antiweakness
attitude. Prejudiced individuals thus tend to display "nega-
tive identification" with the weak along with their positive
though superficial identification with the strong. (Pg. 387.)
His orientation in interpersonal relations is thus toward
getting power by associating with the powerful and influential,
or at least toward participating in the power of those who
have it. Admiration for the strong and contempt for the weak
accompany this attitude. High scorers show predominantly what
may be called hierarchical conception of human relationships
whereas those who score low conceive of an equalitarian mutual-
ity in such relationships. (Pg. 413.)
Rusalem (1950), Koenig (1949), and Dubrow (1965) hold essentially
the same view as Adorno, et al. Chesler (1965) has provided empirical
evidence supporting the above theoretical approach, I.e., that physi-
cally disabled people are subject to prejudice (which he defines as
attitudinal and behavioral predispositions) in much the same way as
are ethnic minorities. He constructed an Intergroup Relations Scale,
which Is a modification of the Anti-Semitism Scale (Levinson and San-
ford, 1944) and the Negro and Minority Subscales of the California
Ethnocentrism Scale (Adorno, et al., 1950), and found that it corre-
lated -.52 with the ATDP. The ATDP correlated -.45 with the Race sub-
scale, -.40 with Religion, -.43 with Nationality, and -.46 with Social
Class, all of which were significant at the .01 level.
4. Theories and studies attributing negative attitudes to a lack
of ego-strength. Siller (1959, 1962, 1963a, 1963b) believes that the
degree and fixity of the individual's balance of self to object cathexis
is a major determinant of reaction to personal disability and the dis-
abled. This balance is represented in his theory by degree of ego
strength. Thus, a positive self-image and attainment of stable object
relationships are necessary for the acceptance of the disabled. Con-
versely, low ego-strength, poor self-acceptance, insecurity, and anxiety
are negatively related to acceptance of disabled people.
In attempting to verify this theory, he administered the Atti-
tudes Toward Disabled Persons (ATDP) Scale, the Gough Adjective Check-
list, his own Social Distance Scale and Feeling Checklist, and several
other scales. Although his results are fairly vague, he did find that
college subjects are more accepting than non-college subjects. There
were trends indicating that security, affiliation, and ego-strength are
positively related to acceptance of the disabled. Siller also felt
that his data lent support to the hypothesis that a negative self-
Image and disturbed object relations are conducive to an aversive re-
action to the disabled.
Although It was completed before Siller began his work, a study by
Steingisser (1954) does lend support to Siller's general hypothesis.
He found that individuals who were well adjusted (as measured by a dis-
crepancy between ideal and actual self) had more positive attitudes to-
ward the blind than a poorly adjusted group.
5. Theories and studies attributing negative attitudes to body-
anxiety and concern. The fifth group of theories and studies which
attempt to arrive at some explanation of negative attitudes toward the
disabled are those which are concerned with the body feelings, concepts,
and/or image of the non-disabled. That is, the theorists of this point
of view believe and attempt to show that non-disabled people react nega-
tively to visibly disabled people because of problems related to their
own bodies. In the classic treatise on body-image, Schilder (1950)
contends that a disabled person's physical difference creates uneasi-
ness because It does not fit with a well-ordered body image. A person's
unconscious body image may be threatened by the appearance of someone
with a deformity or missing part, since he identifies to some extent
with this person. He states:
It is obvious that interest in particular parts of one's
own body provokes interest in the corresponding parts of
others. Between one's own body and the bodies of others,
there exists a connection. (Pg. 225.)
There is a continual interchange between our own body-
image and the body-image of others. What we have seen in
others we may find out in ourselves. What we have found
out in ourselves we may see in others. (Pg. 227.)
Menninger (1949) states that unconscious awareness of one's self
as a complete unit includes a belief that all parts of the body are
sound and function normally. Since the loss or the crippling of a
part of the body signifies not only a physical wound but also a signifi-
cant psychological wound, the sight of a disabled person evokes an image
of this loss, thus constituting a threat to the non-disabled person.
In orthodox psychoanalytic theory, the castration complex is
cited as the explanation of the negative, hostile reactions to visibly
disabled persons. Maisel (1953) states that the loss of any part of
the body, or the sight of such a loss, may stir up archaic castration
fears. The Oedipal taboo is recalled, along with the father's potential
revenge--that of cutting off or mutilating the phallus.
Several studies have attempted to investigate the body-image of
the non-disabled as a factor in rejection of the disabled. Masson (1963)
attempted to find whether the definitness of the non-disabled's body
boundaries was related to acceptance or rejection of the visibly dis-
abled. Using the Fisher-Cleveland (1958) system of obtaining body bar-
rier and penetration scores from the Rorschach, he hypothesized that
persons with vague, indefinite body boundaries would be less accepting
and would manifest greater anxiety concerning the disabled than would
persons with definite, firm boundaries. Acceptance and rejection of the
disabled were measured by subjects' responses to the Granofsky Pictures,
a TAT-like instrument in which visibly disabled people were depicted.
His hypotheses were not supported. As was stated earlier, this instru-
,ent will be dealt with in much greater detail in Chapter IV, as it is
intended to serve as one of the criterion instruments in the present
Kaiser and Moosbrucker (1960) demonstrated empirically the relation-
ship between attitudes toward disabled people and concommitant physical
reactions in the non-disabled by correlating the ATDP with GSR reac-
tions to photographs of the disabled. They found that physically normal
college subjects scoring more than I standard deviation below the mean
on the ATDP showed more extreme GSR reaction to the photographs than
did subjects scoring more than 1 standard deviation above the mean.
Centers and Centers (1963b) investigated whether the presence of
amputation represents a threat to the bodily integrity of the non-
amputee. They found that peer group children expressed more rejecting
attitudes toward amputee classmates than toward non-amputee classmates.
Also, the amputee children were often considered the saddest children
in the class. The same authors (1963a) found similar results with
parents of malformed children. There were substantial correlations be-
tween responses to their children's bodies and responses to their own
bodies on the part of the parents of these children. Little or no cor-
relation in this respect existed for parents of normal children.
Epstein and Shontz (1962) investigated the relation between body
satisfaction and dissatisfaction, as measured by Secord and Jourard's
Body-Cathexis Scale (1953), and attitudes toward the disabled. The
authors of this study constructed their own test to measure "vital
interpersonal relationships" between non-disabled and disabled persons.
They found that satisfaction with one's own body was related to accept-
ance of the disabled.
On the basis of the literature review, there seem to be five
major personality constructs, each representing a distinct theoretical
approach, which have been considered to be of primary importance in
attempting to explain negative attitudes toward the physically disabled.
From each theoretical position a hypothesis can be generated stating a
relationship between any one personality variable and attitudes toward
physically disabled people.
These five initial hypotheses are over-generalized and admittedly
predict nothing about the interrelations or dynamic interaction of the
personality variables which contribute to the formation and existence
of the negative attitudes. Their primary value is to provide a frame
of reference from which these interactions can be investigated.
Thus, the initial sample of subjects will be tested on the basis
of these five general hypotheses. In analyzing the data, more subtle
relations between the various predictors and the criterion will be
noted. The interrelations which are found will provide a basis for
the subsequent refinement of hypotheses and cross-validation of the
The procedure will be explained in more detail in the METHOD
section. Let it suffice for now to present the initial hypotheses
which are based on theoretical positions stated in the literature and
which provide a means for investigating more thoroughly the components
of attitudes toward the disabled.
These initial hypotheses will be presented in the same order in
which the corresponding theoretical positions were presented in the
literature review section.
A. Hypothesis I--The Field-Dependency Hypothesis.
The first theoretical approach purporting to account for the nega-
tive feelings and attitudes which many non-disabled people have toward
the visibly physically disabled is that which attributes this behavior
to the Inconsistency between the non-disabled's expectation and his
actual experience. Thus, because of the somewhat strange, unusual
sight of a disabled person, the non-disabled's perceptual field (and
current experience) is in disagreement with what he has come to ex-
pect on the basis of his everyday experience. As a result, he is un-
sure how to react to what he sees, the confusion leading to anxiety
and aversion on his part.
This approach is represented by such diverse theories as those of
Hebb (1946), Winkler (1931), and Helder (1958), which have the basic
communality of emphasizing the effect of the incongruity of the indi-
vidual's perceptual field on his behavior. It seems that these theories
are all dealing with, although from differing viewpoints, the concept of
field-dependence. Witkin, et al. (1954), have defined field-dependence
as the manner of relating to or construing the environment in such a
way that the individual's behavior is determined almost exclusively by
his perceptual field. There is extensive investment in the field, and
behavior is strongly affected by environmental context.
Field-dependent people have been described as being more attentive
to the appearance of others, and tend to be better at recognizing people
they have seen only briefly before (Witkin, et al., 1962). This hypothe-
sis is supported by the studies of Konstadt and Forman (1965), Crutch-
field, et al. (1958), and Messick and Damarin (1964). They express
"impressions" of others in terms of physical features and gross be-
havior rather than in terms of personality characteristics. Thus, the
amount of psychological disturbance aroused by a visibly disabled person
might be a function of the extent to which the individual in which the
disturbance is aroused is field-dependent. Individuals who are able
to function with relative freedom from their environment should not be
as strongly affected by an inconsistency in this environment as would
individuals whose frame of reference is external and whose behavior is
largely determined by the environment.
B. Hypothesis II--The Cultural-Conformity Hypothesis.
A second approach toward explaining negative attitudes toward the
disabled deals with the concept of cultural conformity. Irrational
feelings and attitudes, according to this point of view, are a function
of conforming to the prevailing attitudes of the society. Kolb (1959),
Trippe (1959), Jaques (1960), and Hanks and Hanks (1948) are among
those who would contend that negative attitudes toward the disabled are
the result of the emphasis in our society on such things as physical
beauty, health, wealth, and the value of work.
Individuals who tend to conform to society's rules, conventions,
attitudes, values, etc., would, therefore, be predicted to react nega-
tively to visibly disabled people, since this is passively condoned by
the values of our culture. Those people who are not as unquestioning
or conforming would not be predicted to behave as aversely toward the
disabled, since they do not feel as strongly the need to behave in a
manner consistent with society's conventions.
C. Hypothesis Ill--The Authoritarianism Hypothesis.
Dembo, et al. (1956), Gellman (1959, 1960), and Adorno, et al.
(1950) are most representative of the third group of theorists--those
who would attribute negative attitudes toward the disabled to the
dynamics of authoritarianism. Since the disabled people are a minority,
and are visibly associated with "the weak," they arouse hostile, nega-
tive reactions in people who, in order to enhance themselves and main-
tain psychological equilibrium, must associate themselves with "the
strong," the successful, and the majority.
Thus, these theorists would contend that individuals who are highly
authoritarian ethnocentricc, anti-minority, power-oriented, etc.) would
react in an Irrationally negative, hostile manner toward visibly dis-
abled persons. Individuals who could be described as equalitarian, or
who are not as authority-oriented, would not be as threatened by "the
weak" or the minority that disabled people represent. Therefore, there
should be a minimal arousal of hostility, devaluation, and aversion on
D. Hypothesis IV--The Ego-Strength Hypothesis.
Siller's (1959) theory of ego-strength as the crucial factor in
acceptance of the visibly disabled people is typical of the fourth
theoretical approach attempting to account for negative reactions to-
ward them. He contends that an individual with satisfactory ego-
strength (stable object relations, a positive self-image, minimal
anxiety, a sense of security) will be acceptant of visibly disabled
people. Conversely, people with negative self-images, feelings of in-
security, etc. would, through what Siller terms narcissistic regression,
react negatively toward disabled persons.
The depth of this narcissistic regression is supposedly dependent
on the balance of self to object cathexis in the individual. Attitudes
become more negative as the regression becomes deeper.
E. Hypothesis V--The Body-Concern Hypothesis.
Finally, there are those theorists who feel that the body-image
and body-feeling of the non-disabled are of utmost importance in their
attitudes toward the disabled. Schilder (1950), and Menninger (1949)
are the leading proponents of this point of view. These theorists
would hold that individuals who are high in body preoccupation and
anxiety (thus investing a large amount of energy, attention, and inter-
est in their bodies) should attend to and be more affected by the sight
of a disabled person than an individual who is not as body-oriented.
Since it is not as likely that body anxiety or conflict would be
aroused in the less body-preoccupied person, he should not be as
threatened by a visibly disabled person, and thus should be more ac-
ceptant of him than an individual with high body-preoccupation. Also,
individuals who are low in body-satisfaction would be less acceptant
of the disabled than persons who are highly satisfied with their
bodies. Since people who are highly satisfied with their bodies have
been found to be more secure (Jourard and Secord, 1955 ; Secord and
Jourard, 1953) than those who are low in satisfaction, it is reasonable
to predict that they will feel less threat and less need to reject and
avoid disabled people.
In summary, there are five initial hypotheses being tested in this
study. They are:
(1) Negative attitudes toward visibly physically disabled people
are a function of the extent of field-dependence on the part of
the non-disabled. The more field-dependent an individual is, the
less acceptant of the disabled he will be.
(2) Negative attitudes toward physically disabled people are a
function of the need for social conformity on the part of the non-
disabled. The more willing an individual is to acquiesce to the
conventions and values of society, the less acceptant of the dis-
abled he will be.
(3) Negative attitudes toward the disabled are a function of the
extent of authoritarian attitudes on the part of the non-disabled.
The more authoritarian an individual is, the less acceptant of
the disabled he will be.
(4) Negative attitudes toward physically disabled people are a
function of the degree of ego-strength on the part of the non-
disabled. The less ego strength an individual possesses, the
less acceptant of the disabled he will be.
(5) a. Negative attitudes toward physical disability are a
function of the degree of conscious body-satisfaction possessed
by the non-disabled. The less satisfied he is with his own
body, the less acceptant he will be of the disabled.
b. Negative attitudes toward physical disability are a
function of the amount of unconscious concern and anxiety the
non-disabled person harbors for his own body. The more somatic
anxiety and concern that are present, the less acceptant he will
be of the disabled.
To repeat what has been stated earlier, and perhaps clarify this
somewhat, these initial hypotheses will serve three main purposes.
First, by testing these hypotheses it will be possible to evaluate the
relative validity of each of the five theoretical constructs and their
respective approaches as an explanation of negative reactions to dis-
abled people. Secondly, some measure of the interrelation of the pre-
dictor variables will be forthcoming. Finally, and most importantly,
they will provide a vehicle for investigating the more subtle relations
between the predictors and criterion. This will make It possible to
formulate more specific hypotheses and determine and predict how these
variables interact to produce negative attitudes toward physically
Each of the theoretical approaches, and related hypotheses, em-
phasizes the importance of a particular personality construct in the
formation and existence of negative attitudes toward the physically
disabled. Thus, there are five independent variables, or predictors,
and one dependent variable, or criterion. This section will describe,
and provide pertinent data on, the instruments used to measure these
variables. They will be presented in the same order in which their
related hypotheses and literature were presented.
A. Predictor Instruments.
1. Field-Dependence. In this experiment, field-dependence was
measured by the Barrett-Frutcher Chair-Window Test (Barratt, 1955).
This test is actually constructed to be a measure of field-independence
rather than dependence. It is a paper-and-pencil test of 32 items in
which the subject's task is to determine through which of the five
windows one would have to look to see a chair from the angle shown in
each picture. The subject's score is the number of items to which he
responds correctly. Thus, scores can range from 0 to 32, the higher
scores indicating field-independence.
Barratt (1955) reported a reliability coefficient of .89 with this
instrument. The type of reliability was not reported. Young (1959)
found that the Chair-Window Test correlated -.49 with Jackson's
Embedded Figures (a short form of Witkin's original test) and -.35,
-.33, and -.48 with three forms of the Rod and Frame test of Witkin.
All coefficients were significant at the .01 level.
These negative correlations were predicted since high scores on
Witkin's instruments are indicative of field-dependence whereas on
the Barratt-Fruchter they indicate field-independence. The Barratt-
Fruchter is particularly valuable because it can be administered to
groups, and unlike the majority of Witkin's tests, does not require
2. Social Conformity. Bass' Social Acquiescence Scale was used
as a measure of social conformity (Bass, 1956). This is an instrument
composed of 56 statements in the form of proverbs, adages, sayings, etc.,
to which subjects are to indicate whether they agree, disagree, or are
uncertain about each statement. Since the individual's score is equal
to the number of times he has indicated agreement, scores can range
from 0-to 56, with the higher scores indicating social acquiescence or
Bass defines social acquiescence as the tendency to agree with or
to accept a wide variety of generalizations. An individual who obtains
a high score is described by Bass as "outward-oriented, unsensitive,
none-intellectual, socially uncritical--and unquestioning conformer to
social demands placed on him."
In Bass' original study of this scale he reported a split-half
reliability of .92 with 100 college students. The proportion of 300
proverbs accepted by a subject was used as a gross measure of the
criterion against which the final, 56-item scale was developed by item
analysis. Two hundred college students were ordered according to these
criterion scores. The performance of the upper 25% of the subjects was
contrasted with that of the lower 25% on each of the 300 statements.
Fifty-six items emerged which were accepted by at least 40% more of
the upper than of the lower criterion group.
In the same study, Bass reported that Southern salesmen, who are
most likely to display Fromm's "marketable personality," show more
social acquiescence than any other group (Southern college students,
Midwestern college students, Northern salesmen). Also, Southern College
students, who are considered to be more readily acceptant of the tradi-
tional and conventional, had a greater tendency to acquiesce than Mid-
Shaw (1961) found that the Social Acquiescence (SA) Scale correlated
positively and significantly with the F Scale and the Individual Promi-
nence scale. There was a significantly negative correlation between
the SA scale and the Concept Mastery Test.
Producing additional data for the validity of this scale, Fisher
(1964), with 49 men and 49 women subjects, found a significant, posi-
tive correlation between the SA scale and religiosity. The Religious
scale of the Vernon-Allport-Lindzey study of values, self-ratings of
religiosity, and frequency of church attendance were taken as indices
3. Authoritarianism. Authoritarianism was measured by means of
the California F scale (Adorno, et al., 1950). This scale came about
as the result of an attempt to measure prejudice without appearing to
have this aim and without mentioning the name of any minority group.
The F scale is a 30-item, Likert-type scale on which the subject
indicates the degree to which he agrees or disagrees with each item on
a 6-point continuum. Responses can range from "strong support" (+3)
to "strong opposition" (-3), with higher positive scores Indicating
increasing authoritarianism. These responses are converted into scores
by a uniform scoring system, ranging from a score of 1 for -3 to a
score of 7 for +3. The number 4 is assigned when an item is omitted.
The sum of an individual's scores on the single items is then divided
by 30 (number of items on the scale) to arrive at one overall score
for each person on the scale. Scores may range from 1.00 to 7.00.
Adorno, et al. (1950), reported an average test-retest reliability
of .90 for this scale, with a range from .81 to .97 for various groups.
The authors interpreted this to mean that the scale could place indivi-
duals along a dimension with a small margin of error. They stated that
the score attained by an individual could be relied on in the sense that
chance errors of measurement have been minimized. The original vali-
dation of the F scale was done primarily by case studies, interviews,
and projective techiques. Clinical syndromes were painstakingly con-
structed and related to scores on the scale. Although this method ad-
mittedly was largely intuitive, numerous empirical studies have since
supported the original findings and substantiated the validity of the
F scale (e.g., Flowerman, et al., 1950; Scodel and Mussen, 1953;
Campbell and McCandless, 1951; Milton, 1952; Radke-Yarrow and Lande,
1953; Kates and Diab, 1955).
Some criticism has been made in recent years concerning the validity
of this scale. Since each item on the scale is worded so that agreement
indicates authoritarian attitudes, it was felt by some (Cohn, 1953; Bass,
1955; Chapman and Campbell, 1957; Jackson and Messick, 1957; Zuckerman
and Norton, 1961) that a large part of the variance could be attributed
to an acquiescent response set, and the scale was, therefore, invalid.
The presence of acquiescence was usually demonstrated by constructing
a "reversed" F scale, in which authoritarian attitudes would be indi-
cated by disagreement with the items, and not finding the expected nega-
Christie, et al. (1958), have challenged this approach (construct-
ing reversed scales) on the grounds that the reversed items may be
grammatically opposite, or logically opposite, but very seldon are
psychologically opposite, and thus should not be expected to yield a
negative correlation. Thus, according to Christie, et al., studies
making use of a reversed scale are most often not a fair test of the
The original authors of the scale (Adorno, et al., 1950) realized
that the items were negatively worded, and admitted that the argument
against using all negative items is that it might produce a "set" or
mechanical tendency to consistently agree or disagree. They answered
this argument on several grounds:
I. Most individuals show variability of response, as indicated
by the item intercorrelation averaging .3 to .4.
2. There is a tendency to vary in order to avoid an extreme
3. Very similar results were obtained when an all-negative
scale was inserted randomly into a longer series containing
4. Since the "set" argument implies that high scorers are not
necessarily anti-Semitic nor lows anti-anti-Semitic, the final
test is the validity of the scale. The authors felt that this
had been convincingly demonstrated.
The authors expressed two major reasons for using only negative
items: (1) they tend to be more discriminating (this has recently
been supported by Gage and Chattcrjee, 1960); (2) they can express sub-
tle hostility without seeming to offend the democratic values which
most prejudiced people feel they must maintain.
Another group of studies defending the F scale has contended that
acquiescence set, in itself, is an indicator of authoritarian attitudes,
and thus does not detract from the validity of the scale. Leavitt, et
al. (1956), state that the F scale confounds form and content variables,
but in the "right" direction. Authoritarian people, as measured by the
scale, agree more with authoritative statements; therefore, a portion
of the discriminatory power of the scale derives from its form, in
addition to its content. Gage, et al. (1957), hold a very similar
view. Messick and Jackson (1958) state that the acquiescent response
set operates to increase the discriminatory power of the scale, rather
than render it invalid. Further support for this point of view comes
from Messick and Frederikson (1958), Zuckerman and Eisen (1962),
Chapman and Block (1958), and Weatherley (1964).
In view of this research, it appeared that the California F scale
does provide a valid and reliable measure of authoritarian attitudes,
despite the presence of acquiescent response sets, and thus could be
used in this study.
4. Ego-strength. To arrive at a measure of ego-strength, the
Barron Ego-Strength Scale (Barron, 1953) was used. This scale is com-
posed of 68 items drawn from the MMPI. An individual's score on this
instrument is equal to the number of responses made in a predetermined,
"healthy" direction. Scores can range from 0 to 68, with high scores
being indicative of ego-strength.
Barron states that this instrument "appears to measure the various
aspects of effective personal functioning which are usually subsumed
under the term ego-strength." He contends further that high scorers
are characterized by the "capacity for personal organization." Among
the characteristics which are collectively referred to as ego-strength,
and which he found to correlate with the prediction scale, are physio-
logical stability and good health, a strong sense of reality, feelings
of personal adequacy, permissive morality, lack of ethnic prejudice,
emotional outgoingness, spontaneity, and intelligence.
The items for this scale were selected on the basis of significant
correlations with rated improvement in 33 psychoneurotic patients who
had been treated for six months in a psychiatric clinic. Responses
were obtained before therapy began.
The sample of 33 patients was divided into two groups: 17 patients
judged to have clearly improved and 16 who were judged to be unimproved.
Each case was Intensively studied, and two skilled judges who had
thoroughly acquainted themselves with the course of therapy were in
considerable agreement (r of .91) in their independent ratings of the
degree of improvement.
The improved and unimproved group could be distinguished on the
basis of the scale at the .01 level. Odd-even reliability of the
scale was .76, test-retest reliability after three months was .72.
The ability of the Barron Ego-Strength to predict outcome in
psychotherapy has been further validated by Wirt (1955), Barron and
Leary (1955), and Sunnett (1962). It has also been found to con-
sistently differentiate between various psychiatric patients and
normals by Gottesman (1959), Quay (1955), Taft (1957), Kleinmutz (1960),
Worrell and Hill (1962), and Tamkin and Klett (1957). Korman (1960)
found that psychiatric inpatients with high scores on the scale were
able to resolve a discrimination conflict more quickly than a low ego-
strength group, and Herron (1962a) has shown that, among schizophrenics,
extreme ego-strength scores (high and low) were significantly in agree-
ment with psychiatric ratings on prognosis. Himelstein (1964) has
found that self-referrals tend to obtain higher scores on the ego-
strength scales than those earned by referred counselees.
Recent evidence (Silverman, 1963) has shown that Barron's scale
can be used without giving the complete MMPI. Silverman found that the
two forms correlated .91 for normals and .85 for schizophrenics. The
ego-strength scale administered alone differentiated between normals
and schizophrenics at the .02 level of significance.
5. Body-concern and Body-satisfaction. Two instruments were used
to measure body-concern and body-satisfaction. The first of these was
the Homonym Test of Secord (1953). This is a list of 100 words, 75
of which are body homonyms, i.e., words which can refer either to
body parts and/or processes or to non-body concepts. An example of a
body homonym is "digit" which could be responded to with "finger" (a
body response) or -number" (a non-body response).
The list of words is read by the examiner at the rate of one every
five seconds. Subjects are instructed to write on an answer sheet the
first word which comes into their minds after the examiner reads each
stimulus word. Twenty-five words on the list are not homonyms, but
"dummy" words, which are included to destroy any set which might dev-
elop. Each subject's score is the number of body responses that are
given to the list of stimulus words.
Secord hypothesized that the greater the number of body responses
that were given by a subject, the more preoccupied and/or anxious he
was concerning his body. Using a group of 145 college students, he
extracted the 15 subjects making the highest scores and the 15 making
the lowest scores as high and low criterion groups. Secord and one
other rater were able to predict from Rorschach protocols whether or
not a person belonged to high or low homonym groups, Secord being cor-
rect on 22 out of 28 (p<.002), and the other rater correctly predicting
20 of 28 (pc.003).
The author reported an inter-rater reliability in scoring of .99,
and split-half reliabilities In two different samples of .81 and .73.
In this same study, a correlation of -.42 ( p<.01) was reported between
homonym scores and the total score on a Body-Acceptance Test, and of
-.54 (p<.01) between homonym scores and a subtest of Acceptance of Body
The second instrument, a measure of body-satisfaction, was the
Body-Cathexis Scale of Secord and Jourard (1953). This is a Likert-
type scale in which the subject must rate his satisfaction or disatls-
faction with 40 body parts or processes. Each body part is rated on a
five-point scale rating from I (complete dissatisfaction) to 5 (complete
satisfaction). The rating given to each body part is added, and the
total divided by 40 to arrive at the overall score for each subject.
The higher an individual scores, the more satisfaction is expressed
with his body. Body-cathexis was defined by the authors as the degeee
of a person's feelings of satisfaction or dissatisfaction with various
parts or processes of the body.
Secord and Jourard (1953) reported split-half reliabilities on
this scale of .78 for males and .83 for females. For 56 college females,
the Body-Cathexis (BC) Scale correlated.-.41 (p<.01) with Homonym Test
scores, indicating that body satisfaction Is negatively related to body
preoccupation and anxiety. The BC scale also correlated -.37 with the
Maslow test of insecurity. Jourard and Secord (1955), in a further
study, found that with males the BC scale correlated .58 with a Self-
Cathexis (SC) Scale; with females the correlation was .66 (both p<.01).
Johnson (1956), in cross-validating several of these findings,
found an inverse relationship between attitudes toward the body and the
number of symptoms reported. More specifically, with males there was
a correlation of -.33 (p4.O5) between Body-Cathexis and symptoms checked
on the Cornell Index. With females this relationship was -.40 (p<.01).
B. Control Instrument
One of the primary dangers in administering a group of personality
inventories or attitude scales is that the results may be confounded
by various response styles on the part of the subjects (Jackson and
Messick, 1958). That is, a subject's responding on these instruments
may be attributable to some extent to such tendencies as his wanting
to appear liberal, needing to avoid extremes, etc., thus detracting
from the results that might have been obtained had the subject been
less confounded in his responding to the item content.
Taylor (1961) has warned that in studies where there are several
samples of self-attitudes or attitudes toward others, a tendency to
find positive correlations between various measures of personality
traits may accrue largely because of the willingness of a subject to
ascribe socially desirable attitudes to himself, or to present himself
in such a light. Edwards (1953, 1957) has voiced similar opinions,
stating that social desirability may be viewed as a characteristic of
test items, and is often a major determiner of a subject's responding
to a personality inventory.
Although a recent study by Siller and Chipman (1964) found that
influence of response sets are usually negligible in this type of ex-
periment, the possibility existed that the social desirability variable
could confound and interfere with the results obtained in this experi-
ment (due to the fact that several of the instruments were self-rating
or attitude scales). It was,therefore, decided to include a measure
of this variable in the test battery as a "control." That is, it was
felt that the results of the study could be evaluated more accurately
if some estimate of the extent of social desirability response sets
A social desirability scale developed by Crowne and Marlowe (1960)
was chosen as the instrument to include for this purpose. The authors
considered their instrument to be superior to that developed by Edwards
(1953) primarily because of the fact that the item content of their
scale was much broader than that of Edwards, which referred almost ex-
clusively to the presence or absence of symptoms and complaints.
Edward's scale was described by them as a measure of the extent to which
an individual is willing to admit to symptoms of maladjustment. Content
of this nature makes it difficult, if not impossible, to determine
whether subjects' responses are due to social desirability, or to a
genuine absence of symptoms. The achievement of high social desirability
scores may reflect the low frequency of pathological symptoms rather
than the need of subjects to present themselves in a favorable light.
This problem is made more acute by the fact that the true frequency of
the various symptoms is not known.
In contrast to this pathology-oriented type of item, the content
of the Marlowe-Crown Social Desirability Scale refers to behaviors which
are culturally sanctioned, but highly improbable. Thus, the rationale
here is similar to the MMPI Lie Scale, although the items are less
extreme. The ambiguity which arises from the failure to consider the
actual incidence of traits,represented in the test items are avoided.
Social desirability is defined by Marlowe and Crowne (1961) as a need
for social approval and acceptance, and the belief that these can be
attained by means of culturally acceptable and appropriate behavior.
Thus, rather than seeing the social desirability response tendency as
a function of the test item as did Edwards (1957), Marlowe and Crowne
define it as a motivational variable.
The Marlowe-Crowne Social Desirability Scale (SDS) is composed
of 33 items which discriminated at the .05 level or better between
high and low total scores of 76 college students. Of these items,
18 are keyed 'true" and 15 are keyed "false," with the subject's score
being the number of responses made in the keyed direction. The larger
the number of keyed responses, the stronger the social desirability
Crowne and Marlowe (1960), using Kuder-Richardson formula 20, re-
port an internal consistency coefficient of .88 with this instrument,
Also, a test-retest correlation of .89 was reported in the same study.
Correlations of .35 and .56 with the Edwards Social Desirability Scale,
and of -.54 between the Social Desirability Scale and the Barron's Inde-
pendence of Judgment Scale (Barron, 1953b), have been reported (all
p<.01). Marlowe and Crowne (1961), with 57 male college students,
found that subjects with a high score on this scale expressed more favor-
able attitudes toward a boring experimental situation. Neither Edwards'
nor Barron's scales were able to make this differentiation.
Although the social desirability variable has been labeled a
"control" variable, it was treated as a predictor in the present study,
and included in that category when the results have been obtained.
However, unlike the rest of the predictors, no hypotheses were made
with respect to its relationships to the criterion, as it has not
been considered theoretically important in contributing to attitudes
toward disability. Rather, its primary importance was to serve as a
check on the relationships of the predictors to the criterion.
C. Criterion instruments
1. The Granofsky Pictures Test. The dependent variable in this
experiment, attitudes toward the physically disabled, was measured by
two instruments, the first developed by Granofsky (1956), and the
second by Yuker, et al. (1960). Granofsky's instrument is a TAT-type
projective instrument on which subjects are instructed to tell stories
to stimulus pictures in which physically disabled people are depicted.
There are 12 scenes in all, three within each of four diagnostic
groups (wheelchair, leg amputation, facial disfigurement, arm amputation).
In order to facilitate the writing of the stories, three questions
appear beneath each picture: "What is happening in this picture?"
"How did it come about?" "What will happen next?" More detailed in-
structions were given on the cover of the booklet in which the pictures
The pictures are printed at the top of 8-1/2 x 1-inch sheets
of paper. Each picture is 4-1/2 x 6 inches, and the rest of the space
is provided for the subject to write his story about the picture. It
has been found (Granofsky, 1956; Masson, 1963) that this space is quite
adequate for recording the stories. However, subjects are instructed
to use the back side of the same sheet if more space is needed.
Since the responses elicited by this instrument are largely un-
structured, it was important that a system of quantification be devel-
oped so that they could be objectively evaluated and scored. Thus, a
scaling method was employed in which reactions to the disabled were
placed on an 8-point continuum. Different degrees of strength of nega-
tive reactions toward the disabled were represented by scores of -5,
-3, and -1, with -5 being the extreme negative reaction. Positive
reactions were scored +1, +3, and +5, with +5 being the most positive,
accepting reaction to the disabled. Neutral responses, which are
usually indicated by a factual, colorless, or descriptive response,
are scored zero. An ambivalent score, where the evidence from the re-
sponse is too contradictory to support the scoring of a negative or
positive alternative, is scored .
The components of a -5 and a +5 reaction were described and speci-
fied by Granofsky to provide weights with which to better judge the
middle range of responses. A -5 score suggests "absolute rejection,
expressions of ridicule, horror, hostility, or repugnance, and/or iso-
lation of the disabled as social outcasts." A +5 response indicates
"the admittance of the disabled to all varieties of social interactions
with the non-disabled. More than mere tolerance, it involves a
deliberate and well considered approach to the problems of accepting
The scoring of responses was further facilitated and clarified by
the conception and description of three theoretical qualitative cate-
gories of aspects of attitudes toward the disabled. These categories,
an evaluation category (concepts of appraisal and estimation of dis-
abled persons), a cognizance theory (reactions to the appearance of
disability), and a social interrelationships category (concepts re-
garding the degree of social rapport with the disabled), were differ-
entiated for both sides of the scale. That is, detailed descriptions
were given of positive and negative evaluation, cognizance, and social
interrelationships. It must be remembered that only one score is
given to each response. These categories are but aspects of the re-
sponse which can be objectively stated, thus facilitating scoring.
The categories are alaborated as follows:
a. Negative evaluation: Feeling of pity and over-
solicitousness toward the disabled. Regarding the dis-
abled as objects of curiosity. Pre-conceived lower
social status roles for the disabled. Over estimation
of psychological concommitants of a disability. Exagger-
ation and misconception regarding limitations set by an
b. Positive evaluation: Objective appraisal and realis-
tic sympathetic evaluation of the disabled. Assessment
of psychological concommitants of a disability in a
practical way. Realistic estimation of limitations of
a disability. Acceptance of the disabled on an equal
social status footing with the non-disabled.
a. Negative cognizance: Undue awareness and occupying
one's self with the appearance of disabilities. Call-
ing particular attention to the disabilities. Over-
estimation of unsightliness of a disability.
b. Positive cognizance: De-emphasis of properties of
disabilities, assignment of minor role to appearance of
3. Social interrelationships
a. Negative social interrelationships: Setting apart
of the disabled from other people. Resistance or reluc-
tance to admit them to participation in various kinds of
social relationships. Feelings of uncomfortableness, un-
easiness, and anxiety and embarrassment in the presence
of the disabled.
b. Positive social interrelationships: Admittance of
disabled to a variety of social interactions with the
non-disabled. Expressions of feeling at ease in the
presence of the disabled.
Thus, subjects' responses to each of the 12 pictures received a
single score depending on the direction and degree of the response.
In order to facilitate tabulation of the data, the original algebraic
scores were converted to continuous numbers by uniformly adding five
points to each of the scores. The conversion system, then, was as
follows: -5 = 0, -3 = 2, -1 = 4, 0 and t = 5, +1 = 6, +3 = 8, +5 = 10.
A subject's total score could range from 0 to 120 points. Granofsky
reported an inter-scorer reliability of .76 for the Pictures Test,
indicating that the judges were evaluating the expressed attitudes to-
ward the disabled in an objective, consistent, and stable manner.
In order to arrive at a suitable criterion against which to vali-
date his instrument, Granofsky contacted 82 new members of the 52 Associ-
ation, a non-profit, volunteer organization in New York City which pro-
vides placement, counseling, and entertainment services for disabled
veterans. These subjects were chosen because their reactions to
physically disabled people could be observed and rated by supervisors
while in the actual situation. To arrive at an empirical measure of
this reaction, Granofsky developed a behavior-rating scale, whose ob-
jective was to secure a set of items which would sample various as-
pects of manifested behavior toward the disabled. Each of the Items
was considered as descriptive of a possible characteristic trend in the
person which would be expressed often enough behaviorally to permit ob-
servation by the supervisors. Fifteen items were eventually decided
upon, five in each of three subscales (evaluation, cognizance, social
interrelationships). An objective scoring system was decided upon
which resulted in one total score for each individual on the scale.
The validity of the Pictures Test, determined by the correlation
between the subjects' scores on this instrument and their score on the
behavior-rating scale, was found to be .56. A validity coefficient of
this degree (.56), while not extremely high, is considered to be sub-
stantial enough to be of use in research of this type (Bell, 1948;
Correlations of the pictures of each of the four diagnostic sub-
groups with the criterion produced the following coefficients: Wheel-
chair group, .58; leg amputation group, .56; facial disfigurement, .40;
arm amputation, .36. Application of Hoyt's (1941) analysis of variance
technique, which computes the proportion of total variance that is true
variance, resulted in a reliability coefficient of .84.
Masson (1963), in a study investigating the relationship between
body-image (as measured by the body-barrier and penetration system of
Fisher and Cleveland, 1958) and attitudes toward the disabled, made
use of Granofskys instrument. Although he was unable to find a sig-
nificant relationship between the two variables, he did report an
inter-rater reliability of .84 with the Pictures Test, thus providing
further evidence that the pictures can be rated in a reliable, con-
2. The ATDP Scale. This instrument is a Likert-type scale in
which the subject responds to the statements by expressing his degree
of agreement or disagreement on a six-point scale ranging from "I agree
very much" to "i disagree very much." To date, three different forms
of the scale have been developed, the original scale being composed of
20 items, and two subsequent scales of 30 items each. Form B, one of
the 30-item scales, is considered superior by its authors in terms of
its correlation with other scales (Block and Yuker, 1966).
In order to bolster the criterion, thus insuring that attitudes
toward disability are being adequately measured, Form B was added to
the battery as a second criterion instrument.
Each statement in the scale suggests that disabled persons are
either the same as physically normal persons, or that they are dif-
ferent. The statements cover two aspects of this problem. Approxi-
mately one-half are worded to point out similarities or differences in
"personality" characteristics, while the others suggest the need or
lack of need for "special treatment" for the disabled. The following
statements are illustrative of each type: (personality)--"Most dis-
abled persons worry a great deal"; (special treatment)--"You should
not expect too much from disabled persons."
The reliability and validity of the three scales have been deter-
mined for both disabled and non-disabled persons, the original disabled
group consisting of 248 workers at Abilities, Inc. (a manufacturing
firm employing severely disabled persons), and the non-disabled group
consisting of 625 Hofstra College students.
Validity was examined in terms of interrelationships that might
be predicted to exist between ATDP scores and other measures of be-
havior. Thus, evidence that this instrument measures what it purports
to measure was obtained through construct validation. With disabled
persons, the ATDP score was found to be positively related to satis-
factory work performance (p<.05), job satisfaction (p<.01), and intel-
ligence (p<.05). Females scored higher (were more acceptant) than
males (p<.05). Scores on this scale were found to be negatively re-
lated to anxiety (p<.01), and age (p<.05). The authors state that a
high score on this instrument by disabled persons indicates self-
acceptance, whereas with non-disabled people it would indicate accept-
ance of the disabled (a low score, conversely, would indicate a preju-
dice of the disabled).
In investigating the performance of non-disabled persons on this
scale, It was found that they score significantly lower (less accept-
ant) than the disabled (p<.001). It was also found that the ATDP was
positively related to the amount of contact that the non-disabled person
had had with disabled people (p<.001). No correlation was found be-
tween the ATDP scale and the Edwards' Social Desirability Scale, sug-
gesting that responses to this scale are not significantly determined
by such a response set. The authors also provide evidence that the
ATDP scale is difficult to fake in a given direction.
Since Form B (one of the 30-item scales) of the ATDP was utili-
zed in this experiment, only that scale's reliability data will be
presented. Split-half reliabilities for Form B have been found to be
.814 (N = 139) and .792 (N -50). The following correlations were re-
ported between Form B and Form A (Equivalent Form reliability):
.831 (N = 57), .721 (N =84), .412 (N = 58). In two different samples
Form B has correlated .762 (N = 40) and .572 (N = 81) with the original
20-item form of the ATDP. All of the above data has been collected
with non-disabled subjects. Although no test-retest reliability
coefficients were reported for Form B, a coefficient of .70 was found
with the original form with a period of four months between testing.
In this study, the directions for taking the ATDP scale were
slightly modified. This was done by explicitly defining to what types
of disability the term "disabled persons" refers. Since a pilot study
indicated that subjects felt that this term was too vague (could refer
to anything from deafness to quadriplegia), it was felt that specifying
the disability to include only amputation, wheelchair confinement, and
facial disfigurement should lead to increased reliability. Bell (1962)
and Siller and Chipman (1964) have previously discussed this apparent
weakness In the ATDP.
It was felt that adding the ATDP scale to the Pictures Test would
result in a measure of the criterion which would be broader and more
inclusive than either measure alone. Since the two instruments appear
to tap quite different levels of functioning (Granofsky Pictures--
unconscious feelings and reactions; ATDP--conscious opinions, beliefs),
the probability that attitudes toward physically disabled people were
being reliably and validly measured should be increased by taking both
instruments into consideration.
In summary, then, there were six instruments used to measure the
prediction variables, one control instrument (thus resulting in seven
predictors), and two instruments to measure the criterion variable.
The five initial hypotheses, stated in terms of the instruments used
to measure the variables, are repeated below.
1. The Barratt-Fruchter Chair-Window Test was used as a measure
of field-dependence. A low score on this instrument should be associ-
ated with negative attitudes toward the disabled, as indicated by low
scores on the Granofsky Pictures and the ATDP scale.
2. The Social Acquiescence Scale was used to measure social con-
formity. A high score on this scale should be associated with negative
attitudes toward the disabled.
3. The California F Scale was used as a measure of authoritarian
tendencies. A high score on this scale should be associated with nega-
tive attitudes toward the disabled.
4. The Barron Ego-Strength Scale was used to measure ego-strength.
A low score on this scale should be associated with negative attitudes
toward the disabled.
5. The Homonym Test and the Body-Cathexis Scale were used to
measure body-concern and body-satisfaction respectively. A high score
on the Homonym Test and a low score on the Body-Cathexis Scale should
be associated with negative attitudes toward the disabled.
The subjects used in this experiment were female college students
who were enrolled in either of two beginning psychology courses.
College students were selected as subjects for three basic reasons:
1. They possess sufficient intellectual and verbal ability
to understand the directions to the various tests and the items
contained in each, and to respond adequately to the demands for
expression inherent in several of the instruments.
2. It was felt that they represented a reasonably healthy,
statistically normal population which, as a group, would not
be expected to harbor attitudes in either an extreme positive
or negative direction.
3. They were accessible due to the fact that they were required
to participate in a certain number of psychological experiments.
In order to keep the subject population as homogeneous as possible
with regard to sex, thus avoiding any additional variance that might
result from differences in performance due to sex of the subject, it
was decided to limit the study to subjects of one sex. Since the
Granofsky Pictures Test, one of the criterion instruments in the study,
was originally validated on female subjects, it was felt that females
would be more appropriate subjects than males.
The class rolls for Psychology 201 and 202 were obtained, each
girl's name in the two classes was recorded, and an attempt was made
to find the mailing addresses of as many of these potential subjects
as possible. A total of 295 mailing addresses were eventually obtained,
and a letter was sent to each student.
Since a copy of the general form letter appears in Appendix A, it
will be stated here only that the letter informed them of the following:
1. Their selection as participants in the study.
2. The time and place that they were expected to attend.
3. The general nature of the experiment.
Of the 295 subjects who originally received letters, 41 did not
participate in the study. The majority of these 41 informed the in-
vestigator that they had fulfilled their experimental requirements for
their particular course and did not wish to participate in this study.
A small minority did not attend the experiment and failed to inform the
examiner in advance that they would not participate.
Of the remaining 254 subjects, four neglected for unknown reasons
to take at least one of the instruments that were included in the
battery. In addition, the tests of ten other subjects were disregarded
due to the incompleteness and/or lack of seriousness with which various
tests (in most cases the Granofsky) were taken. The test data gathered
from the remaining 240 subjects were considered adequate and valid for
the purposes of this experiment.
Although the subjects ranged in age from 17 to 44, 92% of them
(N=221) were aged 18 through 20. The mean age was 19.55 years, and
the mode was 19 (N = 80).
When the data collection was completed, a second letter was sent
to each participant explaining the purpose and rationale of the study
and thanking them for their cooperation. A copy of this letter
appears in Appendix A.
B. Collection of Data.
The data from the 240 subjects were collected by dividing them in-
to large groups and administering the complete battery to each group In
one evening. Four sessions of group testing were needed to complete
the administration of tests to all of the subjects. The sessions were
held approximately one week apart, with an average of 60 participants
per session. Although the testing time ranged from 105 to 165 minutes,
depending primarily on how long it took subjects to complete the Gran-
ofsky, most subjects completed all nine of the tests In approximately
A definite order was followed in the administration of the instru-
ments. This order was as follows:
1. Secord's Homonym Test.
2. Barratt-Fruchter Chair-W-indow Test.
3. Barron's Ego-Strength Scale.
4. Bass' Social Acquiescence Scale.
5. California F Scale.
6. Marlowe-Crowne Social Desirability Scale.
7. Granofsky Pictures Test.
9. Body-Cathexis Scale.
Secord's homonym Test was administered first to insure that sub-
jects' responses to the stimulus words would be as naive and free of
set as possible. Since several of the instruments centered around,
or were concerned with, aspects of the body, it was felt that administer-
ing this test later in the battery would influence the results obtained
Because it requires rather intense concentration and sustained
attention, the Chair-Window Test was administered as early as possible,
while the subjects were still very alert. It was felt that immediately
following the Homonym Test was the best time to administer this instrument.
The next four instruments were administered in their particular
order for no definite reasons. However, since it was considered import-
ant to administer the Homonym Test and the Chair-Window as early as
possible, and because it was felt that the Granofsky Pictures Test
would drain most of the subjects' motivation to perform, it seemed logi-
cal to place them between the first two tests and the Pictures Test.
Also, since they are all relatively short and do not require a great
deal of effort from a subject, it was felt that the subject could still
be reasonably fresh by the time she was to begin responding to the
Pictures Test. Approximately 60 minutes had passed by the time the first
six instruments had been administered.
The ATDP and the Body-Cathexis Scale were kept until last be-
cause the experimenter felt that the focus of these tests (disabled
people and the state of one's body respectively) might possibly inter-
fere with subjects' performances on the Granofsky had they been ad-
ministered prior to It.
C. Scoring of Data.
I. Objective Tests. All instruments in this battery, excepting
the Granofsky Pictures Test, were objective tests of various kinds.
The method for scoring these tests has been previously discussed in
the INSTRUMENTS chapter.
These tests were scored by the experimenter and an undergraduate
assistant. It was not felt that each test must be scored by both in-
dividuals because of the clearness and simplicity of the various scoring
procedures. However, 20 protocols of each instrument, originally scored
by the assistant, were rescored by the experimenter to serve as a
measure of the accuracy and reliability of the original scoring. On
these 160 protocols (8 tests, 20 protocols per test), containing approxi-
mately 8,000 items, only 14 errors were found. It was felt on the basis
of this negligible percentage of errors (less than .002%), that the
reliability of scoring the objective tests has been demonstrated to be
sufficient for the purposes of this study.
2. Projective Tests. The Granofsky Pictures Test, as has been
previously discussed in the chapter on INSTRUMENTS, is a relatively
unstructured, open-ended test, requiring a paragraph-length response
from the subject. Since the subject must invest his own attitudes,
feelings, and beliefs into his response rather than merely responding
Yes or No, True or False, etc., to a statement, this instrument can be
considered to be a projective technique.
Since responses of a subjective, open-ended nature are elicited
by the test, a problem of translating the material into measures en-
abling quantitative analysis arises. Although the general method of
scoring these protocols was discussed in the INSTRUMENTS chapter, and
thus will not be presented again, it is necessary to demonstrate, simi-
larly to the objective tests, that this test was reliably scored in the
present study. Unless the Pictures Test can be shown to have been
scored in a consistent, communicable manner, its value as a measure of
the criterion (attitudes toward disabled people) as seriously lessened.
In order to show that the basis upon which subjects' responses to
pictures of visibly disabled people was consistent, undimensional, and
capable of being communicated to other scorers, the scores arrived at
by the experimenter were correlated with those of another judge who
was following the same general directions for scoring. Since it is
necessary in scoring these responses not to take them at face value,
but rather to look for deeper, dynamic attitudes and feelings, it was
felt that the second judge should be experienced in interpretations on
this level. Ideally he should be of at least as advanced a level of
training and experience as the experimenter. For this reason, a fellow
intern in Clinical Psychology was selected for the task.
The two judges first separately scored subjects' responses to the
pictures on 15 randomly selected protocols (180 stories) with their
only guideline being the "Instructions to the Raters" (Appendix B),
outlined in Granofsky's (1956) original study. There had been no dis-
cussion between the raters concerning scoring procedures prior to this.
On this first training session of 15 protocols, the judges agreed
on 79 of 180 possible responses (43%), with a correlation (Pearson
product-moment) of .67 in their scoring (complete data on inter-scorer
reliability is found in Appendix C). Although this was encouraging
evidence that the system for scoring these responses could be communi
cated from one rater to another, it was felt that the correlation would
have to be considerably higher if this instrument was to be a useful
Following this initial scoring of protocols, the judges conferred
and discussed their rationale and frame of reference leading to their
scoring of the various responses. Particular attention was given to
those responses which varied greatly In the scores given to them by
A second training session of 15 protocols was then randomly selected
and separately scored In the same manner as the first group. This time
there were 92 agreements (51%), and a correlation of .76 between scorers.
Although this Is as high as the interscorer reliability recorded by
Granofsky (1956) in his original study, it was felt that it was possible
to arrive at a still higher level of scoring consistency, thus further
insuring that the criterion would be represented as reliably and
consistently as possible.
The two judges then conferred again, following much the same pat-
tern as the previous discussion. After this second meeting, it was felt
that the judges were ready for a final, 'official" scoring session.
This time 40 responses were randomly selected from the protocols and
separately scored by the judges. On this third session, agreement was
found on 24 (60%) of the responses, and a correlation of .84 was re-
corded. This is the same inter-rater reliability that Masson (1963)
reported, and was considered sufficiently high for the purposes of this
The experimenter, having demonstrated that the responses were
being consistently and undimensionally scored, then proceeded to score
all of the protocols. The subjects were randomly divided into initial
and cross-validation samples after all of the tests, both objective
and projective, had been scored.
D. Analysis of Data.
The analysis of the data that had been collected and scored could
be viewed as a two-step procedure. First, it was necessary to arrange
the data in such a way that the appropriate statistical procedures could
be performed. Secondly, it was necessary to apply these statistical
methods to test the initial hypotheses and enable the experimenter to
derive and evaluate the more subtle interactions that were present in
Since the plan of the study was to test the initial hypotheses that
had been developed on the basis of previous literature, extract from
the data the more subtle interactions occurring between variables,
develop more refined hypotheses based on these interactions, and test
these hypotheses, It was felt that two distinct samples would be needed.
The Initial sample would be instrumental in testing the initial hypothe-
ses and formulating the second-level, interactive ones. The second, or
cross-validation sample would be used to evaluate this second group of
hypotheses. Thus, the first step in the analysis of the data was to
divide it into two large groups, or samples.
This division was done in random fashion by assigning a number to
each subject, placing these numbers in a container, and drawing them
blindly until the samples were complete. All data from all subjects
had been collected and scored before this division into the two samples
was made. It was decided to include 150 subjects in the initial sample,
and 90 In the cross-validation sample.
Since sample size is inversely related to the size needed in a
statistical relationship for that relationship to be significant, it
was felt that it would be appropriate for the initial sample to be
larger than the cross-validation sample. That is, since the initial
sample was seen primarily as a means of developing and refining hypothe-
ses, making its size very large would result in more statistically sig-
nificant relationships to evaluate in the second sample. The cross-vali-
dation sample, being smaller, would require statistical relationships
to be larger if they were to be significant. Thus, any relationships
which might be found to be significant solely by chance in the initial
sample, because of the size of the sample, would most probably be dis-
carded in the cross-validation sample because of the more stringent
criteria of significance due to the smaller sample size. Those re-
lationships which were "true" or meaningful (significance due to actual
relationship of the variables rather than to chance or sample size)
would be more likely to hold up upon cross-validation.
Thus, the initial sample was seen as a means of developing, stating,
and producing a number of relationships which would hopefully be meaning-
ful as well as significant. The cross-validation sample was viewed as
a means of supporting or contraindicating the findings of the first
sample. The relative sizes of the two samples were chosen with these
goals in mind.
Since one of the statistical procedures which was to be used re-
quired that subjects in a particular sample be divided into high,
medium and low on the basis of certain variables, the number of subjects
in that sample would have to be sufficiently large to permit adequate
tests of the hypotheses after the divisions were made. Since a sample
size of 30 is generally considered large enough for this, an overall
N of 90 was felt to be sufficient to provide tests of the three subdivi-
sions of data.
The initial sample was made larger than this minimum of 90 because
of the previously stated purpose for which that sample was intended.
That is, because the primary purpose of the initial sample was to dev-
elop hypotheses to be retested with more stringent criteria for signifi-
cance, the larger size of that sample made it more likely that any
meaningful relationships present in the original data would be brought
to light. At this stage of developing hypotheses, it was felt that it
was better to include more statistically significant interactions than
might be meaningfully warranted, than to be overly stringent and thus
unnecessarily risk the exclusion of some relationships which might be
meaningful and which might be upheld in the cross-validation sample.
Thus, an initial sample of 150 subjects was decided upon, resulting in
subdivisions of 50 for certain statistical analyses.
The five basic, original hypotheses were tested by simply correl-
ating the predictor variables with the two criterion instruments. By
also correlating the predictor and criterion variables with each other,
an intercorrelation matrix was arrived at which provided measures of
the extent to which each variable (or instrument) was related to every
other variable. The correlation coefficients in this matrix were based
on a sample of 150. Although these correlations provide only a limited
amount of information in themselves, they do provide data which, with
more refined procedures, can be viewed, analyzed, and interpreted In
such a way as to generate more refined hypotheses for testing.
In this study, a procedure using moderator variables (Saunders,
1956), was utilized to break down the data In such a way that inter-
active effects could be studied and hypotheses developed on the basis
of these interactions. The particular situation in which the use of
moderator variables is applicable Is that in which the relationship
between two or more given variables is found to vary as a function of
changes In the value of a character of one or more other variables.
For example, if Variable A Is found to minimally correlate with
Variable B, subgroups may be isolated within this overall relationship
which show highly differential patterns of validity or degrees of rela-
tionship. The best way to isolate these subgroups, thus in effect
sorting heterogeneous aggregates of individuals into homogeneous ones,
may be through the use of a third, moderator variable (e.g., Variable
C). Thus, if Variable B is divided into subgroups of high, medium and
low on the basis of the relative performance of subjects on Variable C,
and then the scores of the subgroups are correlated with their corres-
ponding scores on Variable A, some highly varying relationships can be
extracted. These interactions can then be tested out on a subsequent
sample to help determine whether a significant, meaningful relationship
exists between the variables in question.
Thus, if Variable A Is found to correlate .10 with Variable B, we
have only the information that these variables do not appear to be
related to any significant degree. However, if we divide Variable B
into subgroups of high, medium and low on the basis of subjects' per-
formances on Variable C, and then correlate scores between Variable A
and Variable B within the three subgroups, we might find that the re-
lationship varies greatly under these different conditions. For example,
In the subgroup of individuals who scored "high" on Variable C, Variable
A and Variable B may correlate .30. In the "medium" group the correl-
ation might be .03. In the low group it might be -.48. Thus, the
relationship between A and B could possibly vary dramatically under
different conditions, providing a great deal more useful information
than the initial overall correlation between the variables. This was
the technique used in this study to study interactive effects between
the variables, and to break down the overall correlations into relation-
ships which often varied considerably. All Interactions found to be
significant in the initial sample were then stated as hypotheses and
re-examined in the cross-validation sample.
In Banas' (1965) discussion of moderator variables, he states that
they can be chosen on either an empirical or a rational basis, with
rational selection being preferred. Since each of the predictors in
this study has been thought to be of theoretical importance in the
dynamics of attitudes toward disabled people, it is likely that the
resultant attitude would be a function of interactions among them.
Thus, it was felt that utilizing the predictors as moderators for each
other would provide the greatest amount of pertinent information. To
this extent, selection was rational in the present experiment.
The attempt was made in this study to divide the samples into equal
subgroups, that is, into subgroups of 50 subjects each in the initial
sample and of 30 subjects each in the cross-validation sample. However,
this was not always possible due to tie scores in many instances.
For example, if a subgroup in the initial sample consisted of 48
subjects, and the next score was attained by eight subjects, those eight
subjects would begin the next subgroup. Rather than arbitrarily placing
two of those subjects into the preceding subgroup, it wad decided to
make the divisions "natural" and as close to even as possible. Thus,
a typical division might be 52, 49, 49 for high, medium and low, rather
than 50, 50, 50.
The final statistical methodology used in this experiment was
originally described by Hotelling (1936) and termed the "canonical
correlation" by him. This concept is defined by Cooley and Lohnes
(1962), as the maximum correlation between linear functions of two
sets of variables, such as two sets of measurements made on the same
subject. In this experiment, the two sets of variables or measurements
would be the predictor and criterion instruments.
Thus, a linear function is developed for each set of variables.
The problem Is to find two sets of weights which, when applied to the
variables, maximizes the correlation between the linear functions.
Thus, a correlation coefficient is derived which provides the infor-
mation of whether one set of variables (e.g., the predictors) is sig-
nificantly related to the second set (e.g., the criteria). In addition,
through the process of weighting, some measure of the importance of the
variables in each set is arrived at. It is thus possible in the present
study to attain some estimate of the relative importance of the pre-
dictors, and thus of the theoretical approach which each represents in
contributing to attitudes toward disability.
Since both multiple predictors and criteria were present in this
experiment, the canonical correlation was considered most appropriate
for the purposes discussed above. If there had been only one criterion,
the problem would have been one of multiple regression.
Due to their complexity and length, all analyses in the present
study were done by means of electronic computers.
RESULTS OF INITIAL SAMPLE
The maximum canonical correlation of the initial sample was .47,
which is significant at the .001 level. Therefore, there Is at least
one significant way in which the predictors relate to the criteria.
The personality constructs resulting from the various theoretical ap-
proaches can be said then to be significantly related to attitudes to-
ward visibly disabled people as they are measured in this study. Since
the number of possible pairs of linear combinations is equal to the
number of predictor or criterion measures (whichever is smaller), a
second combination was developed in this study, which had two criterion
measures. However, the correlation in this instance (.20) was not
CANONICAL VECTORS OF THE INITIAL SAMPLE
-.68 Authoritariantsm .87 ATOP
.46 Social Desirability .26 Granofsky Pictures
.02 Social Conformity
The contribution which the individual variables made to the sig-
nificantly related canonical variates is presented in Table 1. The
loadings reveal that authoritarianism is the primary variable in pre-
dicting attitudes toward disability. The tendency to respond in a
socially desirable manner is the most important positive predictor,
suggesting that some of the relations between predictors and criteria
in this study may be heavily influenced by this response bias,, with
field-dependence and body-satisfaction also appearing relatively high
in predictive importance.
The ATDP scale appears considerably more important than the Gran-
ofsky Pictures Test as a measure of the criterion in this sample, as
its relative weight is much greater than that of the latter instrument.
On the basis of the nine-variable correlation matrix (Table 2)
derived from the initial sample (N = 150), most of the initial hypothe-
ses were supported to some degree. Hypothesis three was most strongly
supported as authoritariansim, as measured by the California F Scale,
correlated in a negative, highly significant (p(.01) manner with both
measures of attitudes toward disabled persons. This would indicate,
consistent with the hypothesis, that highly negative attitudes toward
disabled people are directly related to the degree of authoritarianism
on the part of the non-disabled.
Hypothesis 5, which dealt with the relationship between the non-
disabled individual's feelings about his own body and his attitudes
toward physically disabled people, received somewhat inconsistent support.
Body-satisfaction, as measured by the Body-Cathexis Scale, was found to
correlate significantly (p,.05) with both criterion measures, thus
supporting Hypothesis 5a, but no significant relationship between
anxiety and concern about one's own body, as measured by the Homonym
Test, and attitudes toward the disabled was found, thus not supporting
Hypothesis 5b. Thus, the data would suggest that some positive re-
lationship exists between the degree of conscious satisfaction an in-
dividual has with his body and the manner in which he reacts to disabled
people; the greater the self-satisfaction, the more acceptant and posi-
tive the reaction to disability. On the basis of the initial data,
however, there appears to be no relationship between any type of more
basic, unconscious somatic concerns and attitudes toward disabled
people. Conscious attitudes and feelings about one's body, then, would
appear to be more important in determining one's reaction to this type
Hypothesis 1 (the more field-dependent an individual is, the less
acceptant of the disabled he will be) and 4 (the less ego-strength an
individual possesses, the less acceptant he will be) were also supported
by the data, although in neither case could the support be considered
strong. A significant, positive relationship (p<.05) was found between
field-independence, as measured by the Chair-Window Test, and one of
the criterion measures for attitudes toward disability (ATDP scale),
but the second criterion measure (The Granofsky Pictures) was virtually
unrelated (r = .02) to field-independence as it was measured here.
A very similar situation is found with the relationship between
ego-strength and attitudes toward disabled people. Once again, the
CO -." C0
- O O LO
0 U 4.J
\D IFCO m1
c 4 E 0
Ec i c u
) 0 O -c ro
C. 4- 4. 4-
O C L CO)
0 0 cL wo 0 41c
V \7 I V)
Q. m. "0 3 0 ( u I
; U10 -0
0 o < U
U. fL < ,, m
- N m --f L1n vo3 r- wo cm>
- N r .- .\
ATDP correlated significantly (p<.01) with the predictor variable,
as measured by Barron's Ego-strength Scale, but only a minimal cor-
relation could be found between the Granofsky and the Ego-Strength
Scale. Thus, there is slight, tentative support for the possibility
of positive relationships between ego-strength and attitudes toward
disabled people. The relationship between ego-strength and attitudes
toward disability, as well as that between body-satisfaction and atti-
tudes toward the disabled, is made to seem ever more tentative, how-
ever, by the fact that social desirability correlated .22 (p4.01) with
the ATDP, .19 with the Ego-Strength Scale (p<.05), and .17 (p<.05)
with the Body-Cathexis Scale, suggesting that some degree of the cor-
relations of these predictor variables with the ATDP scale might be
attributable to this response style.
No support was given to Hypothesis 2 (the more socially conforming
an individual is, the less acceptant of the disabled he will be).
Although the relationship between social conformity, as measured by
Bass' Social Acquiescence Scale and the two criterion measures was in
the right direction (i.e., a negative correlation between social con-
formity and attitudes toward the disabled), It was not of such a degree
as to be considered significant at the .05 level.
Thus, as a way of summarizing the findings with respect to the
initial hypotheses, they are restated below with a note added to each
describing whether each particular hypothesis was supported by the data,
and if so, the degree of the support.
The more field-dependent an individual is, the less acceptant of
the disabled he will be. Slightly supported by the data, as a cor-
relation of .20 (pc.05) was found between the Chair-Window Test (a
measure of field-independence) and the ATDP scale.
Hypothesis 2. The more socially conforming an individual is, the
less acceptant of the disabled he will be. Not supported by the data.
Hypothesis 3. The more authoritarian an individual is, the less
acceptant he will be of disability. Strongly supported, as the Calif-
ornia F Scale correlated -.331 and -.251 (both p<.01) respectively
with the ATDP and the Granofsky Pictures.
Hypothesis 4. The more ego-strength an individual possesses, the
more acceptant of disability he will be. Slightly supported, as a
correlation of .23 (p4.01) was found between the Ego-Strength Scale
and the ATDP.
Hypothesis 5. (a) The more consciously satisfied an individual
is with his body, the more acceptant of disabled persons he will be.
Strongly supported, as the Body-Cathexis Scale was found to correlate
.20 and .16 (both p<.05) with the ATDP and the Granofsky Pictures, re-
spectively. (b) The more anxious and concerned an individual is about
his body, the less acceptant of the disabled he will be. Not supported
The highest intercorrelation between any two of the predictor
variables was that of authoritarianism with social conformity, which
yielded a coefficient of .46 (p<.01). This would suggest that conformity
to society's dictates is positively related to the degree of authori-
tarian attitudes present within an individual, i.e., highly conforming
people would also tend to be highly authoritarian.
This finding is consistent with the views of Adorno, et al.
(1950), who, in discussing the relationship between authoritarianism
and conformity, state:
Prejudiced subjects tend to report a relatively harsh
and more threatening type of home discipline which was ex-
perienced as arbitrary by the child. Related to this is a
tendency apparent in families of prejudiced subjects to base
interrelationships on rather clearly defined roles of domi-
nance and submission in contradistinction to equalitarian
policies. Family relationships are characterized by fearful
subservience to the demands of the parents and by an early
suppression of the impulses not acceptable to them.
The goals which such parents have in mind in rearing
and training their children tend to be highly conventional
The status-anxiety so often found in families of prejudiced
subjects is reflected in the adoption of a rigid and ex-
ternalized set of values: what is socially acceptable and
what is helpful in climbing the social ladder is considered
"good," and what deviates, what is different, and what is
socially inferior is considered "bad". . (Pg. 385.)
Recent studies by Bass (1956), Shaw (1961), and Vaughn and White
(1966) have also reported significantly positive relationships between
these two variables.
Authoritarianism, in this matrix, was found to be negatively
related (r = -.30) to ego-strength (p<.01). This also is consistent
with the views expressed by Adorno, at al., who state:
Low scorers (i.e., low authoritarians) often tend toward
a more successful integration of the various aspects of their
personalities, they tend to remain less immature and less in-
fantile. They thus turn out to have more capacity for sus-
tained effort, more ability to postpone pleasure for the sake
of internalized values, more ability to assume responsibility,
and more emotional maturity. The absence versus the presence
of any or all of these characteristics may be summarized
as a "weak" versus a "strong" ego.
The fact that low scorers manifest relative strength of
the rational tendencies as compared to the irrational may be
due to their attempt to master and sublimate rather than
escape the unconscious. Thus, the low scorer's adaptation
to reality is more flexible in spite of the more open con-
flict and anxiety which accompanies the greater awareness
of existing problems. (Pg. 457.)
Barron (1953a) has reported findings which also support the re-
sults of the present study.
Although a significant (p<.01) positive relationship is found
between body-satisfaction and ego-strength (r = .32), the relatively
high correlation of both of these variables with the socially desir-
able response set (both p<.05) suggests that the strength of this
relationship may be due to a large extent to the fact that they are
both susceptible to this bias.
Since four out of the five hypotheses were to some degree sup-
ported (although the possibility existed that a socially desirable
response set may have contributed somewhat to some of the significant
relationships), and all of the correlations between predictors and
criterion were in the predicted direction, it was felt that a closer
examination of the data was certainly warranted. A more detailed look
at the data would also provide a great deal more information and clari-
fication of the interrelationships of the nine variables.
As has been previously explained, a method of arranging and analyz-
ing the data utilizing moderator variables was the procedure by which
this examination would take place. By means of this procedure, 126
correlations between the various predictors and each criterion, thus
resulting in a total of 252 correlation coefficients, could be deter-
mined. Each predictor variable was correlated with each criterion
variable when each of these criterion variables was divided into high,
medium and low on the basis of subjects' performances on each of the
predictors, which in effect served as moderators for each other. For
example, a variable such as social conformity was correlated with
attitudes toward disabled people when this criterion was divided into
high, medium and low on the basis of subjects' relative standing on
each of the other predictors, such as body-satisfaction, ego-strength,
authoritarianism, etc., in turn, until each predictor had served as a
moderator for every other predictor.
The relationship between each predictor and each criterion instru-
ment could be viewed under 18 different conditions (six other predictors
serving as moderators, each under high, medium and low conditions), in
addition to the initial overall correlations previously discussed.
The information concerning these Interactions is presented in
Tables 3 to 9. Of the 252 coefficients attained by this method, 49 are
significant at the .05 level. Twenty of these significant correlations
were significant at the .01 level. Since only 13 correlations would
be expected to be significant by chance at the .05 level, and only
three at the .01 level, the number that were found to be significant
represents a genuine departure from chance. Since only one significant
correlation was in a direction opposite to that which was predicted,
evidence is provided for the meaningfulness, in addition to significance
of the relationships.
The finding (on the basis of the initial correlation matrix and
canonical correlation) that the single most Important predictor of
attitudes toward disability was the degree of authoritarianism of the
non-disabled was supported by the subgroup correlations. Of the 49
statistically significant relationships isolated by this technique
between predictor and criterion instruments, 17 had authoritarianism
as the predictor. (See Table 5.)
In several instances the use of a moderator variable did not sig-
nificantly change the overall relationship between predictor and
criterion. This was most apparent when either social conformity
(Table 2) or body-concern (Table 8) was the predictor. In others the
relationship was considerably clarified. For example, the overall
relationship between ego-strength and attitudes toward the disabled
was one which originally appeared minimally statistically significant
and lacking in Informational value. When subgroups were Isolated by
the use of moderator variables, the relationship was seen to vary
considerably depending on the condition under which it was examined.
An example of this can be seen by consulting section C of Table 6.
As shown in this table and in Table 2, the overall correlation
between ego-strength and the respective criterion measures was .23
(p<.01) with the ATDP scale and .09 with the Granofsky Pictures Test.
When authoritarianism was the moderator, as shown in Table 6, Section
C, a definite pattern in the relationship between the predictor and
criteria emerges. Thus, for subjects who are highly authoritarian,
LL. U .1-
U l O ,
0 0 M
O U. -..
) 0 1
O \D0 0m M O
CM 04 0 000
O ** 0I
- M 0\D
- O 00--
LL C %
LI '0 Q
0 u m
V) *- -
>. L. -.
0 U Ca
1 Mi .
Ul L. -,
.0 C U1
r -.- -t
00 r-.. cl
CO I. 1
I I I
4- CO. -O
U SB U)
>. u ----
0 4 U)%
IN f .4'
*' 4- 0
*- .3' ft
- t ft
C.I I I-
CM r. e
o D 0A N
0 c 0
U do CO 3" CO 0 -
0 L-ut -O o D oo
( * * *
% >. .M
0-- In w
-J C C_ CMD 100
C) I I
In u >0 to C *,n *
I- 0 Uo FCo OO 01
S *c** *-* *
In 0 4f I
Q 1U in
Im- C U m1p mo
U 0 rA o C1 m a 0 c
SL- N ON -CL
S.V n r *
-. l- 1'
I 1O=- L. ) r G0 0 4.
04-- * *
> Z. CC a C
w 0 0
in 0 o- -
z <*- c:- com-s- c1- .0
o L.-o 04-0
I- 4 l0. )
'U 0 0
o0 o -D D
* * * *
CoM0 C L
CON- nC -0
* . *
0 M 0
o V m mN CO N
l>. U CU0 QT aDCOr-
10 e ca 04 0 0 0 N
L1 -- * * *
in 1 I I
-. a "-l r, 1
1 r_ 0- 0 C C4 OM Lt
0 m % . .. .
V I I I I
1/ I u I I
OI oCa N A C-o -CO a L.
S- Lo Or CD 0- L
I 4-' W CL I I SI
Q I I II I
M Eu --
i - U . . .
I 0 I I I
4JL I I0
I U E --0 OO
w 'a 1 4, -
0 0 m-
mI i . l
0 I0 4- C
a 3- N 0 0
oL- I I i -
o 0 0 Z
v o o o
S- C L.
> I- Z
correlations of .28 (p<.05) between the ATOP scale and ego-strength
and of .15 between the Granofsky and ego-strength are found. For
subjects who are moderately authoritarian, the relationship changes to
a negative one (-.09 correlation with ATDP, -.19 with Granofsky). In
the final subgroup, composed of low authoritarians, the relationship
reverts to one that is very similar to that of the high authoritarian
group. In this instance, correlations of .27 (p(.05) and .17 are
found between the ego-strength scale and the ATDP and Granofsky respec-
Thus, the moderator in this instance was instrumental in eliciting
a pattern in the relationship between the predictor and criteria that
added considerable informational value to the original overall corre-
Since the plan of the present study was to develop more refined
hypotheses of the relationships between the various predictors and
criteria on the basis of thoriginal sample, any discussion or attempted
explanation of the findings at this point will be very limited. Rather,
the current emphasis will be on the presentation of the hypothese re-
sulting from this sample, with comments of an elaborative or explana-
tory nature being presented as appears appropriate in the course of
stating the refined hypotheses.
The hypotheses, which are listed below, represent significant inter-
actions between various predictors and at least one of the criterion
instruments in the initial sample. That is, the correlations that
were found to be statistically significant have been selected from the
data and restated in the form of hypotheses to be re-evaluated in the
cross-validation sample. They are as follows:
1. Social conformity (Table 4) is negatively related to atti-
tudes toward disability when the non-disabled individual is characteri-
zed by either of the following: (a) Moderate body-concern, (b) Moderate
2. Body-satisfaction (Table 7) is positively related to attitudes
toward disabled people when the non-disabled Individual is characterized
by the following: (a) High ego-strength, (b) Moderate body-concern,
(c) Moderate field-independence, (d) Moderate social conformity, (e) High
3. Ego-strength (Table 6) is positively related to attitudes to-
ward disabled people when the non-disabled individual is characterized
by the following: (a) High field-independence, (b) Low authoritarianism,
(c) Low social conformity, (d) Moderate body-satisfaction, (e) High body-
concern, (f) High authoritarianism. Also, a significantly positive
relationship between ego-strength and attitudes toward disability is
associated with low social approval needs (as shown by low scores on
the social desirability scale).
4. Authoritarianism (able 5) Is negatively related to attitudes
toward disabled people when the non-disabled person is characterized by
the following: (a) Moderate field-independence, (b) Low field-inde-
pendence, (c) Low ego-strength, (d) Moderate body-satisfaction, (c) Low
body-satisfaction, (f) Moderate body-concern, (g) Moderate social con-
formity, (h) Low social conformity, (i) High ego-strength.
5. Body-concern and anxiety (Table 8) is negatively related to
attitudes toward disability when the non-disabled individual is char-
acterized by the following: (a) Low ego-strength, (b) Moderate body-
stisfaction, Body-concern was found to be positively related to atti-
tudes toward disability when the non-disabled was characterized by
high social approval needs (reflected in the tendency to score highly
on the social desirability scale).
6. Field-independence (Table 3) is positively related to attitudes
toward disability when the non-disabled individual is characterized by
the following: (a) Moderate ego-strength, (b) Moderate body-concern
(c) High Authoritarianism. Low social desirability response tendencies
are also associated with a positive relationship between this predictor
and the criteria.
The social desirability response tendency (Table 9) was found to
be positively related to attitudes toward disability when the non-dis-
abled individual was characterized by: (a) Moderate social conformity,
(b) Moderate field-independence, (c) Low field-independence, (d) Low
ego-strength, (d) High authoritarianism. Thus, a positive, significant
correlation between social desirability and attitudes toward disability
would seem to be associated with the basically "unhealthy" poorly inte-
grated individual who behaves in an acceptant manner because of his
need for social approval rather than for any genuine interest in the
welfare of the disabled.
it appeared that one subgroup of each of the moderator variables
was particularly effective in enhancing the relationship between the
predictors and criterion. For example, the characteristic of being
moderate in degree of field-independence was associated with a signi-
ficant negative correlation between social and conformity and attitudes
toward disability, and significant positive correlations between this
criterion and body-satisfaction, authoritatianism, and social desira-
bility motives. Other medium subgroups which proved effective were
those of social conformity, body-concern, and body-satisfaction. Ego-
strength and social desirability were the variables whose low subgroups
resulted in more significant relationships than their high or medium
subgroups. Authoritarianism was the only variable in which the high
subgroup was most important in precipitating significant correlations.
RESULTS OF CROSS-VALIDATION SAMPLE
As has been previously stated, the primary purpose of the second
cross-validation sample was to evaluate the hypotheses which had been
developed on the basis of the initial sample. Thus, the major focus
of this chapter is the presentation of data pertinent to the inter-
active hypotheses which were stated in Chapter VI.
Rather than presenting the complete moderator analysis of the
cross-validation sample in this chapter, as was done for the initial
sample in the previous chapter, only those interactions which were
found to be significant in the initial analysis and restated as
hypotheses are examined. The complete analysis can be found in
In this second moderator analysis, 37 subgroup correlations of a
possible 252 were found to be significant at the .05 level. Of these
8 were significant at the .01 level. Although these figures are some-
what smaller than those presented following the analysis of the initial
sample, they are considerably larger than the number of significance
that would be expected by chance (13 at the .05 level, 3 at the .01
level). The smaller number of significance was in fact predicted
because of the more stringent criteria for significance due to the
smaller sample size in this sample.
The data comparing the relationships that were found to be
significant in the initial sample with those same relationships in
the cross-validation sample ae found in Table 10. Of the 40 hypothe-
ses resulting from the 49 significant correlations found in the first
sample, 12 were supported in the second sample. Eight of these sup-
ported hypotheses had authoritatianism as a predictor, thus reinforc-
ing the initial sample's finding that it was the most important of the
5 personality constructs, and thus of the 5 respective theoretical
approaches in predicting attitudes toward disability. Since an assump-
tion of this study was that the two criterion measures were viewed as
two aspects of the same thing (attitudes toward disabled people)
rather than as 2 independent, separate criteria, a significant corre-
lation between a predictor and either of the 2 criterion instruments
in the initial sample was considered to be supported if that predictor
was found to be significantly related to either of those instruments
In the second sample. An example of this is found in Table 10,
Section A. In the initial sample, social conformity was found to be
significantly related to attitudes toward disability, as measured by
the Granofsky Pictures, when the individuals were characterized by
medium body-concern. In the cross-validation sample, social conformity
was once again found to correlate significantly with attitudes toward
disability under that particular moderating condition, but in this
Instance the ATDP Scale was the criterion instrument with which it was
significantly related. Although the predictor correlated significantly
with different criterion Instruments in the two samples, the hypothesis
generated in this instance was considered to be supported.
It can be seen upon examining Table 10, Section A, that both
hypotheses concerning the relationship of social conformity to atti-
tudes toward physically disabled people were supported. Thus, it can
be stated with some degree of confidence that the degree of social con-
formity which is present in individuals who are moderately concerned
about their bodies or who are neither strongly nor minimally dependent
on their perceptual field for behavioral direction, is inversely re-
lated to their atittudes toward disability. That is, for these indiv-
iduals, low social conformity needs would be associated with positive
attitudes toward the disabled, and strong needs for conformity would
be indicative of aversion and discrimination of them.
In addition, 8 subgroup correlations involving social conformity
were found to be significant in the second sample that were not sig-
nificant in the Initial one. Since they cannot be given as much weight
as those correlations found to be significant in both samples, they
will not be examined in detail. This finding is mentioned primarily
to emphasize the extent to which this variable increase in predictive
importance in the second sample. The complete list of correlations
can be found in Appendix D.
In contrast to these results, none of the 5 hypotheses relating
body-satisfaction to attitudes toward disability were supported by
the cross-validation data. This is shown in Section B of Table 10.
In fact, the overall positive relationship that was found between the
Body-Cathexis Scale and the criterion in the initial sample, and which
is shown in the correlation matrix of that sample (Table 2), was
/) Z z z z z z
*- 4 D
CI 0 C
I I I
0 C C
ci.- e i
Z Z Z Z VZ ) ) V) Z ) V) Z
- LA N 3-
I I I I
(/) tI Z 4 )/ Z Z 2 z
0 %0 1 1 eI1 %0 %I %1
- N 0M n -
Ln M C4 CM
II I I
r. r. -
Z z z z z z
00 0- 0
-e0 f MN OfCN N-
mO -(o (MO -0
N COG%1 0
CN N 4 N C
N, rd -~
- LA 00co
0 OD -
-3- I- (M M 0 -
--0- 0 0
I I I I
- c rCl _-r L %D r- CO m
. 0 .
- 41 0
\O r-oo 0
C) .- I-
F *- .c
0) u 4-
0 0 0
) 01 <
- 4N c _.; LA
reversed in the second sample. It can be seen in Table 11, which
contains the correlation matrix of the cross-validation sample, that
body-satisfaction correlated -.08 with the Granofsky Pictures and -.07
with the ATDP Scale in that sample. This is contrasted with its cor-
relations of .20 and .16 with the respective criterion instruments in
the initial sample. This finding suggests that conscious satisfaction
with one's body is either not related in any meaningful way to one's
attitude toward disability or that an individual's rating of his body
is confounded by so many factors that it is unlikely that an adequate,
relatively "pure" measure of this relationship can be obtained. Since
the results will be discussed in detail in Chapter VIII, any attempts
to account for the results of this study in the present chapter will
be minimal. However, It can be stated on the basis of the Instruments
used that no relationship appears to exist between conscious satis-
faction with one's body and attitudes toward the physically disabled.
A similar conclusion is easily reached with regard to the re-
lationship between ego-strength and the criterion, as only one of nine
hypotheses developed from the initial sample was supported by the cross-
validation sample (Table 10, Section C). Thus, ego-strength, as repre-
sented by Barron's scale, appears to be of minimal value in predicting
attitudes toward disability. On the basis of the cross-validation
findings with the respective instruments, it would appear that only
in the case of highly field-independent individuals would degree of
ego-strength be an effective predictor of acceptance of disability.
Consistent with previous findings in the initial sample (correlation
matrix, canonical loadings, moderator analysis) authoritarianism re-
mains the most reliable predictor of attitudes toward disability,
that is, high authoritarianism has been consistently fonnd to be nega-
tively related to attitudes toward disabled people. Of the 11 hypothe-
ses developed from the initial sample relating this predictor to the
criterion measures, 8 were supported by the cross-validation data
(Table 10, Section D). These data suggest that Individuals who are
characterized by low body-satisfaction and a moderate degree of
anxiety and concern about their bodies, a tendency toward field-in-
dependence, a tendency toward having low social conformity needs, and
high ego-strength, would express attitudes toward disability which
would be inversely related to the degree of authoritarianism that they
could be characterized by. Thus, for these individuals, a low degree
of authoritarianism would be associated with positive attitudes toward
the disabled. Conversely, high authoritarian tendencies would suggest
negative, aversive attitudes on their part.
Of the three hypotheses relating body-concern, as measured by
Secord's Homonym Test, to attitudes toward disability, one was supported.
Thus, this variable, as in the cases of body-satisfaction and ego-
strength, appears minimally important as a predictor of attitudes. Only
when individuals are moderate in the extent to which they are consciously
satisfied with their own bodies, is unconscious somatic concern negatively
related to attitudes toward individuals with a physical disability.
(Section E of Table 10).
None of the four hypotheses generated by the final predictor
(field-independence) were supported by the data of the cross-valida-
tion sample, suggesting that no consistent relationships exist between
this variable as it is measured here and attitudes toward disability.
Also, none of the statistically significant subgroup correlations
between social desirability and the criteria that were found in the
initial sample were supported in the second sample. On this basis, it
would appear that there is an absence of any consistent interactive
relationships between this control instrument, the predictors, and
the criterion instruments. It should be noted, howe,.er, that the
social desirability scale was found to have a significant overall cor-
relations (r = .22, p.05) with the Granofsky Pictures Test in the
cross-validation sample (Table II). Thus, whereas in the initial
sample it appeared as if the ATOP scale might have been susceptible
to this response bias since it correlated significantly with the
Marlowe-Crowne scale in that sample, the Granofsky appears more likely
to have been Influenced by it in the second sample.
As in the initial sample, the Barron Ego-Strength Scale was found
to correlate significantly with the Social Desirability Scale. As is
shown in the matrix of the cross-validation sample, the correlation in
this instance was .24 (p.O05). Thus, once again this variable's re-
lationships with other variables must be evaluated in light of its
apparent vulnerability to the social desirability response tendency.
Since the Granofsky Test and Barron's E-S Scale are the only two instru-
ments to correlate significantly with social desirability in the second
sample, it would appear as if the influence of this control variable is
not extensive in that sample.
Other than the significant correlation found between social
desirability and the Granofsky Pictures, only two of the predictors
were found to be significantly related to attitudes toward disability.
The strongest relationships between any predictor and criterion were
once again found to involve authoritarianism as the predictor. In
the cross-validation sample, correlations of -.32 (p<.01) between the
California F Scale and the ATOP Scale and of -.26 (p(.05) between it
and the Granofsky were found. These results are extremely consistent
with the overall correlations between the measure of authoritatianism
and the two criterion instruments that were reported in the initial
Social conformity was also found to correlate significantly over-
all (p(.05) with both criterion instruments (r = -.24 with the ATOP;
r = -.21 with the Granofsky). The findings of the correlation matrix
of the cross-validation sample together with the findings derived from
the moderator analysis of that sample suggest that this instrument may
possess considerably more predictive power than was shown on the basis
of the Initial sample.
Body-satisfaction, ego-strength, and field-dependence, all of
which were to some degree significantly related to the criteria in the
initial sample, were not found to possess overall significant corre-
lations with either criterion instrument in the cross-validation sample.
Thus, the tentative importance attributed to these variables as predic-
tors of attitudes toward the disabled in the initial sample must be