Locus of control

MISSING IMAGE

Material Information

Title:
Locus of control its relationship to religious denominations and leader-follower perceptions of behavior
Physical Description:
ix, 148 leaves : ; 28 cm.
Language:
English
Creator:
Gaskins, Leslie Elwood, 1922-
Publication Date:

Subjects

Subjects / Keywords:
Control (Psychology)   ( lcsh )
Leadership   ( lcsh )
Psychology, Religious   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis--University of Florida.
Bibliography:
Includes bibliographical references (leaves 144-147).
Statement of Responsibility:
by Leslie E. Gaskins.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000081986
notis - AAJ7312
oclc - 05112693
System ID:
AA00003907:00001

Full Text









LOCUS OF CONTROL: ITS RELATIONSHIP TO RELIGIOUS DENOMINATIONS
AND LEADER-FOLLOWER PERCEPTIONS OF BEHAVIOR








By

LESLIE E. GASKINS


A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY








UNIVERSITY OF FLORIDA


1978













ACKNOWLEDGEMENTS

The author wishes to express his appreciation to the members of his

supervisory committee Professors H. Joseph Reitz, Jerald W. Young, and

Richard M. Swanson, not only for their patience and expert assistance

in completing this dissertation, but for their encouragement, guidance,

and friendship over the past four years.

Similarly, the author wishes to thank the members and ministers of

the churches who gave so generously of their time to provide the data

for this study.

An additional debt of gratitude is owed to the many members of the

staff and faculty of the University of Florida for their special con-

sideration and support, especially to Doctor Jack Feldman who served as

my advisor during my first years at the University.

Finally, this dissertation is dedicated to my wife, Anne, in grati-

tude for her love, devotion, encouragement, and sacrifice over the past

35 years.















TABLE OF CONTENTS


ACKNOWLEDGEMENTS . .

LIST OF TABLES . .

LIST OF FIGURES . .

ABSTRACT . . .

SECTION

I. INTRODUCTION . .

II. BACKGROUND AND RELEVANT LITERATURE

LOC and Religion .
LOC and Leadership .

III. METHOD . .

Subjects .
Rokeach's Values Scales .
The Consideration and Structure S
Biographical Data .

IV. RESULTS . .

LOC and Religion .
LOC and Leadership ...

V. DISCUSSION . .

LOC and Religion .
LOC and Leadership .

APPENDIX A: INSTRUMENTS . .

APPENDIX B: CORRESPONDENCE AND NOTICES .


REFERENCES .

BIOGRAPHICAL


c


SKETCH . .


Page

11

. iv

. vi

. vii


:ales
* .
. .

. .
* .
. .

. .
. .


1

6

6
19

31

31
41
43
43

45

45
62

82

82
87

102

139

144

148













LIST OF TABLES


Table


Page


18


1. Results of Factor Analyses of LOC Scales From
Past Research . . .

2. Median Differences Between Baptists and Other Denomi-
nations on Rokeach's Value Scale (Significance is Chi
Square Median Test) . . .

3. Relationships Between the Follower's LOC and His Per-
ceptions of the Leader's Behavior From Prior Research

4. Sample Population Characteristics . .

5. Comparison of LBDQ Form XII Scores for Ministers and
Other Leaders . . .

6. Correlation Matrix (Whole Sample) Rotter's Scale Versus
Modified Scale . . .

7. Correlation Matrix (Methodist Sample) Rotter's Scale
Versus Modified Scale . . .

8. One-Way Analysis of Variance LOC by Denomination .

9. Correlations Between LOC Subfactors and Rokeach's Values

10. Significant Multiple-Range Comparisons Between
Denominations on Value Variables . .

11. Hierarchial Regression of LOC with Possible
Moderating Variables . . .

12. Four-Factor Solution to LOC Scale . .

13. Comparison of Present Study Factor Analysis With
Previous Research Findings . .

14. Multiple-Range Comparison of Denominations By
LOC Factor I . . .

15. Hierachial Regression of LOC Factor I with Possible
Moderating Variables . . .


.









Table

16. Standardized Discriminant Function Coefficients for
the LOC Factors and Rokeach's Values . .

17. Denominational Centroids Based on Discriminant Functions

18. Confusion Matrix Comparison of Accuracy of Discriminant
Function With Prediction by Chance . .


19. The Relationship Between the Follower's LOC Factors
and Consideration/Structure at the Individual
Differences Level (Level Three) . .

20. Four-Factor Principle Component Solution for Combined
Scales and Consideration and Structure .

21. Three-Factor Solution for Combined Scales of
Consideration and Structure . .

22. Correlations Between Follower LBDQ Factors .. ...

23. The Relationship Between the Follower's LOC and
LBDQ Factors at Two Levels . .

24. Relationship Between the Follower's LOC and LBDQ
Factors at Two Levels . .

25. Correlation Coefficients of Supervisor and Follower
LOC Factors With the Follower's Perceptions of the
Leader's Behavior . .

26. Correlation Coefficients of Leader LOC With His
Perceptions of His Own Behavior . .

27. Correlation Coefficients of Follower LOC With the
Leader's Perceptions of His Own Behavior .

28. Correlation Coefficients of Leader and Follower LOC
With Leader and Follower Perceptions of Leader
Behavior . . .


. 68


. 70


. 71

. 72


* 75



* 78


* 80


. 81



92


Page


. 58

. 59


. 63


I .













LIST OF FIGURES


Figure Page

1. Plot of Denominational Group Centroids . 61








Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy

LOCUS OF CONTROL: ITS RELATIONSHIP TO RELIGIOUS DENOMINATIONS
AND LEADER-FOLLOWER PERCEPTIONS OF BEHAVIOR

By

Leslie E. Gaskins

December, 1978

Chairman: H. Joseph Reitz
Major Department: Management

While a great deal of research has been conducted on the personality

trait Locus of Control (the extent to which individuals perceive that

their rewards are contingent on their own behavior versus the extent to

which they feel that their rewards are controlled by fate, chance, or

powerful others), very little has been done to investigate the ante-

cedents of this trait. It appears possible that there might be a rela-

tionship between Locus of Control and religion in that many religions

differ in their teachings as to the degree of control man has over his

outcomes. If such a relationship between Locus of Control and religion

were found, there would be justification for conducting further research

to determine the direction of causality.

Because of the conflicting results reported in past research on

Locus of Control in leadership situations, other areas of interest were:

1) the relationship between the leader's and the follower's Locus of
Control and the follower's perceptions of the leader's behavior, and 2)

the relationship between the leader's and the follower's Locus of Control

and the leader's perceptions of his own behavior.

Two pilot studies were conducted to determine what modifications to

the existing Locus of Control scales might be necessary in order to

vii








reflect the possible relationship between religion and Locus of Control.

The results indicated that additional responses reflecting belief in God

as a controlling force in human events was required.

This modified Locus of Control scale, the consideration and struc-
ture subscales from the Leader Behavior Description Questionnaire -

Form XII, and Rokeach's values scales were combined into a survey question-

naire which was administered to a volunteer sample population of adult

church members and ministers from two churches in each of the six major

Protestant denominations (Baptist, Congregational, Episcopalian, Lutheran,

Methodist, and Presbyterian).

A factor analysis of the Locus of Control, consideration, and struc-

ture scales produced four Locus of Control factors and three LBDQ factors.
Zero-order correlations, multiple-range tests, and multiple regression

analysis were used to examine the relationships described above. The

results indicated that: 1) the God external Locus of Control factor
best discriminated between the Protestant denominations (2 < .00001)

while the other three Locus of Control factors did not. 2) The emergence

of the God external factor in the Locus of Control scale, and its

effectiveness in discriminating between the denominations, appears to

support the contentions of other researchers that the existing Locus of

Control scales do not tap the full dimensions of the Locus of Control

construct. 3) There were significant relationships between Locus of

Control and several of Rokeach's values. 4) Significant, complex rela-

tionships existed between the leader's and follower's Locus of Control

factors and the leader's and the follower's perceptions of the leader's

behavior. For example, both the leader's and the follower's Locus of

Control were related to the follower's perceptions of leader behavior on

viii








two of the three Leader Behavior Questionnaire factors, but not to the

third factor. In the case of the leader's perceptions of his own behavior

as measured by the same factors, neither the leader's nor the follower's

Locus of Control was related to one of the three factors, only the

leader's Locus of Control was related to another factor, and both the

leader's and follower's Locus of Control were related to the third factor.
The results of this study indicate that further research may be

justified to determine whether or not religion may be an antecedent of

Locus of Control, and that further research also appears justified in

order to determine the full implications of Locus of Control for leader-
member interactions.













SECTION I

INTRODUCTION

Rotter (1966, 1975) has found evidence that individuals differ in

their expectations about internal as opposed to external control of

reinforcement. Specifically, individuals differ in the extent to which

they perceive that their rewards are contingent on their own behavior

versus the extent to which they feel that their rewards are controlled

by fate, chance, or powerful others. Rotter has further maintained that

these expectations, which he labelled Locus of Control (LOC), form a

relatively enduring personality trait closely associated with the con-

cept of powerlessness. This trait, in combination with other variables,

is said to affect the individual's attitudes and subsequent behaviors.

Despite the extensive research that has occurred since Rotter's

1966 findings, at least two problem areas regarding this trait still

exist. First, while there has been some speculation as to the antecedents

of LOC, little research has been directed toward explaining the causes

of individual differences in LOC (Joe, 1971; Lefcourt, 1966). Second,

some research has been conducted on the relationship between LOC and

the class of behaviors generally labelled "Leadership." However, the

results have been inconsistent. The research effort described in this

paper is directed toward these two problem areas.

A review of the literature reveals four possible reasons for the

persistence of these two problem areas: shortcomings in the LOC scale,

the use of varying and conflicting criteria, the use of restricted
sample populations, and lack of research effort in these two areas.
1





2

Rotter (1966) originally proposed that the LOC scale was unidi-

mensional. However, subsequent researchers have found evidence that

the LOC scale is multidimensional (Duffy et al., 1977; Gurin, Gurin,

Lao, & Beattie, 1969; Joe, 1971; Mirels, 1970; Shraugher & Silverman,

1971). In his 1975 review, Rotter finally agreed that the LOC scale

was multidimensional. Sanger and Adler (1972) cited the possibility

that a multidimensional approach to LOC was required because the popu-

lation was becoming increasingly aware of the concept of control and

concommitantly more critical of attempts by others to limit that control.

While Duffy and Shiflett (1977) claimed that the predictive value of

the subscales was no greater than that of the entire scale, others (Gurin

et al.,1969; Mirels, 1970; Sanger & Adler, 1972) claimed that using the

total scale score might mask important differences in the subject

population.

Those proposing the use of the subscales present a very logical

argument. They proposed that those who feel that their outcomes are

controlled by chance, or a world whose rules are too impossibly complex

to cope with, are likely to exhibit passive behavior because there is

no way they can hope to gain control over their outcomes. Those who

feel that their outcomes are controlled by the socio-political system,

or powerful others, may be quite active and aggressive because they

perceive the possibility that such behavior may enable them to gain

control of their outcomes.

Along with growing evidence supporting the multidimensionality of

the LOC scales came the realization that the scale might not be tapping
the full construct of LOC. This shortcoming appears to be at least a

partial cause for Rotter's failure to find a relationship between LOC








and religious affiliation. While he found no significant relationship

between subjects' LOC scores and religious affiliation, interviews with

the subjects at least suggested to Rotter that religion might have a role

in the development of LOC and that further research in this area was needed

(Rotter, 1966). Benson and Spilka (1973) commented:

Rotter's (1966) scale defines external control in terms of luck
fate, and chance. While it seems reasonable to argue that one
who places his fate in God's hands is externally controlled, he
may find options phrased in luck and chance terminology
irrelevant. (p. 308)

That the use of varying criteria may be responsible for the con-

flicting research findings becomes more evident as one studies the

research methodologies utilized in the various studies pertaining to

the two areas of interest. Rotter (1966), using his LOC scale and

scores on the McLean scale which measures belief in the literalness of

the Bible found no significant relationship between religion and LOC.

However, Shraugher and Silverman (1971) used the LOC scale and religious

affiliation to explore the relationship between LOC and religion. They

found significant differences among members of the Jewish, Catholic,

and Protestant faiths moderated by frequency of church attendance.

Similarly, in the leadership area, Durand and Nord (1976) used a

version of Gurin et al.'s (1969) LOC subscales to predict follower

perceptions of their leader's behavior as measured by the structure and

consideration subscales of the LBDQ Form XII (Stogdill, 1963). Pryer

and Distefano (1972) used the same LBDQ Form XII subscales, but used
the total score on the LOC scale. Evans (1974) used the total score on

the LOC scale, but used three subscales of the LBDQ Form XII to measure

structure and three subscales to measure consideration. Goodstadt and








Hjelle (1973) used the total LOC score, but measured leadership behavior

in terms of coercive power and persuasive power. The conflicting

results of these studies becomes evident when one examines Table 3.

For the Relationship Between LOC and Structure, two researchers found

a positive relationship, two found a negative relationship, and one

found no significant relationship. For LOC and Consideration, three

researchers found a negative relationship, one found a positive rela-

tionship, and one found no significant relationship.

The use of restricted sample populations and the paucity of research

in these two areas appear to contribute to the conflicting findings

plus casting doubt on the generalizability of the findings. Only four

studies have been accomplished on the relationship between LOC and the

leader's behavior, and LOC and the follower's perceptions of the leader's

behavior. Of these four studies, two have been conducted using college

students. Of the applicable studies reviewed in the area of religion,

only two of twelve studies used an adult, non-college sample population.

Cherlin and Bourque (1974) commented that the LOC research appeared to

lack generalizability because the sample population has largely con-

sisted of subjects under thirty years old, with a greater than high

school education, and from the middle socio-economic class.

Section II of this study will review the relevant literature in an

effort to:

1. Develop a rationale supporting the contention that there is a

relationship between an individual's LOC and his religious affiliation.

(This will be the first step in determining whether or not religious

socialization is a possible antecedent in the formation of an individual's

LOC.)





5


2. Clarify the relationship between LOC and the leader's behavior

as perceived by the leader.

3. Clarify the relationship between LOC and the follower's

perceptions of his leader's behavior.

4. Form testable hypotheses concerning the relationships dis-

cussed in 1, 2, and 3 above.

Section III will develop the methodology for testing the hypotheses

using adult members of religious organizations as the sample population.

Section IV will report on the results of the study, and Section V

will contain the discussion and recommendations resulting from the

analysis of the results.













SECTION II

BACKGROUND AND RELEVANT LITERATURE

LOC and Religion

In the following paragraphs it will be shown that religious denomi-

nations vary significantly on factors which have a close similarity to

the constructs of LOC. Specifically, a rationale will be developed for

predicting significant differences among Protestant denominations on

the LOC dimension. If such differences are found, the study may con-

tribute to a better understanding of LOC as little research appears to

have been done to determine the antecedents of LOC (Phares, 1976). The

research that has been done has concentrated on parental influence

(Chance, 1965); ethnic background (Jessor et al., 1968), and social class

(Gruen & Ottinger, 1969). The results, although not conclusively con-

sistent, have provided a degree of support to speculate that religion

may be major causal variable for differences in individual LOC, and that

parental behavior, ethnic background, and social class are either inter-

vening or moderating variables.

If the hypotheses predicting the relationship between LOC and

religious denominations are supported by this study, it would appear

that longitudinal studies to determine the direction of causality would

be justified. If, as many researchers claim, LOC has important impli-

cations for individual coping behavior, then a knowledge of its ante-

cedents could provide important information for altering the individual's





7


coping behavior. Such a capability may contribute both to developing

individual potential, and indirectly, the effectiveness of organizations

(Phares, 1976).

The development of the hypotheses will proceed from a discussion

of some general findings and comments of other researchers to a dis-

cussion of the constructs of LOC and how these constructs relate to the

specific findings of Rokeach (1969) in such a manner as to lead to the

prediction of significant differences in LOC between the members of

different Protestant denominations.

A General Discussion of the Rela-
tionship Between LOC and Religion

The Bible itself appears to present evidence that the Judaeo-

Christian religions vary in their beliefs as to the degree of control

man has over his fate. Proverbs emphasizes man's ability to control his

destiny through his own abilities and performance rather than depending

upon the grace of God. Ecclesiastes, however, takes the opposite view,

emphasizing the powerlessness of man, who must place himself in God's

hands because he is unable to do anything for himself. Life's outcomes

are unpredictable, controlled by a God whom man can never know

(Rylaarsdam, 1974).

Shraugher and Silverman (1971) commented:

Many religious doctrines appear to have implications for the
development of attitudes about one's potential for control over
what happens to him. These doctrines often appear inconsistent,
however; some imply that the individual is responsible for his
rewards and punishments, and others imply that one's fate rests
with sources outside himself. (p.11)

To explore the relationship between LOC and Religion, Shraugher and

Silverman administered the LOC scale to 465 undergraduate students and

gathered such other data as demographic, frequency of church attendance,








and religious affiliation (Jewish, Catholic, or Protestant). Analysis
of variance for the LOC scores indicated a significant main effect for
religious affiliation with the Jewishsubjects being significantly more
external than Protestants (Sheffe test; p <.05), while Catholics did not

differ significantly from either group. There was an interaction between
frequency of attendance and religious affiliation with regularly attending
Catholics being significantly more external than Protestants (Sheffe
test; E<.05).

After conducting a study of the religious beliefs of college students,
Poppleton and Pilkington (1963) concluded that religious attitudes differ
sharply within the Protestant denominations and that studies which

"lumped together" various Protestant denominations obscured important
differences between them.
Benson and Spilka (1973) conducted a study using Catholic high
school students to determine whether or not self-esteem and LOC were
related to the individual's image of God. They found the following

correlations between an individual's LOC and his image of God: A Loving
God (-.30, p <.01), A Vindictive God (+.23, p <.01), and an Impersonal
God (+.18, p <.05). The findings appear to support their contention

that a demanding, powerful, and controlling God is consistent with the
perception that one is not in control of his outcomes while a perception
of a God who is freeing rather than restricting, and who is uncontrolling
rather than controlling, is consistent with a perception of internality.
Nunn (1964) conducted a study investigating parents use of the
concept of God as a means of controlling their children. He found that
parents who felt powerless tended to form a coalition with God in which
God was invoked as a third party to exercise control over the children.








In the course of the investigation, he came to the conclusion that,
powerless people who form a coalition with God to control
their children tend to participate in a religious system that
especially denies the competence of the actor and exalts the
direct control of man's affairs by God. These people tend to
prefer sect and transitional churches, while those who do not
feel powerless tend to prefer denominational type churches.
(p. 420)
He classified transitional churches as Church of Christ, and Southern

Baptists, sects as Jehovah's Witnesses and Baptist sects, denominational

churches as Methodists, Presbyterians, "and other Protestants."

Despite the above findings, the question still remains unanswered
as to whether religion acts indirectly on the personality of the child

through the way it affects the attitudes and behavior of the parents in

their interactions with the child; or whether the influence of religion

is more direct through the effects of religious education experienced by
the child. It is quite possible that the effects are a combination of
both.
Allport et al. (1948) in investigating the religion of post-war

college students, concluded that early religious training was likely to
be a principle psychological influence upon the individual's later

religious life. They further found that 50% of the college students

surveyed reported that they had turned against the religious teachings
of their parents. However, they reported considerable reason to suppose
that college students are in the least religious period of their lives,

and that religion becomes more important as they mature and face increased
responsibilities. This finding appears to indicate, as maintained pre-

viously, that college populations are not representative of the general
population in their religious beliefs. This may explain previous failures
to find a relationship between religion and LOC when college students








were used as subjects. The further finding that 43% of the students still
accepted their religious training argues for considering the possibility

that religion may well influence such personality traits as LOC.
As mentioned previously, Shraugher and Silverman (1971) found that

the LOC of members of the Jewish, Catholic, and Protestant faiths differed
significantly as moderated by church attendance. Although they did not

investigate the differences between Protestant denominations, Glock and

Stark (1966), and Poppleton and Pilkington (1963) maintain that there are
more differences between the Protestant denominations than between the

three major faiths, and that to attempt to group the Protestant denomina-
tions together, as has been done in the past, can obscure important
differences.
From this discussion it can be seen that Juadeo-Christian writings
conflict as to the degree to which man possesses the ability to control

his outcomes. Religious and parental teachings may result in individual
differences in one's image of God which has been found to correlate with

LOC, and research has indicated that there are significant differences
between the Jewish, Catholic, and Protestant faiths in the LOC of their

members. Also, several researchers have found more theological differ-
ences in the way that doctrine presents the image of God and the degree

of control man has over his outcomes.
Dimensions of LOC
In this subsection it will be shown that significant differences

have been found between Protestant denominations on variables which
appear to be associated with the construct of LOC. The construct of LOC

will be examined and related to research findings on the differences
between the Protestant denominations. Hypotheses will be developed








predicting differences in the LOC of the members of the major Protestant

denominations.

Rotter (1966) at first maintained that LOC was unidimensional.

However, subsequent research (Collins, 1974; Duffy et al., 1977; Gurin
et al., 1969; Joe & Jahn, 1973; Mirels, 1970) indicated that the construct

is multidimensional, and in his latest report (1975) Rotter himself

accepted the multidimensionality of LOC.
An examination of Table 1 reveals that although there is general

agreement as to the multidimensionality of LOC, the findings as to the
dimensions of the subfactors are inconsistent. There appear to be three

possible reasons for the inconsistency. First, the methods of factor
analysis and the criteria for determining a significant factor loading
have varied among the researchers. Second, different LOC scales have
been used, and used in varying combinations with other scales.
Mirels used the original Rotter scale, while Joe and Jahn used the

original scale, but asked the subjects to indicate the degree of agree-
ment between the internal and external pairs on a six-point, agree-

disagree format. Using a slightly different approach, Collins (1974)
split each question of the Rotter scale (the internal and external choices)

into two separate questions (for a total of 46 questions) and used a

Likert scale format. While Joe and Jahn (1973) did not report on the
correlation between their instrument and the Rotter scale, Collins
reported a total score correlation of .82 indicating that the instruments

were very similar. He did not, however, report the correlation between
the individual items in the two scales after his scale items had been

randomly mixed with additional questions. Gurin et al. (1969) noting the

inconsistencies between their factors and those of other researchers,















U)
C
*r

1 4J c0 0W
<.1 A"C S1. 1 ..
u IA .- 4) c .5 U

C-n, C Uu C8

W z 0j, M-. 0 4a E (A
U0 0 Q U"- 10 "
o "> o .0 0L ( Ju- o u- ,- -
4j 4 C *f- ui W ui
U U a- O. 5. x '-C '
m > ,- 00nooo 2 C
U. o w t .- 1 I0 CL,- .3- I 4--
3- .0i.) I u 4- 1. M. CA 0 1m c
4- 0. o1- to a 4- >> oeiu tu u (u
o V03 .-4-' f- < > =) = a- U) S.
l- > o C4- 4%- 40i4- ir *.- 4 S- 'o
c do 0- 43- 0 0i
Cc U iao oo s-0 4.uEj -u 00.. o
o ). -m OCg Z0.C45'eU Ae 4- +
4.- 0U 3 L I o ( t 0a
a >C O C3 C U V) 4a

-- 41JU)4n 1 -C 4- U4fC- C: 04
4.. -- C.1- .13 4-- L )W 0 24-
S4-' .0 Q 0W '- .- =3 0 C

V 41
SC- 3 m 0





































SI U4.
a C E 3 RR WR 42 U 0

... 4-


0 0 to Uo


4--. )
0. .- wi f 1 a a 4-








BCU


o a) .
L I U)







LL m I -

Q -a i- -
L% JC C'c 2 03
U *






C:)
0a3 4






















0) Im c-
4- .- 03






S0 M- Q on c
1 Ic C.0
,- 3a





n 4n a M a) C a

4 -i 0 e- 00 0)




) 0 .. 0- L0 tn .N
4- *y Va'J EU Ev
41 Lu; r <

to to U)
U- cm



- CC 4) V o
C-! 'AC 03

4 41 o .-C c c0 r 01 1..
U u (MJC m B.n0 o. 0 4

UI4- .= +13
'2 3 O I *o u
V r- J 1n 43 4-
1-1 C) 0 10 -- CL U
U) i-fJ *s. 1 0 0 :3 o a3
a S. .-0* o 'a P, c a


In .o .1fU ..V,' -c 3<3 0a

EU 03
9A 4.-*
tu 01 41-a)
0) U1 4w4
Z U1 LI U) n "0 3

-ru 01.- 0 34.. o o




cl *) 0 U0
Lu G- 0- -
U) -O *r- C
03 U 453 *
.0 13 c


C -, 23
E u la .0 Ei Li u .,- .cS.

t -- crJi SC O*-- r-.

l > S -








commented that the inclusion of other items in, or in conjunction with

the LOC scale might have sensitized the respondents and affected the

composition of the factors. Third, Rotter (1975) claimed that the
differences in the factor structure reported by the researchers were not
related to the construct of LOC, but to the nature of the sampled popu-

lations. It may be that the development of factors for specific popu-
lations could improve the predictive validity of the scale for specific

populations if one is willing to accept the loss in generality.
Despite the lack of consistency in the factors reported by the various
researchers, an examination of Table 1 does reveal common concepts or

subfactors, although the items contained in the factors vary. The most
consistently reported factor was that involving the individual's per-

ception of the degree of control one has over the political system, or

powerful others (Mirels, & Joe & Jahns factor 2, Gurin et al.'s factor 3,
& Collins & Duffy's factor 4). The other common theme refers to the
individual's perception of internal control versus a lack of control

because the world has no underlying order in the sense that outcomes are
controlled by luck or chance. In a very similar manner the individual

may perceive that although the world has an underlying order, that order

is too complex to be understood. This construct is common to the first
factor reported by all the researchers in Table 1.
The research indicating that individuals appear to perceive them-

selves as either in control of their outcomes or not in control of their
outcomes because of the lack of order in the world, the complexity of

the world, or because the outcomes are controlled by powerful others, can

lead to other logical implications. Seeman and Evans (1962) referred to

Rotter's scale as the "alienation scale." Seeman (1959) identified five








alternative meanings of alienation: powerlessness, meaninglessness,

normlessness, isolation, and self-estrangement. He interpreted power-

lessness, "This variant of alienation can be conceived as the expectancy

or probability held by the individual that his own behavior cannot

determine the occurrence of the outcomes or reinforcements he seeks." (p.784)
However, he limits powerlessness further to the individual's influence

over the political system. It can be seen that Seeman's powerlessness
is identical to the previously discussed first factor. However, LOC

emobodies still another dimension of powerlessness as a perception by

the individual of the degree that his outcomes are controlled by himself
versus chance (in the sense that the world is unordered, or too complex

to cope with). It is also logical that those who feel that they are

controlled by external forces are unlikely to value a sense of accom-
plishment or competency as they would not perceive a connection between
their behavior and the resulting outcomes (Collins et al., 1973; Gurin
et al., 1969; Runyon, 1973; Szilagyi & Sims, 1975). It appears then that
there are at least three factors constituting the construct of LOC:

powerlessness, competency, and a sense of accomplishment.
LOC and Rokeach's Study of Values
in Religious Denominations

This subsection will examine a study conducted by Rokeach (1969) in
which it was found that the members of the Protestant denominations varied
significantly in the importance they attached to certain values, among
which were competence, a sense of accomplishment, and salvation. By

interpreting and relating these values to the factors of LOC, hypotheses
will be developed predicting significant differences in the LOC of members

of six Protestant denominations.







Using a National Area Probability Sample of 1,400 Americans over 21

and a sample of 300 college students, Rokeach administered two scales, a
"Terminal Values Scale" and an "Instrumental Values Scale." In both of
these scales the respondents were asked to rank a list of 18 values in
order of their perceived importance (see Appendix A for the instruments).
Rokeach defined Terminal Values as, "Preferred end-states of existence,"
and Instrumental values as, "Preferred modes of behavior."
The most highly significant Terminal Value distinguishing among
denominations was "Salvation" (median test = p <.0). The denominations
were ordered as follows on the value they attached to "Salvation." (The
ranking attached to "Salvation" by denomination is indicated in the
parentheses with 1 equalling the highest value and 18 the lowest value):
Baptist (3), Lutheran (9), Methodist (10), Episcopalian (10), Presby-

terian (11), and Congregational (11). Rokeach theorized that, "those
who valued salvation have an other-worldly orientation which would appeal
to those who feel powerless and that they exert little or no influence
in affecting the course of political and social events in their society."(p.5)
Note how similar this description of those who value salvation is to
Seeman's (1959) concept of powerlessness as the individual's sense of a
lack of influence over the political system. Nunn (1964) also maintained
that powerless people would tend to seek a religion which especially
denies the competence of the actor and exalts the direct control of man's
affairs by God. He identified such churches as Baptist sects and Southern
Baptists (among others). He held that those who do not feel powerless

would seek such churches as the Methodist and Presbyterian. Although
Nunn did not explain why this should be, it appears logical that the

denominations vary from those who emphasize the powerlessness of man








(Baptist) to those who emphasize man's ability to control his own life
through the exercise of free will, as may be the case with the Methodist
and Presbyterian churches.
Rokeach also found a significant difference between denominations on

their ranking of the terminal value "A Sense-of-Accomplishment" (median
test = P<.012). The rankings were: Espicopalians (4), Congregation-
alists (6), Lutherans (7), Presbyterians (9), Methodists (12), and
Baptists (12). In a previous discussion it was asserted that externals
would not value a "sense of accomplishment" as they would not perceive
a connection between their behavior and the resulting outcomes.
In his discussion of the Instrumental Values scale, Rokeach identi-
fied seven items as "competence" values. He defined these values as
"preferred modes of behavior which, when violated, lead to shame about
competence rather than guilt about wrongdoing" (Rokeach, 1973, p. 142).
He identified the seven values as: ambitious, broadminded, capable,
imaginative, independent, intellectual, and logical. The following are
the rankings of the three competence values that differed significantly
among denominations according to median tests.
Bapt. Method. Epis. Pres. Luth. Cong.

Broadminded 7 2 3 3 5 2
Capable 14 9 6 5 10 13
Logical 14 17 15 16 15 16
An examination of the rankings indicates that the Baptists tended to rank
competency values lower than the other denominations while the Episco-
palians and Presbyterians tended to rank the competency values the highest.
Runyon (1966) contended as a result of his studies on LOC that those who
feel that the world is controlled externally perceive little opportunity
to demonstrate competency.








Table 2 was constructed by measuring the differences in medians

between the Baptists and the other denominations for salvation, sense of

accomplishment, logical, capable, and broadminded as reported by Rokeach

in his 1973 study. For example, the median differences shown in the

table for salvation were derived as follows. The median scores for each

of the denominations on the value of salvation were: Baptists 4.4,

Methodists 8.7, Episcopalians 13.0, Presbyterians 10.0, Lutherans 8.8,
and Congregationalists 9.3. The median of 4.4 for the Baptists was used

as the zero reference point and the differences between the Baptists and

the other denominations were computed by subtracting the median score of

the Baptists from the median score of each of the other denominations.

Hypotheses Related to LOC and
Religious Denominations

If externality is positively correlated with placing a high value

on the value salvation, and if externality is negatively correlated with

high values on a sense of accomplishment and the competence values, it

appears logical to form the following hypotheses:
1. Protestant denominations will vary significantly on the

dimensions of LOC. The direction of the differences in LOC from the

most internal will be as follows: Baptist, Methodist, Congregational,

Lutheran, Presbyterian, and Episcopalian.
2. Internals will place a higher degree of importance on the values

sense of accomplishment, broadminded, capable, and logical than will

externals.

3. Individuals who place a high degree of importance on the value

of salvation will be more external than those who place a low degree of

importance on the value of salvation.







18




0 N0 ,--
f) CD 0 r- 0
00000


0 0 0 0
*





4.)









0 V,,, V, V V ,
C 5. C j N N Qj 0
3- C -













O
r*- 3 ,
S<-













.C mRI D W
OU
( ..) *
















cc 4 (A

Cr V r--


t-
0 5-



















C- ca -

(.0 (
:- 4-)
1U


















CL L CD CDL0 D (
EU

o) a






















*-



0 r ("-U- 1 C"
nU) > CM CMJ CM i- in 'ct































00 0 ",
*i- (










r C (n
r_ CC 4
.0

C.. 10 0
o' E
4M *4- 0







*r
EU 0. d
-r

So to
oU U3 U-
J +0 Ci 2
- di 0 0-










C C
U) 4. ) r..

*r- -O CO C, J CO r- CM .0
4> Q 0 *
1UP d1






CO 4.)








4 *-- U












CO C. -IJ (I Cd)








4. Members of the various Protestant denominations will vary

significantly as to the degree of importance they place upon the value of

salvation. The order of the differences, from the denomination which
places the highest importance on salvation to the one which places the
lowest importance on salvation, will be as follows: Baptist, Methodist,
Lutheran, Congregational, Presbyterian, and Episcopalian.
LOC and Leadership

This subsection will examine relationships between 1) LOC and

the leader's behavior (as perceived by the leader), and 2) the follower's
LOC and his perceptions of his leader's behavior. The possible reasons
for the inconsistencies in past research will be discussed, and a ration-
ale will be developed to provide for testable hypotheses directed towards resolving
these inconsistencies.
Inconsistencies in Past Research
Examination of Table 3 indicates that the findings of the studies on
the relationship between: 1) LOC and the leader's behavior, and 2) the
follower's LOC and his perceptions of his leader's behavior, are incon-

sistent. Durand and Nord (1976) found that external followers (those
scoring high on Rotter's scale) tended to perceive that their leaders
exhibited more structuring behaviors than did internal followers. Evans
(1974) reported that external followers tended to perceive that their
leaders exhibited less structuring behaviors than did internal followers,

while Pryer and Distefano (1972) reported finding no significant rela-
tionship between the follower's LOC and their perceptions of the leader's
structuring behavior.













Table 3

Relationships Between the Follower's LOC and
His Perceptions of the Leader's Behavior
From Prior Research


Researcher
Structure


Variables


Durand and Nord

Evans

Pryer and Distefano


Consideration


-.25*

-.32*


Relationships Between the Leader's LOC and
His Leadership Behavior From
Prior Research


Researcher
Structure


Durand and Nord

Goodstadt and Hjelle






<.05
**p <.01


Variables


-(ANOVA)


Consideration


-(ANOVA)








Evans and Pryer and Distefano both reported that external followers

tend to perceive their leaders as less considerate than did internal

followers, while Durand and Nord found a nonsignificant relationship,

but in the same direction as the other researchers just discussed.
As to the relationship between LOC and the leader's behavior, only

the two studies referenced in Table 3 appear to have been published, and

the criteria used in the two studies are not directly comparable. This
author's interpretation (to be discussed later in detail) is that the

findings of the two studies are inconsistent. Durand and Nord (1976)
reported that external leaders were perceived to be both more structuring

and more considerate than internal leaders. Goodstadt and Hjelle (1973)
reported that external leaders tended to exercise more coercive power on
their followers than did internal leaders, and internal leaders tended to

exercise more persuasive power on their followers than did external
leaders.

The reasons for these inconsistencies may be the results of differ-

ences in the criterion variables and the instruments used to measure both
the predictor and the criterion variables. All four studies used Rotter's

LOC scale as the predictor variable. However, Goodstadt and Hjelle, Evans,

and Pryer and Distefano used the total score on the scale while Durand
and Nord used the subscales reported by Gurin et al. (1969) with one

modification. From the discussion in the previous subsection, it will be
recalled that the general themes underlying the concept of LOC are some-
what consistent; i.e., the external forces which may control one's life

are complexity, lack of underlying order in the world, or powerful
others. There does not, however, appear to be a general consensus as to

exactly what items make up those themes. Furthermore:, although Durand








and Nord used Gurin et al.'s two factors, Control Ideology and Personal
Control, they did not use the two questions that Gurin et al. reported
as comprising the third factor, System Modifiability. Rather, they

lumped the remaining six questions in the Rotter scale into a single
factor which they identified as "I-E Residual." Of the six questions not
included in Gurin et al.'s System Modifiability, only three seem to be

logically acceptable. Question 3 (See Rotter's scale in Appendix A)
deals with the individual's ability to influence government decisions,
question 22 deals with the individual's control over politicians, and
question 5 which deals with whether or not teachers are unfair to their
students, could be construed to tap the dimension of control by powerful

others which is characteristic of the System Modifiability factor. The
other three questions (2, 4, and 21) do not appear to fit into this
factor (see Rotter's instrument in Appendix A for the questions).
The logic of Durand and Nord's use of the subscales seems question-
able. Gurin et al. admitted that their factors might not be representa-

tive of the white population as their subjects were black. It would
appear that unless the factor structure could be proven to be character-
istic of the sample population, its use as a predictor variable is of

dubious value. Of all the factor analyses of the Rotter scale, only one
has been conducted using a non-college population. It should also be
pointed out, that of the four significant relationships reported by
Durand and Nord, three were based on the predictor variable I-E Residual.
Pryer and Distefano (1972), Evans (1974), and Durand and Nord (1976)

used various combinations of Stogdill's (1963) Leader Behavior Descrip-
tion Questionnaire (LBDQ Form XII) to measure the criterion variables.
Both Pryer and Distefano, and Durand and Nord used the two subscales of








the LBDQ Form XII "Consideration" and "Structure" while Evans summed

three subfactors of the LBDQ Form XII to arrive at scores for consider-
ation and structure. Consideration consisted of the sum of the considera-
tion, tolerance of freedom and integration subscales while structure

consisted of the structure, role assumption, and production emphasis
subscales. The use of the subscales may be justified, but the amount of
unique variance contributed by each of the subscales may well contribute
to the reason for the lack of agreement between Evan's findings and
those of the other researchers.
Discussion of Past Research Findings
Rather than using the follower's perceptions as a measure of the

leader's behavior, Goodstadt and Hjelle (1973) used the experimenter's
observations of the leader's behavior as their measure. The criterion
were French and Raven's (1959) forms of power, reward power, expert power,
and coercive power. As stated previously, they found that external
leaders used significantly more coercive power than did internal leaders
(P <.05), and that internal leaders used significantly more persuasive
power (encouragement, praise, setting new standards, and admonishment)

than did external leaders (P <.05).

Evan's (1974) used the previously discussed dimensions consideration

and structure as perceived by the followers as the predictor variables
with LOC and found that external followers perceived their leaders to be
both less structuring and less considerate than did internals. While
Goodstadt and Hjelle's predictor variables are not necessarily identical

to those used by Evans, it may be reasonable to assume that coercive
power is the opposite of considerate behavior. Stogdill (1963) describes
consideration as "Regards the comfort, well-being, status and contributions








of the followers" (p. 3). Similarly, it may be argued that persuasive
power is similar to two of the three subscales of the LBDQ Form XII

structure and production emphasis which were summed by Evans for the

structure score. Structure is defined by Stodgill as, "clearly defines

own role, lets followers know what is expected" and production emphasis
as, "applies pressure for productive outputs." Schriescheim and Stogdill's

(1975) analysis of these scales maintains that the scales contain a

minimum of punitive concepts (which may be taken as a minimum degree of

coercive power). If one accepts these definitions, the two studies
appear to report consistent findings.
Durand and Nord (1976) reported that followers who were external (as

measured by the I-E Residual subscale) tended to perceive their leaders
as more structuring that did internal followers (p < .05). They reported

a nonsignificant relationship between LOC and consideration. Durand and

Nord claimed that their findings were consistent with those of Goodstadt
and Hjelle, a questionable contention. Durand and Nord equate structure
as measured by the LBDQ Form XII with Goodstadt and Hjelle's coercive
power. However, as discussed previously, Schriescheim and Stogdill's

(1975) analysis suggests that the LBDQ Form XII measures a minimum
degree of arbitrary or punitive leader behavior. Further Evans (1974)

postulates that rewarding behavior and consideration are very similar
constructs. Therefore, Durand and Nord's findings do not appear to be
consistent with either of the previously discussed studies.
Pryer and Distefano (1972) reported no significant relationship

between LOC and structure. At one hierarchial level of the three investi-
gated, they found that internal followers perceived their leaders as more
considerate than did external followers (p < .01). The finding of no








significant relationship between the follower's LOC and structure con-

flicts with the findings of both Evans and Durand and Nord, while the

finding of a significant relationship between the follower's LOC and

consideration agrees with that of Evans.

At this point in the discussion, it is important to emphasize the

difference between Goodstadt and Hjelle's study and the findings of the

other researchers which have been discussed thus far. From Goodstadt and

Hjelle's methodology (of observing the leader's behavior and administering
the LOC scale to the leader) it can be seen that the effect on the

follower's LOC on the perceptions of the leader's behavior have been

minimized (admitting that there may still be an interaction between the

follower's LOC and his behavior which would in turn affect the leader's

behavior). While both Durand and Nord, and Goodstadt and Hjelle report
findings on the possible relationship between the leader's LOC and his

behavior, Durand and Nord measure the leader's behavior in terms of the

follower's perceptions of that behavior and therefore, the relationship
between the leader's behavior and his LOC is confounded by the follower's

LOC as it may affect his perceptions. As Durand and Nord combined the

various LOC factors used to measure the LOC of the leader and the

follower to find the best predictor of the leader's consideration and

structuring behaviors, it is possible to gain some insight into the
relative importance of the effects of both the leader's and the follower's

LOC on the follower's perceptions of the leader's behavior. In predicting

the leader's structuring behavior, they found the leader's Personal Con-

trol Factor accounted for 14% of the variance in the leader's perceived

behavior and the follower's I-E Residual factor of the LOC scale

accounted for an additional 10% of the variance, suggesting that the








leader's LOC may be slightly more important to the relationship. When

predicting the perception of the leader's consideration behaviors, they

reported that the leader's I-E Residual factor accounted for 13% of the

variance while none of the follower's LOC factors made a significant
contribution to the prediction of leader consideration. While these

regressions provide some indication of the relationship between the

leader and the follower's LOC and the leader's behavior as perceived by
the follower, it is still not possible to determine the relationship

between the leader's LOC and his true behavior from Durand and Nord's
data.

From this discussion of the leadership literature, it appears that
the effectiveness of the leader depends upon a complex circular rela-
tionship including many moderating variables and interactions. The
leader's personality, combined with situational factors such as the

nature of the organization and the constraints it places upon the leader,
combine to cause the leader to behave in certain ways (say in terms of

consideration and structure). These true behaviors, as perceived by the
follower, are distorted to an unknown degree by the personality and

expectations of the follower who responds with behaviors which are in

turn perceived by the leader, and to some degree, moderated or distorted
by the leader's personality. The leader then modifies his behavior in
an attempt to influence the follower's behavior in a manner which the

leader hopes will gain his goals. This leader behavior is then perceived
and reacted to by the follower and the cycle repeats itself. The main

thrust of this portion of the study is to attempt to clarify how much
of this distortion in the follower's perception of the leader's behavior
can be attributed to the follower's LOC and how the leader's behavior is

affected by his own LOC.








Comparing the findings of Durand and Nord with those of Goodstadt

and Hjelle still reveals conflict. If it were assumed that the follower's

perceptions were veridical, the external leaders in the Durand and Nord

study were found to be both more structuring and more considerate than

internal leaders. Goodstadt and Hjelle (who did not rely on the follower's

perceptions to measure leaders behavior) found, however, that external

leaders were both less structuring and less considerate than internal

leaders.
From the foregoing discussion it can be seen that, with the possible

exception of Goodstadt and Hjelle's study, the relationship between the
leader's LOC and his behavior is still unclear. Some clarity might be

achieved if the effect of the follower's LOC and his perceptions of the

leader's behavior could be isolated. Some increased understanding of the

relationship may be attained if one considers how LOC as a generalized

expectancy should affect the follower's perceptions of the leader's

behavior.
In his 1975 discussion of LOC, Rotter makes a very important state-

ment when he points out the different roles that generalized and

situation-specific expectancy play in the prediction of behavior. The

more novel and ambiguous the situation, the more impact generalized

expectancy will have on the individual's attitudes and behaviors. As

the situation becomes more routine and unambiguous, the importance of

generalized expectancy decreases and that of situation-specific expectancy
increases. In this discussion Rotter assumes that experience in the

situation derived from environmental clues is responsible for the shift

from general to situation-specific expectancy. However, research indicates

that externals are far less likely to perceive, and utilize, the








environmental clues to modify their attitudes and behaviors. It would

appear that previous research has virtually ignored this particular aspect

of LOC. It would seem that if LOC were a strong enough personality

trait, it might be that externals would be so oblivious to the environ-

mental clues that only minor shifts toward situation-specific expectancy

would occur over relatively long periods of time.

Pryer and Distefano's findings that LOC was significantly related

to the follower's perceptions of leadership behavior at only one of three

hierarchial levels is particularly interesting in the light of the roles

generalized and situation-specific expectancies are postulated to play

in the determination of the follower's attitudes and behaviors. The

average tenure of the subjects at the hierarchial level in which the

significant relationship between the follower's LOC and his perceptions
of the leader's consideration behaviors was 94 months versus 15.4 and

43 months for the hierarchial levels in which LOC had no significant

relationship to perceived leader behavior. Rotter's theory would pre-

dict that LOC as a generalized expectancy would be less: ambiguous. It

is apparent that more research is required to determine the moderating
effects of tenure on the relationship between the follower's LOC and his

perceptions of the leader's behavior.

The Hypotheses
From this review of the theories concerning the behavioral implica-

tions of LOC, it appears that one result of external LOC might be a passive

approach to leadership behavior. If the individual as a leader does not

perceive that outcomes are contingent upon his behavior, either because

the outcomes are controlled by chance, or because of a sense of powerless-

ness, it seems possible that he would resort to a laissez-faire style of








leadership. In this case, the leader should be perceived as being low in

both consideration and structuring behaviors.
Similarly, if the sense of powerlessness results from low self-esteem

and a feeling of inferiority, the leader may be driven to strive for power

over others (Phares, 1976). As previously discussed, in the discussion
of Goodstadt and Hjelle's study, the drive for power over others should
result in low consideration behavior. Similarly, as the structure scale

of the LBDQ Form XII contains only a minimum measure of coercive, or
punitive behavior, the external leaders should score lower than the

internal leader on this scale. Thus we would expect to find a significant
negative correlation between the leader's total score on the LOC scale

and his score on the consideration and structure subscales of the LBDQ -

Form XII.
If, as Phares (1976) argues, externals have a feeling of powerless-

ness, and as Runyon (1973) maintains, are less motivated and show less

job involvement, it seems likely that they would perceive their leader
as more structuring than would internal followers. Part of the perception
may be moderated by the follower's LOC and part may be veridical in that

the leader must actually behave in a more structuring and less considerate

manner in order to obtain satisfactory performance.
Durand and Nord (1976) predicted that external followers would per-

ceive the leader to be low in consideration because the follower would be
reluctant to attempt to influence the leader by making his needs known.
The leader, being unaware of these needs, would make no effort to satisfy

them, and thus would be perceived as low in consideration.
The negative relationship between LOC and the follower's perception

of considerate behavior on the part of the leader appears to be indirectly








supported by the findings that externals express lower job satisfaction

than internals (Duffy et al., 1977; Lichtman, 1970; Mitchell et al.,

1975; Organ & Green, 1974). Stogdill's (1974) review of the leadership

literature reports that 12 of 14 studies found a positive correlation

between follower satisfaction and leader consideration behavior (of the

remainder, one finding was nonsignificant, and the other reported a

negative correlation). As these were correlational studies, causality

cannot be determined. It is possible that the external followers tend

to perceive their leaders as low in consideration which in turn leads

to the reported lower job satisfaction of externals as compared to

internals.

From the foregoing discussion, it seems reasonable to predict the

following hypotheses:

5. Internal leaders will perceive themselves to behave in a more

structuring and considerate manner than will external leaders.

6. External followers will perceive their leaders to be more

structuring than will internal followers.

7. External followers will perceive their leaders to be less

considerate than will internal followers.

8. Follower perceptions of the leader's behavior will be moderated

by the length of time they have been associated with that leader.













SECTION III

METHOD

Subjects

To test the hypotheses related to the relationship between LOC and

religious denominations and LOC and the follower's perceptions of their

leader's behavior, a subject population was drawn from two churches in

each of the six major Protestant denominations located near the

University of Florida. The major denominations are Episcopal, Lutheran,

Methodist, Presbyterian, Congregational, and Baptist. The churches were

selected from each of the denominations based upon the recommendations

of the Department of Religion, University of Florida that these churches

were the most representative of their denominations. If access to the

first choice of churches was denied, the next most representative church

on the list was to be selected until access was gained or that denomina-

tion eliminated from the study for lack of access. Two churches were

chosen from each denomination to provide for comparison of within and

between denomination variance on the variables of interest.

The researcher's first step was to make an appointment with the

minister of each church, explain the purpose of the survey, and gain

access to the congregation. In no case was access denied; however, in

three cases additional approval of the church's governing body was

required (one Lutheran and two Presbyterian). In the case of the Lutheran

church, a formal written request was required (see Appendix B). In the

other two churches, the ministers requested approval verbally.








At the initial meeting with the minister, the purpose of the survey

was explained, the minister was given a copy of the questionnaire with a

cover letter of explanation and instructions for the subjects. In
addition, a short written notice was provided from which the minister

could make a verbal or written announcement of the survey to the congre-
gation (see Appendix B). A request was also made to administer the

questionnaires to members of adult Sunday school classes during their
Sunday sessions so as to provide the maximum control over the conditions

under which the survey was conducted and to increase the response rate.
The first church, a Methodist church, gave this permission. The success

of the proposal was attested to by the fact that a 100% response rate
was achieved. Sixty-two subjects responded before the survey of that
church was terminated. Unfortunately, this procedure was either not

acceptable to the other churches because the time involved detracted from

their classes, or because classes were not held during the summer months
in which the survey was conducted.
With the exception of the one Methodist church, the questionnaires

were delivered to the minister who made them available to his congre-

gation on a voluntary basis. The completed questionnaires were picked

up at the church by the researcher and there was no direct contact

between the researcher and the subjects. As the data collection progressed,
it became apparent that the response rate from the Lutheran and Presby-
terian churches was going to be lower than desired. In the Presbyterian
churches, an additional attempt was made to increase the response by

delivering an additional 36 questionnaires. When this attempt was un-
successful, a third church was contacted and access was approved. In
the case of the Lutherans, one of the selected churches was too small,








consisting of only 40 members of whom six responded. As in the case of

the Presbyterians, a third church was contacted and access was granted.
In the case of the Congregationalists, there was only one church in the

area consisting of 130 members.
Table 4 provides a description of the sample population. A total

of 557 questionnaires were distributed over a period of two and one half

months and 261 were returned for an overall response rate of 47%. Within

the denominations, the response rate ranged from a low of 30% for the

Presbyterians to a high of 77% by the Methodists. The sample contained

slightly more females than males (57% female), and the average age of

the population was 43. Fifty percent of the sample attended church and/or
Sunday school at least twice a week and 40% of the sample attended

church or Sunday school once a week. Only 10% of the sample were infre-
quent church goers. To determine if the differences in sex, age, or
attendance were significantly different between churches, a chi-square
test of the differences was conducted. All differences were non-
significant with the exception of age which differed significantly

(p <.0074, Eta square = .12). This significant difference will be dis-
cussed later in the analyses.
To test the hypotheses related to LOC and leader behavior, each of the

13 ministers of the churches involved in the survey was personally

briefed on the purpose of the survey and given a questionnaire containing

the leader's version of the LBDQ Form XII and the LOC scale. The
values scale was omitted in hopes of increasing the response rate. To

increase the size of the leaders sample, 30 additional questionnaires
were mailed to Protestant ministers selected at random from the yellow
pages of the telephone book. These ministers were sent a copy of the

























U
to



r)


C.
u


















0

04
4-)



(n












CC
rtp


5-
4)

03


)n L" LO











000
C) CM Cn










U-


0
C


4-) U U- C
SL-(ra 0)
4r 3 S-
JUU C-
C0 0
co 0


00



0


OO 0
C O



00



L-,-
n OO



00




'vt o0


0 I,--O





















c r-. C


'O 0



O 0



00



C0D)

CO d



CY C)






CM C


000



000



000



000



OO0.




CD CCOC


%t ce)


r- Cf) m-fl


r- CMI C )


0











CD





a LO
0











0n

r b


.4- oDin
q:ar in 4::r


o3


Or

0





0

0











CO CM
U-CO
nDC',










00u,
tLo'
1.13L


)in nc.Olm CM OJCM o0o0co r-


a)


r- CM O -Cs MN 0) 0)
S= J= l.C.C C 0
U UUU UU U- Ur 4 UU- 3
o -- 0- S- o S- .-0 0- -.0 S-c -a c 0
U 3 = 3 3 a)3Z 4J 0 4 3 3 -0 4- V)
= MCC0 =XCO == 0 lC C 0
u uC1-- C+3L. U. 41u I- X QUG-- E- r
( S-t =
=L am L U0
81i E CL V)


I-















I-
I-







0



0 U)











0
0
r.
Co
C*


C
0
*r

Ca
*r
E
o
C
0


o 1 id


C0 q0
SCM 04O








LBDQ Form XII and the LOC scale along with a letter requesting his

participation in the survey (see Appendix B). In addition, a self-

addressed, stamped envelope and a post card which could be returned

separately (to protect subject's anonymity) if the results of the survey

were requested.
Two weeks after the mailing, only two questionnaires had been

returned whereupon the researcher attempted to contact each of the 32

ministers by telephone. In the subsequent three-week period, contact was

made with 12 of the 32 (38%). Of these contacted, only one openly
refused to participate. The final sample consisted of 20 ministers for a

47% response rate.
For those who may question the use of ministers as suitable subjects

for the study of leadership, it must be admitted that the research on

the leadership characteristics of ministers is sparse. Glasse (1968)

maintained that the minister's position in the church had the characteris-

tics of a top-level organizational leadership position, and Stogdill

(1963) included ministers in his studies of the Leader Behavior Descrip-

tion Questionnaire. An examination of his reported data in Table 5 does

not appear to indicate any substantial differences between the scores

on consideration and structure subscales for the ministers and other

professions such as army and highway patrol offices, community leaders,

etc.








Table 5

Comparison of LBDQ Form XII Scores for
Ministers and Other Leaders

Structure

Entire Sample Ministers

Mean 38.2 38.7
S.D. 5.2 4.7
Range of Means 36.6 39.7

Consideration

Mean 40.1 42.5
S.D. 5.3 5.8
Range of Means 37.1 42.5

The follower's perception of ministers as high in consideration

behaviors appears to be associated with leaders who require the support

of their followers to hold their positions of leadership as the other

two professions highest on consideration behavior are community leaders

and labor union presidents.

Instruments

The LOC Scale

Benson and Spilka commented:

Rotter's scale defines external control in terms of luck, fate,
and chance. While it seems reasonable to argue that one who
places his fate in God's hands is externally controlled, he
may find options phrased in luck and chance terminology irrele-
vant. (1973, p. 308)

This comment appears to be particularly relevant to this study which

seeks to determine if there is a relationship between LOC and religious

denominations. It was therefore decided to devise and test a modified

LOC scale which would include the concept of God as an external source

of control.








Two pilot studies were conducted to determine whether a modified

LOC scale adding the concept of God as an external alternative would

change the nature of the results from the original scale and add to the

construct of LOC.

In pilot study number one, a modified LOC questionnaire was

developed which was very similar to Rotter's scale, but inserted God as

the external alternative (see Appendix A). For example, Rotter's

external alternative to question number two is, "Many of the unhappy

things in people's lives are partly due to bad luck." The modified

alternative to this question was, "Many of the unhappy things in people's

lives are partly due to the will of God." The two questionnaires

(Rotter's scale and the modified questionnaire) were administered to a

sample of 75 undergraduate business students and 32 members of adult

Sunday school classes at a local Methodist church. Half of the subjects

were administered Rotter's scale first, and half were administered the
modified scale first. A correlation matrix of the whole sample (Table 6)

revealed moderate correlations between like-numbered items on the two

scales, averaging .30 for 14 of the 23 questions (p < .05) while the

correlation matrix for the Methodist sample (Table 7) revealed that only

four of the like-numbered items correlated significantly.

These results appeared to indicate that individuals belonging to

religious groups discriminate between luck, fate, chance, and God as

external forms of control more than do groups of students. A t-test for

correlated groups indicated a significant difference between the total

scores for the two instruments for the whole sample (p < .001). The

correlation between the two test scores was .32 (p < .001) indicating a

shared variance of 10% between the two instruments. The results of this































NU


C)

















NN N


0%




m UC.
( co r.



COCO





m ino
C(M fj


I D


Ci nC ce W



mcn C-
rM oN N


01)0





'-'S,


C.,




h..h


O 0
c- CJ


c. nC)
cN C.4


0








rJ
CJ


0 CO




)0
to




Cin
cJ



r-


C--)
1





ic r
'.o t


i- CI


cm-i


VI
C

*r





0
0
U














0
4-























-)
tu














































O
U
'




4

























Cr
an











4J
















































01



fu


C
a












































S-.
C4
CZ
0






















1-)
0u
a,

.u




















'4-
10-
in
^ o
Cio
u)





















<*! U









































o
CMi







rC-

cm
CD













OJ























cl


in in

co



J* C C* U) .*O -. o .. o e- CV C.) 0 Un 40 1. G 0) rC- Cj mn
C r- C- -j 'J CM Ci


o\ M to to Cq
w uj, (i1A M



in COM
^' m en


C
0



4.

o






I-


f.
0









(U











,-














E
U
u











































c-
0

0





























43

















r-




C







E-
0C



c
0




43















a,





C0--





*rC



JC <








study appeared to indicate that the addition of God as an external source

of control resulted in a very different LOC score which shares little in

common with the original LOC scale. Further, part of the difference (as

expected) could be due to the fact that the subjects were given only an

external alternative (God) in the modified scale which did not tap the

whole of the LOC construct.

The second pilot study was designed to provide an instrument which

might measure the full dimension of the LOC construct by including the

original external alternatives fate, luck, chance, and powerful others,

plus the concept of God as still another external alternative. This

new instrument was reduced in length by removing the six filler questions

contained in the Rotter scale and the three questions dealing with the
student's perceptions of control by teachers which were considered

irrelevant to adult, non-student populations. The new scale also used a

Likert scale and split Rotter's internal-external alternatives into two

separate questions (20 internal and 20 external) and added 20 additional

items presenting God as the external alternative which were drawn from

the scale used in the first pilot study. The new scale consisted of

60 questions, 20 internal control choices, 20 external control choices

from Rotter's scale, and 20 external alternatives in which God is the

external control. The total score on the modified scale was computed

by reversing the scores on the internal alternatives and summing the

scores on the 60 questions (see Appendix A).
This scale and Rotter's scale were administered to 35 undergraduate

business students, alternating the order of administration in the same

manner as in the first study. The correlation of Rotter's scale with
the portion of the new questionnaire derived from Rotter's scale (the 20








internal and 20 external alternatives now using a Likert scale) was .76

(p < .001) which compares favorably with Collin's finding of .82 (1974).
The correlation between Rotter's scale and the total score of the new
scale containing the internal and both the external alternatives (60

questions) was .56 (p < .001).

The internal reliability (Cronbach's Alpha) was tested in both

pilot studies and in the final study. In the first study the Alpha for

the Rotter scale was .73, for the modified scale .80, and for both scales
combined .82. In the second study, the Alpha for Rotter's scale was .76,

for the modified scale without the God alternative .86, and for the
complete new scale, the Alpha was .91. The external alternative subscale

with God as the external alternative had an Alpha of .94. These findings
compare very favorably with those of other researchers who found the

internal reliability of Rotter's scale to vary between .65 and .79 with

the majority being in the .70's (Joe, 1971). From the results of the

pilot studies it was concluded that the addition of God as an external
alternative does have a possibility of adding another relevant dimension

to the construct of LOC, and that the new scale should be used in the

follow-on study.
Rokeach's Values Scales

Rokeach's (1969) Value Survey Instrument measures the individual's

value structure through the use of a "Terminal Values and an Instrumental
Values Scale." The subjects are presented with an alphabetical list of
18 terminal values and 18 instrumental value and are asked to rank the

18 values in each scale in order of importance. Rokeach maintains that
a hierarchy of individual values can be ascertained from these rankings
and used to determine the beliefs and attitudes of the subject population








and how these values provide a basis for the development of beliefs and

attitudes. The Terminal Values Scale refers to the preferred end-state

of existence while the Instrumental Values Scale measures the individual's

preferred modes of behavior. In the present study, certain Terminal and

Instrumental values will be measured and their relationship to LOC will

be examined.

Robinson and Shaver (1973) report a test-retest reliability of

". in the 70's" for the values scales. The two full scales contain

18 values in each scale, but for the purposes of this study, only eight
values were chosen from each of the scales in order to reduce the size

of the survey questionnaire. For the Terminal Values scale, salvation and

a sense of accomplishment were chosen as two of the variables based upon
the previous discussion of their possible relationship to LOC. Six other

values were chosen at random from the same scale to serve as a basis for

comparison. Similarly, in the Instrumental Values scale, the values

broadminded, capable, and logical were chosen because they are the compe-

tence values identified by Rokeach as varying significantly between the

Protestant denominations, and because of their previously discussed possi-

ble relationship to LOC. The five other values from the same scale

(forgiving, helpful, imaginative, independent, and obedient) were chosen

at random to serve as a basis for comparison. These eight instrumental
and eight terminal values were alternated randomly in forming the scale

used in this study with odd numbered questions containing the instrumental
values and the even numbered questions containing the terminal values.

Within this framework, the order in which the values appeared in each
question was determined randomly (see Appendix A).








The score for each value was computed by developing a skewed

symetrical matrix for the instrumental and terminal values. The columns

in the matrix were then summed and divided by eight minus any missing

values.
The Consideration and Structure Scales

The consideration and structure scales were taken from the similarly

identified subscales of Stogdill's (1963) Leadership Behavior Description

Questionnaire (LBDQ Form XII). Stogdill reported that the reliability

for the consideration subscales ranged from .76 to .87 with an average

reliability of .81, and that the reliability of the structure subscale

ranged from .70 to .80 with an average reliability of .76. Two versions

of these subscales were used in this study. The consideration and struc-

ture subscales using Stogdill's wording were used to measure the follower's

perceptions of the leader's behavior. The wording on both of the scales

was changed from the third person to the first person to obtain the two

scales used to measure the leader's perceptions of his own leadership

behavior (see Appendix A). The score for each of the subjects was ob-

tained by averaging the responses on each scale to minimize loss of data.

The subject's scores were utilized if a minimum of seven of the 10 items

were completed on each of the subscales. The internal reliability for

the scales in this study were .92 for consideration and .88 for the

structure subscale.
Biographical Data

The biographical data sheet was included in each of the follower's

questionnaires to gather data on variables of interest to this study (see

Appendix A). With the exception of religious denomination, all variables

contained in the data sheet were anticipated intervening or moderator





44


variables: age, sex, frequency of participation in church activity, and

length of association with the minister.













SECTION IV

RESULTS

LOC and Religion

The analysis of the relationship between LOC and religious denomina-

tions included an examination of the differences between churches within

each denomination, the differences between denominations, an examination

of possible moderating or intervening variables, a factor analysis of the

modified LOC scale, and a discriminant analysis to determine which

variables best predicted group membership.

Differences between the individual churches within the denominations

were examined through t-test on the variables LOC, salvation, a sense of

accomplishment, broadminded, and capable. None of the churches within

the denominations varied significantly on the LOC dimension, but the two

Presbyterian churches did vary significantly on the values capable and a

sense of accomplishment (p < .05).

Differences between the denominations were then examined by one-way

analysis of variance. An examination of the literature indicated that two

tests were appropriate for a posteriori contrasts when there were unequal

cell sizes, the LSDMOD and Scheffe (Nie et al., 1975). However, these

methods are the most conservative of the multiple range tests. The results

using both methods are as indicated by Table 8. The overall F-test

(p < .0001) indicates that, in this sample, the denominations differed

significantly on the LOC dimension. These findings support hypothesis

number one in that the denominations differ significantly on the LOC








dimension; however, the differences were not in the order predicted.

The predicted order, from the most external to the most internal, was

Baptist, Methodist, Lutheran, Congregational, Presbyterian, and Episco-

palian. The order as indicated in Table 8 is Presbyterian, Lutheran,

Episcopalian, Baptist, Methodist, and Congregational which is almost the

reverse of the order predicted.
Table 8

One-Way Analysis of Variance
LOC by Denomination


LSDMOD

Subset 1 CON
Subset 2 METH BAP EPIS
Subset 3 BAP EPIS LUT
Subset 4 LUT PRES


Scheffe

Subset 1 CON
Subset 2 METH BAP EPIS
Subset 3 BAP EPIS LUT PRES


Note: Overall F-test (p < .000) d.f. 5,235

The hypothesis was based upon Nunn's (1964) finding that powerless

individuals tend to prefer Baptist and non-sectarian churches, Rokeach's

(1969) findings that religious denominations varied significantly as to

the degree of importance they placed on certain values, and this author's

predictions of the relationship between LOC and Rokeach's values.

Because of these conflicting findings, the relationship between LOC

and these values was examined by product-moment correlation with results

as indicated in Table 9. The findings partially support hypothesis number

two in that the correlation between LOC and a sense of accomplishment was







































r- C C C
I'










0 tO1.0 n)
CM I- I- C
* I


4'
4'
4'
I-
C'.


(A CYJI'
C*j
E *
ws


I-
U

0



4-


t,-


L-J






u)




a,
cm
















LiL
*r-




















C



*r-
LU





o






5c

0



4-
U-
L10


4. )


- '-4- M > E )
i- e-- *- 0 t/1


a,, r'f oo tn'
* *




I I





. 4 0 rn
















Sc



















ln U J (A U)
C-CC

















4'Uur
.4' .4'


0)


- ,-1 -4-> E. A1
'-- 0 4/I
U


IV) W V) V
r-E


t)
L.

4-)


c-





cU
0










I-
t--



I
*r-










4-)
I-














4-)
I)





03
0I
0














*-
.la


r





0)
4-
0

4,-
-o







0
5--
S-
-o













)
(U











0(4








-.16 (j < .01), and between LOC and capable was -.17 (j < .01). However,

there was no significant relationship between LOC and the values broadminded

and logical. Hypothesis number three was supported by the finding that

LOC correlated .33 (e < .001) with salvation. These findings confirm

that externals place a higher degree of importance on salvation and a

lessor degree of importance on the competency variables sense of accomplish-

ment and capable than do internals.

As indicated in Table 10, hypothesis number four was partially sup-

ported. The denominations did differ significantly in the importance they

placed on salvation. However, the difference was significant only between

the Congregationalists and the rest of the denominations (p < .0001).

Table 10 also indicates that the denominations vary significantly on six

values other than salvation: broadminded, equality, an exciting life,

imaginative, obedient, and pleasure. There does appear to be a tendency

for the denominations to vary on a dimension that could be described as

liberal-conservative. The Baptists consistently placed a low value on
the variables boradminded, equality, exciting life, and pleasure, while

they placed a high value on obedient. The Presbyterians and the Episco-

palians occupied the liberal end of the scale by placing a high value on
broadminded, equality, an exciting life (except for the Episcopalians

who differed on this one item), and a low value on obedient.

The effect of other variables on LOC was examined by hierarchial

regression. As the author had no specific theory to support the order in

which the variables were to be added to the regression, the variables

were entered based on the probability of their having a causal relation-

ship with the other variables in the regression, i.e.; sex, age, frequency

of church attendance, and the interactions of sex by denomination, and













Table 10

Significant Multiple-Range Comparisons Between
Denominations on Value Variables


Broadminded (p <.0000)
Group 1 BAP LUT MET
LUT MET PRES EPIS CON

Equality (p <.02)
Group 1 BAP MET LUT
Group 2 LUT EPIS PRES CON

Exciting (p <.0000)
Group 1 LUT BAP EPIS MET
Group 2 BAP EPIS MET PRES
Group 3 PRES CON

Imaginative (.< .03)
Group 1 LUT MET BAP EPIS PRES
Group 2 MET BAP EPIS PRES CON

Obedient (p< .0000)
Group 1 CON EPIS PRES
Group 2 EPIS PRES MET LUT
Group 3 PRES MET LUT BAP

Pleasure (p <.0013)
Group 1 EPIS BAP PRES MET LUT
Group 2 LUT CON

Salvation (p< .0000)
Group 1 CON
Group 2 PRES EPIS MET BAP LUT


Note: Multiple-range tests were conducted by LSDMOD.








age by denomination (Nie et al., 1970). The results as shown in Table 11
support the hypothesis that LOC was related to denominational differences

as denomination accounted for 15% of the variance in LOC. The variables

sex, age, and frequency of church attendance made a small but significant
contribution to the variance in LOC (1%, p < .01) while the interactions

of sex by denomination and age by denomination contributed an additional

5.5% (p < .01).
Table 11

Hierarchial Regression of LOC with Possible
Moderating Variables


Variable R R Square Change in R Square

Denomination 15

Sex 06 15 0056

Age -08 15 003

Attendance 06 15 004

Sex x Denomination 17 02

Age x Denomination 20 035

Note: All variables are significant p < .01.

Note: Decimal points omitted.

A factor analysis of the LOC scale was conducted to determine if it

contained subfactors similar to those reported by past researchers and to

determine if these subfactors could clarify the relationship between LOC

and religion. The initial analysis indicated that 16 factors met the

Kaiser criterion of having an eigenvalue greater than one (Thorndike, 1978).

As the largest number of factors reported by past researchers was five

(Duffy et al., 1977), and as the author had reason to suspect that the








inclusion of the God external alternative might have added still another

factor, the factor analysis was repeated successively limiting the solu-

tion to 6, 5, 4, and 3 factors. Each of these four solutions was examined

and those items loading greater than .40 on each of the factors were

selected to represent the subfactors. The first factor remained stable

for all four solutions, and both the three and four-factor solutions had

a high face validity. However, the four-factor solution was chosen as

the best because the three-factor solution deleted the "system modifi-

ability" factor (Gurin et al., 1969) or the "politically responsive-

unresponsive world" (Collins 1974; Duffy et al., 1977). The items

comprising the four-factor solution and the item loadings on the four

factors are shown in Table 12.

A comparison of this factor analysis with those of past researchers

is as shown in Table 13. Factor I of the present study consists solely

of the 20 God external alternatives. Factor II contains questions 19,

23, 28, 29, 31, 37, 49, and 59, all but two of which are internal alter-

natives denying the importance of, or the existence of luck. The two

questions (23 and 29) are external alternatives dealing with accidents

as a controlling alternative, or "getting to be boss depends upon luck."

This factor is almost an exact duplication of Collins' (1974) Factor III

and contains all of Duffy et al.'s (1977) Factor I. It seems reasonable

to accept Collins' description of this factor as the "predictable-

unpredictable world" factor.
Factor III in this study was dominated by external alternatives,

containing only one internal and four external alternatives describing

Collins' "politically responsive-unresponsive world." Factor IV contains

six internal alternatives describing the belief that the individual's













Table 12

Four-Factor Solution to LOC Scale


Factor Factor
Question F a hrt Question III IV
I II III- IV I II III IV


-05 -05
-03 03
11 -12
11 02
06 -06
06 -12
24 -03
19 -18
06 14
16 01
23 00
07 -01
19 -15
30 -06
28 05
02 02
-02 02
07 04
27 -04
20 20


58
44
62
51
52
53
69
56
-02
15
-09
00
-13
05
06
04
07
-08
06


-10 24
10 -22
-04 27
24 -09
-03 30
-24 02
-19 04
13 10
42 04
52 -08
45 38
55 02
51 13
01 44
01 44
10 45
-05 45
30 42
-04 47


74 -11
66 -10
68 00
44 -10
74 11
46 07
53 15
62 07
70 06
55 25
50 14
72 03
55 02
59 10
59 -14
69 -02
64 06
76 06
44 02
46 05














Table 13

Comparison of Present Study Factor Analysis
With Previous Research Findings


Present Study Collins Study Duffy et al.'s Study


Factor II


Factor III


Factor I


Factor III


Factor IV


Factor III


Factor II


Factor IV


Factor II








rewards are obtained by being competent, and all but one of the items is

couched in general terms, i.e.; "people" or "we" and is similar to Gurin

et al.'s (1969) "control ideology." This study's factors did not include

the "easy-difficult" world reported by Collins and Duffy et al.

It is possible that the differences in the sample populations could

account for the failure to replicate this factor. Collins' population

consisted of 300 college students and Duffy et al.'s consisted of 275 Army

Special Forces personnel (Duffy used the same Likert scale format as

Collins). Collins' population appears to be limited in age and experience

while Duffy et al.'s male Army Special Forces personnel do not appear to
be especially representative of the general population. The sample

population in the present study are undoubtedly older than the other popu-

lations studied (the average age was 43), and had females in the sample

which Duffy et al. did not. Except for the possible confounding effects

of religious commitment, this study's population appears to be more

representative of the general population. However, other than for this one

factor, the factors in this study appear to be well supported by the

findings of the other two studies as only three items, number 29 in

Factor II and items 40 and 46 of Factor IV are not included in the factors

reported by Collins and Duffy et al. The degree of independence between

the four factors was also examined by product-moment correlation revealing
that the factors had a high degree of independence except for a moderate

correlation between Factors I and III (.29, p < .001).

The relationship between the LOC factors and religious denominations

was examined by multiple-range comparison (LSDMOD) which indicated that

only Factor I was significantly related to religious denominations (p <

.00001). However, the ordering of the denominations was essentially the

same as that for the total LOC scale (Table 14).





55


Table 14

Multiple-Range Comparison of Denominations
By LOC Factor I


Group 1 CON
Group 2 MET EPIS BAP
Group 3 EPIS BAP PRES
Group 4 PRES LUT

Note: Overall F-test (p < .000).

The relationship between the LOC factors and Rokeach's values (Table

9) does provide for a clearer understanding of the relationship between LOC

and the values. It appears that those who feel that God controls their

lives also place a low value on an exciting life, pleasure, capable,

logical, and a sense of accomplishment while they place a high value on

salvation and obedience. Strangely, they also place a low value on

equality.

Those who feel that their outcomes are controlled by luck, or that

the world is unpredictable (Factor II), predictably place a high value

on an exciting life and being imaginative, while placing a low value on

capable, obedience, and salvation. Unexpectedly, they do place a high

value on a sense of accomplishment which may be partially explained by

Rotter's (1975) comments on "defensive externals." It is possible, that

while the defensive externals value a sense of accomplishment, they use

luck as an external control to explain their failure to achieve.

Those who feel that they have little influence on world affairs or

the political system (Factor III) place a low value on equality, an

exciting life, and a sense of accomplishment. These individuals do place

a high value on salvation which appears to support Rokeach's (1969) state-

ment that, "those who value salvation have an other-worldly orientation














Table 15

Hierachial Regression of LOC Factor I with
Possible Moderating Variables


Variable r R Square Change in R Square

Denomination 21 21

Sex -09 21 003

Age 12 22 005

Sex x Denomination 24 01

Age x Denomination 26 025


Note:

Note:


All variables are significant p < .01

Decimal points omitted.








which would appeal to those who feel powerless and that they exert little

or no influence in affecting the course of political social events in

their society."
The relationship of other variables to LOC Factor I was examined by

hierarchial regression in the same manner as that for the total LOC

scale. The results in Table 15 indicate that the religious denominations

now account for 21% of the variance in LOC Factor I (p < .01) which is

6% more than that accounted for by denominations in the total LOC scale.

As in the case of the total LOC scale, sex and age made small but

significant contributions to the variance in LOC Factor I while the inter-

actions of sex by denomination and age by denomination only accounted for

an additional 3.5% of the variance (p < .01).

Discriminant analysis of the variables accounting for the differences

in denominations. In view of the numerous significant intercorrelations

between the LOC variables and the value variables (see Table 9), it seemed

necessary to heed Tatsuoko's (1970) warning that the danger of getting a

distorted picture of group differences tends to increase as the corre-

lations among the variables get larger. Accordingly, a discriminant analysis

was conducted to determine the differences between the denominations by

submitting the four LOC factors and the values involved in the original

hypotheses (a sense of accomplishment, broadminded, capable, logical, and

salvation) to a step-wise discriminant analysis (Nie et al., 1975). Three

discriminant functions were significant (p < .05) which, using Tatsuoka's

(1970) formula for determining the multivariate omega squared, indicated

that approximately 39% of the total variability in the discriminant

functions was attributable to denominational differences. The nature of

the three functions can be determined by an examination of the standardized








discriminant function coefficients in Table 16. Only two of the LOC

factors, Factors I and IV, were included in the significant discriminant

functions, and while the weights of Factor I remained above .50 in all

three discriminant functions, Factor IV only attained a maximum weight

of -.31 in the second function indicating that Factor IV (control

ideology) does not make a large contribution to the differences between

the denominations.

Discriminant function 1, accounting for 61% of the explained

variance between the denominations, was about equally dominated by sal-

vation and LOC Factor I with weights of .60 and .50 respectively. Dis-

criminant function 2, accounting for 26% of the explained variance, is

dominated by LOC Factor I with a weight of -.82 and some discrimination

is added by the values broadminded (-.49) and salvation (.41). Discriminant

function 3 which only accounts for 7% of the explained variance, is

dominated by the values salvation (-.85) and broadminded (-.84).
Table 16

Standardized Discriminant Function Coefficients for
the LOC Factors and Rokeach's Values

Function 1 Function 2 Function 3

LOC Factor I 50 -82 46
LOC Factor IV -12 -31 -06
Broadminded -22 -49 -84
Capable 02 02 -03
Salvation 60 41 -85
Percent of Variance 61 26 7


Note: Decimal points omitted.







Table 17

Denominational Centroids Based on Discriminant Functions


Group Function 1 Function 2 Function 3

Baptist .34 -.82 .15
Congregational -1.82 .03 .37
Episcopal -.11 -.25 -.33
Lutheran .69 -.26 .26
Methodist -.03 .37 -.08
Presbyterian -.10 -.67 .04

Figure 1 contains. the plots of the group centroids of each

denomination using coordinates obtained by combining discriminant functions

1 and 2, 1 and 3, and 2 and 3. Remembering that function 1 is a salva-

tion-LOC dimension, and that function 2 is primarily a LOC function, and

that function 3 is primarily a values function, looking at the figure

reveals that quadrant I may be designated as the high salvation-low LOC

quadrant. Those denominations falling in that quadrant are characterized

by placing a high value on salvation, and being internal on the LOC

dimension. Quadrant II is characterized by low LOC scores and low sal-

vation scores. Quadrant II is characterized by containing those denomi-

nations with high LOC scores (externals) who place a low value on sal-

vation. Quadrant IV is characterized by those who place a high value on

salvation and are external.

From the plots of the centroids on functions 1 and 2, it can be seen

that the Baptists and the Presbyterians are the most unlike on those

dimensions in that the Baptists are more internal and place a higher

value on salvation than the Presbyterians. Using the Baptist centroid as

a reference, it can be seen that they are most like the Methodists

followed by the Lutherans and the Episcopalians.


















LOW LOC
HIGH SALVATION


-.8




.8



.6


+
.4



.2








.2


HIGH
SALVATION


SALVATION
G5 AS


.3
3
A3


-31


HIGH LOC
LOW SALVATION
F =FUONC I x2
0 =FUNC Ix3
A =FUNC 2 3


I = Baptist
2= Congregytional
3= Episcopol


4= Lutheron
5= Methodist
6= Presbyterion


Figure 2. Plot of Denominational Group Centroids.


5 A2


L 6


----


.2 .4


-.4 -.2


Ae 1








When the centroids are plotted using functions 1 and 3, it is

possible to discern the similarities and dissimilarities when LOC is

deemphasized and the values salvation and broadminded are emphasized.

Now, the Baptist are most unlike the Episcopalians with the Baptists

placing a relatively lower value on salvation and broadminded than the

Episcopalians. The Baptists are almost equidistant from the other

denominations, but slightly closer, or more similar to the Lutherans. It

is important to note that all the centroids are now much closer together

confirming the previous findings that the differences between the

denominations is greater for the LOC dimension than for the values

dimensions.

When the centroids are plotted utilizing functions 2 and 3 which

again emphasize LOC Factor I and the values, the large difference between

the Baptists and the Presbyterians reappears and the Baptists again

become most like the Methodists and the Congregationalists, and least like

the Presbyterians and Episcopalians. It thus becomes evident that the

LOC Factor I dimension best discriminates between the denominations.

However, examination of the "confusion matrix" (Table 18) which permits

a comparison between the probability of correctly predicting group member-

ship with the discriminant function versus the probability of correctly

prediting group membership by chance based upon size, indicates that the

discriminant functions do not predict any better than chance on some

denominations, for example, the discrimant functions only correctly

classified 7% of the Baptists as belonging to the Baptist denomination

versus an expectation of correctly classifying 17% correctly by chance

based upon group size. Similarly, the discriminant functions accurately

predicted group membership for the Episcopalians with a 19% accuracy








while a chance prediction based upon group size would predict with an 18%

accuracy. However, the discriminant functions do classify the remaining

four denominations better than chance.
Table 18

Confusion Matrix Comparison of Accuracy of Discriminant
Function With Prediction by Chance

Predicted Group Membership
Actual Group N 1 2 3 4 5 6 Chance

1 41 7 2 5 12 66 7 17
2 14 0 64 7 0 29 0 06
3 43 5 2 19 7 51 16 18
4 32 6 3 3 31 47 9 13
5 79 6 3 5 4 81 1 33
6 33 12 9 12 18 18 30 14

Note: Decimal points omitted. Predicted group membership by percent.

In summary, with the exception of the Baptists and the Episcopalians,

the discriminant functions do classify subjects correctly by denominations

better than chance with the major discriminating variables being LOC

Factor I. When the denominations are classified by values as in functions

1 and 3, the centroids of the denominations become much closer together.

The values contributing the most to the discriminant functions are sal-

vation, and broadminded.

LOC and Leadership

This subsection deals with the investigation of the relationship

between: 1) The follower's LOC and his perceptions of the leader's be-

havior, 2) the leader's LOC and his perceptions of his own behavior, 3)

the follower's and the leader's LOC and the follower's perceptions of the

leader's behavior, and 4) the relationship between the follower's LOC and

the leader's perceptions of his own behavior. The appropriate model for








analysis will be discussed, and then the above relationships will be

examined first through zero-order correlation, and then by multiple

regression using total scale scores and factor scores for each of the

three dimensions LOC consideration and structure.

The Model

Dansereau and Dumas (1976) maintain that much of the conflicting and

inconsistent findings reported in past leadership studies may be due to

faulty assumptions as to the basic nature of leader behavior and the

follower's perceptions of that behavior. From this author's interpre-

tation of Dansereau and Dumas' paper, it appears that three basic assump-

tions variously made by past researchers are: 1) that leaders behave

homogeneously towards subordinates regardless of the effects of the

organizational setting, or the behavior of the subordinates. 2) That

leader behavior is affected by the organizational setting and the be-

havior of the subordinates as a group, but the leader behaves homogene-

ously towards subordinates within the group, or 3) that leaders behave in

a heterogeneous manner towards individual subordinates in the unit.

Dansereau and Dumas maintain that past researchers have often measured

leadership variables at one level of analysis and drawn conclusions based

upon assumptions which apply at another level.

In reviewing the past literature, it appears that this may well be

the case. For example, Durand and Nord (1976) assumed that leader

behavior was heterogeneous and conducted their analysis at the dyadic

level by measuring the relationships between the leaders and the followers

within one organization. This analysis disregarded any possible effects

of the organizational setting on the leader-follower relationship. In

his subsequent research, which entailed more than one organization,








Nord (Note 1) found that the signs of the relationships between the LOC
factors and consideration and structure differed between organizations.
It is possible, that while the leader-follower interactions may result in

heterogeneous leader behavior, still other factors at the organizational

level are operating to cause a certain amount of homogenity in the
leader's behavior.

Similarly, Evan's (1974) analysis appears to have been based upon

assumptions that leader behavior is homogeneous in that he conducted his
analysis of the relationship between LOC and leader behavior by averaging
responses across units, thus ignoring the possible effects of the organi-
zation and the leader-follower interactions on leader behavior. Goodstadt

and Hjelle (1973) based their analysis on the dyadic relationship, but
drew their conclusions based upon the assumption that leader behavior was
homogeneous. Pryer and Distefano (1972) apparently assumed that leader
behavior was homogeneous within the organization as their analysis was

based upon three organizational levels within a hospital. Unfortunately,

they did not specify how many leaders were being described by the fol-

lowers at each of the three levels. It is possible that at one level,

the findings were based upon group-level data, while at another level,

the data may have been based upon the dyadic relationship.

Referring to Table 3, it can be seen that Durand and Nord's findings
based upon a dyadic analysis have been compared with Evan's data based
upon an analysis which did not take into account the possible effects of
the organization, or the leader-follower interactions on the relationship
between LOC and leader behavior, and Pryer and Distefano's findings which


Personal communication, September 18, 1978.


INord, W.







considered only the effects of the organization (and possibly, the leader-

follower interaction) on the relationship, all of which may account for

the conflicting findings.
The analyses in the present study were based upon Dansereau and

Dumas' (1976) recommendations which are as follows: 1) the first, or

broadest level of analysis was conducted by computing zero-order corre-
lations between the variables using the difference between the individual
scores and the grand mean. If significant correlations between LOC and

leader behavior were found, it might be assumed that some general rela-
tionship existed which was not affected by the organizational setting or
the leader-follower interaction. 2) The next step in the analysis was to
examine the relationships between the variables utilizing the differences

between the mean of each group and the mean of the group means as the
basis for the analysis. If significant correlations were found at this

level where none were found at the previous level of analysis, it would
appear to indicate that factors related to the organizational setting, or

to the collective influence of the followers might be responsible for the

LOC-leader behavior relationship. 3) The last level of analysis was con-

ducted utilizing the differences between the individual follower's score
and the group mean score as the basis of analysis. Significant relation-

ships at this level of analysis would appear to indicate that the leader's
behavior was heterogeneous in regards to the subordinates. It is, of
course, possible that significant relationships might be found at all
three levels of analysis indicating that, while leader behavior tends to
be homogeneous, factors in the organizational setting and/or the leader-
follower interactions cause the leader to behave (or be perceived by the

followers to behave) in a heterogeneous manner.








It might be possible for example that external leaders would tend to

behave in the same manner based upon their LOC. However, constraints or

other factors operating at the organizational or dyadic level, may result

in heterogeneous behavior towards the group as a whole, or towards sub-

ordinates on a dyadic leader-follower basis. For ease of discussion, the

above steps in the analysis will be referred to in the remainder of this
study as level one, level two, and level three.

The relationship between the follower's total LOC scale score and
consideration and structure was examined using the above method. The

correlations between the total LOC scores and consideration and structure

at level one were not significant. However, consideration and structure
were moderately and significantly correlated (r = .49, p <.001). This is
similar to the findings of other researchers who report moderate to high

correlations between the two scales (Kerr, Schriescheim, Murphy, &

Stogdill, 1974; Schriesheim, House, & Kerr, 1976).

The relationship between LOC and the two scales was then examined at

level two (the group level) with the finding that while LOC's relation-

ship approached significance with structure (r = .41, E <.095), there was
no significant relationship between LOC and consideration. Similarly,

no significant relationships were found between the scales at level three

(the individual differences level). Hypotheses six and seven were not sup-
ported by these analyses.
The Relationship Between the Follower's
Four LOC Factors and Consideration
and Structure
A product-moment correlation of the four LOC factors with the con-

sideration and structure scales at level one indicated that no significant

relationships existed between the four LOC factors and the consideration








and structure scales. At level two (the group level), a significant

relationship was found between LOC Factor II (luck external) and struc-

ture (r =-.62, p < .03). In this sample population, there appears to be

a general tendency for those who feel that their outcomes are controlled

by luck to perceive their leaders as less structuring than those followers

who feel that their outcomes are not controlled by luck.
Table 19

The Relationship Between the Follower's LOC Factors and
Consideration/Structure at the Individual
Differences Level (Level Three)

Consideration Structure
LOC Factors I II III IV I II III IV
Groups
11 -40*

31 -38* 35*

42 84* -84*

51 26*

61 -71***

< .05
** < .01

*** < .001
Note: Decimal points omitted

Moving to level three (the individual differences level), signifi-

cant relationships were found in five of the twelve groups as indicated

in Table 19. The pattern of these relationships, and the fact that they

existed in only five of the twelve groups, appears to indicate that

either: 1) that leaders behave in a heterogeneous manner towards their

subordinates, 2) that the existence of other factors operating within








the group interact with the dyadic relationship to produce individual
differences in the follower's perceptions of the leader's behavior, or 3)

that because of the large number of relationships examined, the significant

relationships occurred by chance.

Factor Analysis of Consideration
and Structure Scales

As it was possible that the consideration and structure scales might

contain subfactors which are differentially related to the LOC subfactors

and thus mask or confound the relationship, a factor analysis of the con-

sideration and structure scales was conducted. Both the scales were com-

bined and factor analyzed by varimax rotation. The solution identified

four factors having an eigenvalue greater than one, and accounting for 62%

of the variance (Table 20). Further each of the first two factors con-
tained a mixture of consideration and structure items indicating, that

for this sample population, the use of the original classification of the

individual scales into consideration and structure scales was inappro-

priate. Factor IV contained two items with loadings on that factor of

.49 and .52, but these items also had high factor loadings on Factor I of

.65 and .70 indicating a degree of redundancy between Factors I and IV.

Because of this redundancy, it appeared that a three-factor solution

might be more parsimonious.

The three-factor varimax solution provided the factors as shown in

Table 21. This solution appears to be clearer and more parsimonious in

that the two items loading on Factor IV in the principle components solu-

tion moved to Factor I with loadings of .71 and .76. Factor I contains

four consideration and three structure items. Factor III contains seven

of the original structure items and one consideration item (item C3 was
eliminated because of its nearly equal loadings on Factors III and IV.














Table 20
Four-Factor Principle Component Solution for Combined
Scales and Consideration and Structure


Factors I II III IV


Item


C7 82
C8 78
C9 65 49
S8 76
59 85
S10 59
C10 70 52
C3 51
C5 56
S1 46
S2 49
S3 55
S4 50
55 48
56 56
S7 56
C1 70
C2 72
C4 76
C6 50
Percent of
Variance 45 17 10 8


Note: Decimal points omitted.













Table 21

Three-Factor Solution for Combined Scales of
Consideration and Structure


Factors I II III IV

Item

C7 81
C8 81
C9 71
C10 76
S8 70
S9 85
S10 57
C1 67
C2 68
C3 43 46
C4 74
C6 53
C5 52
Sl 46
S2 49
53 54
S4 49
55 50
56 57
57 55

Percent of
Variance 63 21 16


Note: Decimal points omitted.








Factor I is difficult to assess because of the nature of the items

comprising the factor. Factor III, containing only one item from the

consideration scale, may logically be called the "structure factor," and

Factor II appears to consist of items that emphasize the interpersonal

relationships dimension of the consideration scale. There is also a

moderate correlation between the three subscales (Table 22), but the

correlations are not as high as that between the original consideration

and structure scales (that R was .49).
Table 22

Correlations Between Follower LBDQ Factors


Factor 1 2 3

1 34** 37**
2 34** 43***
3

**P < .01
*** < .001
Note: Decimal points omitted.

The Relationship Between the
Follower's LOC and LBDQ Factors

The analysis of the relationship between the four LOC factors and

the three LBDQ factors at level one found that LOC Factor II (luck exter-

nal) was significantly and negatively related to LBDQ Factor III (struc-

ture) (r = -.14, p < .05). At level two (group level), LOC Factor II

(luck external) was found to be negatively and significantly correlated

with LBDQ Factor I (mixed consideration and structure) (r = -.53, p < .038),

and with LBDQ Factor III (structure) (r = -.57, p < .027). At the indi-

vidual differences level (level three), significant relationships were

found in five of the twelve groups. Four of the five groups were the








Table 23

The Relationship Between the Follower's LOC and
LBDQ Factors at Two Levels
LOC LBDQ Factors
Factors I II III

At the Group
Level Three
I
II -54* -57*
III
IV

At the Individual
Differences Level

I -35*
Group II
11 III -49*
IV

I 72*
Group II 75*
12 III
IV

I
Group II -40*
31 III
IV 38*

I 85*
Group II
42 III
IV

I
Group II
61 III
IV 61** -53* -65**

< .05
** p < .01
Note: Decimal points omitted








same groups in which significant relationships were found in the previous

analysis. The findings at this level lead to the same alternative con-

clusions as in the previous analysis: 1) that leaders behave in a hetero-

geneous manner towards their subordinates, 2) that the existence of other

factors operating within the group interacts with the dyadic relationship

to produce individual differences in the follower's perceptions of the

leader's behavior, or 3) because of the large number of relationships

examined, the significant relationships occurred by chance (see Table 23).

Table 24 presents the relationship between the LOC and LBDQ factors

from a different perspective. In this table it can be seen that only

LOC Factor II has a significant (negative) relationship at the group level

of analysis. However, at the individual differences level, only two

significant relationships were found between LOC Factor II and the LBDQ

factors (one positive and one negative). LOC Factor I had no significant

relationships with the LBDQ factors at the group level, but significant

relationships were found within the groups for LBDQ Factors II and III.

Here again, the direction of the relationships were not consistent. LOC

Factor IV similarly had no significant relationships with the LBDQ

Factors at level two (group level), but had significant relationships with

the LBDQ factors at the individual differences level and again, the direc-

tion of the relationships were inconsistent.

From these findings, it appears that: 1) there is a general

tendency for luck external followers to perceive their leaders as less

structuring than do luck internal followers. Further this same relation-

ship becomes stronger when measured at the group differences level and

becomes inconsistent at the dyadic level in that significant relationships

were only found in two of the twelve groups and the direction of the








relationship differed between the two groups, and 2) luck external

followers tend to perceive their leaders as exhibiting less of the mixed

consideration and structure behavior than do luck internal followers. As

no significant relationship was found between these two factors at level

one, it would appear that variables operating at the group, or at the

dyadic level would be operating to form this relationship. As no

significant relationships were found at the dyadic level, it seems

logical to argue that characteristics within the leader, or follower,

and variables at the organizational level are combining to produce homo-

geneous perceptions of the leader's behavior by the followers.

Table 24

Relationship Between the Follower's LOC and
LBDQ Factors at Two Levels

LOC LBDQ Factors by Level Three
Factors I Level II III I Lev III III

I ++

II +-

III

IV + +- -

Note: At the (Level Three) level, the signs indicate the direction
of each significant relationship found within a group.

The Relationship of Tenure to the
LBDQ Factors

Tenure was entered as a predictor variable into a series of multiple

regression equations with the differences between the leader's and the

follower's scores on each of the three LBDQ factors serving as the cri-

terion variable. The findings were that tenure accounted for small, but

significant amounts of variance in the cases of LBDQ Factors I and II








(r = .005, p < .01), and (r = .007, p < .05). An examination of the
simple correlations indicated that, as tenure increases, the followers

tend to perceive their leaders as exhibiting more mixed consideration-

structuring behaviors, and to be more interpersonally involved. While

the amount of variance accounted for by the effects of tenure were small,

the analyses supported hypothesis eight.
The Relationship Between LOC and the
Leader's Perceptions of His Own
Behavior
Because of the finding that the consideration and structure scales

were comprised of three factors, the relationship between the leader's
LOC factors and each of the original consideration and structure scales

was not examined. A product-moment correlation of the four LOC factors

with the three LBDQ factors indicated that LOC Factor II (luck external)
correlated significantly with LBDQ Factor II (interpersonal relations)
(r = -.46, p < .02), and with LBDQ Factor III (structure) (r = -.57,

S< .005).

These findings provide partial support for hypothesis number five in

so far as LBDQ Factor II can be assumed similar to consideration, and

LBDQ Factor III can be assumed to resemble structure. The findings

indicate that internal leaders perceive themselves to be more structuring
and more interpersonally involved than do external leaders.

Leader's Behavior and the Leader's
Description of His Own Behavior

The degree of agreement between the follower's perceptions of the

leader's behavior and the leader's description of his own behavior was

examined by correlating the differences between the follower's group mean
scores and the mean of the groups scores with the difference between the

leader's score and the mean of the leaders' scores. No significant








relationships were found indicating that there is no relationship between

how the leader thinks he behaves towards the group and the followers'

perceptions of their leader's behavior.
Regression Analysis of the Relationship
Between LOC and Leadership

As the most consistent relationships between the LOC and LBDQ were

found at the group level, it appeared appropriate to move from an examina-

tion of the zero-order correlations between the LOC and LBDQ factors to
a regression analysis at the group level. The following are the findings

which resulted from entering the follower's LOC and LBDQ factors measured
as the differences between the group means and the mean of the group means

and the leader's LOC and LBDQ factors measured as the difference between
the individual leader's scores and the mean of the scores for the leaders

on each of the variables. In all the following analyses the stepwise
regression method was utilized in which the computer chose the most appro-
priate variable to enter into the equation at each step (Nie et al., 1975).

The first analysis used the follower's LBDQ factors as the criterion
variables and the leader and follower's LOC factors as the predictor

variables in order to determine which of the LOC factors best predicted

the follower's perceptions of the leader's behavior. The results of the
first regression analysis with LBDQ Factor I as the criterion variable are

as shown in Table 25. The follower's LOC Factor I was the first variable
selected, and while the relationship between LOC Factor I and LBDQ

Factor I was not significant by itself, the inclusion of the leader's LOC
Factor IV in the equation produced an equation which accounted for 56% of
the variance in the follower's perception of the leader's behavior as

measured by LBDQ Factor I (e < .05). The third variable to enter the

equation was the leader's LOC Factor II which accounted for an additional








20% of the variance (p < .01). The leader's LOC Factor I accounted for

an additional 9% of the variance (p < .01), and the follower's LOC Factor

I plus the leader's LOC Factor III accounted for an additional 1% of the

variance (P < .05). Thus, a combination of the leader and follower's LOC

factors accounted for 86% of the variance in the follower's perceptions

of the leader's behavior as measured by LBDQ Factor I.
Table 25

Correlation Coefficients of Supervisor and Follower LOC
Factors With the Follower's Perceptions of the
Leader's Behavior


Criterion Variable: LBDQ Factor I
Predictor Variables R Square

Follower LOC Factor II 28ns
Leader LOC Factor IV 56*
Leader LOC Factor II 76**
Leader LOC Factor I 85**
Follower LOC Factor I 86*
Leader LOC Factor III 86*

Criterion Variable: LBDQ Factor I
Predictor Variables R Square

Leader LOC Factor II 56**
Leader LOC Factor III 61*
Follower LOC Factor III 67*


Change in R Square Simple R
28 -53
28 44
20 -18
09 12
01 -08
26


Change in R Square Simple R

56 -75
05 -22
06 -15


I


< .05
** P < .01
Note: Decimal points omitted.

An examination of the simple correlations indicated a case of "net

suppression" (Cohen & Cohen, 1975) in the case of the follower's LOC

Factor II and the leader's LOC Factor IV. The simple correlation between

the follower's LOC Factor II and LBDQ Factor I was -.53 and that of the

leader's LOC Factor IV with LBDQ Factor I was .44. Each of the variables

accounted for 28% of the variance in the leader's behavior as perceived








by the follower's on the dimension of LBDQ Factor I. Followers who

believe that luck controls their lives tend to perceive their leaders as

exhibiting less of the mixed consideration-structure behavior than do

followers who do not believe that luck controls their lives. However, this

effect appears to be offset by the indications that leaders who feel that

competency will not gain the rewards (competency externals), tend to be
perceived by their followers as exhibiting more of the mixed consideration-

structure behavior than leaders who are competency internals.

This same regression analysis indicates that leaders who are luck
externals tend to be perceived as exhibiting less of the mixed consideration-

structure behavior than leaders who are luck internals and leaders who
are God externals tend to be perceived as exhibiting more of the mixed
consideration-structure behavior than leaders who are God internals. As

indicated in the table, the follower's LOC Factor I and the leader's LOC

Factor II made a minimal, but significant contribution to the prediction
of the follower's perceptions of their leader's behavior as measured by

LBDQ Factor I.

The results of the analysis using the follower's LBDQ Factor II are
as indicated in Table 25. The first predictor variable to enter the

equation was the leader's LOC Factor II (luck external) which accounted

for 56% (p < .01) of the variance in the leader's behavior on the LBDQ
Factor II (interpersonal relations) as perceived by the followers. The
leader's LOC Factor III (system external) accounted for an additional 5%

of the variance (p < .05), and the follower's LOC Factor III for 6%. The

leader's two LOC factors plus the follower's LOC Factor III thus accounted
for 67% of the variance in the leader's behavior as measured by LBDQ
Factor II as perceived by the followers.








An examination of the simple correlations between the predictor and

criterion variables indicates that: 1) leaders who are luck externals

tend to be perceived by their followers as less interpersonally involved

with their subordinates than are luck internal leaders. 2) Leaders who

are system externals tend to be perceived by their followers as less

interpersonally involved than are system internal leaders, and 3) similarly,

system external followers tend to perceive their leaders as less inter-

personally involved than do system internal followers.

The leader's perceptions of his own leadership behavior was examined

in the same manner as in the previous analysis with the leader's self-

reported behavior as measured by the three LBDQ factors serving as the

criterion variables and his LOC factors serving as the predictor variables.

The results in Table 26 appear to indicate that: 1) the leader's self-

reported behavior on the LBDQ Factor I and II dimensions were not signifi-

cantly related to his LOC factors, 2) 61% of the variance in the leader's

self-reported behavior on LBDQ Factor III was accounted for by his LOC

Factor IV (p < .01), an additional 4% of the variance was accounted for by

his LOC Factor II (p < .01), and a small, but significant amount of

variance was accounted for by his LOC Factor I.
Table 26

Correlation Coefficients of Leader LOC With His
Perceptions of His Own Behavior

Criterion Variable: LBDQ Factor III
Predictor Variables R Square Change in R Square Simple R
Leader LOC Factor IV 61** 61 -78
Leader LOC Factor II 64** 03 -47
Leader LOC Factor I 65* 01 -17

p < .05
** P < .01
Note: Decimal points omitted.








An examination of the simple correlations appears to indicate that:
1) competency external leaders (LOC Factor IV) tend to perceive themselves

as less structuring than do leaders who are competency internals, and 2)

luck external leaders (LOC Factor II) tend to perceive themselves as less
structuring than do leaders who are luck internals.
The relationship between the leader's perceptions of his own behavior
and the LOC of his followers was also examined in a similar manner with
the leader's self-reported behavior on the three LBDQ dimensions serving
as the criterion variables and the follower's LOC factors serving as
predictor variables. The results in Table 27 appear to indicate that:
1) a combination of the follower's LOC Factors I and III account for 52%

of the variance (p < .05) in the leader's self-reported behavior as
measured by LBDQ Factor I, 2) the follower's LOC did not significantly
predict the leader's self-reported behavior as measured by LBDQ Factor II
(interpersonal relations), and 3) the follower's LOC Factor IV (competency
external) accounted for 52% of the variance (. < .01) in the leader's self-

reported behavior as measured by LBDQ Factor III (structure), LOC Factor
I (God external) accounted for an additional 14% (p < .01), and the
follower's LOC Factor III (system external) for still another 4% (p < .05).
An examination of the simple correlations appears to indicate that:

1) leaders who followers are God externals, or system externals, tend to
describe themselves as exhibiting more mixed consideration-structure
behavior (LBDQ Factor I) and more structuring behavior (LBDQ Factor III)
than do leaders whose followers are God or system externals, and 2)
leaders who followers are competency externals tend to describe themselves
as less structuring than do leaders whose followers are competency
internals.








Table 27


Correlation Coefficients
Leader's Perceptions


Criterion Variable: I


of Follower LOC With the
of His Own Behavior


LBDQ Factor I


Predictor Variables R Square Change in R Square Simple R
Follower LOC Factor I 31ns 31 55
Follower LOC Factor III 52* 21 03

Criterion Variable: LBDQ Factor III
Predictor Variables R Square Change in R Square Simple R


Follower LOC Factor IV
Follower LOC Factor I
Follower LOC Factor III


52**
66**
72**


52 -72
14 64
06 13


* e < .05
** < .01


Note: Decimal points omitted.













SECTION V

DISCUSSION

LOC and Religion

Summary of Findings

In summary, an examination of the relationship between LOC and

religion found:

1. That the six Protestant denominations differed significantly on

the LOC dimension, and that the "God external" subfactor accounted for

this difference.
2. While the denominations differed significantly on the LOC

dimension the ordering of the denominations was not as predicted by

hypothesis number one.
3. As predicted by hypothesis number two, externals were found to

place a higher degree of importance on the value salvation and lower

degree of importance on the values ofa sense of accomplishment and capable

than did internals, but contrary to the prediction, no significant rela-

tionship was found to exist between LOC and the values broadminded and

logical.
4. While hypothesis four, predicting a significant difference

between the denominations on the importance of salvation, was supported,

the only significant difference existed between the Congregationalists and

the rest of the denominations.
5. The factor analysis of the LOC scale produced a factor structure

very similar to that found by Collins (1974) and Duffy et al. (1977).








6. Demographic variables sex and age made only small, but signifi-

cant contributions to the variance in LOC.
7. Discriminant analysis indicated that LOC Factor I accounted for

most of the differences between the denominations, and that there was

very little difference between the denominations on the values.

Discussion of the Findings

The use of the modified LOC scale which added the concept of God as

an additional form of external control, did discriminate significantly

between the six major Protestant denominations while the use of the

unmodified scale did not. This finding appears to support the contentions

of Glock and Stark (1966), and Poppleton and Pilkington (1963) that

Protestant denominations differ among themselves on important dimensions,

and that to lump the Protestant denominations together as has been done
in some past research, is inappropriate.
Further, the results of this study appear to support Benson and

Spilka's (1973) contention that the Rotter scale may not tap the full

dimensions of the LOC construct in that individuals who are external, in

that they feel that God controls their lives, would not be identified by

their responses to the Rotter scale. In the Rotter scale, subjects are

only provided with a choice between luck, chance, and powerful others as

external controls over their outcomes. This contention appears to be

further supported by the finding that the LOC scale used in this study

correlated only moderately (r = .56, p < .001) with Rotter's scale indi-
cating the possibility that the new scale added an additional dimension

to the LOC construct. The failure of the Rotter scale to tap the full
dimensions of LOC by including the concept of God as an external alter-

native may also explain why the efforts of past researchers (using








Rotter's scale) failed to find differences between the religious denomi-
nations on the LOC dimension.
Several factors involved in this study may contribute to an explana-

tion for the failure to predict the ordering of the denominations on the
LOC dimension. First, the sample population was skewed in the direction
of frequent church attendance as evidenced by the fact that 90% of the
sample reported attending church, and/or Sunday School more than once a
week. Further, the members of all the denominations except the Congre-
gationalists placed an equal value on the variable salvation. Rokeach
(1969) reported that many of the value differences between the denomina-
tions disappeared for those subjects who attended church frequently, and
that the importance of the values examined by hypothesis number one de-
creased linearly with a decrease in frequency of church attendance.
A replication of Rokeach's findings was attempted by dividing the

present sample population into two groups, those attending church, and/or

Sunday School once a week or more, and those attending less frequently
than once a week. The N of the former group was 217 and that of the
latter group was 24. T-tests for significant differences between these
two groups on the four LOC factors and the values contained in the values
scale revealed significant differences only on the values sense of accom-
plishment (p < .04), imaginative (p < .01), and pleasure (p < .01), all
of which were more important to those who attended church and/or Sunday
School less frequently. Therefore, the unexpectedly high frequency of
church and/or Sunday School attendance appears to have dampened or dimin-
ished the denominational differences on the values dimensions whose
hypothesized relationships with LOC were used as a basis for the predic-
tions of denominational differences.







It is also possible that administering the questionnaires through

the church organizations caused the subjects to bias their responses

reflecting their denominational theology rather than values. If this is

the case, the ordering of the denominations in Table 12 may be explained

by the differences in the denominational theology rather than by the

hypothesized relationship between the values and LOC. The overlapping

groupings of the denominations in this study ranging from the most inter-

nal to the most external on the LOC dimension were Congregational, Metho-

dist, Episcopalian, Baptist, Presbyterian, and Lutheran. The Lutherans
and Presbyterians were on the external end of the continuum because their

doctrine emphasizes that God is all powerful and predestines, or controls,

individual destinies (Calvinist doctrine). Further, salvation is a free

gift to Grace which is given, or denied, individuals for reasons known

only to God. All that man can do to achieve his rewards is to have

faith.

Moving towards the internal end of the continuum, the Baptist doc-

trine explains that, while salvation is a free gift of God, man must do

his part by actively accepting the gift and turning towards obedience

and repentance. The Episcopalians accept basically the same doctrine as

the Baptists except that there is not quite the same emphasis on individual

responsibility. Rather it is a group responsibility to accept the free

gift of salvation and its attendant obligations. This emphasis on the

two-way convenant gradually reduces as one proceeds from the Episcopalians

through the Methodists to the Congregationalists who deemphasize the

value of salvation and the controlling influence of God, and increase the
emphasis on the humanistic and ethical aspects of Christianity. This








explanation (Note 2) is consistent with the findings of this study that

the ordering of the denominations on the LOC dimension from the most

internal to the most external was Congregational, Methodist, Baptist,

Episcopalian, Lutheran, and Presbyterian.

From the results of this study, it appears that there is a moderate,

significant relationship between religion and LOC. However, this study

had access to a limited population sample and the findings may have been

confounded by regional culture. Sims and Baumann (1972) found evidence

to suggest that Southerners believe that God controls their lives more

than do Northerners. Still unanswered is the question of whether religious

socialization may be an antecedent of the LOC personality trait, or
whether individuals tend to seek a religion which meets the needs formed

by their personalities.

The next step in the examination of the relationship between LOC and

Religion should be to expand the sample population to include a popula-

tion which will include more possible regional cultural differences. The

involvement of theologians will also be necessary for a more detailed, and

in-depth analysis of the implications of the relationship between LOC and

Religion.

From the standpoint of generalized research into LOC, it appears that

more attention will have to be given to the construct. While there is some

general agreement as to the nature of the LOC factors, these factors are
not stable and may vary from one sample population to another. As the


The author is indebted to Dr. Hill of the Religion Department,
University of Florida, for his personal communication explaining the
findings discussed above, September 21, 1978.








composition of these factors varies, it would seem to be important to
determine the specific factor structure for each population under study.

While larger sample sizes are required to support a factor analysis, to

assume that a particular factor structure exists without testing that

structure does not appear to be tenable. Further, the use of the total
scale appears to mask important relationships as was demonstrated in this

study.
LOC and Leadership

Summary of the Findings
An examination of the relationship between LOC and the follower's

perceptions of the leader's behavior, and LOC and the leader's perceptions
of his own behavior found:
1. No significant relationship between the three scales LOC, con-

sideration, and structure at any of the three levels of analysis. There-

fore hypothesis six and seven were not supported.
2. A significant relationship between LOC Factor II (luck external)

and structure (r = -.62, p < .03) at the group level of analysis (level

two). Also, at the individual differences level (level three), signifi-

cant correlations were found between the LOC factors and consideration
and structure in five of the twelve groups. However, the relationships
at this level were inconsistent across the groups. For example, LOC
Factor I correlated positively with consideration in one group, but no

other significant relationships between LOC Factor I and either consider-
ation or structure were found in any of the other groups (see Table 19).
3. A factor analysis of the combined consideration and structure

scales resulted in three factors, two of which contained items from both

of the scales indicating that the use of the consideration and structure








scales as entities were not appropriate for analysis of this sample

population.

4. An analysis of the relationship between the four LOC factors and

the three LBDQ factors at the three levels of analysis revealed that:
a. At level one, LOC Factor II (luck external) was negatively

and significantly correlated with LBDQ Factor III (structure).
b. At level two (group level), the follower's LOC Factor II

(luck external) correlated negatively with LBDQ Factor I (mixed
consideration-structure), and negatively with LBDQ Factor III (structure).
c. At the individual differences level (level three), the same

inconsistent pattern of relationships was found to exist between the LOC

and the LBDQ factors (see Table 23).
d. Tenure did not relate significantly with the follower's per-

ceptions of the leader's behavior as measured by either the consideration

or structure scales, but did relate significantly and negatively with LBDQ

Factor I (mixed consideration and structure) at the group level (level

two).
5. No significant relationships were found between the leader's

description of his own behavior and the follower's description of that
behavior.
6. A multiple regression analysis of the relationship between the
leader and follower LOC factors and both the leader and follower's per-
ceptions of the leader's behavior found that:
a. In relation to the follower's perceptions--
1. Luck and God external followers (those who believe that

their outcomes are controlled by luck or God) tend to perceive their
leaders as exhibiting less mixed consideration-structure behavior than do

luck or God internal followers.







2. System external followers tend to perceive their leaders

as less interpersonally involved than do system internal followers.
b. In relation to the way leaders are perceived by their followers--

1. Luck external leaders tend to be perceived by their fol-

lowers as exhibiting less mixed consideration-structure behaviors and to

be less interpersonally involved than are luck internal leaders.
2. System external leaders tend to be perceived by their

followers as exhibiting more mixed consideration-structure behaviors and
to be less interpersonally involved than are system internal leaders.
3. God external leaders tend to be perceived by their fol-

lowers as exhibiting more mixed consideration-structure behaviors than

are God internal leaders.
4. Competency external leaders tend to be perceived by their

followers as exhibiting less mixed consideration-structure behaviors than

do competency internal leaders.
c. In relation to the way the leaders perceive themselves--

1. Luck external leaders tend to perceive themselves as more

structuring than do luck internal leaders.
2. Competency external leaders tend to perceive themselves

as less structuring than do competency-internal leaders.
3. God external leaders tend to perceive themselves as less

structuring than do God internal leaders.
d. In relation to the effect of the follower's LOC on the leader's

self-perceptions--
1. Leaders whose followers are competency externals tend to

perceive themselves as less structuring than do leaders whose followers

are competency internals.








2. Leaders whose followers are God externals tend to

describe themselves as exhibiting more mixed consideration-structuring

behaviors and to be more structuring than do leaders whose followers are
God internals.
3. Leaders whose followers are system externals tend to

describe themselves as exhibiting more mixed consideration-structuring

behaviors, and as more structuring than do leaders whose followers are

system internals.

Discussion of the Findings
An examination of the results of the regression analyses revealed

the expected complexity of the relationships between the leader's and

the follower's LOC and the way in which the followers perceived their

leader, and the way in which the leader perceived his own behavior. It

appears, for example, that the way the follower's perceive their leader's

behavior is differentially related to a combination of the leader's and the

follower's LOC factors.
Table 28 indicates that the follower's perceptions of the leader's

behavior as measured by LBDQ Factor I (mixed consideration-structure) is
most closely related to the follower's LOC Factor II (luck external), and

the leader's LOC Factor IV (competency external). To the extent that a

follower believes that luck controls his life, he tends to describe his
leader as low in LBDQ Factor I behaviors. However, to the extent that the

leader believes that competency behavior will not achieve his desired out-

comes, the followers tend to describe the leader as high in LBDQ Factor I

behavior. Yet, when the leader describes his own behavior on the LBDQ
Factor I dimensions, his own LOC appears to have no relationship. The

major significant relationship with the leader's self-description appears





























S.
a,




-J


0
(.
0>-
Cc



0





..4J
5.-i
0 0







-I- r
O


00
*0
0 <0






i 0.
0 I
O-

cJ
S-)
c(UC




O
.o 3
*r- U












O
0)
I-r
t-0
(U -


1-4
s.
0


+Y
u

"r
0
-J
CI)



~s-
0)
-0
01
ci
-J


M"lli


0
0

0
-J

I-
3
I--
rI-
O0
LL


I-.

I-.

0
U
10
LLi

0'
e


00
(ci
~IT
(U


aU
43
*.-
E
O

4)
C,





E
0
*r-






4.)
E
U




*


_ _1


H^siF"I H-i H PI1"


)II