God Image as a Function of Self-Esteem and Locus of Control

Author(s): Peter Benson and Bernard Spilka

Source: Journal for the Scientific Study of Religion , Sep., 1973, Vol. 12, No. 3 (Sep.,
1973), pp. 297-310
Published by: Wiley on behalf of Society for the Scientific Study of Religion

Stable URL: https://www.jstor.org/stable/1384430

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Wiley and Society for the Scientific Study of Religion are collaborating with JSTOR to
digitize, preserve and extend access to Journal for the Scientific Study of Religion

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
 God Image as a Function of
Self-Esteem and Locus of Control
PETER BENSON

Department of Psychology
Eastern Michigan University

BERNARD SPILKA

Department of Psychology
University of Denver

A cognitive consistency framework was adopted to predict that a believer's level of self-
esteem and his location on the locus of control dimension influence his description and
definition of God. On a sample of 128 Catholic subjects with approximately identical reli-
gious backgrounds, self-esteem was positively related to loving-accepting God-images and
negatively to rejecting images. Locus of control was unrelated to controlling beliefs. Statis-
tical and methodological controls were utilized to offer an interpretation that self-esteem
may be a major determinant of God-images. It was proposed that these findings have
important implications for understanding the dynamics of personal religion.

he purposes of this study are twofold. First, it is intended to investigate

theorized relationships between self-regarding attitudes and God-images. And
second, it attempts to substantiate the interpretation that the hypothesized corre-
lations can be explained as a process of such attitudes influencing the definition
of God.
Cognitive consistency theory provides the basis for predicting and explaining
these relationships. Consistency theory suggests that information which implies
the reverse of one's usual level of self-regard tends to create dissonance (Bramel,
1968; Smith, 1968). In order to avoid the discomfort of such dissonant cognitions,
techniques like selective perception, distortion, and denial can be used to keep
information consonant with one's self-image. For example, a person with high self-
esteem probably finds failure and interpersonal rejection inconsistent with his
beliefs that he is basically successful and likeable. Hence, he may adopt perceptual
and behavioral strategies to counter these kinds of experiences. The more inter-
esting prediction is that an individual with low self-esteem who believes he is a
failure and unlikeable will find success and social approval unpleasant. Conse-
quently, he will attempt to disconfirm or avoid success and approval in order to
keep his self-image and information about himself consistent. These hypotheses
have gained some research support (Bramel, 1962; Fitch, 1970; Jacobs, Berscheid
& Walster, 1971; Walster, 1965).

297

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
298 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

Accordingly, persons with different levels of self-esteem may find it difficult

to share the same religious beliefs. A theology predicated on a loving, accepting
God is cognitively compatible with high self-esteem, but it could be a source of
discomfort for a believer low in self-esteem. It does not make good cognitive sense
to be loved when one is unlovable. Consequently, the latter person can march to a
different theology, one that is more consistent with his self-image.
Dissonance theory can also be utilized to predict how feelings of personal
control influence the interpretation of the causes of one's actions. If an individual
views himself as the cause of his own behavior (internal control), belief in a God
who intervenes in his life would be dissonant. Hence, God then would be per-
ceived as freeing rather than restricting, and as uncontrolling rather than con-
trolling. Alternatively, a demanding, powerful and controlling God should be
consistent with the cognition that one is not the master of his own fate (external
control).
In summary, it is argued that self-regarding attitudes influence God-images.
A believer selects a "god" that is consistent with power-related (locus of control)
and evaluative (self-esteem) self-images. In correlational terms, it is hypothesized
that:
1. Self-esteem will be positively related to loving God-images, hence
2. Self-esteem will be negatively related to rejecting or nonloving images.
3. External control will be related positively to controlling God-images.

METHOD

If the proposed associations are substantiated on a heterogeneous religious

sample (where "heterogeneous" designates existing variations in religious back-
ground among subjects), three explanations are plausible: (a) As hypothesized,
self-regarding attitudes influence God-images. (b) God-images influence self-
regarding attitudes. A case for this interpretation depends on the argument that
initial variations in God-images originate in differences in religious training.
Consequently, these training-produced beliefs may differentially affect self-
concept variables. (c) The relationships between self-regarding attitudes and God-
images are due to a third variable or set of variables correlated with both.
The case for (a) depends on how well alternative explanations (b) and (c) can
be dismissed. Therefore, -several methodological strategies were adopted as at-
tempts to eliminate these competing interpretations. Explanation (b) is plausible
if it can be posited that belief variations have origins in religious training factors.
If a group of subjects (Ss) is selected who have essentially identical religious
training, and if noteworthy correlations remain between self-regarding attitudes
and God-images, it may be argued that these relationships are due to (a) or (c).
A partial correlation analysis can be used to control a number of potential corre-
lates of both self-regarding attitudes and the religious measures. If substantial
correlations remain after this analysis, it may further be argued that alternative
(c) above has been seriously questioned as the primary explanation for the data.
Explanation (a) then would remain as a reasonably compelling interpretation of
the results.

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
 GOD IMAGE AND SELF-ESTEEM 299

Subjects

Two hundred and five male Ss attending a Catholic high school were initially
tested. In order to obtain a highly homogeneous religious sample, Ss were selected
from this subject pool who met the following criteria: S considered himself Cath-
olic, both parents were Catholic, S had never been a member of a different denom-
ination, had never associated with a non-Catholic religious organization, and had
been a member of a local Catholic parish for at least the last ten years. This final
criterion was adopted as a control for geographical differences in theological
emphasis. In addition, only Ss who regarded religion as personally important
(above the midpoint on a nine-point semantic differential item) and who desig-
nated that they believed in God (above the midpoint on a seven-point item) were
used. These two items were utilized to eliminate Ss who might be religious
"detractors."
One hundred and twenty-eight Ss met these criteria. The final sample had the
following educational distribution: 44 Freshmen, 31 Sophomores, 19 Juniors,
and 34 Seniors. Mean S age was 15.4 years.

Tests

The first 23 items were selected from Coopersmith's (1967) original 50 items
to form a self-esteem scale. On a sample of 55 Catholic youth, this short form
correlated highly with the other 27 items (r = .78). Consistent with Coopersmith's
scoring procedures, each high self-esteem response was multiplied by four. Thus,
the maximum score on this scale was 92. The original scale has been widely used
with children and adolescents. It is concerned with self-attitudes in four areas
found to be important at these age levels: peers, parents, schools, and personal
interests. The short-form used in this study contains items related to each of
these areas. Rotter's (1966) 23-item scale was used as a measure of internal-
external control. One point accrued for each external option marked, so that the
maximum score on the scale was 23. External control was designated as a high
score on the scale, and internal control signified a low score.
Single items were used to obtain information regarding age, grade, father's
occupation, and religious behavior. These were included to measure possible
extraneous influences on the proposed personality and religion relationships.
Father's occupation was used as an indicant of socioeconomic status (SES), based
on procedures developed by Edwards (Miller, 1964). Occupations were categor-
ized into one of six SES levels. From high to low status, these levels were: pro-
fessional or technical, business managers or proprietors, clerical or sales workers,
craftsmen or foremen, operatives, and unskilled. On this variable, a score of one
was assigned to the highest status and a six was given to the lowest status. The
three religious behavior items were used to measure frequency of devotions at
home, frequency of religious discussions at home, and hours spent per month in
church or church-related activities. Each question had six options scored 0 to 5
with 5 corresponding to the highest level of activity.

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
300 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

A thirteen-item semantic differential scale was developed to measure loving

and controlling God-images. Scores on the following five pairs of adjectives were
summed to yield a Loving God index: rejecting-accepting, loving-hating, damning-
saving, unforgiving-forgiving, and approving-disapproving. Each item was scored
O to 6. The maximum Loving God score was 30. Scores on the following five pairs
of adjectives were summed to give a Controlling God index: demanding-not
demanding, freeing-restricting, controlling-uncontrolling, strict-lenient, and
permissive-rigid. The same scoring procedure was used, with 30 representing the
maximum Controlling God score. On a sample of 50 Lutheran Ss, scale homo-
geneity was .72 for the Loving God scale and .60 for the Controlling God measure.
A 64-adjective Q-sort developed by Spilka, Armatas, and Nussbaum (1964)
was used to obtain other God-image measures. Scores were obtained on six
factors selected from their eleven factor orthogonal solution for a sample of 200
Catholic girls. The Vindictive God factor measures a God-image that is based on
such Old Testament descriptions of God as wrathful, avenging, and damning.
The Stern Father factor defines God as unyielding, punishing, and restricting.
Three impersonal factors measure different facets of a God who is "out there,"
as opposed to a God that is closely related to human affairs. The Allness factor
defines this impersonal dimension in terms of God's infinite, absolute, and un-
changing power and wisdom; the Distant factor defines this impersonal nature
more in terms of God's inaccessibility; and the Supreme Ruler factor deals more
with a majestic and sovereign God. The Kindly Father factor measures a more
personal image, based on such New Testament notions as mercy, forgiveness,
and patience. The Kuder-Richardson reliabilities (K-R 20) for these six factors
are: Vindictive God, .60; Stern Father, .71; Impersonal Allness, .77; Impersonal
Distant, .51; Impersonal Supreme Ruler, .56; and Kindly Father, .61.
Ss were tested at their school during classroom hours, and in order to elicit
frank answers and assure that responses were confidential, they were not required
to sign or otherwise code their test battery.

RESULTS

Self-esteem and God-images

As predicted (Hypothesis 1), self-esteem was significantly related to loving

God-images (see Table 1). This relationship held for both the Loving God scale
(r = .51, p <.01) and the Kindly Father factor (r = .31, p <.01). These two vari-
ables may be, to some degree, measuring the same beliefs. The scales are corre-
lated (r = .32, p <.01) and both include "loving" and "forgiving" adjectives.
It was also hypothesized that self-esteem would be negatively related to
rejecting God-images (Hypothesis 2). The obtained relationship between self-
esteem and the Vindictive God factor (r = -.49, p <.01) confirmed this prediction.
Self-esteem also correlated negatively with three other God measures, all of
which were positively related to Vindictive God. These self-esteem relationships
were as follows: to Impersonal Allness (r = -.23, p <.01), to Controlling God
(r = -.35, p <.01), and to Stern Father (r = -.21, p <.05). The Vindictive God
correlations to these belief measures were: to Impersonal Allness (r = .28, p <.01),

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
GOD IMAGE AND SELF-ESTEEM 301

TABLE I

MATRIX OF CORRELATIONS (N = 128)

7 8 9 10 11 12 13 14 15 16

1. Age 02' 07 -16 14 -06 -06 14 02 -04 -01

2. Grade 06 06 -15 19* -10 -04 11 -04 01 -05

3. Frequency of devotions 12 -30** 34** 01 -05 -03 -09 -12 15 -08

4. Frequency of religious
discussions 19* -18* 23* -14 -02 -01 -05 04 06 01.

5. Church hours per month 09 -19* 19* -04 -06 -04 -07 -10 10 -07

6. SES -05 25** -02 -11 14 02 04 -01 -14 08

7. Self-esteem -32** 51** -35** -49** -21* -23** -17* 31** -18*

8. Internal-external control -28** 02 23** 10 18* 16 -14 12

9. Loving God -30** -30** -21* -32** -30** 32** -06

10. Controlling God 30** 31** 09 -17 -22* 02

11. Vindictive God 45** 28** 13 -33** 17

12. Stern Father 13 17 -07 -16

13. Impersonal: Allness 07 -46** 20*

14. Impersonal: Distant -13 -07

15. Kindly Father 47**

16. Impersonal: Supreme

Ruler

1. Decimal points not presented

*Significant at .05 level
**Significant at .01 level

to Controlling God (r = .30, p <.01), and to Stern Father (r = .45, p <.01). These
intercorrelations suggest that impersonality, rigid control, and vindictiveness may
have been to some degree perceived as dimensions or components of nonlove. Ac-
cordingly, they appeared antithetic to loving images. The negatively significant
correlations between Loving God and Impersonal Allness (r = -.32, p <.01), Im-
personal Distant (r = -.30, p <.01), Controlling God (r = -.30, p <.01), Stern Father
(r = -.21, p <.01), and Vindictive God (r = -.30, p <.01) support this contention.
Self-esteem was also correlated negatively with Impersonal Distant (r = -.17,
p <.05) and Impersonal Supreme Ruler (r = -.18, p <.05). Overall, self-esteem
was significantly related to all eight God measures. The pattern suggests that
self-esteem is related positively to loving images and negatively to rejecting-
impersonal-controlling views, thereby supporting both Hypotheses 1 and 2.

Locus of control and God-images

The data do not support Hypothesis 3. Locus of control was unrelated to the
Controlling God scale (r = .02). It was however, significantly related to three

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
302 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

other measures: with Loving God (r = -.28, p <.01), Vindictive God (r = .23,
p <.01), and Impersonal Allness (r = .18, p< .05). In all three instances, the direc-
tion of the relationship is opposite that obtained with self-esteem. One possible
explanation is that a high score on the locus of control measure, which is an exter-
nal orientation, is a measure of low self-esteem. The negative and significant
relationship between self-esteem and locus of control (r = -.32, p <.01) tentatively
supports this interpretation. It is also possible that the Controlling God scale is
more sensitive to a loving-nonloving dimension than to a controlling-noncontrol-
ling one. The intercorrelations among the scales support this contention. For
example, the Controlling God index correlated negatively with both Loving God
(r = -.30, p <.01) and Kindly Father (r = -.22, p <.05), and positively with
Vindictive God (r = .30, p <.01) and Stern Father (r = .31, p <.0 1).

Self-esteem, locus of control, and religious behavior

The three religious behavior items were related to internal control and one was
associated positively with self-esteem. Frequency of religious discussions corre-
lated with both self-esteem (r = .19, p <.05) and internal control (r = -18, p <.05).
Frequency of devotions (r = -.30, p <.01) and hours spent in church activities
(r = .-19, p <.05) were both significantly related to internal control.

Controlling for demographic variables and religious behavior

The reported results were obtained on a sample of Ss with highly similar

backgrounds. Many theorists would argue that such Ss should espouse parallel
beliefs. However, as the data indicate, beliefs vary with self-regarding attitudes.
The structure of the sample however, suggests that self-regarding attitudes may
be more responsible for these variations than are teaching or training factors.
It could be that self-regarding attitudes and God-images are the common
effects of a third variable or set of variables. Age, SES, or educational level may
play such a role. Also, the quantity (frequency) or location of religious exposure
can influence the relationships. In order to add support to the interpretation that
self-regarding attitudes probably influence God-images, these variables are con-
trolled via partial correlational means. Table 1 indicates the variables that should
be partialled. Four that correlated with both self-esteem and a number of religious
measures are: frequency of devotions, frequency of discussions, hours spent per
month in church or church-related activities, and locus of control. Of these, fre-
quency of discussions (r = .19, p <.05) and locus of control (r = -.32, p <.01)
correlate significantly with self-esteem. While the relationships to church hours
(r = .09) and frequency of devotions (r = .12) did not attain the .05 level of signifi-
cance, they may still account for some of the variance shared by self-esteem and
the religious variables. Partialling.the above four variables affords a look at self-
esteem and religious variable relationships when the influence of these correlates
has been removed. It should also be noted that these indices do not correlate with
all eight religious measures. Rather, the three behavior items correlated only
with Loving God. It is on this variable that partial correlations are most crucial

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
GOD IMAGE AND SELF-ESTEEM 303

for questioning explanations that the personality and religion relationships are
spurious. However, the removal of these four correlates from self-esteem in those
cases where their relationship to religious measures is minimal is quite useful. In
these situations, partial correlation techniques clarify the amount of variance that
is uniquely shared by self-esteem and the religious variables. In essence, a partial
correlation in such cases reduces to a part correlation. Age, grade, and SES did not
correlate significantly with self-esteem. While it is doubtful that extracting these
variables will lead to noteworthy changes in the self-esteem-religious variable
correlations, doing so reveals whether these measures function as "suppressors."
Since in a few cases grade and SES correlate negatively (though nonsignificantly)
with the religious variables, they may possibly suppress the self-esteem-religion
relationships. Age and grade are highly correlated in this sample (.95); thus
controlling for grade level essentially controls for age.
Table 1 also reveals that the locus of control and religious variable relation-
ships could be influenced by six variables. Four that correlated with both locus of
control and with some of the religious variables are frequency of devotions, fre-
quency of discussions, hours spent in church, and self-esteem. These were also
submitted to a partial correlation analysis. A fifth variable, SES, correlated signifi-
cantly with locus of control, but with none of the religious measures. A partial
analysis of this variable functions primarily to remove SES from locus of control.
Grade level was also partialled out for reasons similar to the case with self-esteem
discussed above.
In summary, six variables were partialled from self-esteem and the religious
measures, and from internal-external control and the same indices. If relationships
between the self-regarding attitudes and the religious measures exist after the
partial correlation analysis, it can be argued that the explanation that the associa-
tions are spurious has been seriously questioned. However, a note of caution is in
order. Two factors make it invalid to dismiss the "spurious" explanation in any
absolute sense. First, the partial correlation does not simply represent the corre-
lation between two variables when a third variable has been held constant. Rather,
it can be interpreted "as a weighted average of the correlation coefficients that
would have been obtained had the control variable been divided into very small
intervals and separate correlations computed within each of the categories"
(Blalock, 1972: 436). Therefore, the resulting partial correlation is affected by the
ranges of the scores obtained on the controlled variables. In the present study,
these ranges may have been restricted by the characteristics of the subjects sam-
pled. Therefore, if relationships exist after the partials, it can only be said that
the variables were controlled within certain restricted boundaries. Second, since
only six variables were partialled in each analysis, other uninvestigated variables
could possibly affect the self-regarding attitude and religion associations. Ob-
viously, it is impossible to absolutely dismiss a "spurious" explanation of the
data, since it is impossible to control for all correlates. At best, it can be argued
that the variables actually controlled in a study were the ones that could have the
greatest influence, based on theoretical or empirical information.

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
304 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

Self-esteem and God-images

Tables 2 and 3 contain the results that emerge from these statistical controls.
Before the partial correlation analysis, self-esteem was related significantly to all
eight God measures. After all six variables (internal-external control, frequency
of discussions, frequency of devotions, church participation rates, SES and grade)
were removed, self-esteem remained signiflcantly related to six measures. Rela-
tionships to Impersonal Distant and Impersonal Supreme Ruler did not attain the
.05 level of signiflcance. In both instances, however, the amount of shared variance
declined negligibly. This suggests that these relationships were in fact minimal
before the partial analysis was undertaken. The Loving God and self-esteem
relationship changed from .51 to .47, indicating that the amount of shared variance
fell from 26 percent to 22 percent. The correlation remains significant at the .01
level.
In all other cases the reduction in shared variance was 3 percent or less. Over-
all, it appears that the partialling of these six variables did not appreciably affect

TABLE 2

CORRELATIONS BETWEEN SELF-ESTEEM AND RELIGIOUS VARIABLES WHEN THE

FOLLOWING VARIABLES HAVE BEEN REMOVED: FREQUENCY OF DEVOTIONS AT HOME;
FREQUENCY OF RELIGIOUS DISCUSSIONS AT HOME; HOURS SPENT PER MONTH IN
CHURCH OR CHURCH-RELATED ACTIVITIES; I-E, SES, GRADE

After After
After Devotions Devotions
After Devotions Discussions Discussions
Partials Discussions Attendance Attendance
Before on all and Atten- and I-E I-E and SES
Religious Variable Partials Variables dance Removed Removed Removed

Loving God .51** .47** .49** .46** .46**

(.26)1 (.22) (.24) (.21) (.21)

Controlling God -.35** -.37** -.33** -.34** -.34**

(.12) (.14) (.11) (.12) (.12)
Vindictive God -.49** -.46** -.49** -.46** -.46**
(.24) (.21) (.24) (.21) (.21)

Stern Father -.21* 1.19* -.21* -.19* 19*

(.04) (.04) (.04) (.04) (.04)

Impersonal: Allness -.23** -.20* -.22* -.19* -.19*

(.05) (.04) (.05) (.04) (.04)
Impersonal: Distant -.17* -.14 -.17* -.15 -.14
(.03) (.02) (.03) (.02) (.02)

Kindly Father .31 ** .29** .30** .29** .29**

(. 10) (.09) (.09) (.09) (.09)
Impersonal: Supreme Ruler -.18* -.15 -.18* -.15 -.16
(.03) (.02) (.03) (.02) (.03)
1. Squared correlation coefficient
*Significant at .05 level
**Significant at .01 level

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
 GOD IMAGE AND SELF-ESTEEM 305

the original relationships. This could have been expected for all associations other
than that of selfesteem and Loving God. In these other situations, the original
correlations between the God measure and the partialled variables were small.
Potentially, the greatest change could have involved Loving God since four of the
six partialled variables correlated with this variable. Nevertheless this relationship
remained intact.
Self-esteem remains positively correlated with Loving God (r = .47, p <.01)
and Kindly Father (r = .29, p <.01) and negatively with Controlling God (r = -.37,
p <.01), Vindictive God (r = -.46, p <.01), Stern Father (r = -.19, p <.05), and
Impersonal Ailness (r = -.20, p <.05). These data, then, are consistent with the
hypothesis that self-esteem influences these God-images. The influence can be
understood in terms of both loving and nonloving images. It can be argued that this
is a reasonable explanation since the data were obtained on subjects with highly
similar backgrounds after a number of correlates were controlled.

TABLE 3

CORRELATIONS BETWEEN INTERNAL-EXTERNAL CONTROL AND RELIGIOUS

VARIABLES WHEN THE FOLLOWING VARIABLES HAVE BEEN REMOVED: FREQUENCY
OF DEVOTIONS AT HOME; FREQUENCY OF RELIGIOUS DISCUSSIONS AT HOME;
HOURS SPENT PER MONTH IN CHURCH OR CHURCH-RELATED ACTIVITIES;
SELF-ESTEEM, SES, GRADE

After After
Devotions, Devotions,
After Discussions, Discussions,
After Devotions, Attendance Attendance,
Partials Discussions, and Self-esteem,
Before on all and Atten- Self-esteem and SES
Religious Variable Partials Variables dance Removed Removed Removed

Loving God -.28** -.07 -.18* -.05 -.06

(.07)' (.00) (.03) (.00) (.00)
Controlling God -.02 -.07 -.01 -.10 -.08
(.00) (.00) (.00) (.01) (.01)
Vindictive God .23** .07 .22* .10 .07
(.05) (.00) (.05) (.01) (.00)

Stern Father .10 .04 .10 .04 .04

(.01) (.00) (.01) (.00) (.00)
Impersonal: Allness .18* .10 .15 .10 .09
(.03) (.01) (.02) (.01) (.01)

Impersonal: Distant .16 .09 .13 .08 .09

(.03) (.01) (.02) (.01) (.01)

Kindly Father -.14 .02 -.09 -.01 .02

(.02) (.00) (.01) (.00) (.00)
Impersonal:Supreme Ruler .12 .04 .10 .05 .04
(.01) (.00) (.01) (.00) (.00)

1. Squared correlation coefficient

*Significant at .05 level
**Significant at .01 level

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
306 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

Locus of Control and God-images

The partial correlation analysis was more influential in the case of locus of
control and God-belief relationships. Before the analysis, locus of control was
significantly correlated with three of the measures, although not with Controlling
God as predicted. After the analysis, all three relationships failed to reach the .05
level of significance. The original associations with Loving God, Vindictive God,
and Impersonal Allness reduce to .10 or less. Table 3 shows that the removal of
self-esteem after the initial removal of the three religious behavior items was most
responsible for these changes. This step accounted for a 3 percent reduction in the
variance originally shared by locus of control and Loving God, and a 4 percent
reduction in variance shared with Vindictive God. Self-esteem may therefore be
an important component of locus of control. In summary, it does not appear that
locus of control influences God-images, at least as they are measured in this study.

DISCUSSION

The major finding of this study is that self-esteem is related positively to loving
God-images and negatively to rejecting-impersonal-controlling definitions of God.
Furthermore, it is argued that the data provide some support for the explanation
that self-esteem is a major determinant of God-images. This interpretation was
made after a consideration of the alternatives. By obtaining the results on Ss with
nearly identical religious training, it was posited that a "belief influencing person-
ality" position was seriously questioned as the primary explanation for the find-
ings. And by partialling variables dealing with internal-external control, SES,
grade, and frequency of participation in religious activities, it was further sug-
gested that explanations based on the premise that both self-esteem and God-
images were the common effects of other variables were considerably weakened.
By the processes of elimination, then, it seems appropriate to argue that self-
esteem influences God-images. Several other explanations of the data need to be
investigated, and refuted, in order to maximize the credibility of this position.
Other religious training factors need to be controlled. Differences in parish empha-
ses, teachers, clergy, and reading material could possibly affect beliefs, which in
turn influence self-esteem. Parents and peers may influence both God-images and
self-esteem. For example, rejecting parents might induce low self-esteem, and as
previous research suggests (Siegman, 1961; Strunk, 1959), Ss may define God in
terms similar to the way they view their parents (which could include a rejecting
image in this case). It may also be true that those with low self-esteem engage in
problem behaviors which elicit from parents and other authorities fear-provoking
religious concepts used in an attempt to control the behavior. Accordingly, these
Ss could receive a different religious "treatment" than those with high self-esteem.
Thus, low self-esteem could be paired with rejecting God concepts.
While such processes may influence God-images and self-esteem in certain
individual cases, it is proposed that they are not of sufficient general strength to
account for any large portion of shared variance between self-esteem and God-
images in a relatively large sample.

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
 GOD IMAGE AND SELF-ESTEEM 307

Other issues need to be raised. While the reported data suggest a broad inter-
pretation of self-esteem influencing God-images, the results do not provide many
details about the relationship. Table 4 presents frequencies and percentages of
Loving God scale scores at three levels of self-esteem. The average scale score
among all 128 Ss was 25.3. Ss were divided into low, medium, and high levels of
self-esteem. The first row shows that 79 percent of the "highs" scored above the
mean (scale score of 26-30) while only 30 percent of the "lows" scored similarly.
A chi square performed on scores above and below the mean suggests that "highs"
tend to perceive God as significantly more loving than "lows" (X 2 = 21.45, p<
.001). Additionally, scores below 20 occurred only in the "low" group. Twenty-
eight percent of the "lows" scored between nine and 20 points. Twenty-three per-
cent of the "highs" and only 5 percent of the "lows" scored the maximum 30
points.

TABLE 4

FREQUENCIES AND PERCENTAGES OF LOVING GOD SCALE SCORES

AT THREE LEVELS OF SELF-ESTEEM

Level of Self-esteem

Loving God Scale Score Low Medium High

(X = 25.3) N= 43 N= 42 N= 43

Above 25.3 13 26 34
(30%)' (62%) (79%)
Below 25.3 30 16 9
(70%) (38%) (21%)
30 (Maximum scale score) 2 5 10
5%) (12%) (23%)
Below 20 12 0 0
(28%)
Mean Scale Score 22.40 26.21 27.23

1. Percentage of total N in the corresponding self-esteem level

%2 = 21.45, p <.001. Computed on the first two rows.

The combination of these figures and the previously cited correlations suggests
that a majority of Ss tend to perceive God in a way that is congruent with self-
image. Some God-images, however, do not fit this consistency pattern. Twenty-
three percent of the "lows" score above 26 points. This may indicate that some
Ss are motivated to attain approval rather than consistency. Persons with high
self-esteem can meet consistency and approval needs simultaneously, while per-
sons with low self-esteem find these needs in conflict. A highly accepting God
communicates approval, while a more rejecting God preserves consistency. In this
context, it would be important to explicate what factors determine which option
is chosen, and whether or not these needs operate for the same S in different
situations.
Interpreting how individuals actually develop these God-images is difficult.
Several possibilities exist. "Highs" may internalize the preached axioms of a given

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
308 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

theology. Accordingly, "lows" then tend to change or modify this theology to con-
form more consistently with their self-images. It could also be, however, that
individuals are exposed to a variety of religious teachings. These teachings could
variously present God as loving and punishing, supernatural and immanent, warm
and impersonal. Thus, the believer may select the set of preached God-images
which is most consistent.
The nonsignificant correlations between locus of control and God beliefs
merit discussion. Rotters (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 that options phrased in luck
and chance terminology are irrelevant. Relationships between locus of control
and God-images could have been masked by this measurement problem.
The consequences of holding a controlling God-image warrants further
research. Belief in a God who controls the destiny of human events may lead an
individual to view his own behavior as inconsequential for social change. Seeman's
(1959) work on alienation implies that external control and powerlessness are
conceptually similar. This may partially explain why some church members refuse
to engage in social action and conform to approved expectations in order to gain
some feeling of control.

SOME IMPLICATIONS

It has often been charged that religious individuals use religion for purposes
of comfort-at the expense of challenging and changing the secular world (Glock,
Ringer & Babbie, 1967). If this criticism is valid, where should the blame be
placed?
One possible explanation lies in the conservative emphasis of religious institu-
tions. As O'Dea (1966) has indicated, organized institutions tend to be concerned
with stability and continuity. In the service of self-maintenance, such institutions
attempt to avoid conflict. Consequently, members may learn that a life of faith is
personal and not social. Theology can be presented to justify such a nonsocial
orientation.
The data from the present study, however, suggest that the problem may also
have its source in the psychological processes of the individual. Low self-esteem
may lead one to resist a loving God and a compassionate ethical orientation, even
when both are emphasized in a particular institutional setting. Assuming that
self-esteem influences the content of religious beliefs, we can hypothesize that
religious belief functions in part to maintain or preserve one's level of self-esteem.
Belief justifies good or bad self-images, providing relief from the struggle to find
meaningful explanations for one's identity. One's self-esteem level affects what
can and cannot be accepted as a tenet of faith. Church members with high self-
esteem can understand and appreciate a loving, accepting God-and may also
apply principles of love and acceptance to various social settings. Persons with
low self-esteem, however, find both God and ethics based on love and acceptance
as uncomfortably inconsistent. Consequently, their needs lead them to replace

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
 GOD IMAGE AND SELF-ESTEEM 309

these images with a more rejecting God who does not command a loving orienta-
tion to the world. Thus, some of the social complacency attributed to church
members may reflect personal inhibitions rather than church policy. This problem
may be immense, if as previous research indicates, self-concept variables corre-
late negatively with religious attitudes and involvement (Cowen, 1954; Strunk,
1958).
While such an inference implies that level of self-esteem is an antecedent to
specific belief content, it should be recognized that theology might (if known by
the member) also influence self-concept. Perhaps self-concept and belief are an-
tinomously related, where there is continual interaction between these variables.

CONCLUSION

These findings and inferences must be regarded as tentative pending further

investigation. It should be recognized that the present study is only a beginning
in establishing and explaining personality-religion relationships. Future research
should strive to utilize theoretical models and methodological designs which
make both the causes and consequences of religiosity more understandable. By
explicating when and how personality influences religion and when and how
religion influences personality, the scientific study of religion will have produced
findings that are of both theoretical and practical use for understanding the
dynamics of personal religion.

REFERENCES

Blalock, Hubert M. Dilemma of the Contemporary

1972 Social Statistics. New York: McGraw- Church. Berkeley: University of
Hill. California Press.
Bramel, D. Jacobs, L., E. Berscheid, and E. Walster
1962 "A dissonance theory approach to 1971 "Self-esteem and attraction." Journal
defensive projection." Journal of of Personality and Social Psychology
Abnormal and Social Psychology 64: 17: 84-91.
121-129. Miller, D. C.
1968 "Dissonance, expectation, and the 1964 Handbook of Research Design and
self." Pp. 355-366 in R. P. Abelson, Social Measurement. New York:
E. Aronson, W. J. McGuire, T. M. McKay.
Newcomb, M. J. Rosenberg, and P. O'Dea, Thomas F.
H. Tannenbaum (eds.), Theories of 1966 The Sociology of Religion. Engle-
Cognitive Consistency: A Source- wood Cliffs, N.J.: Prentice-Hall.
book. Chicago: Rand McNally. Rotter, J. B.
Coopersmith, S. 1966 "Generalized expectancies for inter-
1967 The Antecedents of Self-Esteem. nal versus external control of rein-
San Francisco: Freeman. forcement." Psychological Mono-
Cowen, E. L. graphs 80 (1, Whole No. 609).
1954 "The negative concept as a person- Seeman, Melvin
ality measure." Journal of Consulting 1959 "On the meaning of alienation."
Psychology 18: 138-142. American Sociological Review 24:
Fitch, G. 783-790.
1970 "Effects of self-esteem, perceived Siegman, A. W.
performance, and choice on causal 1961 "An empirical investigation of the
attributions. Journal of Personality psychoanalytic theory of religious
and Social Psychology 16: 311-315. behavior." Journal for the Scientific
Glock, C. Y., B. B. Ringer, and E. R. Babbie Study of Religion 1: 74-78.
1967 To Comfort and to Challenge: A

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms
310 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

Smith, M. Brewster Strunk, Otto, Jr.

1968 "The self and cognitive consistency." 1958 "Relation between self-reports and
Pp. 366-373 in R. P. Abelson, E. adolescent religiosity." Psychological
Aronson, W. J. McGuire, T. M. New- Reports 4: 683-686.
comb, M. J. Rosenberg, and P. H. 1959 "Perceived relationships between
Tannenbaum (eds.), Theories of Cog- parental and deity concepts." Psycho-
nitive Consistency: A Sourcebook. logical Newsletter 10: 222-226.
Chicago: Rand McNally. Waister, E.
Spilka, Bernard, P. Armatas, and J. Nussbaum 1965 "The effect of self-esteem on roman-
1964 "The concept of God: A factor- tic liking." Journal of Experimental
analytic approach." Review of Reli- Social Psychology 1: 184-197.
gious Research 6: 28-36.

This content downloaded from

36.68.217.87 on Tue, 11 Jun 2024 13:27:40 +00:00
All use subject to https://about.jstor.org/terms