political orientation and church attendance
TRANSCRIPT
Church Attendance and Political Orientation
By: Keenan M. Afram
2
Section 1: Introduction
This is a study of the relationship between church attendance of people in the United States
and their political orientation (e.g. conservative, liberal, and moderate). The benefits of this
study, and why the field of Political Science needs this contribution, can be found in Section 2.
According to a review of the current literature, found in Section 3 and 4, there is a general
consensus that individuals who attend church more regularly are more conservative and vice
versa. The aim of this study is to explore the extent that church attendance correlates with
political orientation.
To elucidate this relationship, three hypotheses will be tested:
H1: Those who regularly attend church will generally self-identify as conservative.
H2: Those who sometimes attend church will have no political preference.
H3: Those who rarely attend church will generally self-identity as liberal.
In section 5, using data from the National Election Studies, these hypotheses will be explored
through cross-tabulation and bivariate regression analysis. Following this, in Section 6, insight
and conclusions about our study will be summarized. And finally, Section 7 contains the
methodological appendix, including details on data gathering and analysis.
Section 2: Why Political Science Needs this Contribution
Political Science is the study of governments, public policies and political processes,
systems, and political behavior. Our study elucidates a key area of Political Science: political
behavior. If correlations can be drawn between church attendance and political orientation, an
important demographic of the United States can be better understood; this understanding can
3
elucidate the nature of the church attendance voting bloc, which helps to understand the overall
political climate and nature of the American political framework.
This study serves as a starting point for other studies - by determining what the
relationship is between church attendance and political orientation, further studies can explore
why the relationship is the way it is. This may lead to further studies that examine the
consequences of this relationship, strategies for mobilization, etc. The field of Political Science
needs this study’s contribution to provide insight into the American political framework and to
provide a catalyst and foundation for future studies.
Section 3: Literature Review
Michael J. Perry, in his essay titled Religion in Politics, lays the foundation for our study
by enumerating that, “an overwhelming 70% of American adults are members of a church”
(Perry 3). Perry, however, does not specify which God or religion is being discussed, nor which
church is being attended. Perry’s contribution is nonetheless significant because it reveals, “the
role of religion is anything but marginal to American politics” (Perry 3). According to Perry,
religion and church attendance do indeed play a role in American politics; if that is the case,
since 70% of Americans belong to a church, a large majority, the exploration of the effects of
church attendance in American politics is vital to explore. This imperative forms the basis of our
study: what effect, if any, does church attendance on American politics?
Before the question of what effect church attendance has on American politics can be
asked, Perry’s claim that church attendance has any effect at all must be confirmed and verified.
Kenneth D. Wald, Dennis E. Owen, and Samuel S. Hill, authors of Churches as Political
Centers, substantiate Perry’s claim by exploring the ways in which churches can act as a political
4
nexus. According to the authors, churches function as political communities capable of swaying
and impressing political opinions upon those who are attending. The reason is due to the nature
of churches – they are a community of sorts built on a consensus of theology and beliefs (Wald,
Owen, and Hill 533). This community of consensus creates an amiable and collusive
environment to share and express ideas; in addition, within the church community, there is a
deferential hierarchy. Attendees listen to the one preaching and generally are not able to
communicate or interact - this prevents discussion or exploration of beliefs, resulting in the
attendees submitting to the preacher’s words. Because of the lack of discussion, the collusive
environment, and the community of consensus, the church environment maximizes behavioral
contagion and is ideal for dissemination of common views and theologies (Wald, Owen, and Hill
540). Perry, as well as Wald, Owen, and Hill, concur that church communities and preachers
frequently dabble into political matters. Thus, churches do indeed act as political communities by
directly and noticeably affecting the attendees’ political outlook, societal attitudes, and other
such variables that affect political orientation.
So far, the literature review has concluded that 70% of Americans attend some kind of
church, and that church communities are capable of noticeably impacting political behavior.
What, then, are the results of these impactful institutions? For that answer, we look to Geoffrey
Layman’s essay titled Religion and Political Behavior in the United States. In his essay, Layman
documents and explores the rise of a religious fundamentalism and orthodoxy within the United
States – this fundamentalism entails frequent church attendance and adherence to
orthodox/fundamentalist beliefs (Layman 300). According to Layman, this rise in
fundamentalism within the United States has created a new religious cleavage that emphasizes
theology and beliefs in determining political behavior and choices. Using data from the
5
American National Election Studies from 1980 to 1994, Layman tests his hypothesis that the rise
of doctrinal orthodoxy, religious commitment, and increased church attendance, has produced a
growing cleavage of conservative church goers in the United States.
Taken collectively, the various authors make three key observations. The first
observation, Perry’s main contribution, is the sheer number of people, 70%, that have indicated
they attend a church – by making this point, the topic of religion and politics potentially becomes
incredibly relevant as it permeates the entire American society. Perry’s observation is confirmed
by Wald et al., and the authors introduce a second key point: churches function as political
committees. Due to the collusive environment of theology, churches are capable of impressing
theological views on attendees – however, as Perry and Wald et al. have noted, the preachers in
these church communities often introduce politics. Thus, churches are a prime environment for
introducing and disseminating political values. Layman consolidates and coalesces the two
previous points and introduces us to our third point, the effect of these political centers
(churches) on over 70% of the American population. According to Layman, there is a rise in
orthodoxy that has produced a strong conservative movement within the United States. Those
who attend church and practice this orthodox faith identify strongly with conservative political
orientation.
What exactly is a conservative political orientation? The sheer volume of ideas,
institutions, beliefs, philosophies, and histories that fall under the sphere of the American
conservative movement is too vast to explore individually; instead, a broader, holistic approach
to identifying political conservative orientation will be taken. Accordingly, this study looks at
self-identification as the primary means of discerning political orientation because of the
efficiency it can provide salient information.
6
The last article, Closing the God Gap, by Mark Tooley, explores the voting breakdown in
the 2008 presidential election between John McCain and Barack Obama; this article provides a
real world example of this study’s hypotheses in action. This article will be used extensively in
the concluding section, found in Section 6.
Section 4: Problems with Existing Literature
While all three authors deal with different facets of the original topic, they each
contribute to an overall understanding: first, that many believe in God, second that churches act
as political centers, and third, that the political philosophy taught in these political centers is
trending towards conservatism. However, there are some definite holes and problems with this
existing literature.
Collectively, all of the authors focus generally on Christian denominations; while
Christianity is the largest sect in the United States, there could be important variations between
church attendance and political orientation among other non-mainstream religions, such as
Hinduism, Buddhism, Islam, etc. However, because the literature does not specify the particular
denomination, religion, or church being attended, our study must necessarily refrain from
distinguishing between denominations as well.
In addition, much of the existing literature uses statistical data from 1980 to 1994 and
thus may not properly reflect the current demographics of the United States. However, according
to Layman, the new religious cleavage began to take shape during those years, and thus the
statistics will provide salient and relevant information (Layman 302). Our study utilizes data
from 1948 to 2004 presidential election to avoid this pratfall.
7
Lastly, Tooley’s article, Closing the God Gap, defines “religiously observant” as
attending church; there are indeed other facets to being religiously observant other than attending
church, but they are difficult to track, categorize, and analyze. Thus, this study utilizes the
criterion as Tooley does.
Section 5: Data and Analysis
Section 5.1: Hypothesis 1
Our first hypothesis (H1) is as follows: Those who regularly attend church will generally
self-identify as conservative. Based on the literature reviewed, a confirmation of this hypothesis
would be logical. To test this hypothesis, this study utilizes data provided by the National
Election Studies from 1948 to 2004 will be utilized; specifically, the variables VCF0804 (which
this study designates political Orientation) andVCF0130 (which this study designates Church
Attendance).
Figure 1: Regular Attendance and Political Orientation Crosstabulation
Regularly Attend * Orientation recoded Crosstabulation
Count Orientation recoded
Total Liberal Moderate Conservative
Regularly Attend 1577 2823 4617 9017
Figure 1 is a crosstabulation summary of the number of people who regularly attend church and
identify as liberal, moderate, or conservative. The large percentage of conservative respondents
8
is readily apparent; to explore this relationship and test H1, the data was run through a linear
regression analysis. The results are found in Figure 2 and Figure 3.
Figure 2: Regularly Attend and Political Orientation Graphic Representation with Linear
Line
Figure 3: Regularly Attend and Political Orientation Linear Regression Values
Model
Unstandardized
Coefficients
B Std. Error
1 (Constant) 2.035 .008
Attendance
Regularly DUMMY
.652 .011
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
Liberal Moderate Conservative
Orientation recoded
Regularly Attend
Regularly Attend
Linear (RegularlyAttend)
9
Figure 4: Interpretation of Pearson Product-Moment Correlation Coefficients
(Stockburger)
Correlation Negative Positive
None -0.09 to 0.0 0.0 to 0.09
Small -0.3 to -0.1 0.1 to 0.3
Medium -0.5 to -0.3 0.3 to 0.5
Strong -1.0 to -0.5 0.5 to 1.0
Figures 2 and 3 reveal a strong correlation between regularly attending church and self-
identifying as conservative. Confirming this, the Pearson Product-Moment Correlation
Coefficient found through the regression analysis was .652 (Figure 3). Based on the
interpretation chart (Figure 4), this represents a strong correlation.
Thus, H1 (Those who regularly attend church will generally self-identify as conservative)
is a definitively confirmed hypothesis with a strong Pearson Product-Moment Correlation
Coefficient of .652 – this result was expected based on the extensive literature on the rise of the
religious conservative cleavage documented Geoffrey Layman’s essay, Religion and Political
Behavior in the United States.
Section 5.2: Hypothesis 2
Our second hypothesis (H2) is as follows: Those who sometimes attend church will have
no political preference. Based on the literature reviewed, the outcome of this hypothesis is
unknown; we know there is a strong correlation between regularly attending and conservatism,
but the effects of sometimes attending cannot easily be garnered from the literature or previous
10
test. To test this hypothesis, this study utilizes data provided by the National Election Studies
from 1948 to 2004 will be utilized; specifically, the variables VCF0804 (which this study
designates political Orientation) andVCF0130 (which this study designates Church Attendance).
Figure 5: Sometimes Attend and Political Orientation Crosstabulation
Attendance Sometimes * Orientation recoded Crosstabulation
Count Orientation recoded
Total Liberal Moderate Conservative
Attendance Sometimes 816 1135 1207 3158
Figure 5 is a crosstabulation summary of the number of people who sometimes attend church and
identify as liberal, moderate, or conservative. There is not a very significant amount of
differentiation from the mean, 1052.66, for each category. The number of liberals only deviates
from the mean by 236.66; moderates by 82.34; and conservatives by 154.34. The deviations are
generally similar, thus potentially reveal an insignificant correlation between sometimes
attending church and political orientation. To explore this relationship and test H1, the data was
run through a linear regression analysis. The results are found in Figure 6 and Figure 7.
Figure 6: Sometimes Attend and Political Orientation Graphic Representation with Linear
Line
11
Figure 7: Sometimes Attend and Political Orientation Linear Regression Values
Model
Unstandardized
Coefficients
B Std. Error
1 (Constant) 2.035 .008
Attendance
Sometimes
DUMMY
.089 .011
Figure 8: Interpretation of Pearson Product-Moment Correlation Coefficients
(Stockburger)
Correlation Negative Positive
None -0.09 to 0.0 0.0 to 0.09
0100200300400500600700800900
100011001200
Liberal Moderate Conservative
Orientation recoded
Sometimes Attend
Sometimes Attend
Linear (SometimesAttend)
12
Small -0.3 to -0.1 0.1 to 0.3
Medium -0.5 to -0.3 0.3 to 0.5
Strong -1.0 to -0.5 0.5 to 1.0
Figure 6 and 7 reveal a very weak or non-existent correlation between sometimes attending
church and a particular political orientation. Confirming this, the Pearson Product-Moment
Correlation Coefficient found through the regression analysis was .089 (Figure 7). Based on the
interpretation chart (Figure 8), this represents a nonexistent or small correlation.
Thus, H2 (Those who sometimes attend church will have no political preference) is a
definitively confirmed hypothesis with a non-existent or small Pearson Product-Moment
Correlation Coefficient of .089 – this result was neither expected nor unexpected based on the
literature of the topic, as no literature extensively explored the political orientation of those who
sometimes attend church.
Section 5.3: Hypothesis 3
Our third hypothesis (H3) is as follows: H3: Those who rarely attend church will
generally self-identity as liberal. Based on the literature reviewed and the results of H1, this
hypothesis will likely be confirmed; because the correlation between regularly attending and
conservatism is very high, the opposite is likely true. To test this hypothesis, this study utilizes
data provided by the National Election Studies from 1948 to 2004 will be utilized; specifically,
the variables VCF0804 (which this study designates political Orientation) andVCF0130 (which
this study designates Church Attendance).
Figure 9: Sometimes Attend and Political Orientation Crosstabulation
13
Rarely Attend * Orientation Crosstabulation
Count Orientation recoded
Total Liberal Moderate Conservative
Rarely Attend 3147 3729 3511 10387
Figure 9 is a crosstabulation summary of the number of people who rarely attend church and
identify as liberal, moderate, or conservative. There is not a very significant amount of
differentiation from the mean, 3462.33, for each category. The number of liberals only deviates
from the mean by -315.33; moderates by +266.67; and conservatives by +48.67. The deviations
between liberal and moderate are very similar – however, conservatives are closest to the mean.
This data seems to suggest there is no real correlation between rarely attending church and
identifying as liberal, seemingly disproving H3.To explore this relationship and test H1, the data
was run through a linear regression analysis. The results are found in Figure 10 and Figure 11.
Figure 10: Rarely Attend and Political Orientation Graphic Representation with Linear
Line
14
Figure 11: Sometimes Attend and Political Orientation Linear Regression Values
Model
Unstandardized
Coefficients
B Std. Error
(Constant) 2.035 .008
Attendance Rarely
DUMMY
.115 .011
Figure 12: Interpretation of Pearson Product-Moment Correlation Coefficients
(Stockburger)
Correlation Negative Positive
None -0.09 to 0.0 0.0 to 0.09
Small -0.3 to -0.1 0.1 to 0.3
0
500
1000
1500
2000
2500
3000
3500
Liberal Moderate Conservative
Political Orientation
Rarely Attend
Rarely Attend
Linear (Rarely Attend)
15
Medium -0.5 to -0.3 0.3 to 0.5
Strong -1.0 to -0.5 0.5 to 1.0
Figure 10 and 11 reveal a very weak correlation between rarely attending church and liberal
political orientation, as H3 predicted. Instead, rarely attending church is more identified with
being of moderate and conservative political orientation, but this correlation is so weak as to be
non-significant. Confirming this, the Pearson Product-Moment Correlation Coefficient found
through the regression analysis was .115 (Figure 11). Based on the interpretation chart (Figure
12), this represents a nonexistent or small correlation.
Thus, H3 (Those who rarely attend church will generally self-identity as liberal) has been
disproved– there is a nonexistent or small Pearson Product-Moment Correlation Coefficient of
.115 between rarely attending church and identifying as liberal. In order to have confirmed this
hypothesis, the correlation coefficient should have been a negative coefficient between -1.0 to -
0.5. This result was unexpected based on the literature of the topic; the literature extensively
confirmed H1, thus potentially revealing the behavior of the opposite demographic in H3 – this
was certainly not the case.
Section 5.4: Results and Additional Data
To summarize, based on the above data and the correlation coefficients:
H1 and H2 are confirmed.
H3 is disproved.
16
However, the correlation values fail to take into account the overall number of liberals,
moderates, and conservatives, found in Figure 13 below
Figure 13: Total Number of Respondents who Identified as Liberal, Moderate, or
Conservative.
Attendance recoded
Total Rarely Sometimes Regularly
Orientation
recoded
Liberal 3147
(56.8%)
816
(14.7%)
1577
(28.5%)
5540
Moderate 3729
(48.5%)
1135
(14.8%)
2823
(36.7%)
7687
Conservative 3511
(37.6%)
1207
(12.9%)
4617
(49.5%)
9335
Total 10387 3158 9017 22562
As Figure 13 shows us, conservative respondents outnumber liberal respondents almost 2
to 1; it would make sense, then, that more moderates and conservatives rarely attend church
simply because there are more moderates and conservatives than there are liberals. Thus, the
results of the data above are technically true; if you examine all of the Americans that
participated in this survey, you would find that there was little correlation between sometimes
attending church and rarely attending church and identifying as liberal or moderate. And, since
conservatives were the largest group within the survey, the respondents were spread within the
17
various attendance categories. This, however, still does not reveal the true nature of church
attendance and political orientation, largely because of the skewed and disproportionate number
of people that identify as conservative.
Figure 14: A Graphical Representation of the Attendance and Orientation as a Percentage
of the Total Number
To successfully determine the correlation between liberalism and church attendance, the
number of respondents would have to be even distributed between liberals, moderates, and
conservatives; to do that, each respondent was taken as a percentage of the whole, rather than as
an absolute number (since there are disproportionately more moderates and conservatives than
liberals). Figure 14 reveals a more accurate breakdown that compares the percentage of liberals,
moderates, and conservatives, that rarely attend, sometimes attend, and regularly attend church.
We cannot, however, perform a regression analysis, as the numbers are not absolute, nor do they
0
10
20
30
40
50
60
Liberal Moderate Conservative
Rarely Attend
Sometimes Attend
Regularly Attend
18
properly reflect the data. Nonetheless, powerful observations can be gleaned that can be used to
explore our hypotheses more accurately than if we used absolute data.
H1: Those who regularly attend church will generally self-identify as conservative.
Respondents who regularly attend church:
Identify as Conservative 49.5% of the time
Identify as Moderate 36.7% of the time
Identify as Liberal 28.5% of the time.
By looking at the relative percentages, we can see those who attend church regularly generally
identify as conservative (49.5% of the time) rather than moderate (36.7%) or liberal (28%).
H1 is thus definitively confirmed.
H2: Those who sometimes attend church will have no political preference.
Respondents who sometimes attend church:
Identify as Conservative 12.9% of the time.
Identify as Moderate 14.8% of the time.
Identify as Liberal 14.7% of the time.
By looking at the relative percentages, we can see there is very little differentiation between
people who sometimes attend church and their political orientation. Generally, the same amount
19
of conservatives, moderates, and liberals attend church sometimes. Thus, there is no political
preference for those who sometimes attend church.
H2 is thus definitively confirmed.
H3: Those who rarely attend church will generally self-identity as liberal.
Respondents who rarely attend church:
Identify as Conservative 37.6% of the time
Identify as Moderate 48.5% of the time.
Identify as Liberal 56.8% of the time.
By looking at the relative percentages, we can see there is a significant differentiation between
people who rarely attend church and their political orientation. Those who rarely attend church
are far more likely to identify as liberal and moderate. Thus, we cannot say that those who rarely
attend church will generally identify as liberal, since those who identified as moderate was a high
percentage (48.5%), we can, however, conclude the null hypothesis of H1, meaning, those who
rarely attend church will generally not identify as conservative.
H3 is thus partially confirmed.
Section 7: Methodological Appendix
Section 7.1: Data and Analyses
20
The data in this study was provided by the National Election Studies from 1948 to 2004;
the two main variables analyzed are variable “VCF0804” (which this study designates Political
Orientation) and variable “VCF0130” (which this study designates Church Attendance).
In order to analyze each level of church attendance separate, we separated VCF0130,
Church Attendance, into 3 compressed categories: sometimes, rarely, and regularly. This
matches the spirit of the original categories, which are as follows:
1. Every Week
2. Almost every week
3. Once or twice a month
4. A few times a year
5. Never
The regularly category is compromised of #1 and #2.
The sometimes category is compromised of #3.
The rarely category is compromised of #4 and #5.
In order to properly analyze effect of church attendance on political orientation, we
analyzed VCF0804, the compressed version of the liberal-conservative scale found in VCF0803,
“Liberal-Conservative Scale.” This allowed us to approach the correlation in a holistic manner
that corresponded to the collapsed version of the church attendance variable.
The correlations in our study were acquired by utilizing linear regression analysis and
finding a Pearson product-moment correlation coefficient using SPSS.
21
The graphical representations and linear lines were found using Microsoft Excel’s
graphing functions.
Section 7.2: Figures and Table
Figure 1: Regular Attendance and Political Orientation Crosstabulation
Regularly Attend * Orientation recoded Crosstabulation
Count Orientation recoded
Total Liberal Moderate Conservative
Regularly Attend 1577 2823 4617 9017
Figure 2: Regularly Attend and Political Orientation Graphic Representation with Linear
Line
Figure 3: Regularly Attend and Political Orientation Linear Regression Values
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
Liberal Moderate Conservative
Orientation recoded
Regularly Attend
Regularly Attend
Linear (RegularlyAttend)
22
Model
Unstandardized
Coefficients
B Std. Error
1 (Constant) 2.035 .008
Attendance
Regularly DUMMY
.652 .011
Figure 4: Interpretation of Pearson Product-Moment Correlation Coefficients
(Stockburger)
Correlation Negative Positive
None -0.09 to 0.0 0.0 to 0.09
Small -0.3 to -0.1 0.1 to 0.3
Medium -0.5 to -0.3 0.3 to 0.5
Strong -1.0 to -0.5 0.5 to 1.0
Figure 5: Sometimes Attend and Political Orientation Crosstabulation
Attendance Sometimes * Orientation recoded Crosstabulation
Count Orientation recoded
Total Liberal Moderate Conservative
Attendance Sometimes 816 1135 1207 3158
23
Figure 6: Sometimes Attend and Political Orientation Graphic Representation with Linear
Line
Figure 7: Sometimes Attend and Political Orientation Linear Regression Values
Model
Unstandardized
Coefficients
B Std. Error
1 (Constant) 2.035 .008
Attendance
Sometimes
DUMMY
.089 .011
Figure 8: Interpretation of Pearson Product-Moment Correlation Coefficients
(Stockburger)
0100200300400500600700800900
100011001200
Liberal Moderate Conservative
Orientation recoded
Sometimes Attend
Sometimes Attend
Linear (SometimesAttend)
24
Correlation Negative Positive
None -0.09 to 0.0 0.0 to 0.09
Small -0.3 to -0.1 0.1 to 0.3
Medium -0.5 to -0.3 0.3 to 0.5
Strong -1.0 to -0.5 0.5 to 1.0
Figure 9: Sometimes Attend and Political Orientation Crosstabulation
Rarely Attend * Orientation Crosstabulation
Count Orientation recoded
Total Liberal Moderate Conservative
Rarely Attend 3147 3729 3511 10387
Figure 10: Rarely Attend and Political Orientation Graphic Representation with Linear
Line
25
Figure 11: Sometimes Attend and Political Orientation Linear Regression Values
Model
Unstandardized
Coefficients
B Std. Error
(Constant) 2.035 .008
Attendance Rarely
DUMMY
.115 .011
Figure 12: Interpretation of Pearson Product-Moment Correlation Coefficients
(Stockburger)
Correlation Negative Positive
None -0.09 to 0.0 0.0 to 0.09
Small -0.3 to -0.1 0.1 to 0.3
0
500
1000
1500
2000
2500
3000
3500
Liberal Moderate Conservative
Political Orientation
Rarely Attend
Rarely Attend
Linear (Rarely Attend)
26
Medium -0.5 to -0.3 0.3 to 0.5
Strong -1.0 to -0.5 0.5 to 1.0
Figure 13: Total Number of Respondents who Identified as Liberal, Moderate, or
Conservative.
Attendance recoded
Total Rarely Sometimes Regularly
Orientation
recoded
Liberal 3147
(56.8%)
816
(14.7%)
1577
(28.5%)
5540
Moderate 3729
(48.5%)
1135
(14.8%)
2823
(36.7%)
7687
Conservative 3511
(37.6%)
1207
(12.9%)
4617
(49.5%)
9335
Total 10387 3158 9017 22562
Figure 14: A Graphical Representation of the Attendance and Orientation as a Percentage
of the Total Number
27
0
10
20
30
40
50
60
Liberal Moderate Conservative
Rarely Attend
Sometimes Attend
Regularly Attend
28
References
Layman, Geoffrey C. Religion and Political Behavior in the United States: The Impact of
Beliefs, Affiliations, and Commitment From 1980 to 1994. The Public Opinion Quarterly.
Vol. 61 No. 2. American Association for Public Opinion Research. pp288 – 316. Web.
Perry, Michael J. Religion in Politics. U.C. Davis L. Rev. 729. 1995-1996. Web.
Stockburger, David. Correlation. Missouri State Education. Web.11 May. 2012.
Tooley, Mark D. "Closing the God Gap; Did Obama's religious outreach pay off?" The Weekly
Standard 7 Nov. 2008. Academic OneFile. Web. 26 Apr. 2012
Wald, Kenneth, D. Dennis E Owen, and Samuel S Hill Jr. Churches as Political Communities.
The American Political Science Review. Vol 82. No. 2. American Political Science
Association 1988. pp 531-548.Web.