Citation
A test of the southern subculture of violence theory

Material Information

Title:
A test of the southern subculture of violence theory
Creator:
Landy, Christian Adam
Publication Date:
Language:
English
Physical Description:
59 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Violence -- Southern States ( lcsh )
Crime -- Sociological aspects -- Southern States ( lcsh )
Crime -- Cross-cultural studies -- United States ( lcsh )
Homicide -- Southern States ( lcsh )
Crime ( fast )
Crime -- Sociological aspects ( fast )
Homicide ( fast )
Violence ( fast )
Southern States ( fast )
United States ( fast )
Genre:
Cross-cultural studies. ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )
Cross-cultural studies ( fast )

Notes

Bibliography:
Includes bibliographical references (leaves 58-59).
General Note:
Department of Sociology
Statement of Responsibility:
by Christian Adam Landy.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
53362558 ( OCLC )
ocm53362558
Classification:
LD1190.L66 2002m .L36 ( lcc )

Full Text
A TEST OF THE SOUTHERN SUBCULTURE OF VIOLENCE THEORY
by
Christian Adam Landy
B.A., Arizona State University, 2000
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Arts
Sociology
2002
'T
AL >
{ 1:


This thesis for the Master of Arts
degree by
Christian Adam Landy
has been approved
by
Richard Anderson
Virginia Fink
Date


Landy, Christian Adam (M. A., Sociology)
A Test of the Southern Subculture of Violence Theory
Thesis directed by Associate Professor Richard Anderson
ABSTRACT
Regional variations in violent crime rates across the United States have
spun several cultural and subcultural theories. However, empirical support for
this theory has been inconsistent at best. The purpose of this thesis is to test the
Southern Subculture of Violence Theory by examining attitudinal questions that
describe violence used in defense or in a retaliatory manner. The data examined
was drawn from the 1994 General Social Survey. The results illustrated little
evidence of attitudinal answers varying by regions.
This abstract accurately represents the content of the candidates thesis. I
recommend its publication.
m
Richard Anderson


DEDICATION
I dedicate this thesis to my parents for their undying support and understanding
during the completion of my studies. Without their help this would not be
possible.


ACKNOWLEDGEMENT
I'd like to give special thanks to my advisor, Richard Anderson, for his guidance
and support during the past two years. I would also like to thank all of my
instructors who all assisted me in the completion of this program.


CONTENTS
Tables.......................................vii
CHAPTER
1. INTRODUCTION.................................1
Why is this Important?.....................4
2. LITERATURE REVIEW............................8
3. THEORETICAL CONSIDERATIONS..................21
4. DATA AND METHODOLOGY........................29
5. RESULTS.....................................42
6. DISCUSSION..................................53
REFERENCES..................................57
vi


TABLES
Table
5.1 Crosstabs of Approval of Capital Punishment and Region............42
5.2 Crosstabs of Approval of Capital Punishment and Political View...43
5.3 Crosstabs of Approval of Capital Punishment and Fundamentalist..44
5.4 Crosstabs of Approval of Capital Punishment and Interracial Marriages....45
5.5 Crosstabs of Approval of Capital Punishment and Neighborhood Race... .46
5.6 Logistic Regression for Capital Punishment......................49
5.7 Logistic Regression for Police Attack...........................50
5.8 Logistic Regression for Police Abuse............................52
vii


CHAPTER 1
INTRODUCTION
A crime can be defined as any act or omission permitted by public law for the
protection of the public, and made punishable by the state in a judicial
proceeding in its own name. (Wolfgang 1976, 14)
Crime is present not only in the majority of societies but in all societies.
One of the most important issues in criminology is the search for causes. The
majority of criminological research looks for causes of crime. The conclusions
that are drawn as a result of the research can mainly be seen as hypotheses.
These hypotheses continue to require further testing before an actual cause can
be established. This clearly is one of the objectives of this paper. I wish to look
at a current theory and test it again so as to compare my results with the results
of the past.
Perhaps the most intriguing type of crime in the world is homicides.
Homicides can be defined as an act in which one person takes the life of another
person. Either because of the severity of the crime or because of media
l


exposure, homicides continue to fascinate the entire world. Especially in the
United States, homicides have a common stigma about them. In the United
States, local and national homicides continue to receive extensive media
exposure. In the past, cases like the Jon Benet Ramsey murder and the O. J.
Simpson murder trial have captivated the entire country. Nowhere else in the
Industrialized world are more homicides committed then the United States.
According to the Bureau of Justice Statistics, the United States homicide rate
peaked in 1980 at 10.2 per 100,000 populations and in 1999 it has leveled off at
5.7 per 100,000 populations. (Bureau of Justice Statistics 2001)
It has been known that in the United States, there is a wide variation in
the rates between the North and South regions. These variations have produced
a number of theories that examine these rates. Culture and subculture theories
claim that certain groups of people contain subcultures that are more lenient
towards specific types of violence. These theories were then applied to the high
homicide rates in the South and the Southern Culture of Violence theory was
created. Formulated by Marvin Wolfgang and Franco Ferracuti in 1967, the
culture of violence theory claims that subcultures among groups provide greater
support for violence when upholding honor, courage, and manliness. This
theory only claims that violence is accepted only when its being used in self-
defense. A common misconception is that the theory claims that there is a
2


general, overall, acceptance of violence. However, this is not the case. (Smith &
Zahn 1999)
In this thesis, I will test the Southern Culture of Violence theory, which
claims that violent attitudes are linked to region of residence. I believe
examining this theory again has many advantages. Not only will I contribute to
the existing pool of research, but I will also test this theory using different
methods from most previous studies.
3


Why is This Important?
One of the most interesting trends in homicide research is the homicide
rate of the United States. When compared to the rest of the industrial world, the
United States has a much higher rate. This topic has received volumes of
research devoted to it in hopes of discovering some answers. One way people
have studied this phenomenon is by looking at regional patterns. In the past, the
South has long been viewed as place of warmth and romance. Southerners have
been seen as kind and caring. Even the term Southern Hospitality was coined
to describe this caring personality. However, there have also been claims that
the South has a much darker side. Foreign visitors as well as northerners have
shared this belief. Feuds, Duels, and lynchings have been stereotyped as
common occurrences in the South. Even in classic literature Southerners have
been stereotyped as violent. In Mark Twains bookPuddnhead Wilson, the
character Pembroke Howard was a man who was always ready to stand up to
anyone with any weapon if any act or word he said was doubted. Between 1865
and 1915, some Southern homicide rates were reported as high as 130 per
100,000 populations. (Nisbett & Cohen 1996)
A similar, less drastic trend can be seen today. When examining the
homicide rates in the United States by regions, researchers found that the rates
4


were consistently higher in the South as compared to the rest of the country.
Soon, theories were bom which claimed that the high rates in the south were to
blame for the high rates of the United States in general. The focus of many
researchers continued on regional differences. If these high rates in the south
are the reason why the United States leads the industrial world in homicides then
we must know whats affecting these rates. Many studies have either used
region as a control variable or they have directly examined regions to explain
why. Despite how it is studied, regional differences have been given a great
deal of attention by homicide researchers.
Why has this subculture of violence developed in the South? The
historical origins of the southern subculture of violence remain inconsistent.
One explanation traces the roots of southern violence to life in the antebellum
South, especially the settling of the backwoods frontier. Another explanation
emphasized the emergence of a tradition of southern violence in the years
following the Civil War. This explanation claims that the military defeat of the
South by the northern forces gave rise to high levels of tolerance towards
interpersonal violence. Subsequent political and economic domination and
exploitation also contributed. Its also important to look at how these tolerant
attitudes towards violence are fostered or maintained. Reed and others have
claimed that childhood socialization and peer group influence may be at work.
5


He claimed that playground fights might mold attitudes towards violence at a
young age. He points out that these young playground fights are quite common
in the South. Another possibility to consider is the long tradition military
services have had in the South. This tradition and respect for military customs
may also affect attitudes towards interpersonal violence.
I have chosen to look at the culture of violence theory for two reasons.
First, of the past theories many of the findings and results have been
inconsistent. Some researchers have found a culture of violence in the south,
such as Messner (Messner 1983), Hackney (Hackney 1969) and Gastil (Gastil
1971) and Reed (Reed 1971). These studies found that regional variables
produced a significant independent effect on homicide rates. However, there
have been several studies that have found inconclusive evidence of this
subculture such as Loftin and Hill (Loftin and Hill 1974) and Jo Dixton and
Alan Lizote (Dixton and Lizotte 1987). In the literature review that follows, I
will review key studies that supported this theory such as those conducted by
Steven Messner, Sheldon Hackney, Raymond Gastil, John Reed, and
Christopher Ellison. I will also review the key studies that did not show support
for this theory. Included are studies conducted by Jo Dixon and Alan Lizotte,
Colin Lofton and Robert Hill, and Sandra Ball-Rokeach. In addition to this
many studies have failed to test for defensive violence. The Southern
6


Subculture of Violence clearly states that the tolerance towards violence is only
violence that is used in defense. Second, I believe a theory must be supported
by research before any other applications can be made. Before this theory is
applied to other problems, I feel its imperative that the theory should be tested
enough to the point where the results are conclusive. In my example, in order to
state that the Southern Subculture of Violence Theory explains the high
homicide rates of the United States, its necessary to find evidence that a
Southern Subculture of Violence exists. Therefore, I feel it is necessary to test
this theory again. The vast majority of research on this topic has chosen to use
surveys as their method of measuring violence in a particular culture. I will also
be using data that was gathered through surveys as well, however, I will be
looking at some variables past studies have not.
By conducting this study I hope to answer two important questions.
First, is region significantly associated with attitudes towards capital
punishment? And second, is region significantly associated with attitudes
towards defensive violence.
7


CHAPTER 2
LITERATURE REVIEW
As I mentioned above in the introduction, many studies have examined
and tested this theory. While most of these studies used similar methods, some
of them found conflicting results. Im going to look at a couple studies that
supported the theory as well as those that did not. Before I review these studies,
I want to first define the Southern Subculture of Violence theory. Formulated
by Marvin Wolfgang and Franco Ferracuti in 1967, the culture of violence
theory claims that subcultures among groups provide greater support for
violence when upholding honor, courage, and manliness. This theory only
claims that violence is accepted only when its being used in self-defense. A
common misconception is that the theory claims that there is a general, overall,
acceptance of violence. However, this is not the case. (Smith & Zahn 1999)
In past studies there have been many different measures of this theory.
Hackney and Gastil conducted regression analyses of state homicide rates that
indicated significant regional differences. Both authors claimed that when a
regional variable for the South was entered into a regression equation along with
socioeconomic and demographic variables, it produced a significant independent
8


effect on the state homicide rate. Loftin and Hill, as well as Messner, examined
official homicide rates for 204 SMSAs. These rates were examined along with
a structural poverty index and two regional variables. Ellison and Ball-Rokeach
used different measures to examine this theory. They examined survey data at
the individual level that included questions that pertained to the respondents
tolerance of and participation in violent behavior. Ive chosen to use similar
methods for two reasons. First, I feel a study using individual level would be
more valuable because I would be able to make inferences about micro-level
processes much easier than if aggregate level data was used. This is often
referred to as aggregation bias. As Loftin and Hill stated, Arbitrarily
aggregating data into state units introduces a serious threat of bias and prevents
us from interpreting the results of the regression analysis as meaningful
estimates of the models parameters. (Loftin and Hill: 1974, P 723)
I now want to look at a study conducted by Steve Messner in 1983. The
purpose of his study was to evaluate the thesis by examining the relationships
among region, racial composition, and the homicide rate for a sample of 204
Standard Metropolitan Statistical Areas or SMSAs. The dependent variable
used for his analysis was the official homicide rate for 204 SMSAs. He
computed a three-year average from the years 1969-1971. The homicide
figures were obtained from the FBIs Uniform Crime Report or UCR's. These
9


reports contain homicides known to the police per 100,000 populations. The
first independent variable used in his analysis was a structural poverty index.
This was included because in Loftin and Hills study, they found that a poverty
index was the strongest predictor of homicide rates. This structural poverty
index consisted of five indicators that were all chosen from Loftin and Hills
study. The first indicator was infant mortality rates. These were obtained
through the U.S. Department of Health, Education, and Welfare from 1975. The
second indicator was the percent of persons 25 years and older with less than
five years of schooling. These data were gathered from the U.S. Bureau of
Census from 1973. The third indicator was the failure rate on the Armed Forces
Mental Test. These data were obtained from the National Center for
Educational Statistics from 1973. The final indicator was the percentage of
children living with one parent. These data were drawn from the U.S. Bureau of
Census in 1973. Messner also analyzed two regional variables. The first was a
Confederate dummy that contained all SMSAs located in the former
Confederate States. The second regional variable was a southemess index,
which was determined by assigning SMSAs the Southemess score of their
respective states. Messner also added five control variables. These variables
were included because they all have emerged as significant determinants of the
homicide rate in other studies. The first is the percentage of the population that
10


is black. These data were drawn from the U.S. Bureau of Census from 1978.
The second control variable was the percent aged 20-34. These data were drawn
from the Census Bureau as well. The third control variable was the Gini index
of income inequality, which was drawn from the Census. The fourth control
variable was the natural log of the population. These data were drawn from the
Federal Bureau of Investigation from 1970. The final control variable was the
natural log of the population per square mile. These data were drawn from the
Census Bureau of 1973.
Through regression analysis, Messner found that both southern region
and the relative size of the black population exhibit significant partial effects on
the homicide rate even with controls for the theoretically important
sociodemographic and demographic variables. He also discovered that the
relationship between racial composition and the homicide rate varies
significantly by region. This means that the proportion of the population that is
black has a strongly positive effect on the homicide rate outside the South, but
its effect is negligible inside the south. Overall, his results showed support for
the Southern Subculture theory. (Messner 1983)
Hackney and Gastil conducted similar studies in 1969. They questioned
several non-cultural explanations claiming that they dont hold up under
examination. They argued that the South does have a distinctive subculture that
ll


can be described as violent. He supported his claim with two types of evidence.
First, they cited features of southern history that contained violent orientations.
For example, he claimed violence emerged during the Civil War era. The defeat
of the South and its overall exploitation by the North generated collective
grievances and low thresholds for aggression. Second, he indicated various
regional differences in homicide rates with regression analyses. He found that
when a regional variable for the south was entered in a regression equation, it
had a significant effect on homicide rates. He claimed that this effect was
reflected by the violent characteristics of the southern culture. (Messner 1983)
Next, Im going to look at an article written by John Shelton Reed. The
purpose of this article was to examine regional differences in attitudes and
behaviors of those concerning private ownership and use of guns and those
concerning corporal punishment of children. He used data gathered by the
Roper Public Opinion Research Center. He used this data to first look at the
direction and magnitude of the differences between the South and the rest of the
United States. Second, he looked at the extent to which these differences are
due to economic and social differences between the South and non-South. Next,
he examined the changes in these attitudes and behaviors in the recent past.
When looking at guns he found that southerners are more likely to own guns
when compared with other regions. He proposes that the high percentage of
12


southerners that have guns could be result of a greater need due to hunting or
because of relaxed gun laws. When looking at how parents discipline their
children, he found that southerners were more likely to report that themselves
were spanked as children and that they approve of spanking their own children.
Finally, the author used the technique of test-factor standardization which
allowed him to construct a hypothetical Southern population that was
industrialized, urban, and well educated. Attitudinal studies control on test
factors by inspecting the association within each of the test factor categories. A
test-factor standardization provides a summary measure of what population rates
would be if certain population characteristics we held constant. For more
information on test factor standardization see Rosenberg. (Rosenberg 1962)
Then reed would test the arguments that the differences noted above are because
southerners are less educated and more rural. Here he found insignificant
differences in their answers. This means that the differences between regions
that still remain are due to something else. Overall, his results showed support
for the Southern Subculture of Violence Theory. (Reed 1971)
I now want to look at a study that was conducted by Christopher Ellison
in 1991. The purpose of his study was to test the Southern Subculture of
Violence thesis. More specifically, his research tests the hypothesis of regional
differences in levels of individual support for violence by focusing on defensive
13


and retaliatory forms of violence. In order to conduct his analysis, Ellison used
data from the 1983 General Social Survey. His study was designed to explore
regional differences in the levels and determinants of support for defensive
violence. First, he regressed an index of support for defensive violence with
regional and social factors. The defensive violence index consisted of three
questions that were combined. The first asks, Would you approve of a man
punching a stranger in the face if the stranger punched his child after the child
accidentally damaged his car? The second asks, Would you approve of a
man punching a stranger in the face if the stranger was beating up a woman?
The third question asks, Would you approve of the use of violence against
someone who broke into your home? Ellison continued his analysis by
inserting a series of interaction terms into the full ordinary least squares
regression model in order to investigate whether the predictors of violence vary
amongst regions. Ellison found that Southerners were more supportive of
defensive violence in hypothetical situations than are nonsouthemers. This
finding is consistent with the Southern Subculture of Violence theory. (Ellison
1991)
Im going to now look at a book that was written in 1996 that examined
violence in the South. In their book, Richard Nisbett and Dov Cohen attempt to
explain the high homicide rates in the South. The authors argue that these high
14


rates arent due to socioeconomic problems, population density, past slavery, or
higher temperatures. They claim that these rates are due to a culture of honor in
which a man's reputation is central to his economic survival. By conducting
laboratory research as well as field studies, the authors found several trends that
supported the culture theory. First, by examining social policies, surveys, and
historical data, the authors found that it in the south, it is more acceptable to be
violent in response to insults or in order to protect homes and property.
Another trend the authors found had to do with Southerners responses to
insults. The authors found that when Southerners and Northerners were both
insulted in laboratories, Southerners not only displayed more anger but they also
displayed more physiological changes that are associated with stress and
aggression. The authors also looked at hiring patterns of Southern employers as
opposed to Northern employers. They found that Southern employers are more
likely to hire people who have killed in an honor-related crime than are Northern
employers. Finally, the authors looked at how both Northern and Southern
newspapers described killings. More specifically, they looked at how the each
described honor-related killings and they found that Southern newspapers
described honor-related killings in ways that were more sympathetic to the
killers. These newspapers would often give the killers the benefit of the doubt.
Overall, I thought this book did an excellent job examining the Southern Culture
15


of Violence. By conducting field and lab experiments, the authors were able to
present support for this theory. (Nisbett & Cohen 1996)
Im going to continue by examining a study conducted by Jo Dixton and
Alan Lizotte that did not find any support for the theory. In their study, they
attempted to find a relationship between regions and subcultures of violence
while determining the relationship of both to gun ownership. They analyzed
data from the General Social Surveys using regressions. They looked at male
respondents from the years 1976, 1980, and 1984. The questions used in Ball-
Rokeach were used in this study as well. These questions asked the respondents
whether they approve of the use of violence in general situations. The authors
found no evidence that a southern subculture of violence existed. After
controlling for economic and demographic factors, they failed to find any
significant differences in the violent attitudes of those in the South and those of
other regions. When looking at gun ownership, again they found no significant
relationship between gun ownership and violent attitudes. (Dixon & Lizotte
1987)
Loftin and Hill conducted another study in 1974. In response to the
Gastil-Hackney model of homicide, these authors argued that the structural
variables used by those researchers are poorly measured and highly correlated
with region, so that it is impossible to separate the effects of each on homicide
16


rates. In this study, the researchers analyzed cross-sectional data on state
homicide rates with a structural poverty index that combines several
socioeconomic variables. These variables were percent of persons twenty-five
years old, percent of the population illiterate, percent of families with income
under $1,000, percent of children living with one parent, armed forces mental
test failures, and infant mortality rates. In addition the authors included a Gini
index as a measure of income inequality. They found that the structural poverty
index was the most powerful predictor of a states homicide rate. They also
discovered that racial composition is not significant in a regression analysis once
the poverty index has been included. Overall, their findings claimed that
poverty, not subcultural values, accounts for the high rates of homicide in
southern states and in states with large proportion of blacks. (Smith & Zahn
1999)
Sandra Ball-Rokeach also tested this theory in 1973. More specifically,
she conducted two independent studies: one on values and attitudes underlying
interpersonal violence in a national probability sample of 1,429 adult Americans
over twenty-one years of age, the other on value differences among 363 men
incarcerated in a Michigan prison for various felonious offenses. Data for the
first study were gathered by the Nation Opinion Research Center in April 1971.
Cochrane collected the prison inmate data for the second study in 1968 and
17


1969. In her study, Ball-Rokeach defined violent behavior with four questions:
Have you ever been punched or beaten?, Have you ever been threatened or
actually cut with a knife?, Have you ever punched or beaten another person?,
and have you ever had to use your fists, a knife or some other weapon to
defend yourself from another person? Attitudes towards violence were defined
by three questions based on hypothetical scenarios. They all ask the respondents
if there are any situations in which they would approve of: a teenage boy
punching or beating another teenage boy, a public school teacher hitting a
student, or a judge sentencing a person to death? Her results suggested that
values play little part or no role at all as determinants of interpersonal violence
in a national representative sample of adult male Americans. Therefore, her data
did not support the subculture of violence theory of violent behavior. (Ball-
Rokeach 1973)
Marion Borg wrote another article that examined this theory. Instead of
looking at attitudes towards defensive violence, she examined attitudes toward
capital punishment. In her study she used data from the 1990 GSS to assess the
argument that a southern subculture of punitiveness exists. Her analysis
included whites respondents only because support for capital punishment differs
significantly between Whites and Blacks. Punitiveness refers to the degree of
punishment judged appropriate for individuals convicted of murder and is
18


indicated by the respondents viewpoints on capital punishment. The first
independent variable was native southerners. They were defined as those who
lived in the South at age 16 and also at the time of the interviews. In the 1990
GSS, the final sample used consisted of 1,074 respondents. There were 217
native southerners and 857 nonsouthemers. Another independent variable used
was racial intolerance. Racial intolerance refers to an aversive or negative
attitude toward African Americans. A racial antipathy scale and a racial
stereotyping scale gathered measures of intolerance. Another independent
variable was membership in a fundamentalist church, which was operationalized
according to the classification scheme suggested by Roof and McKinney. (1987)
Finally, the author controlled for sociodemographic characteristics previously
linked to support for capital punishment.
In her analysis Borg found two conclusions. First, level of support for
the death penalty does not seem to differ significantly between southerners and
nonsouthemers. The vast majority of both subgroups support capital punishment
for those convicted of murder. Second, she found that support for capital
punishment is influenced by the interaction of regional socialization with
religious fundamentalism and racial intolerance. Political conservatism was also
significantly related to support for capital punishment. (Borg 1997)
19


While some research has illustrated support for this theory, there have
been numerous studies that have found little to no evidence of a subculture of
violence. Because of these conflicting results I feel its necessary to test this
theory again. By conducting further research we can learn more about the
theory as well as the reliability and validity of past studies.
20


CHAPTER 3
THEORETICAL CONSIDERATIONS
When exploring causes of crime, scholars have examined several
different perspectives. However, three prevalent perspectives exist when an
application is made to violent crime. One such perspective is often referred to
as Opportunity Structure theories. Originally formulated by Robert Merton in
1938, these theories look at the material and social structures that shape the
values and activities of different social groups. Since legitimate opportunities
for wealth are unequal amongst all groups, some are forced to use illegitimate
means to acquire them. When applied to violent crime, this theory claims that
since some people are unable to achieve their means through societies legitimate
means, they often see crime as their only way of achievement. Merton describes
this theory in detail in his famous article, Social Structure and Anomie. He
begins by describing the two important elements of social and cultural structure.
The first consists of culturally defined goals, purposes, and interests. Some of
these cultural aspirations are related to the original drives of man; however, they
are not determined by them. These are referred to as cultural goals. The second
aspect of the social structure defines, regulates, and controls the acceptable
21


modes of achieving these goals. These are referred to as institutionalized means.
Equilibrium must exist between these two aspects in order for society to
maintain a normative function. If a person believed that they were obeying the
institutional norms, then balance would occur. Merton also claims that it is
important that for all social classes, the culturally desired goals be achievable by
legitimate means. This means that if goals arent achievable equally, than
illegitimate means might be used. Many times there is a greater emphasis on the
goals in compared to the means. This causes some people to achieve their goals
by any means necessary. Merton believes that this leads people to crime. Its an
over emphasis on the goals when compared to the means. Merton illustrates his
point by giving the example of a football team winning a game. He claims that
the point of competitive athletics is to win at all costs as opposed to winning
through circumscribed modes of activity.(Merton 1938) I feel that the notion
of achieving ones goals at all cost creates an environment conducive to violent
crime. The all or nothing attitude can provide reassurance that someone had
no choice but to commit a violent crime in order to achieve their desired goals.
Merton then developed five possible reactions to such a disparity. He
called the first reaction, Conformity. This describes those who accept both the
goals and the means of the society. Conformity is the most common reaction.
The second reaction is Innovation. This describes those who accept the cultural
22


goals but not the institutionalized means. These people will often use deviant
means to achieve their goals. The third type of reaction is Ritualism. This
describes those who accept the institutionalized means while rejecting the
cultural goals. The fourth type of reaction is Retreatism. This describes those
who reject both the cultural goals and the institutionalized means. These people
are a part of society; however, they do not participate in any functioning
methods. The fifth type of reaction is Rebellion. This describes those who are
completely against the current social order because of frustration. Then in 1948,
Merton revised his theory in which he gave a more complex definition of
cultural goals, institutionalized means, and the five reactions. (Merton 1938)
Before continuing on to the next perspective it is important to note that
there have been significant developments in strain theory most notably by
Robert Agnew. Agnew claimed that one of the reasons strain theory failed to
receive much empirical support was because researchers were limited in their
conception of what constitutes strain. He expanded the concept of strain to
include the disjunction between expectations and actual achievements. He also
claimed that strain may develop through a disjunction between just and fair
outcomes. (Curran and Renzetti 1994)
In his article, A Revised Strain Theory of Delinquency, Agnew points
out another major source of frustration and delinquency, the blockage of pain-
23


avoidance. He claimed that if adolescents are forced to stay in painful
environments there is little they can do legally to escape. This blockage of pain-
avoidance can lead to frustration, which can often lead to delinquent behavior.
In this article Agnew tested this expansion of strain theory using a national
sample of adolescent boys. His results indicated that location in aversive
environments has a direct effect on delinquency and an indirect effect through
anger. He also found that these results held true when controlling for social
control and subculture deviance variables. (Agnew 1985)
Another major perspective that is often applied to violent crime is Social
Disorganization. In addition to the other theories, this perspective also takes
into account breakdowns of community social institutions that often result from
social change. In organized neighborhoods, social institutions often work
together to limit crime, especially amongst local youths. If these institutions are
disrupted, social control is diminished and crime may increase. Besides social
control, institutions also provide access to the material and social resources
needed for physical survival.
This perspective dates all the way back to 1920 when William I. Thomas
and Florian Znaniecki wrote about the many social changes that affected Polish
immigrants a century ago. (Lee, Martinez Jr. and Rosenfeld 2001) Since then,
the theory has most often been applied to urban crime and deviance. McKay
24


and Shaw plotted out the residential location of youths that were referred to
juvenile court from different areas of the city. They found several consistent
patterns. The rates of delinquency in the lower-class neighborhoods were higher
in the inner city and decreased outwardly. (Akers 1994) Another significant
finding in this study was that Shaw and McKay failed to find any particular
racial groups as dominant. While racial and ethnic groups changed in the areas
of the years, the rates of delinquency stayed consistent. Therefore, Shaw and
McKay reported that the neighborhoods with the worst delinquency problems
also had the highest rates of other serious problems. These residents were also
the most economically disadvantaged in the city. Shaw and McKay believed
that these problems werent caused simply by economic inequality but instead
they emphasized the impact of social disorganization. These neighborhoods
were often in transition. They were often invaded by immigrant and migrant
groups, which resulted in the fleeing of the current residents. The new groups
lacked resources to live in better areas of the city and they also had to face the
daunting task of adjusting to a new society. At about the same time, business
and industry continued to encroach these areas causing further deterioration.
Soon the instability of these neighborhoods caused traditional social controls to
break down which allowed immoral values to set in. Therefore the solution to
25


the crime problem lies not with the offenders rather with the social controls of
their neighborhood. (Curran and Renzetti 1994)
The final theory Im going to examine is the theory that Ill be testing in
this thesis, the Southern Culture of Violence theory. This theory derived from
past culture theories that claim some subcultures are more tolerant towards
violence used to protect honor, courage, and manliness. Furthermore, Wolfgang
and Ferracuti claimed that individuals socialized in the South learn to approve
violence in a wide range of situations and feel violence is important in
enhancing their honor or reputation. (Wolfgang & Ferracuti 1967) More
recently, Reed advanced two basic concepts. First, Southerners do not approve
of violence in all situations. Instead, only in certain culturally defined situations
is violence accepted. For example, it would be accepted under situations when
ones honor is questioned. Reeds second concept is the understanding of these
situations is shared amongst socialized southerners. These situations are not
known amongst marginal citizens who dont care what mainstream southern
society expects of them. (Reed 1982)
In my study, I will simply test whether a subculture of violence exists in
the South. In order to apply this theory to explain high national homicide rates,
its imperative to know whether or not the theory can be applied to the Southern
United States. If violence in the Southern United States can be explained by this
26


theory, we will have a better understanding as to why the homicide rate of the
United States is so high. Once the theory is supported then it can be applied to
other problems. In a similar study that was reviewed earlier, Ellison found
distinct evidence that southerners were more inclined to condone defensive or
retaliatory violence. The methods of my study closely mirror his study
therefore, by completing this research I hope to give support for the subculture
of violence theory.
Like I stated in the introduction, cultural theories have a long history.
They can successfully be dated back to 1833 when Andre-Michel Guerry
studied the high rates of violence in Southern France. He believed the regional
variations were due to cultural differences left over from tribal settlement
patterns. A generation later, concerns grew in Italy as to what to do about the
large amounts of homicides occurring in the southern region. With the large
presence of barbarians, homicide rates were reported as high as 1000 homicides
a year. Most scholars claimed there was no need to study this problem. They
believed that the high rate was due to the corruption of the Neapolitans,
Calabrians, and Sicilians. Still others blamed social-structure problems like
poverty and social inequality. Then in 1980, Redfield showed the world that
southern homicide was also a problem in the United States. Soon more patterns
27


of cultural influence were discovered and a culture theory was applied to
violence.
In 1986, Ann Swidler developed one of the most persuasive arguments
on how culture influences action. She claims that culture is a collection of
resources that actors use to shape their strategies of action. This and other
cultural explanations try to explain why different people make different choices
in similar situations. In her argument, culture provides people with ideas,
definitions of types of situations, material products, and other factors that may
be combined to develop actions. When people are put in similar situations, they
will act differently because they have different contents in their tool kits. She
calls this concept culture as a tool kit. She continues by claiming that there
are at least two paths by which cultural differences could affect levels of
violence. The first path is knowledge of weapons. Obviously, the greater
knowledge of weapons and how to use them could increase the likelihood of
violence. The second path is that culture provides ways of organizing sensory
experiences and identifying situations. Because of a persons culture, they may
define a particular situation as one in which a physical assault is appropriate or
even demanded. (Smith & Zahn 1999)
28


CHAPTER 4
DATA AND METHODOLOGY
In this study I will be testing two hypotheses. My first hypothesis is that
the respondents in the South region will have more tolerant attitudes towards
capital punishment than those in the nonsouth region. My second hypothesis is
that the respondents in the South region will have more tolerant attitudes
towards defensive violence than those in the nonsouth region. After analyzing
the data I expect to find the respondents in the South will be more tolerant of
both capital punishment and defensive violence.
Since many scholars have argued that these and all hypotheses are
predictors of what we believe will happen, either one or both of our hypotheses
will be supported or maybe none at all. Therefore there are three possible
outcomes anticipated. The first scenario is that both of the hypotheses tested are
supported by the results. The second scenario is that neither of the hypotheses
tested will be supported by the results. The third scenario is that one of the
hypotheses tested will be supported while the other will not be supported. It is
noteworthy that all hypotheses testing using various samples may or may not be
representative of the general public. It is important to understand that if these
29


hypotheses are supported it does not mean that the theory is completely valid.
The same is true if these hypotheses are refuted. There are many different
factors that can influence hypothesis-test outcomes. Factors such as theory,
methodology, and a representative sample all can have effects on outcomes.
Even using the wrong statistical test can affect the outcomes. In other words,
its important to not attach too much importance to any single study. Instead,
its important to compare how these results compare with the findings of similar
studies. (Champion 1994)
For the purpose of this study I have chosen to use General Social Survey
data of a secondary source. Using secondary data has advantages as well as
disadvantages that I feel are worth mentioning. Perhaps the most important
advantage is that they save time and money because ready-made data is already
available for the public. This allows the researcher to save valuable time and
money while conducting their analysis. Another important advantage is that the
information contained by secondary sources often contains information that
pertains to a national population. Therefore, generalizations to larger
populations have greater legitimacy than those based on relatively small samples
of elements. Another important advantage of secondary sources is the
information may be standardized overtime. This means that the data can be
examined at different periods of time with confidence that the results will reflect
30


what the people were thinking or doing at that time. And a final advantage is
that researchers avoid potential ethical problems and harm to human subjects by
studying documents directly instead of people. (Champion 1994)
The use of secondary data is not without its disadvantages. Perhaps most
importantly, this data was collected for other purposes and may only incidentally
be related to the researchers goals and interests. In other words, the data is
being used in a way other than its original intent. Another important
disadvantage is that researchers have no way of reconstructing missing data.
Non-responses may be an important limitation that can affect the reliability of
the data. A final disadvantage is researchers must often speculate the meanings
of certain phrases used by the original researchers. They dont have the
opportunity to obtain further clarifications. (Champion 1993)
The data set I chose is from the General Social Survey. The General
Social Survey is a personal interview survey of United States households
conducted by the National Opinion Research Center (NORC) in Chicago,
Illinois. The principal investigators are James A Davis, Tom W. Smith, and
Peter V. Marsden. These investigations started in 1972 and since then there
have been more than 38,000 respondents who have answered more than 3,260
different questions. The General Social Survey is known for its broad coverage,
use of replication, its cross-national perspective, and its attention to data
31


quality. It is also known as one of the only social science data sets designed to
be analyzed by users instead of the principal investigators and project staff. The
mission statement of the GSS is, to make timely, high-quality, scientifically
relevant data available to the social science research community. This survey
is very general and covers many topics from views of abortion to astrological
signs. The GSS also follows the highest standards in design, sampling,
interviewing, processing, and documentation. Items are designed by leading
specialists in their field and then pretested, full-probability sampling is used, a
high response rate is obtained, and many data quality checks from validation to
verification are employed. These studies also stress replication through design.
Over 1,000 questions have been replicated and some of them date back all the
way to 1972. (GSS Website, August 21, 2001)
In my study, I will be analyzing the attitudes and opinions of people in
the United States in 1998. I decided to use the data set from 1998 because it was
a more recent dataset that contained similar items to those used by Borg. This
survey contains attitudinal items on a respondents approval to the use of capital
punishment. More specifically, it asks the respondents whether or not they
approve of the use of capital punishment for someone convicted of murder.
Borg used the same variable for her test however she used the 1990 dataset. In
her study Borg analyzed White respondents only because support for capital
32


punishment differs significantly between Whites and Blacks. Not only does
their tolerance vary, but the attitudinal factors that prompt Blacks opinions
regarding capital punishment vary considerably compared from those affecting
Whites. Therefore I will also only include Whites in my analysis.
I will be testing four independent variables and three dependent variables
in this study. The first independent variable I will be testing is region and its
operationalized as the region in which the individual lived in while they were
16. This particular age was chosen because I feel it represents a time in an
adolescents life where the socialization process is at its greatest. I feel that
most attitudes have developed by this age, therefore if the respondent has since
relocated to a different region it will have a smaller effect on my analysis.
Wolfgang and Ferracuti claimed in their definition of the Southern Subculture of
Violence theory that individuals socialized in the South learn to approve
violence in a wide range of situations and feel violence is important in
enhancing their honor or reputation. (Wolfgang & Ferracuti 1967) This is
especially noteworthy because they state that the theory is applies to those
socialized in the South. Therefore we must examine those who were socialized
in the South or in other words, those who lived there when they were 16.
In the data set, the respondents are asked which region they lived in
when they were 16 years old. They were given several different regions to
33


chose from. I will recode all of these regions into two regions. The first region
will be called the South Region and it will consist of the South Atlantic region,
East South Central region, and the West South Central region. The second
region will be called the Nonsouth region and it will consist of the New England
Region, Middle Atlantic Region, the East North Central Region, the West North
Central Region, the Mountain Region, and the Pacific region. The final foreign
region was disregarded because it represents respondents not from the United
States. This variable was recoded as a dummy variable in the following
manner: 1 = South and 0 = Nonsouth.
The second and third independent variables that I will be using for my
analysis describe racial intolerance. Since the exact measures of Borgs study
were not available with the 1998 data, I have decided to use different measures
that closely resemble those used in Borgs. In her study Borg used two
questions that asked respondents whether or not they favored or opposed living
in a neighborhood where half of your neighbors are Black and having a close
relative or family member marry a Black person. The first indicator of racial
intolerance that I am using examines the respondents attitudes towards race and
marriage. More specifically, it asks the respondents if they think there should
be a laws against marriages between African Americans/Negroes/Blacks and
whites? I feel this question approximates Borgs measures of racial
34


intolerance. The variable was coded in the following manner: A 1 response =
Yes and a 2 response = No. The no answer response were discarded. The next
measure of racial intolerance that Im using asks the respondents about the racial
make up of their neighborhood. Specifically, it asks, Are there any African
Americans/Negroes/Blacks that live in your neighborhood. Again I feel that this
variable will be a good indicator of racial intolerance. The variable was coded in
the following manner: A 1 response = Yes and a 2 response = No. Again a no
answer response was discarded.
I chose to include measures of racial intolerance for the following
reasons. First we know from past analyses that stronger attitudes of racial
intolerance exist among White southerners when compared to those of Whites
who were bom and live elsewhere. (Ellison 1991) For example Barkan and
Cohn (1994) examined support for death penalty among respondents in the 1990
GSS. They included an index of racial prejudice as well as an index of racial
stereotyping. They found both types of intolerance were significantly linked to
greater support for the death penalty even while controlling for necessary
demographic and attitudinal variables. (Barkan and Cohn 1994) Second, since
racial intolerance has been widely recognized as a dominant feature of the
southern subculture, it should be relevant in accounting for southern attitudes
towards capital punishment. Third, I have included a crosstab analysis of these
35


variables that indicated a statistically significant relationship. Therefore, I feel
that it is beneficial to include these measures in my model.
The fourth and final independent variable is membership in a
fundamental church. Fundamentalist respondents refer to the members of the
following groups: Southern Baptist, Independent Baptist, Evangelical Church of
God in Christ, Brethren, Church of Christ (except United Church of Christ),
Churches of God, Church of God in Christ, Evangelical, Evangelical Reformed,
Full Gospel, Foursquare Gospel, Holiness, Nazarene, Free Methodist, Missouri
Synod Lutheran, Pentecostal, Sanctified/Sanctification, Seventh Day Adventist,
and associated smaller evangelical and fundamentalist groups. Catholic, Jews,
those of other religions and those of fundamentalist Protestant denominations
make up the nonfundamentalist category. The respondents were asked the
fundamentalism/liberalism of their religion. This variable was recoded as a
dummy variable in the following manner: 1= Fundamentalist and 0 =
Nonfundamentalist.
I elected to include fundamentalism for a couple reasons. To start, past
research has illustrated a significant relationship between fundamentalism and
capital punishment. For example, Young (1992) found that fundamentalism
significantly affects greater levels of support for the death penalty. The
relationship reflects several interrelated beliefs among fundamentalists that
36


originate in their literal understanding of the bible. In addition, like the racial
intolerance measure it has been widely recognized as a dominant feature of the
southern subculture, therefore it should be relevant in accounting for southern
attitudes towards capital punishment. (Borg 1997) I have included a crosstab
examination of this variable as well that illustrates this significant relationship.
As this chapter continues I will switch my focus to the dependent
variables in my study. The first dependent variable that I will be using is
Punitiveness. Punitiveness is defined as the degree of punishment judged
appropriate for individuals convicted of murder and is indicated by the
respondents views on capital punishment. The respondents were asked, Do
you favor or oppose the death penalty for those convicted of murder? The
variable was coded in the following manner: A 1 response = Yes and a 2
response = No. Again a no answer response was discarded.
Punitiveness was included in this analysis because I feel it does an
excellent job describing a scenario that fits in the Southern Subculture of
Violence. As early as 1967 researchers examining this theory have stated that
the tolerance towards violence doesnt exist towards all violence. Instead, the
tolerance only exists towards specific situations. For example Ellison (1991)
demonstrated that southerners seemed more incline to condone defensive or
retaliatory forms of violence. Reed (1982) agreed and referred to this tendency
37


as violence for a cause. One context in which southerners seem likely to
advocate violence for a cause is punishing those convicted of breaking the law.
In addition, execution rates in the South also illustrate this notion. Between
1976 (the year capital punishment was reinstated) and 1995, the South executed
263 criminals. That trumps the 50 executions that occurred in that same time
period in the other regions of the US. Although differences in capital crime
rates between these regions may have played a role, its difficult to ignore this
difference. (Borg 1997)
The second an third dependent variables that Im going to test are
indicators of defensive violence. The second variable is police attack and it asks
the respondents whether or not they approved of an officer hitting a citizen who
was trying to attack him. The third variable is police abuse and it asks the
respondents whether or not they approved of an officer hitting a citizen who
was using obscenities towards the officer. Both of these variables ask the
respondents their attitudes towards violence used by a police officer in a
defensive manner. Since the Southern Subculture of Violence Theory clearly
states that the tolerance towards violence only exists if the violence is defensive
or for a cause, I felt that these variable should be tested as well. Both of these
variables were recoded in the following manner: 1 = Yes and 2 = No.
38


Finally Im going to define the other control variables that I will be
including in my analysis. These socioeconomic characteristics will allow me to
see which independent variable is associated with my dependent variable the
most. All of these variables have been previously linked to support for capital
punishment. (Borg 1997) First I will include age measured in years. Next I
will include education measured by the respondents degree. The respondents
were asked whether or not they had a high school degree or less or more than a
high school degree. This variable was coded 1 = less than a high school degree
and 2 = more than a high school degree. I will also include family income,
which was measured by the respondents total family income. The respondents
were asked what their total family income was and given the following
responses to chose from: Under $1000, $1000 to $2999, $3000 to $3999, $4000
to $4999, $5000 to $5999, $6000 to $6999, $7000 to $7999, $8000 to $8999,
$9000, to $9999, $10000 to $14999, $15000 to $19999, $20000 to $24999, and
$25000 or over. The next control variable is sex. This variable is measured by
the respondents gender. A 1 = Male and 2 = Female. I also included a measure
for urban residency. The SRC Belt variable places the respondents into
different urban or rural categories. This variable was recoded the following way:
Central city of 12 largest SMSAs, Central city of remainder of the 100 largest
SMSAs, Suburbs of 12 largest SMSAs, Suburbs of the remaining 100 largest
39


SMSAs, and Other Urban (counties having towns of 10,000 or more), were all
collapsed into one variable called Urban. Other rural (counties having no
towns of 10,000 or more) were left the same as Rural. This variable was
recoded as a dummy variable. A 1 = Urban and 0 = Rural. The next control
variable is political conservatism. Respondents were asked whether they
considered themselves extremely liberal, liberal, slightly liberal, moderate or
middle of the road, slightly conservative, conservative or extremely
conservative. For my analysis, all of the conservative answers were collapsed
into one response called conservative. All of the liberal responses as well as
the middle of the road response were collapsed into one response called
nonconservative. This variable was recoded as a dummy variable as well. A
1 = Conservative and 0 = Nonconservative.
I wish to conclude this chapter by describing how Im going to analyze
my data. To begin, I will run crosstabs of the independent variables by the
capital punishment variable. These crosstabs will include the chi-square
measurement of significance as well as the Phi measure of significance. As my
analysis continues I will include three logistic regression analyses. The first will
test all of the independent variables and control variables against the dependent
variable capital punishment. The second will test the same independents and
controls however this one will contain the dependent variable police attack. The
40


final regression will again contain the same independent and controls however
this one will contain the dependent variable police abuse. For the purposes of
this study, I will be using a significance level of .05; therefore, any numbers
smaller than .05 can be considered statistically significant. For the Phi
measures, I will use the value of .10 as my cutoff between a strong and weak
relationship. Therefore, any relationship with a Phi value of. 10 or higher will
be considered a strong relationship. For my analysis purposes I have only
included White respondents only because support for capital punishment differs
significantly between Whites and Blacks. Not only do their endorsement levels
vary but the attitudinal factors that prompt these differences vary as well.
Bivariate analyses were also run to ensure that multicolinearity was not a factor.
41


CHAPTER 5
RESULTS
I would like to open my analysis by first examining the crosstabs that I
describe earlier in this thesis. The first table that Im going to look at examines
approval for capital punishment by regions and can be found below. Of the
respondents in the South region, 80.2% said that they approved the use of capital
punishment. In the nonsouth region, only 71.2% approved. Chi-square value
was 14.3 with a significance of .000. This means that there is a 0% chance that
the findings are due to coincidence and therefore they are considered statistically
significant. However, the Phi value was .09 with a significance of .000. Since
the Phi value is less than 10 it is considered to be a weak relationship.
Table 5.1
Approval of Capital Punishment by Region
(Reported in Percentages)
Capital Region Total
Punishment Nonsouth South
Favor 71.2% 80.2% 78.3%
Oppose 28.8% 19.8% 21.7%
Total 382 1377 1759
Chi Square 14.3 P =.000, Phi .09 P =.000
42


Next Im going to examine how attitudes towards capital punishment
vary amongst conservatives, liberals, and those in the middle. In table 5.2
below, 82.5% of conservatives approved of capital punishment while 82.2% of
those in the middle and 67.3% of liberals approved. Chi-square value was 52.3
with a significance of .000. Again this means that these differences are
statistically significant. The Phi value was .162 with a significance of .000.
This means that the relationship is significant and can be considered moderate.
Table 5.2
Approval of Capital Punishment by Political View
(Reported in Percentages)
Capital Punishment? Liberal Political View Middle Conservative Total
Favor 67.3% 82.2% 82.5% 78.2%
Oppose 32.7% 17.8% 17.5% 21.8%
Total 545 701 748 1994
Chi Square 52.3 P =.000, Phi .162 P =.000
Im now going to look at how approval for capital punishment varies
amongst fundamentalist and nonfundamentalist. In table 5.3 below, 81.8% of
fundamentalist approved of capital punishment while only 77.3% of
nonfundamentalist approved. The chi-square value was 4.6 with a significance
43


of .032. This means that these differences are statistically significant. While
these results are statistically significant, the relationship is considered weak.
Table 5.3
Approval of Capital Punishment by
Fundamentalist
(Reported in Percentages)
Fundamentalist Total
Capital Punishment Fundamentalist Nonfundamentalist
Favor 81.8% 77.3% 78.5%
Oppose 18.2% 22.7% 21.5%
Total 528 1411 1939
Chi Square 4.6 P = 032, Phi .049 P=.032
Im now going to turn our attention to how approval for capital
punishment varies amongst those who support laws against interracial marriage
and those who dont support laws against interracial marriages. Of the
respondents who were in favor of laws against interracial marriage, 77.0% said
that they were in favor of capital punishment. Of the respondent who werent in
favor of laws against interracial marriages, 72.9% said that they were in favor of
capital punishment. The chi-square value was 23.9 with a significance of .000.
The Phi value was 127 with a significance of .000. This indicates a moderate
relationship that is statistically significant.
44


Table 5.4
Approval of Capital Punishment by
In favor of Laws Against Interracial Marriages
(reported in Percentages)
Capital Favor Law against Interracial Marriage Yes NO Total
Punishment Favor 77.0% 72.9% 73.1
Oppose 12.8% 20.4% % 19.2
Total 187 1257 % 1486
Chi Square 23.9 P = .000, Phi .127 P =.000
Next I will examine approval for capital punishment with whether or not
African Americans live in the respondents neighborhood. Table 5.5 below,
illustrates that 82.8% of the respondents who said that there werent any African
Americans in their neighborhood approved of capital punishment. For the
respondents who did have African Americans living in their neighborhood,
75.5% said that they approved of capital punishment. The chi-square value was
14.1 with a significance of .000. The phi value was -.085 with a significance of
.000. This means that these results are considered statistically significant
however, the relationship is considered weak.
45


Table 5.5
Approval of Capital Punishment by
Are their African Americans Living in Your Neighborhood
(results shown in percentages)
Are There African Total
Americans in Your
Neighborhood?
Capital PunishmentYes No
Favor 75.5% 82.8 78.2%
%
Oppose 24.5% 17.2 21.8%
%
Total 1230 716 1946
Chi Square 14.1 P =.000, Phi -.085 P =.000
As my analysis continues I wish to now examine the logistic regressions.
The first regression examines the effect of the independent variables
simultaneously on capital punishment. To begin analyzing this regression, I
would like to first look at the Model Chi-Square. Also known as the goodness
of fit, the Model Chi-Square functions as a significance test similar to the F test
of the OLS regression model. Model Chi-Square provides the usual significance
test for a logistic model. It tests the null hypothesis that none of the independent
variables are linearly related to the log odds of the dependent. In other words,
the Model Chi-Square tests the null hypothesis that all population logistic
regression coefficients except the constant are zero. Therefore, it is an overall
model test that does not assure that every independent is significant. (NCSU,
46


November 19, 2002) In this regression the Model Chi-Square value is 49.050
with a significance of .000. This indicates that the overall model is significant at
the .05 level. As the analysis continues I will now focus on the individual
independent variables. As table 5.6 indicates below this paragraph, the
neighborhood race variable was strongest associated with attitudes towards
capital punishment. The Wald value 13.961 with a significance of .000. The B
value was -.802. This means that those without African Americans living in their
neighborhood are more likely to approve of capital punishment. This variable
was followed by the church attendance variable. This variable asked
respondents how often they attend religious services. The Wald value for this
variable was 7.998 with a significance of .005. The B value for this variable
was .098. This means that the more often the respondents attended church, the
more likely they are to oppose of capital punishment. Gender was the next
strongest variable with a Wald value of 5.852 and a significance of .016. The B
value was .444. South was the next strongest associated with capital
punishment. The Wald value was 5.468 with a significance of .019. TheB
value for this variable was -.488. This means that those in the South were more
likely to approve of capital punishment then those in the nonsouth. The final
statistically significant variable was conservative. The Wald value was 4.405
47


with a significance of .036. The B value for this variable was .407 indicating a
positive relationship.
I also conducted a stepwise analysis that included all of the significant
variables in this model in order of strength. The strongest variable racial
neighborhood was entered first with a B value of -.446 and a Wald value of
13.983 with a significance of .000. Both the B and the Wald value continued to
grow with each step until finally they reached -.547 and 16.118 respectfully.
The Model of Chi-Square also grew with each step. It began at 14.95 with a
significance of .000 and by the final step it was 63.840 with a significance of
.000. Therefore, the Model Chi-Square improved with each step.
48


Table 5.6
Capital Punishment Logistic Regression
Variables in the Equation
B S.E. Wald Sig. Exp(B)
Age .003 .006 .297 .586 1.003
Church Attendance .098 .035 7.998 .005 1.103
Conservative .407 .194 4.405 .036 1.502
Education -.164 .205 .636 .425 .849
Fundamentalist .311 .216 2.082 .149 1.365
Racial Neighborhood -.802 .215 13.961 .000 .448
Interracial Marriage .453 .314 2.085 .149 1.573
Income -.069 .038 3.338 .068 .933
Gender .444 .184 5.852 .016 1.559
South -.488 .209 5.468 .019 .614
Urban .126 .201 .394 .530 1.134
Constant -1.659 .924 3.221 .073 .190
Model of Chi Square = 49.050, P = .000
Next I would like to examine the results of the second logistic
regression. In this regression Ive included the same independent and dependent
variables as before however this time I have substituted attitudes towards capital
punishment with the second dependent variable: police attack. This variable
asked respondents whether or not they approved of a police officer hitting a
citizen who was attacking him. I would like to begin my analysis again by first
examining the Model of Chi-Square. Again the Model of Chi-square tests the
null hypothesis that all population logistic regression coefficients except the
constant are zero. For this regression, the Model of Chi-Square value was
49


18.765 with a significance of .065. This indicates that the overall model was not
considered statistically significant. As my analysis continues I would now like
to focus on the independent variables. As table 5.7 indicates the only variable
that was statistically associated with the dependent variable was the South
variable with a Wald value of 10.218 and a significance of .001. TheB value
was -1.281. This indicates that those in the South were more likely to approve
of capital punishment than those in the nonsouth.
Table 5.7
Police Attack Logistic Regression
Variables in the Equation
Age -.003 .011 .082 .775 .997
Church Attendance .030 .074 .160 .689 1.030
Conservative .082 .397 .043 .836 1.085
Education .091 .438 .043 .836 1.095
Fundamentalist .473 .493 .920 .337 1.604
Racial Neighborhood -.009 .411 .000 .983 .991
Interracial Marriage -.340 .595 .326 .568 .712
Income -.069 .080 .760 .383 .933
Sex .729 .397 3.374 .066 2.074
South -1.281 .401 10.218 .001 .278
Urban -.439 .484 .822 .365 .645
Constant -1.870 1.808 1.069 .301 .154
Model Chi-Square 18.76, P=.065
As my analysis continues I would now like to examine the final
regression. In this regression I have included the same independent variables
and the same control variables however, this time I have used my third
50


dependent variable: police abuse. This variable asked respondents whether or
not they approved of a police officer hitting a citizen who was using obscenities
at the officer. Again I will begin with the Model of Chi-Square. The Model of
Chi-Square was significant at 26.858 with a significance level of .005. I would
now like to examine the independent variables. As table 5.8 indicates below, the
interracial marriage variable was the most association. The Wald value for this
variable was 9.514 with a significance of .002. The B value for this variable
was 1.476 indicating a positive relationship. The next variable that was
associated with the dependent variable was the respondents total family
income. The Wald value was 4.650 with a significance of .031. The B value
was .154 indicating another positive relationship.
I conducted a stepwise analysis for this model as well. Again I included
both of the significant variables in order of their strength. Interracial marriage
was the strongest variable so it was the first variable entered. When entered
alone, the B value was 1.562 and the Wald value was 28.816 with a significance
of .000. The Model Chi-Square was 24.411 with a significance of .000. Next,
income was entered into the model. The B value for interracial marriage went
down to 1.542 and the Wald value went down to 21.425 with a significance of
.000. The Model of Chi-Square also went down to 23.057 with a significance of
.000.
51


Table 5.8
Police Abuse Logistic Regression
Variables in the Equation
B S.E. Wald Sig. Exp(B)
Age -.009 .012 .583 .445 .991
Church Attendance -.097 .081 1.441 .230 .907
Conservative -.099 .432 .053 .818 .905
Education .235 .515 .209 .648 1.265
Fundamentalist -.669 .503 1.767 .184 .512
Racial Neighborhood -.190 .430 .196 .658 .827
Interracial Marriage 1.476 .479 9.514 .002 4.377
Income .154 .071 4.650 .031 1.166
Sex .641 .417 2.360 .124 1.899
South .221 .501 .195 .659 1.247
Urban -.463 .429 1.163 .281 .629
Constant -1.112 1.775 .393 .531 .329
Model of Chi Square = 26.858, P =.005
52


CHAPTER 6
DISCUSSION
In this final chapter I wish to discuss my results and their implications. I
will begin with the first logistic regression. This regression analyzed all of the
independent variables by the dependent variable capital punishment. In this
regression we found that the neighborhood race variable was the strongest
associated with capital punishment. This racial intolerance indicator asked
respondents whether or not they had members of the opposite race living in their
neighborhood. Again this finding is consistent with the past research that has
illustrated several patterns suggesting that racial prejudice was a relevant factor
in explaining support for the death penalty. Like I stated in chapter 4, Barkan
and Cohn (1994) examined support for the death penalty among respondents in
the 1990 GSS. They included an index of racial prejudice as well as an index of
racial stereotyping. They found both types of intolerance were significantly
linked to greater support for the death penalty even while controlling for
necessary demographic and attitudinal variables. (Barkan and Cohn 1994)
Since this analysis as well as others have found racial intolerance to be
significantly associated with attitudes towards capital punishment I believe that
53


future studies that examine attitudes towards capital punishment should include
racial intolerance measures. The South variable was statistically significant as
well meaning that those in the South approved of capital punishment more than
those in the nonsouth. However, since other variables were stronger only partial
support can be argued in this regression.
I would now wish to turn our attention to the second logistic regression.
Again this regression examined all the same independent and control variables
with a different dependent variable: police attack. The police attack variable
was the first indicator of defensive violence and it asked respondents whether or
not they approved of a police officer hitting a citizen that was attacking him.
My analysis illustrates that South variable was the strongest associated with this
dependent variable. This illustrates that those from the South are more likely to
approve of police violence in this situation. However, the Model of Chi-Square
was not statistically significant. Therefore, this regression only illustrates partial
support for the Southern Subculture of Violence Theory.
The final regression included in my analysis was of the dependent
variable police abuse, which was the other indicator of defensive violence. This
dependent variable asked the respondents whether or not they approved of a
police officer hitting a citizen who was using obscenities towards him. The
regression illustrates that the interracial marriage variable was the strongest
54


associated with this dependent variable. Once again past research has illustrated
a relationship between racial intolerance and support for capital punishment.
(Ellison 1991; Barkan and Cohn 1994) This finding illustrates that the
relationship between racial intolerance and capital punishment may be extended
to attitudes towards defensive violence as well.
As my discussion continues I wish to revisit my two hypotheses. My
first hypothesis stated that the respondents in the South region would have more
tolerant attitudes towards capital punishment. When looking at my first
regression (table 5.6) we found that while region wasnt the strongest variable it
was statistically significant. Therefore we can argue that hypothesis one was
partially supported. Hypothesis two stated that the respondents from the South
region would be more tolerant towards violence used in a defensive manner.
When first looking at the second regression (table 5.7) we find that region was
the highest associated variable that was statistically significant. However, the
Model of Chi-Square was not statistically significant. Therefore, this model
illustrates partial support for this hypothesis. When looking at the final
regression we find that the South variable was not significantly associated with
the defensive violence measure. This model did not illustrate any support for
this theory. Therefore, again we can argue for partial support of hypothesis two.
The partial support found for this theory in this study illustrates that the South
55


has not homogenized as much as expected over the years. Similarities found in
this study along with past studies may illustrate that Southern attitudes towards
capital punishment and defensive violence have stayed consistent over the years.
I will conclude this thesis with a suggestion for future research. Tiiis
study as well as those in the past have oft relied on survey techniques.
Inconsistent and negative results maybe a result of limitations in this
methodology. Surveys test stated values, which may not always coincide with
what is actually guiding behavior. A possible solution to this would be to test
actual culture amongst regions. Violent themes expressed through art, music,
and other leisure activities may illustrate a subculture of violence that cant be
shown in surveys.
56


REFERENCES
Agnew, Robert (1985) A Revised Strain Theory of Delinquency. Social
Forces 64, 151-167.
Ball-Rokeach, Sandra. (1973) Values and Violence: A Test of the
Subculture of Violence Thesis American Sociological Review Vol 38, 736-
749.
Barkan, S and Steven Cohn. 1994. Racial Prejudice and Support for the
Death Penalty by Whites Journal of Research in Crime and Delinquency
Vol 31, 202-209.
Borg, Marian. (1997) The Southern Subculture of Punitiveness? Regional
Variation in Support for Capital Punishment. Journal of research in Crime
and Delinquency Vol 34, 25-46.
Bureau of Justice Statistics, http ://www.ojp.usdoj. gov/bi s Retrieved on
November 29,2001.
Champion, Dean J. (1994) Research Methods for Criminal Justice and
Criminology. Englewood Cliffs, New Jersey: Regents/Prentice Hall.
Curran, Daniel and Claire Renzetti. (1994) Theories of Crime. Needham
Heights, Ma: Allyn and Bacon.
Dixon & Lizotte. (1987) Gun Ownership and the Southern Subculture of
Violence. American Journal of Sociology Vol 93, 383-405.
Ellison, Christopher. (1991) Southern Culture and Firearm Ownership.
Social Science Quarterly Vol 72 267-283.
Ellison, Christopher. (1991) An Eye for an Eye? A Note on the Southern
Subculture of Violence Thesis Social Forces Vol 69, 1223-1239.
General Social Survey Website. Retrieved on August 23, 2001.
http://www.icpsr.umich.edu/GSS/.
57


Land, Kenneth, Patricia L. McCall, and Lawrence E. Cohen. (1990)
Structural Covariates of Homicide Rates: Are There Any Invariances across
Time and Social Space? American Journal of Sociology Vol 95, 922-63.
Lee, M., Martinez, R., & Rosenfeld, R. (2001) Does immigration increase
homicide? Negative evidence from three border cities. The Sociological
Quarterly 42: 559-623.
Loftin, Colin and Robert Hill. 1974. Regional Subculture and Homicide.
American Sociological Review Vol 39, 714-724
Merton, R. (1938) Social Structure and Anomie. American Sociological
Review, Vol 3, 672-682.
Messner, Steven. (1983) Regional and Racial Effects on the Urban
Homicide Rate. American Journal of Sociology. Vol 88, Issue 5. Mar.
997-1007.
North Carolina State University Website. Retrieved on November 20,
2002.http://www2.chass.ncsu.edu/garson/pa765/logi stic.htm
Nisbett & Cohen. (1996) Culture of Honor. Westview Press: Boulder, Co.
Parker, Robert Nash. (1989) Poverty, Subculture of Violence, and Type of
Homicide. Social Forces. Vol 67, 983-1007.
Reed, John. (1971) To Live and Die in Dixi. Political Science Quarterly.
Vol 86, 429-443.
Reed, John. (1982) One South: An Ethnic Approach to Regional Culture.
Louisiana State University Press: Louisiana.
Rosenberg, Morris (1962) Test Factor Standardization as a Method of
Interpretation. Social Forces Vol 41, 53-61.
Smith &Zahn. (1999) Homicide. Sage Publications: Thousand Oaks,
California.
U.S. Department of Justice. 1992. Soucebook of Criminal Justice Statistics
1992. Washington D.C.: U.S. Department of Justice
58


Wolfgang, M & Ferracuti, F. (1967) The Subculture of Violence. Tavistock
Young, Robert. (1985) Perceptions of Race, Racial Attitudes, and
Firearms Ownership. Social Forces Vol 64, 473-486.
59