AN INVESTIGATION OF THE SOUTHERN SUBCULTURE OF VIOLENCE: STRUCTURAL AND CULTURAL PREDICTORS OF AN EXPANDED CONCEPTUALIZATION OF VIOLENCE Except where reference is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory committee. This thesis does not include proprietary or classified information. __________________________ Andrea L. Weston Certificate of Approval: _________________________ __________________________ L. Connor Bailey Greg S. Weaver Professor Associate Professor Rural Sociology Criminology and Criminal Justice _________________________ __________________________ Thomas A. Petee Raj P. Mohan Professor Professor Criminology and Criminal Justice Sociology __________________________ Stephen L. McFarland Acting Dean Graduate School AN INVESTIGATION OF THE SOUTHERN SUBCULTURE OF VIOLENCE: STRUCTURAL AND CULTURAL PREDICTORS OF AN EXPANDED CONCEPTUALIZATION OF VIOLENCE Andrea L. Weston A Thesis Submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Master of Science Auburn, Alabama August 7, 2006 AN INVESTIGATION OF THE SOUTHERN SUBCULTURE OF VIOLENCE: STRUCTURAL AND CULTURAL PREDICTORS OF AN EXPANDED CONCEPTUALIZTION OF VIOLENCE Andrea L. Weston Permission is granted to Auburn University to make copies of this thesis at its discretion, upon request of individuals or institutions and at their expense. The author reserves all publication rights. ____________________________ Signature of Author ____________________________ Date of Graduation iii VITA Andrea Lynn (Williams) Weston, daughter of Dr. Roger Lea and Nina Lynn Williams, was born June 17, 1977, in Huntingdon, Pennsylvania. She moved to State College, Pennsylvania, and graduated from State College Area High School in 1995. She attended Penn State University for one year and then joined the United States Army in 1996 to serve for three years as a Parachute Rigger for the U.S. Army Airborne School in Ft. Benning, Georgia. After serving in the Army, she returned to college in 2000, to pursue her undergraduate education at Columbus State University in Columbus, Georgia, and graduated magna cum laude with a Bachelor of Science degree in Criminal Justice in January, 2004. She then entered graduate School at Auburn University as a Graduate Teaching Assistant in August, 2004. She has one daughter, Kaiya Natalie Weston, born on December 17, 1999. iv THESIS ABSTRACT AN INVESTIGATION OF THE SOUTHERN SUBCULTURE OF VIOLENCE: STRUCTURAL AND CULTURAL PREDICTORS OF AN EXPANDED CONCEPTUALIZATION OF VIOLENCE Andrea L. Weston Master of Science, August 7, 2006 (B.S., Columbus State University, 2004) 114 Typed Pages Directed by Gregory S. Weaver This research investigates the impact of both structural and cultural variables on regional variations in rates of homicide and aggravated assault. Particular attention is given to the influence of conservative Protestantism as a measure of Southern regional culture and to expanding the construct of violence beyond a narrow focus on homicide. Based on 2000 Census data, four separate one least squares regression models for 3,109 U.S. counties suggest that the influence of Southern culture on rates of homicide and aggravated assault is diminishing over time. This research provides statistical evidence that rates of violence in western regions or the U.S. are converging with, if not surpassing those of the South. The theoretical significance of the findings and implications for future research are discussed. v ACKNOWLEDGMENTS The author would like to thank Dr. Greg Weaver for his countless hours of guidance, instruction and support in putting together this thesis. I couldn?t have made it through this process without you and all the time you devoted to helping me, so THANK YOU Dr. Weaver!!. Thanks are also due to my family for their support and encouragement throughout the duration of this project. Also, a very special thanks to my brother, Dr. Nathan L. Williams, for his constant support, patience, and sheer genius in helping me to better understand the entire research process. This paper is dedicated to you, Nate, for always being there for me and believing in me when I never thought I?d make it this far. vi Style manual or journal used: American Sociological Association. Computer software used: SPSS 13.0 for Windows. vii TABLE OF CONTENTS LIST OF TABLES AND FIGURES ix INTRODUCTION 1 AN OVERVIEW OF CULTURAL/SUBCULTURAL THEORIES 2 THE SUBCULTURE OF VIOLENCE 5 THE SOUTHERN SUBCULTURE OF VIOLENCE 7 POTENIAL CAUSES OF SOUTHERN VIOLENCE 10 THE SOCIOHISTORICAL TRADITIONS OF THE SOUTH 10 THEORETICAL PERSPECTIVES 14 THE CULTURAL POSITION 14 THE STRUCTURAL POSITION 16 LITERATURE REVIEW 19 RELIGION 19 ASSAULT 27 CULTURAL INTERPRETATIONS 29 STRUCTURAL INTERPRETATIONS 43 THE INTEGRATED APPROACH 55 PRESENT STUDY 68 HYPOTHESIS 69 METHODOLOGY 70 DATA AND METHODS 70 DEPENDENT VARIABLES 71 INDEPENDENT VARIABLES 72 RESULTS 79 DISCUSSION 85 LIMITATIONS AND FUTURE RESEARCH 89 CONCLUSION 92 REFERENCES 95 APPENDICES 100 viii LIST OF TABLES AND FIGURES TABLE 1 Descriptive Statistics, US Counties, 2000 Census Data 101 TABLE 2 OLS Regression Predicting Homicide Rates (controlling for race) 102 TABLE 3 OLS Regression Predicting Homicide Rates (no control for race) 103 TABLE 4 OLS Regression Predicting Aggravated Assault Rates (no control for race) 104 TABLE 5 OLS Regression Predicting Aggravated Assault Rates (controlling for race) 105 ix INTRODUCTION For centuries, the American South has been regarded as a distinctive region, more violent than the rest of the country. A substantial body of research shows that the South leads the nation in per-capita homicides and certain other violent crimes (Ellison, 1991). The fact that the South has the highest violent crime and homicide rates has proven to be one of the most stable findings in the field of criminology (Huff-Corzine, Corzine, and Moore, 1991). However, the question that remains unresolved is why this region has maintained higher levels of lethal violence than the rest of the U.S. for such a prolonged period of time (Huff-Corzine, Corzine, and Moore, 1986). Scholarly opinion remains sharply divided over explanations for the high rates of homicide that characterize the South. Some investigators attribute the South?s predominance for violence to cultural differences that have manifested through regional historical experiences, while others claim that structural differences, such as socioeconomic inequality and poverty are what create the gap. In order to determine why crime rates differ from place to place or from one time period to another requires an investigation of which variations in social conditions are associated with the differences in rates of criminal violence (Blau and Blau, 1982). One common explanation is the subculture of violence theory, proposed by Wolfgang and Ferracuti in 1967, which suggests that certain portions of society develop and adopt violent values that remain detached from the dominant culture. These values 1 function to normalize and reinforce violent behavior, increasing the chances that hostile incidents will escalate to homicidal ones (Wolfgang and Ferracuti, 1967). One extension and frequent application of this theory is the Southern subculture of violence thesis, which proposes that individuals socialized in the South learn to approve of violence in a wide range of situations and to view violence as important in enhancing their honor or reputation (Ellison, 1991; Reed, 1971). The literature in this line of research typically focuses on whether Southerners are more violent because of differences in structural factors, with most citing indicators of poverty as the cause, or because of the existence of a distinctive culture that fosters violence (Chu, Rivera, and Loftin, 2000). An Overview of Cultural and Subcultural theories In any examination of potential cultural influences on rates of violence, it is imperative to have an understanding of the many forms and facets the term culture may imply. Ann Swidler (1986:273) advances the idea that culture is more of an assortment of resources (?cultural ?tool kit??) belonging to a person that he or she uses when planning or thinking about a particular course of action. Swidler explains that differences in the contents of cultural tool kits may cause individuals to act differently in similar situations. In her view, culture allows people to organize and identify their experiences and situations. Culture is useful in explaining continuities in action that remain stable despite any structural changes that take place (Swidler, 1986). Smith and Zahn (1999:45) offer a similar definition; ?culture provides persons with repertoires of ideas, definitions of types of situations, material products, and other factors that may be combined in numerous ways in developing actions.? Smith and Zahn (1999:46) point out that ?culture?s primary influence on violence, such as homicide, is 2 through ?definitions of the situation,? ?frames,? and/or ?attributional styles? that affect the likelihood that an actor will define a situation as one in which physical assault, perhaps with the intent to kill an opponent, is appropriate or even demanded.? Curtis (1975:7) offers a broader conceptualization of culture, defining it ?as consisting of values, behaviors, outlooks, imagery, expectations, definitions of reality, and meanings specific to a community that shares them.? He suggests that culture can affect the ways that people assign meanings to circumstances, events, and situations, which also influences the ways they interpret interactions and ultimately assign causality or blame for them. Smith and Zahn (1999) support these notions of culture, pointing out the inherent value of attribution theory in linking culture and homicide at a macro level. Subcultural explanations ?account for behavioral variation through the values, norms and beliefs held by members of a group or subgroup? (Smith and Zahn, 1999:28). Cultural and subcultural theories also focus on the role of ideas in causing criminal behaviors (Vold, Bernard, and Snipes, 2002). In a most general sense, the meaning of subculture is a smaller part of a larger culture. Culture is typically defined as, ?a set of shared meanings and patterns of actions that are transmitted from one generation to another;? therefore, a ?subculture would simply be any identifiable part of such a culture? (Douglas, 1982:80). Social scientists prefer a more precise definition of subculture as, ?existing only if this part of the larger culture has some shared and transmitted meanings and patterns of action that are different or distinctive from those of the larger culture? (Douglas, 1982:80). Some of the characteristics of the major components of subcultures include: 1) They are identifiable parts of the larger society, different in some but not all aspects from 3 the larger society; 2) As a part of that society, they are subject to at least some of its rules and laws; 3) As groups with identifiable differences from the larger society, they have their own conduct norms for members; 4) They are functioning unities-that is, for at least some purposes they are capable of acting as a whole; 5) They are conscious of themselves as units in some way separate from the broader society (Douglas, 1982). Subcultures also engage in a transmission of subcultural values, the sharing of which ?means that there has been a learning process that established a dynamic lasting linkage between the values and the individuals? (Wolfgang and Ferracuti, 1967:107). The learning process allows for individual variations, but must have generated common motives, reaction patterns, and differential perceptual habits that will ensure the subculture?s continued survival. A group can only be referred to as a deviant or violent subculture if some of its shared meanings and actions violate some important dominant cultural value. From the viewpoint of the larger dominant culture, the values of the subculture set it apart and prevent total integration, occasionally causing open or covert conflicts (Wolfgang and Ferracuti, 1967). The mainstream culture may directly or indirectly promote this apartness, and normative isolation and solidarity of the subculture typically result. Wolfgang and Ferracuti (1967), emphasize that no subculture can be totally different from or completely in conflict with the society of which it is a part. The conflict stems from a contrast of two or more normative systems, at least one of which follows strong adherence to a set of moral values that are often codified. Larger society tolerates only those values that do not cause disruptive conflict or that do not disturb the dominant 4 normative solidarity. The difficulty in studying subcultures is that there is a great deal of variability in the degree to which the subculture overlaps the larger culture. The Subculture of Violence In 1967, Dr. Marvin Wolfgang and Franco Ferracuti published The Subculture of Violence, in which they initially presented their hypothesis that violent values are uniquely widespread among African Americans. Their thesis was based upon research conducted in the mid 1950?s in inner-city Philadelphia, which attempted to explain the strikingly high rate of black-on-black and black-on-white homicide and violent crime. They proposed that among blacks in the U.S., there is a subculture of violence in which there is, ?a potent theme of violence current in the cluster of values that make up the life style, the socialization process, the interpersonal relationships of individuals living in similar conditions? (Wolfgang and Ferracuti, 1967:140). Wolfgang and Ferracuti point to specific historical reasons as causal factors in the emergence of the subculture and suggest that historical experiences translate into culture which is then transmitted from generation to generation as a set of ideas even after the original, causal social conditions have already disappeared (Vold, et. al., 2002). In any heterogeneous society, there will be differences in ideas and attitudes toward the use of violence and these differences are commonly observed through variables related to social class. Wolfgang and Ferracuti (1967) attempted to demonstrate the existence of a subculture of violence by examining the social groups and individuals who experience the highest rates of manifest violence. This specific subculture is defined by the ready use of violence, immediate defense and counter-aggression expected of its members (Wolfgang and Ferracuti, 1967). For deviant subcultures, violence has proven 5 to be a common subcultural response to certain stimuli. Wolfgang and Ferracuti (1967) insist that in a subculture of violence, it is not unlikely for all members to accept a value set dependent upon violence, demand or encourage adherence to violence, and penalize deviation. They also assert that members of the subculture of violence tend to value honor more highly than people in larger society, and tend to value human life less highly (Wolfgang and Ferracuti, 1967). Wolfgang and Ferracuti (1967:156) describe homicide as ?a situation not unlike that of confrontations in wartime combat, in which two individuals committed to the value of violence came together, and in which chance, prowess, or possession of a particular weapon dictates the identity of the slayer and of the slain.? The overt use of force or violence, either in interpersonal relationships or in group interaction, is generally viewed as a reflection of basic values that are distinct from the dominant culture. Wolfgang and Ferracuti speculate that quick resort to physical combat as a measure of daring, courage, or defense of status appears to be a cultural expression, especially for lower class males irrespective of race (Vold, et. al., 2002). When two individuals interact and share such a cultural norm response mechanism, physical assaults, altercations, and violent domestic quarrels that result in homicide are likely to be common (Wolfgang and Ferracuti, 1967). To support the thesis of a subculture of violence, Wolfgang and Ferracuti (1967:158), offer the following corollary propositions: 1) No subculture can be totally different from or totally in conflict with the society of which it is part; 2) To establish the existence of a subculture of violence does not require that the actors sharing in these basic value elements should express violence in all situations; 3) The potential resort or 6 willingness to resort to violence in a variety of situations emphasizes the penetrating and diffuse character of this culture; 4) The subcultural ethos of violence may be shared by all ages in a sub-society, but this ethos is most prominent in a limited age group, ranging from late adolescence to middle age; 5) The counter-norm is nonviolence; 6) The development of favorable attitudes toward and the use of violence in a subculture usually involve learned behavior and a process of differential learning, association, or identification; 7) The use of violence in a subculture is not necessarily viewed as illicit conduct and the users therefore do not have to deal with feelings of guilt about their aggression. It is likely that no single theory will ever explain the variety and patterns of observable violent behavior; however, the subculture of violence approach offers the advantage of bringing together psychological and sociological constructs to assist in the explanation of the concentration of violence in specific socio-economic groups and ecological areas (Wolfgang and Ferracuti, 1967). The Southern Subculture of Violence Wolfgang and Ferracuti?s subculture of violence thesis generated a vast amount of additional theory and research, especially in regards to explaining high levels of violent crime in the American South and among African Americans. The most frequent application of Wolfgang and Ferracuti?s thesis has been the attempt to explain the regional variations in homicide across the nation in terms of a Southern subculture of violence. Many researchers argue that the South does indeed have a distinctive cultural orientation that is characteristically violent (Messner, 1983). Researchers generally focus on a historical understanding of the influence of Southern regional culture. The Southern 7 subculture of violence is one theory as to why homicide rates among blacks and whites in the South are well above national averages. The Southern subculture of violence thesis is generally evaluated against a social disorganization model of the factors that may drive crime rates (Land, McCall, and Cohen 1990). Social disorganization theories are rooted in the classical Chicago school of thought and were largely developed through initial studies by Wirth (1938) and Shaw and McKay (1942). Wirth?s (1938) study of urbanization showed ?how the large population size, high density, and great racial and ethnic heterogeneity of urban areas produce anonymous, superficial, and transitory human relations? (Farrell and Swigert, 1987). Wirth believed that these types of relationships led to friction among urbanities and consequently, a need for more rules, regulations and formal controls. He suggests that cities and other large urban areas served to stratify social relationship and interactions, making it more and more difficult for shared cultural understandings to survive. Shaw and McKay (1942) extended the work of Wirth by adding that the urban characteristics chronicled in his study, coupled with poverty and population mobility, lead to social disorganization. Shaw and McKay (1942) suggest that these particular features of urban areas weaken social control and lead to increases in crime rates and the formation of delinquent subcultures. Family disintegration, poverty, and other related community-level characteristics are proposed to ?directly or indirectly affect informal social control networks, community attachment, anonymity, and the capacity for surveillance and guardianship? (Land, et. al., 1990:925). When these facets of social order and organization are weakened, higher rates of deviance and lethal violence are typical consequences. The Southern subculture of violence thesis and the rancorous 8 debate surrounding it have roots in the macro theories of early investigators such as Shaw and McKay (1942) and Wirth (1938). The literature on Southern violence largely concerns which factors to include as appropriately defined measures of either cultural or structural regional characteristics. Many researchers and theorists argue that there is a Southern subculture of violence that has its historical roots in, ?the exaggerated sense of honor among Southern gentlemen, the institutionalized violence associated with maintaining a part of the population in slavery, the defeat at the hands of Northerners in the Civil War, the subsequent economic and political exploitation of Southern states by the North, and so on? (Vold, et. al., 2002: 147). Similar to Wolfgang and Ferracuti, theorists in this genre insist that the Southern subculture of violence arose out of these historical roots, but that its continued existence can be attributed to the ideas being passed from generation to generation, although the conditions that initially gave rise to the ideas no longer exist (Vold, et. al., 2002). In addition, two basic conditions must be met in order to demonstrate the existence of a regional subculture of violence. First, ?individuals in one region must manifest attitudes that differ from those of broader society. Second, one must identify the mechanisms of socialization and intragroup contact via which these distinctive regional attitudes toward violence are transmitted across generations and preserved among group members? (Gastil, 1971:908). By these standards, there is a substantial body of research that supports the Southern subculture of violence thesis. 9 Potential Causes of Southern Violence The Socio-historical Tradition of the South While many researchers have labeled the South as a distinct region, culturally separate from the rest of the United States, the origins of Southern violence are more difficult to trace. Some suggest that the first violent tendencies were born in the antebellum South with the settling of the frontier (Ellison, 1991; Reed, 1972; Cohen and Nisbett, 1994). Murder rates were significantly higher in the South in 1850, and even in its most settled regions, the South had proportionately more murders and homicides than the North (Redfield, 1881). Dickinson, (1946) suggests that violence was a natural product of the rugged individualism encouraged by frontier conditions. Other researchers further this idea by explaining that, ?the very proximity of the frontier in time and space created the sort of atmosphere in which violence could flourish among all classes of Southerners? (Ellison, 1991:1236). The Frontier South lacked strong institutions of law and made a virtue of defending one?s honor in personal duels. Reed (1972:27) suggests that ?all of these factors culminate to highlight a peculiarly Southern disposition to use force to settle personal, sectional, and national grievances.? In 1880, H.V. Redfield examined the difference between murder rates in the North and South, identifying a number of different patterns of Southern homicide. Redfield found that most of the time, people were killed because of various kinds of difficulties; ?drunken brawls in the South might not be too different from those in the North, yet in the South they more often led to murders, both because of attitudes and the practice of carrying guns or knives? (Gastil, 1971:417). Murder also occurred more often 10 in the South to redress insult to personal honor, or because of a man?s desire to show off. Many researchers highlight the importance of chivalry and defensiveness that characterizes the South. The ?feudal spirit of the Old South?, in which honor and loyalty were practiced ideals, has been linked to the southern tendency to resort to violence in situations that threatened these (Ellison, 1225). Research on the Southern subculture of violence uncovers extreme notions of chivalry and exaggerated defensiveness unique to Southern males. The oldest explanation for Southern ideas about violence attributes the exaggerated violence to the existence of slavery. The system of Slavery was far more prevalent and prevailed longer in the South than any other region of the U.S. The slave system was based on force and was inherently maintained by force. Corporal punishment was, as Southern whites admitted, the one thing that made slavery and slaves work and the necessity for physical force was endemic to the system (Dickinson, 1946). Arbitrary violence was inseparable from the institution of slavery; it is what made the system function; it set the tone for relationships between slaves and southern whites; and it set the tone as well for the quality of life led by those in captivity (Dickinson, 1946). Race relations only intensified with the implementation of the Ku Klux Klan, which was first initiated in the South (Tennessee), in 1865. It is imperative to point out that the well known pattern of terror associated with Ku Klux Klan attacks on blacks was a white on white pattern as well. Lynching was another approved mechanism of dispersing racial hatred through violence across the South. The ?stealthy night killing by an organized group or mob was common throughout the South,? and that, ?murder of this 11 type might be accomplished by ambush, by a raid on a house, including the wiping out of a whole family? (Gastil, 1971:417). Hackney (1969) and others have attributed the tradition of Southern violence to the military defeat of the South in the Civil War and years following. This theory holds that ?the military defeat of the South and its subsequent political domination and economic exploitation by Northern forces gave rise to high levels of tolerance for interpersonal violence? (Ellison, 1991:1224). In addition, many researchers suggest that the strong regional tradition of military service and respect for military customs bears upon a Southern willingness to approve of or engage in violence, and that a ?martial spirit? is present in the Old South as well as the New (Dickinson, 1946). Another common explanation of Southern violence is the South?s well- documented ?cultural lag? in education, industrialization, and urbanization (Reed 1971:441). Gastil (1971:417) points out, ?It is the less competitive and less commercially minded rural South that has historically been the source of high homicide rates.? The historical foundations and hastened economic development of the South give researchers even more clues as to the region?s penchant for violence. Early settlement of the Southern region is a central explanatory factor in many studies regarding the Southern subculture of violence. One important point is that more whites in the South initially came from a lower status in England than was true of most Northerners. Early on, the Southern economy was originally based on herding; in contrast, the Northern U.S. was settled by farmers (Puritans, Quakers, Dutch, and Germans) who built a civilization that included artisans, trades people, business people and professionals of all sorts. Settlement in the 12 North was also ?to a greater extent continuous and contiguous;? towns were more important, closer together, and cities grew more rapidly (Gastil, 1971:417). In the South, large landowners initially came there for easy wealth and to establish the type of feudal estates that were not available to them in England (Gastil, 1971). Into the 19 th century, their vision remained one of the exploiting, 17 th century rural aristocracy and many were able to obtain these goals through poor indentured white servants and eventually blacks as slaves (Gastil, 1971). Extreme class differences were expected and were widely apparent from the beginning, as was the acknowledged fact that such highly stratified societies as this had to be maintained by violence. Although these theories may appear to be widely divergent, they all have one common theme: historical events unique to the South have built and shaped a culture more conducive to violence. Without denying the possible role of structural factors such as poverty, the argument that the historical experiences of Southerners has created a continuing cultural legacy conducive to interpersonal violence is a strong one. Huff- Corzine (1986:920) and her colleagues insist that the regional effect on homicide levels cannot be ?explained away? simply as the result of socioeconomic and/or demographic characteristics of the South; rather, it appears that it is the socio-historical tradition of the South, to include: militarism, frontier life, slavery, economic development, settlement, and religion that manifests cultural differences in this more rural region. It is hypothesized here that the socio-historical tradition of the South condones, encourages and facilitates the development and existence of a subculture that values violence. 13 Theoretical Perspectives The Cultural Position In any discussion of Southern violence, there are two competing theoretical frameworks, one structural; the other cultural. The cultural perspective is the foundation of the Southern subculture of violence thesis. There are a vast and divergent number of theories, none of which claim to be mutually exclusive, that emphasize culture in an attempt to account for elevated rates of Southern violence and homicide. Cultural explanations include the tradition of slavery in the South, the frontier spirit of the antebellum South, the feudal spirit of the old South, the loss of the civil war, and the hunting and herding life which was far more common in the South than the North (Huff- Corzine, et. al., 1986). These explanations are presumed to have given rise to a culture of honor in the South; a subcultural system that condones, encourages, or facilitates criminal behavior (Chu, et. al, 2000). Gastil (1971:412) argues that the historical experiences of Southerners have produced a cultural tradition conducive to interpersonal violence, and asserts that, ?it is a predisposition to lethal violence in Southern regional culture that accounts for the greater part of the relative height of the American homicide rate.? Advocates of the cultural perspective assert that while culture is a difficult concept to define and measure, it is still certain to exert some influence on rates of violence and homicide independent of various structural factors. The cultural explanation does not ignore the effects of structural factors such as poverty on homicide rates; rather, this position offers a sub-theory which views poverty as a cultural phenomenon as well. This culture of poverty theory suggests that, ?individuals create, sustain, and transmit to future generations a culture that reinforces the 14 various social and behavioral deficiencies,? that perpetuate poverty (Roach and Gursslin, 1967). According to this thesis, the poor are assumed to harbor dysfunctional attitudes and values which create a self-reinforcing environment of deviant behavior that restricts their access to economic success. The poor become aware of and adapt to their own marginal position within a highly stratified and individualistic capitalist society, which offers them no prospect for upward mobility (Duncan, 1999). This theory is a cyclical one, in that, the persistent failures of poor families discourages them from instilling in their children the values that will someday enable them to lift themselves out of poverty. The poor develop into their own specific subculture, one that embodies common behavioral patterns, norms and values that are distinct from that of the mainstream culture. Poverty, in turn, becomes a vicious cycle from which few escape. A cultural explanation of regional homicide rates has inherent value because it allows for an in depth analysis and explanation as to why there is a disparity in violent crime and homicide rates in the Southern U.S. and why other studies have failed to account for this disparity on a structural basis alone. The cultural perspective also highlights the fact that there are cultural differences in norms concerning violence that exist from region to region. Theories on Southern culture and its link to violence have withstood the test of time for well over a century now; Roland (1982:20) reaffirms this point, ?There is cause to believe the region?s unique combination of political, religious, cultural, ethnic and social traits, reinforced as they are by geography and history, myth and folklore, and convention and inertia, will for a good while yet keep it distinctive.? 15 The Structural Position Many researchers highlight the fact that the South is poorer than other regions in the U.S. and consequently point to poverty and its strong association to violence and homicide to explain regional disparities in rates of lethal violence. Advocates of the structural perspective assert that differences in regional homicide rates are a product of divergent socioeconomic and demographic characteristics in the South. Those who support a structural interpretation view violence as a behavior stemming from an adaptive response to the structural conditions under which people are forced to live (Williams and Flewelling, 1988). There are several theoretical links between the poverty and violence relationship that better explain this position. Relative deprivation is viewed by some investigators as the vehicle by which poverty leads to crime (Blau and Blau, 1982). This theoretical perspective assumes that ?some individuals evaluate their socioeconomic position in relative terms and they are bothered by the perception that others have more desired social and economic resources? (Parker, 1989:985). This altered perception fuels frustration in individuals because they are left feeling hopeless in their relative position; violence then arises out of the frustration. The violence is usually directed at those closest to the deprived individual, rather than at those who are perceived to be in a better social standing (Parker, 1989). Another aspect of the poverty-violence relationship is the suggestion that absolute deprivation, independent of other factors, can cause violence. This approach proposes that violence may stem from emotional situations related to absolute deprivation simply because the day to day life of the poor involves many inherent difficulties. Parker 16 (1989:986) suggests that ?perhaps violence is one of the few options available to those without the economic means to deal with problems and crises of everyday life.? Factors associated with absolute deprivation such as marital instability and unemployment have been found in studies of non-lethal violence to be associated with assaults occurring in the family (Parker, 1989). Lastly, the structural theory of poverty holds that poverty is caused by the structure of the larger socioeconomic order. The structural perspective blames macro forces, such as global capitalism, for producing inequalities in the distribution of resources and opportunities in both the economic and social systems (Corcoran, 1995). For many years, policymakers and scholars have ?criticized the way capitalism allocates resources unequally among groups and places, and the way racism has caused persistent disadvantage for black Americans and other minorities? (Duncan, 1999:187). Proponents of the structural theory argue that poverty can be traced to ?several interrelated institutional environments that serve to favor certain groups over others, generally based on gender, class, or race;? therefore, women and minorities suffer disproportionately (Corcoran, 1995:242). Variances over time and place in structural economic factors create barriers such as unequal accessibility to education, childcare, and opportunities in the job market which ultimately result in keeping the poor in the poorhouse. There is a substantial body of evidence that indicates a strong correlation between poverty and lethal violence among the literature on the Southern subculture of violence thesis; however, there is no evidence as of yet that confirms an explanation due solely to structural factors. 17 The following section includes a review of the literature on Southern religion and violence, the theoretical links between assault and homicide, and a good portion of the prior research surrounding all relevant theoretical perspectives (cultural, structural and integrated) involved in the investigation of Southern violence. 18 LITERATURE REVIEW Religion There are many cultural components distinctive of the South, but none is more so than religion. The region always has been and continues to be overwhelmingly Protestant, making this denomination a driving force behind regional culture and action. The present study seeks to define religion as a measure of Southern culture; therefore, it is important to provide a theoretical link between religion and culture. It is the history and development of conservative Protestantism in the South is what has shaped the region?s unique religiosity. Initially, Baptists (a denomination of the conservative Protestant faith) split off into mainly Southern branches during the Civil War, and religious life in the South began to take on a distinctive flavor. Frontier traditions of appeal to emotions and revivalism, and a more simple faith thrived in the smaller towns and rural areas of the South long after other regions had already undergone major religious changes and begun adapting to pluralism and urban life (Roof and McKinney, 1987). Key to this break in mainstream religion is that, ?religious conservatism helped immunize Southerners against cultural as well as religious influences from the outside and thus contributed to the making of a separate and distinct religious subculture? (Roof and McKinney, 1987:129). Roof and McKinney (1987:126) highlight the power of such a longstanding faith; ?A religious community that has been dominant, or at least very prominent, for a long time has many 19 things making it a formidable force: its extensive infrastructure; its visibility; its size and stability; and the mere fact of its being the ranking faith.? In 1980 alone, 86 percent of Baptist church members were from the Southern region. Some Southern states boast a combined Baptist-Methodist membership of more than 80 percent of the confirmed Protestant affiliation, and are by far the majority of the total religious population (Roof and McKinney, 1987:129). These two denominations comprise most of the church membership of the South and consequently set the tone for the region. Roof and Mckinney (1987:129) also suggest that ?a close, comfortable alliance exists between the popular churches and Southern culture, which makes for a congenial blend of religious beliefs with regional attitudes and thought forms.? Conservative Protestants follow traditional morality and focus on the individual relationship with God and salvation. Their theology is highly subjective and is comprised of small-town values and a likewise view of the world. Roof and McKinney (1987:129) suggest that Southern religion is set apart from other regions because it is ?sustained by theological fundamentalism and a puritan moral ethic?, and insist that ?the ?old-time religion? of Dixie is unlike that found anywhere else in the world.? Smith and Zahn (1999) suggest that social institutions such as churches are the pillars of society, which have roots in the social structure and provide support for the surrounding culture. They emphasize the significance of culture in control explanations of homicide, which explain how when social institutions experience some sort of breakdown, it likely disables some of the controls against violence (Smith and Zahn, 1999). Smith and Zahn (1999:38) also assert that ?the functioning of institutions reflects 20 the cultural orientations that prevail in a society.? Religion is one mechanism by which culture, to include ?distinctive attitudes, values, attributional styles, and worldviews are transmitted across generations? (Smith and Zahn, 1999:53). Roof and McKinney (1987) point to religion as the reason why regionalism has not faded out even in contemporary America. Religion has always been and remains a major contributor to distinct ways of life and cultural traditions in different parts of the U.S. Conservative Protestants, more than any other denomination, remain concentrated in the regions in which they were historically strong. Since most of the conservative Protestant denominations are primarily located in the South, ?they continue to be encapsulated by this regional subculture? (Roof and McKinney, 1987:132). However, there are signs of religious growth and change in the South, albeit at a slower pace than the rest of the country. Some conservative Protestant denominations, including Southern Baptists, are growing faster outside of the South, rather than within it. The theoretical threads that tie religion to culture may appear logical and obvious, however; linking religion to violence and/or homicide can be a bit more complicated. Attribution theory provides a theoretical link between Protestant Fundamentalism and high homicide levels. Corzine and Corzine (1994:149) explain that ?studies linking Protestant Fundamentalism, more common in the South than in other regions, to preference for retribution in criminal sentencing are indirectly relevant to the proposition that the attribution of blame affects the direction of lethal violence in the Southern states.? Fundamentalists focus more on the ?role of characterological traits in guiding behavior;? therefore, they ?are more likely to attribute intentionality to the acts of others? (Smith and Zahn, 1999:53). Also, the Fundamentalist view concerning the actions of 21 others may encourage them to be more supportive of retribution when it comes to punishment. Corzine and Corzine (1994:149) extend this idea by arguing that ?adherence to a fundamentalist doctrine would increase the chances of attributing the causes of one?s failures and frustrations to the malevolent acts of others, thus resulting in aggression being directed outward rather than inward.? Grasmick and his colleagues (1992) also rely on attribution theory as a tool for explaining the extent to which support for the ?retributive doctrine of punishment? is linked to Fundamentalist Protestant denominations and their associated beliefs. Fundamentalists follow a literal interpretation of the Bible, which consequently attaches meaning to any incident or action. The Fundamentalist perspective views retribution as a logical punishment philosophy because its members attribute criminality to be a character or dispositional flaw of the actor; therefore, the only way to correct this sinfulness is to bring about a punishment that will restore justice (Grasmick, Davenport, Chamblin, and Bursick, 1992). The researchers suspect that affiliation with Protestant fundamentalism may be a social and cultural determinant of adherence to or support of the retributive punishment and may also strongly influence views on crime and punishment (Grasmick, et. al., 1992). Their findings indicate that religious affiliation is a significant predictor of retributiveness mainly because fundamentalists are more inclined to interpret the bible literally, which they also find empirical support for. Grasmick and his colleagues conclude that differing religious beliefs between Protestant fundamentalists and other denominations are what create variance in support for retribution among the general population. 22 Using data from the 1983 General Social Survey, Ellison (1991:1221) finds empirical support that native Southerners are more inclined to condone and be supportive of certain forms of violence, and that any exposure to non-Southern culture weakens those Southern values that justify or rationalize violence. Ellison (1991) furthers the work of Reed (1972) and others by testing regional differences in levels of support for violence and in the determinants of attitudes toward violence. Ellison considers the effects of socialization and social ties among southerners, but also investigates the role of formal institutions in the socialization process. Ellison (1991) explores one particularly untapped aspect of Southern life: religion. He suggests that the central features of religious culture in the South may produce a Southern willingness to approve of defensive and retaliatory violence, arguing that ?Southern theology typically emphasizes the sinfulness of human nature and the need for individual salvation from punishment at the hands of a patriarchal, vengeful God? (Ellison, 1991:1226). Ellison (1991) point out that religion, especially conservative Protestantism, is a core characteristic of Southern culture, and that the centrality of this type religion may legitimize the use of interpersonal violence under certain circumstances. Ellison?s results offer no evidence that peer socialization or social ties between Southerners play a part in formulating supportive attitudes toward defensive or retaliatory violence. According to Ellison?s statistical analysis, defensive violence receives its strongest support from the oldest native southerners (over 60 years of age), which is ?consistent with suggestions in the literature that regional attitudinal distinctiveness may be declining? (1991:1230). Regional variations in attitudes toward violence and rates of 23 lethal violence are expected to diminish over time as the cultural lag between the South and other regions disappears. Measuring attendance at religious service and hierarchical religious imagery as indicators of religiosity on attitudes concerning violence, Ellison (1991) finds a pattern: church attendance among native Southerners is a positive predictor of supportive attitudes toward violence in instances of defense or retaliation. He argues that Southern attitudes toward interpersonal violence are related to certain elements of Southern theology and religious culture. Ellison?s (1991:1231) data indicates that a, ?Southern public religious culture is linked with distinctive regional views on the acceptability of violence.? His analysis shows no evidence that confines Southern attitudes about violence to any particular racial group. Ellison?s research is noteworthy not only because it supports the case of a Southern subculture of violence, but also because it supports the notion that violence and violent attitudes can also be the products of aggregate structural factors, such as inequality and poverty. In an extension of his original work, Ellison (2003) and his colleagues examine regional disparities in homicide rates focusing specifically on the role of religious culture, particularly Southern conservative Protestantism. Ellison, Burr and McCall argue that the institution of religion in the South may be one mechanism through which violent orientations are legitimated and even transmitted. Ellison and his colleagues (2003) cite research indicating that those who practice conservative theological beliefs have a tendency to view crime and other types of deviance more harshly and endorse more punitive responses toward offenders. Other research findings show that Southerners are less supportive of unpopular or deviant groups than are other Americans. Southern 24 regional religious culture among whites has consistently been interpreted as placing a greater weight on Old Testament themes, evoking more powerful imagery and narratives that translate to acceptance of defensive or retaliatory violence, and linking moralistic fundamentalism to a strong support of punitive actions including capital punishment (Ellison, Burr, and McCall, 2003). Using a scheme developed by Roof and McKinney (1987) that categorizes which religious groups are conservative Protestant, Ellison (et. al, 2003) finds that conservative Protestant church membership is a positive predictor of homicide and that there is a significant estimated net effect of conservative Protestantism on homicide in Southern Metropolitan Statistical Areas (MSAs) but not among non-Southern MSAs. Ellison, Burr, and McCall (2003:339) also find evidence that points to marked differences in some of their structural variables as well; ?On average, Southern MSAs are smaller, but they also have larger minority population shares, family poverty rates, and rates of population change, and somewhat higher divorce rates, than non-Southern MSAs.? The researchers argue that the influence of conservative religious ideas and imagery is broadly dispersed across the South and contributes to a culture that is more likely to be accepting of violence. Ellison and his colleagues (2003:342) point out that religion shapes culture by leaving an imprint on ?(a) social customs and public policies (e.g., regarding corporal punishment, criminal sentencing, etc.), (b) community reactions to local and national events, (c) public discourse (e.g., by civic elites, media, officials), and (d) popular culture (e.g., stories and legends, music, etc.).? Ellison, Burr, and McCall?s findings help to establish a more solid link between religion and high homicide rates in the South that is suggestive of a different ?tone and 25 character of conservative Protestantism? that is distinct to the region (2003:342). In addition, the researchers suggest that future research should be conducted on other forms of violence because the South has higher rates of non-lethal violence as well as homicide; any study that could link religion to these types of violence would further the core arguments they have developed. In another study on the link between violence and religious culture in the South, Ellison and Musick (1993) investigate southern intolerance as a possible fundamentalist effect finding partial support that regional variations are caused by the prominence of conservative Protestantism in the South. Logic follows that low levels of tolerance are a precursor of violence. Ellison and Musick (1993) provide evidence that shows that Southerners are less willing to extend civil liberties to groups that are unpopular or considered deviant (e.g., homosexuals, atheists, etc.). Numerous studies have linked the South?s form of Protestantism, which carries along with it a literal interpretation of the Bible and a higher frequency of church attendance than any other region, to observed regional differences in tolerance. Ellison and Musick (1993:382) assert that fundamentalist Protestant religious groups are ?more culturally homogeneous,? thereby making them less accepting or understanding of different lifestyles or worldviews. Controlling for several possibly confounding variables, Ellison and Musick (1993) find moderate support for the argument that Southern intolerance is a reflection of the abundance of and commitment to fundamentalist Protestantism in the South. They offer along with religion, the general ?cultural homogeneity of the South? may be another factor that helps to explain the remaining variation in regional levels of tolerance (Ellison and Musick, 1993:389). 26 Assault Prior research has narrowly deemed or defined the subculture in the South as violent based explicitly on only one measure of one form of violence (homicide). This research seeks to expand the concept of violence typically used in research concerning the Southern subculture of violence by including aggravated assault. If the South truly does harbor a violent subculture, then it follows that rates of non-lethal violence will be high as well as homicide rates, and that both crimes will share the same regional predictors. Homicides and aggravated assaults are particularly parallel crimes and should be researched as such. To view homicide and aggravated assault as analogous crimes is to assume that they differ only in the outcome: the victim?s death. Research has shown that homicide and assault represent similar behaviors that differ principally in outcome and that most homicides would be better considered as fatal assaults. Not only does a link exist between these crimes, but also both crimes are committed at a higher rate in the South than any other region of the U.S. Gastil (1975) reaffirms this point, proposing that the differences in aggravated assault rates in the South reflect the murder differentials. There is precedent in past research on violent crime in the U.S. that provides evidence of the strong similarities between aggravated assaults and homicides (see Luckenbill, 1977; Weaver, Wittekind, Huff-Corzine, Corzine, Petee, and Jarvis, 2004). The present study conceptualizes homicide and assault ?as different points on a continuum of potential outcomes of violent encounters ranging from death to no injury that are contingent on characteristics of the offender, the victim, the circumstances, the time, and the location? (Weaver et. al., 2004). Prior research also indicates marked 27 similarities between victim/offender demographic characteristics for both types of crimes (Luckenbill, 1977; Weaver, et. al, 2004). An important element in the assault/homicide continuum is ?to recognize that most behaviors have multiple outcomes, only some of which are desired by the actor? (Felson and Messner, 1996:521). Although harm may occur to the victim, it is not typically the offender?s goal or original intention. A bulk of research suggests that the offender does not premeditate or predetermine the outcome of a violent incident (Wolfgang and Ferracuti, 1967; Luckenbill, 1977; Weaver et al., 2004). In fact, Wolfgang and Ferracuti (1967:156), describe homicide as ?a situation not unlike that of confrontations in wartime combat, in which two individuals committed to the value of violence came together, and in which chance, prowess, or possession of a particular weapon dictates the identity of the slayer and of the slain.? Luckenbill (1977:176) argues along these same lines, insisting that homicides are ?situated transactions? in which ?it is expected that the participants develop particular roles, each shaped by the others and instrumental in some way to the fatal outcome.? With this perspective, it can be argued that assault and homicide are initially identical criminal actions that change due to differing circumstances during the course of the action, ultimately precipitating dissimilar outcomes. Luckenbill (1977:177) views homicide as ?an intense interchange between an offender and victim,? a sort of ?character contest,? in which both the offender and victim contribute to the escalation of the event to murder. He also insists that there is some sort of consensus among those involved that violence is a suitable, if not a necessary resort. Luckenbill (1977:186) investigates 28 numerous criminal homicides on an individual basis, finding that homicide does not appear to be a one-sided event; rather, criminal homicide is more of a ?working agreement? between participants that value violence as an acceptable means of resolve between two people who are already ?committed to battle.? Cultural Interpretations Sheldon Hackney and Raymond Gastil were pioneers in the formation and development of the Southern subculture of violence thesis. Expanding on the work of very early investigators such as Redfield (1880) and Brearley (1935), Hackney and Gastil cite several elements of Southern history that they believe are indicative of violent cultural orientations transmitted from one generation to the other. Hackney (1969) and Gastil (1971) provide evidence to advance the claim that the South retains a subculture predisposed to violence that is to blame for the high rates of homicide that typify the region. Both Gastil and Hackney attribute the subcultural component to cultural differences deep-rooted in the Southern region, a theory supported by their regression analyses of state homicide rates. Controlling for several structural variables, Hackney highlights a significant correlation between ?Southernness,? used as a dummy variable, and homicide rates. Hackney proposes that while structural variables may play a role in the elevation of Southern homicide rates, it is the historical experience of white Southerners that has created an on-going culture favorable to high levels of violence, inclusive of homicide. Each of their findings demonstrates that the use of a regional variable for the South in a regression equation alongside several demographic and socioeconomic variables creates a 29 significant independent effect on the homicide rate for the state. Both authors explain this effect as a result of a violent cultural orientation distinctive to the South. Gastil (1971:412) also attributes the South?s high homicide rates to ?a predisposition to lethal violence? embedded within the regional culture. Gastil uses a Southernness Index (SI), ?based on migration patterns of whites from the South in regression models with demographic and socioeconomic variables? and is able to demonstrate that ?degree of Southernness? has a strong correlation to state homicide rates for the entire population and for whites. Gastil (1971) utilizes qualitative historical evidence and multiple regressions in his study, which finds that an index of Southernness explains a great deal of the variance and more than any other variable when it comes to homicide rates. Gastil (1971) attributes the high homicide rates of the South to the persistence of Southern cultural traditions that developed before the Civil War. Similar to Hackney, Gastil admits that structural conditions may account for a good portion of the variance in regional homicide rates, but also insists that the remainder may be due solely to ?Southerness.? Gastil?s regional culture of violence theory is similar to the subculture of violence theory advanced by Wolfgang and Ferracuti (1967), but differs on the concept of culture. Gastil?s (1975:102) conceptualization of culture focuses more on a violent tradition and emphasizes ?persistence over time and intergenerational reinforcement more than the subcultural concept of Wolfgang and Ferracuti.? Gastil writes that any society containing a regional culture of violence will also likely contain: ?(1) more extreme subcultures of violence and/or a larger percentage of the population involved in violence (with less limitation by class, age, or race); (2) lethal violence as a more important subtheme in the 30 general culture of the region; and (3) weapons and knowledge of their use as an important part of the culture? (1975:103). Gastil?s notion of a regional culture of violence can be used as an explanatory tool for why violent subcultures do not develop equally across the U.S. under conditions that appear quite similar. Similar to Gastil and Hackney, Porterfield and Talbert (1952) were also early pioneers who studied diverging crime trends in Southern and non-Southern cities. Their findings indicate that during the ten year period from 1940-1949, Southern cities had higher average crime rates for murder, robbery, aggravated assault, burglary, larceny, and auto-theft than cities in any other part of the United States. Porterfield and Talbert (1952) are quick to point out that the aggravated assault rate for Southern cities was more than five times greater than that of non-Southern cities. In the year 1949 alone, Southern cities have a murder rate four times that of any other region (Porterfield and Talbert, 1952). Their research also substantiates the argument that race, age, and gender composition of the population do not account for the contrast of crime rates between Southern and non- Southern regions. Porterfield and Talbert (1952:68) attribute crime rate variations to the social structure and social worlds within a community, offering that ?patterns of crime are cultural phenomena which develop in the interaction of persons and groups with one another as ways of behaving in conflict situations.? They highlight a cultural divergence between Southern and non-Southern cities as a possible source of crime rate differentials, with the South containing more ?conflict groups and aggressive cultural definitions? than the rest of the United States (Porterfield and Talbert, 1952:68). Porterfield and Talbert (1952) assert that Southern cultural ties are binding and involve patterns that allow 31 aggressive responses. They predict that even with the accelerated rate at which conditions in the South are improving, the region will take a long time to catch up to other regions in the U.S. and an even longer time for diverging crime trends and high crime rates to diminish or altogether disappear. Parker and Pruitt evaluate the claim that regional patterns of homicide have significantly changed in recent years, with focus shifted away from the Southern region and toward the West. Recent research and literature asserts that homicide rates in the West for specific racial groups have converged with those of the South, if not surpassed them (Parker and Pruitt, 2000). One possible explanation for this convergence is black migration, specifically to the West, ?to evade the poverty and income equality of the South? (Parker and Pruitt, 2000:1497). The researchers investigate what impact structural conditions may have on race-specific rates of homicide. Parker and Pruitt employ multiple regression analysis, using 1990 data on U.S. cities that have a population of 100,000 or more and three models: an overall model, a South model, and a West model. They find that the mean homicide rate for whites is highest in the West, while the mean black homicide rates in the West and the South are more comparable (Parker and Pruitt, 2000). However, a t-test statistic reveals that the regional differences in mean homicide rates for both races are not statistically significant (F =2.229, p>.05; F =2.306, p>.05). Past research on regional homicide rates by race has acknowledged that blacks consistently have higher homicide rates than whites, irrespective of region, and the larger percentage of blacks residing in the South is proposed to account for the higher rates of homicide there (Parker and Pruitt, 2000). Parker and Pruitt?s (2000:1495) findings are consistent with this position, in that the 32 mean black homicide rate is over four times that of whites (34.29 and 7.63, respectively). This suggests that when homicide rates are disaggregated by race, the South and West become quite comparable. Parker and Pruitt find that their measure of racial inequality has a significant, positive effect on rates of black homicide only in the Southern region. The overall model shows that homicide rates for blacks are significantly higher in the West than in any other region. Their resource deprivation/segregation component is positively related to black homicides in the West and the overall model, and their measure of family disruption significantly increases black homicide rates in these same two models. A negative relationship appears between job accessibility and black homicide in all of the models. Parker and Pruitt (2000) suggest that it is a combination of social isolation and extreme disadvantage that have the strongest impact on black homicide rates. Their results further indicate that the resource deprivation/segregation faced by blacks in the West is strongly associated with homicides there, although this is not the case in the South. Parker and Pruitt (2000) conclude that structural disadvantage, which includes various economic and social conditions, increases black homicide offending in general, but that the type of disadvantage differs by region. Location in the South has a significant, positive effect on rates of white homicide, yet does not significantly affect black homicide rates. For white homicide rates, the overall and West models reveal that there are positive relationships between the resource deprivation/affluence index and white homicide rates, the percentage Hispanic and white homicide offending, and also between job accessibility and rates of white homicide (Parker and Pruitt, 2000). These results indicate that structural variables do not explain 33 much of the variance in white homicides in the Southern region, leaving Parker and Pruitt (2000:1500) to suggest that ?Southern culture is deserving of continual attention in studies of homicide rates, particularly among whites.? They assert that white homicide offending in the South is not a product of structural predictors. In sum, Parker and Pruitt (2000) provide evidence that the effects of structural conditions on homicide rates differ significantly by race and region. While homicide rates in the West and the South may be converging, the factors driving them vary greatly by region. Chilton (2004) extends the research of Porterfield and Talbert (1952) and Parker and Pruitt (2000) by using updated data to examine regional variations in serious criminal offenses between the South and the non-South. Chilton (2004) also takes the issue of race under careful consideration, focusing on how black and white rates of homicide differ from the South to the non-South and if Southern location (Southernness) plays a role in elevating homicide and assault rates for cities, counties, and states within the region. Chilton (2004) highlights the importance of disaggregating race to determine any regional effects on lethal violence. Using NIBRS data on state-level offender rates for 19 states, Chilton?s (2004) results provide persuasive support for regional differences in rates of serious criminal offenses that all point to a more violent South. Chilton (2004) also finds that regardless of region, black offender rates are much higher than white offender rates; however white offender rates are higher in the South than any other region. Chilton?s results are consistent with theories that support the existence of a Southern subculture of violence; however, he argues that if there is such a subculture, then it is race-specific to whites and involves only white violence. In agreement with past research, Chilton (2004) finds both 34 murder and assault rates to be much higher in the South than the rest of the country, but he attributes this disparity to the larger percentage of blacks living in the South. Chilton?s statistical analysis should be interpreted cautiously because he is only able to select and use a small sample of states who participate fully in NIBRS data and he draws conclusions and makes generalizations based on data from only four Southern states. He does however, bring a new facet to the literature on subcultures and violence in proposing that regional differences in relevant crime rates may be a product of a ?culture of self-help? instead; a culture in which urbanites believe the police to be ineffectual and consequently feel forced to rely on themselves to resolve violent situations (Chilton 2004:55). Clarke (1998) also takes race into consideration, but in a different form. He analyzes the idea that slavery and race issues (particularly lynching) that were endured in the old South are a product of a subculture of violence that facilitates such atrocities through situationally determined cultural factors. Clarke (1998) insists that these cultural factors cannot be measured with structural variables such as the census, and other economic statistics. Clarke reports that after emancipation, and into the 1880?s and 1890?s, over 95 percent of all lynching victims, 73 percent of which were black, were tortured and killed in states that were former slave states (1998:271). He suggests that before emancipation lynching was a reflection of the frontier era, where law enforcement was more removed in time and space from wrongdoing or the need for harsh justice. The majority of lynching victims were white, until 1868, when the Ku Klux Klan ushered in a new era of the brutal lynching of blacks. Clarke (1998:274) contends that ?lynching was a peculiarly Southern phenomenon, a reflection of complex psychological and cultural 35 factors not accurately measured with economic, demographic and political indicators intended for other purposes.? Clarke cites other features of the South?s violent past, emphasizing slavery, which he proclaims is the single most important predictor of a region predisposed to a violent future. He concludes that violent actions and values born of tumultuous race relations are a firmly implanted in a ?regional culture of white supremacy? (Clarke, 1998:274). According to Clarke (1998) this culture of white supremacy was long a part of the Southern landscape and served to unify whites, keeping class conflict among them in balance by condoning violence toward blacks. Clarke (1998:276) insists that the essential elements of a subculture of violence as outlined by Wolfgang and Ferracuti, are present in and describe a regional subculture in the South ?that evolved out of centuries of racial antipathy between blacks and whites.? He attributes the decline of lynching in the twentieth century to its replacement by a more tolerable form of violence: capital punishment. Clarke (1998) suggests that structural explanations of lynching and violence ignore Southern history and culture and function only to alleviate the perpetrators of blame as if they were the victims of situational forces over which they had no control. His research provides support and evidence for each of Wolfgang and Ferracuti?s seven propositions of a subculture of violence that Clarke argues still dominates the South. He draws one imperative conclusion in the puzzle of southern violence that renders the significance of the cultural position undeniable, ?human behavior cannot be understood outside of the cultural and historical contexts within which it occurs; while history has 36 shown that racism in America is not limited to the South, it has found its most savage and enduring expression there? (1998:289). Nisbett and Reaves argue for a different type of cultural interpretation; one that does not place the origins of Southern violence strictly in the historical context of slavery or the Civil War as previous studies have. Alternatively, they suggest ?that the South is but one realization of a structural process in which violent cultures are a natural adaptation to subsistence problems that arise among marginal farmers and herding peoples throughout the world? (Chu, et al. 2000). The Nisbett-Reaves hypothesis is well known and highly cited in the field of Southern violence. Their theory contains two major propositions. The first is that the South?s high rates of violence are due to a distinct cultural code, called the ?culture of honor, that supports, even demands, assaultive behavior in defense of one?s reputation, family, and other sacred values? (Nisbett 1993:442) The second proposition is termed, ?the herding-culture of honor? proposition, and it argues that the culture of honor formed in response to scarce and vulnerable resources that characterized the economy of the frontier south (Nisbett 1993:442). The culture of honor can be viewed as an adaptation to problems intrinsic to communities where central authority is weak and economic resources are both scarce and vulnerable to theft by competitors (Nisbett, 1993). Under these types of circumstances, a fierce demeanor and a low threshold for violent behavior (even homicide), have economic value. In essence, Nisbett (1993) explains the high rates of homicide in the South by the prevalence of the culture of honor and advances this further by explaining the development of the culture of honor in terms of his herding theory. Nisbett and Reaves hypothesize that areas where topography or precipitation limited agriculture to 37 marginal farming and herding were exposed to the culture of honor and therefore continue to have the highest homicide rates (Chu, et al. 2000). In addition, this hypothesis incorporates the assumption that the culture of honor that developed in response to differences in agriculture would still be existent in these rural Southern counties (Nisbett, 1993). Nisbett and Reaves found that the mean homicide offender rate was significantly higher in the herding counties than in the counties where cultivation was the typical form of agriculture. In an extension of his earlier research on the culture of honor, Nisbett and his colleagues present new evidence at both the individual and societal level for a ?culture of honor? that prevails in the South, causing elevated rates of lethal violence within the region. The researchers discuss prior research to include surveys on individual attitudes and behaviors about violence (see Cohen, Nisbett, Bowdle and Schwarz, 1994 and Cohen and Nisbett, 1994) that advances the notion that Southerners still participate in a culture of honor. Researchers have also found evidence at the societal level that suggest ?that Southern states are more likely to have laws and social policies that promote or accept violence in some way- including allowing capital punishment for crime, corporal punishment in schools, and the killing of an intruder or person fleeing with property, as well as having higher rates of production of military men and football players? (Nisbett, Polly and Lang:137). Nisbett, Polly and Lang test several hypotheses concerning the existence of a southern subculture of violence, including the presumptions that homicide rates will be higher for whites in the South, while homicide rates for blacks will be unrelated to region; also, that the white homicide rate will be higher in smaller southern cities that are 38 tied to rural areas, rather than larger southern cities; and lastly, that argument and conflict related homicides will be the most common and highest type of homicide in the South. Nisbett, Polly and Lang (1995:142) caution that ?it is not merely potentially misleading to fail to disaggregate the two races; it is empirically a mistake if one wishes to know the relationship between region and homicide for either race separately.? Consistent with a majority of prior research, Nisbett and his colleagues find no statistical correlation between the percentage black and the homicide rate; however, their results show a relationship between Southernness and homicide for whites that had previously been obscured by the use of total homicide rates. With the exception of their Southerness variable, Nisbett, Polly and Lang?s findings are largely in agreement with a substantial body of results from other studies. Poverty and Southerness emerge as the strongest predictors of the homicide rate. For smaller and larger cities, the white male homicide rate for argument-related homicides is highest in the South, while felony-related homicides remains low as predicted. These results in particular lend credence to the culture of honor thesis initially proposed by the researchers. In their South versus North comparison, Nisbett, Polly and Lang (1995) provide evidence that shows that Southernness as a variable produces a dramatic effect on homicide rates for small and medium size cities in the South, while the larger cities exhibit a relationship between homicide and Southern location only when regression analysis is performed. Nisbett, Polly and Lang (1995) attribute their findings on race, region and the homicide rate to differences in culture, suggesting that black culture may be more invariant from region to region than white culture is. Nisbett, Polly and Lang?s (1995) research reaffirms a cultural interpretation of regional variances in homicide rates. 39 They note that although poverty has a strong effect on homicide rates, explanations that are strictly non-cultural cannot alone account for regional differences because regional effects remain significant independent of poverty. Hayes and Lee (2005) explore a similar line of research as Nisbett, Polly and Lang (1995) in an attempt to investigate the notion of a culture of honor in the Southern region. Hayes and Lee note that the culture of honor theory involves a specific subsection of the population: white males in rural areas of the South. They also point out that the values condoning the use of violence relate to specific instances which involve family, self-defense, honor, and personal property (Hayes and Lee, 2005). The culture of honor theory is valuable because it combines both ideas and past research on the debate over race and region and also takes gender into account. Hayes and Lee (2005) use the General Sociological Survey (GSS) to construct variables regarding approval of violence in different situations, and logistic regression as an analytical tool. Their findings indicate that in three of the six models, Southern white rural males are more supportive of violence. In these three scenarios, overall support from the general public remains low to moderate, whereas Southern white males in rural areas are significantly more likely to be supportive of violence in the same given situations. Hayes and Lee (2005) suggest that white Southern males are more likely to view the context of these scenarios as threats to family, status and honor. They interpret their findings as partially supportive of a culture of honor theory and more broadly, a subcultural explanation. Reed (1971) also suggests that the Southern pattern of violence has cultural origins, but he contends that they that consist of regional differences in childrearing 40 practices, socialization processes, and gun ownership. Reed (1971) explores regional differences in attitudes and behaviors of gun ownership and use in the South, and corporal punishment of children. He argues that these issues are indicative of ?a more general acceptance of violence and use of force? (Reed, 1971:432). His findings indicate that Southerners are more likely to report being ?beaten? during their childhood, and more likely to approve of ?beating? their own children. He also finds a higher number of playground scuffles in Southern schools, and a greater prevalence of corporal punishment. Reed (1971:436) draws the conclusion that, ?the greater acceptance by Southerners of corporal punishment in the schools, then, does not represent a response to perceived disorder, but a peculiarly southern acceptance use of force.? Concerning regional differences in socialization processes, Reed suggests that there is a familial and communal bond that holds much stronger and is much more unique to the South; ?given the tendency toward regional homophily among Southern natives, social embeddedness (friendships, community participation, and voluntary organization ties) might be expected to foster or reinforce attitudes supportive of defensive or retaliatory violence? (Reed, 1983:433). Reed (1971) points out that Southerners are also more likely to own guns, offering an explanation as to why; the South is a more rural population in which firearms are a normative tool of the regional culture that are commonly used for hunting. He argues that the widespread availability of guns in the South is a factor in escalating what would otherwise be verbal or physical fights into shooting incidents. However, past research has shown simple assault rates to be higher in the South, which indicates that Southern 41 murders are not just assaults that turn into homicides because of the availability and easy access to guns (Reed 1971). Reed uses test factor standardization, finding that the differences that result in South and non-South populations cannot be attributed to specific demographic factors. The remaining regional differences must be due to some other factor, which Reed demonstrates to be Southernness. The effects of standardization are slight, indicating that, ?one cannot rely on regional demographic convergence to obliterate the demonstrated regional differences in attitudes toward firearms and corporal punishment; to the extent that these attitudes contribute to or are indicative of a more general ?subculture of violence? in the South, that subculture cannot be expected to disappear? (Reed 1971:440). Reed?s work is instrumental in determining that there is a cultural preference for violence in the South that cannot be explained away by differences in structural characteristics. Reed (1982) suggests that most Southerners who commit violent acts do so for a cause. He insists that Southerners do not condone violence of all types and in any situation; rather, violence is more often used in situations that are culturally defined. Reed (1982) highlights the interpersonal nature of violence that Southerners react to, which includes ?unwarranted aggression or malicious behavior, interpersonal threats, and intentional affronts to honor? (Ellison, 1991:1225). He contends that socialization plays a major role in molding violent attitudes, and that these attitudes toward violence are formed at a very young age. According to Reed (1982) the situational appropriateness of violence is a knowledge that lies among the ?best socialized? Southerners, although he offers no measure of determination for this hypothesis. 42 Structural Interpretations In the best-known challenge to cultural explanations of Southern violence, Loftin and Hill (1974) criticize research that relies predominantly on a cultural basis of regional variations in homicide rates. Their results indicate that differences in U.S. state homicide rates are more strongly related to situational variables and that cultural variables do not play a major role independent of these variables (Loftin and Hill, 1974:722). Loftin and Hill (1974) are credited with developing a more comprehensive index of poverty that they term the Structural Poverty Index (SPI). A bulk of research on the Southern subculture of violence employs their SPI as a more solid, composite indicator of levels of poverty. Their findings are significant because they do not reject the hypothesis that cultural variables are important, nor do they suggest that structural variables take complete precedence. Loftin and Hill (1974:742) create compelling suggestions for future research in the study of Southern violence by showing, ?that a more definitive assessment of the role of cultural and situational variables on interpersonal violence will require specifying a theoretical model which would allow for a full range of cultural and situational variables, and data which would allow one to avoid problems of aggregation bias, multcollinearity, and interdependence among units of analysis.? Smith and Parker (1980) reconfirm the earlier work of Loftin and Hill (1974) with research that shows that social-structural variables, specifically the SPI, are the strongest predictor of differences in state and regional homicide rates. Their regional (South) variable provides no significant effect on types of homicide by region when structural variables are controlled. Smith and Parker (1980) assert that homicide rates, frequently 43 used as a dependent variable, are not homogeneous phenomena. Their study categorizes criminal homicide into two types: primary and non-primary. Smith and Parker (1980) examine possible regional differences in type of homicide, finding that the South retains the highest average rates of both primary and non-primary homicides; however, the regional difference in non-primary homicides are minimal. Smith and Parker?s (1980:145) results show that economic deprivation has the greatest effects on levels of primary homicides; however they caution that there is no implication of causality, rather a suggestion ?that conditions (both social and psychological) associated with poverty in the United States contribute to an increased propensity for economically disadvantaged individuals to be involved in an act of criminal homicide.? The percent urban variable emerges as the only significant predictor of non-primary homicide rates in their study. Parker (1989) extends his earlier work with Smith, pointing out the complex theoretical issues involved in deciphering the extent to which either structural or cultural variables have a discernible effect on the homicide rates. He discusses the difficulties with the related issue of measurement, which include the need to obtain more accurate measures of subculture, homicide, and poverty in order to show more clear and precise empirical support for both the structural and cultural perspectives. Parker expands his previous research by developing a more broad classification scheme to include four types of homicide rather than two. His homicide types include: (1) family intimate homicide, which he presumes to be highly related to economic deprivation rather than subcultural measures; (2) primary non-intimate homicide, which he hypothesizes will be most closely linked to subcultural norms; (3) robbery homicides, which he assumes will have a strong 44 relationship to poverty; and (4) other felony homicides, with which he associates a significant link to subcultural predictors. Parker (1989) suggests that indicators of subculture and poverty will have differing impacts on the various types of homicide analyzed. Parker (1989) insists that an evaluation of the significance of region, subculture, and poverty in explaining levels of homicide hinges on the ability to classify types of homicide according to the victim- offender relationship. Past studies (Parker and Smith; 1979; Smith and Parker 1980) have provided a correlation among poverty, subculture, and primary homicides, although none among these same indicators and non-primary homicides. Parker?s results show poverty to be related to three of the four types of homicide, indicating that poverty is the most significant and consistent predictor of the four-category homicide typology used in his study. Beyond poverty, the only other important predictor that emerges is racial composition, which is related to two types of homicide, family intimate and primary non- intimate. Parker (1989) presents statistical findings that provide no evidence for the Southern subculture of violence thesis; in fact, his results show a negative relationship between some types of homicide and Southern location. Parker (1989) concludes that subcultural indicators are not important predictors of the types of homicide analyzed. Blau and Blau (1982) agree with Parker (1989), among many other researchers, that cultural variables are not significant predictors of homicide rates. Blau and Blau (1982) investigate what social conditions make it more likely for people to commit crimes and attribute the regional variations in rates of criminal violence to differences in racial inequality in socioeconomic conditions. They find no basis in explanations of the Southern tradition of violence or poverty as central sources of violent crime. Blau and 45 Blau (1982) assert that the higher rates of violence in Southern cities are not due to the historical experience of the South, but rather to the greater economic inequality there. Their results imply that socioeconomic inequality between races and within them, are positively related to high violent crime rates in cities, and when they are controlled, poverty is not related to these same rates (Blau and Blau, 1982). Williams (1984) undertakes a critical analysis of research conducted by Blau and Blau (1982) and Messner (1982) whose studies both find that poverty is not positively associated with the homicide rate. Messner?s study actually finds poverty to have a significant negative effect on the homicide rate. Whereas Blau and Blau insist that inequality accounts for the most significant influence on homicide rates; Messner questions any suggested links among poverty, inequality and the homicide rate. Williams (1984) cites a wide body of past research that contains findings contrary to that of Blau and Blau and Messner, and he attempts to replicate their earlier studies, updating them with more recent data. In re-estimating Blau and Blau?s and Messner?s initial equations, Williams (1984:288) finds evidence that suggests ?that the log of the percent poor has a significant positive effect on the log of the homicide rate? in both of the researchers equations. His statistical analysis also reports a high correlation (r =0.778) between the percent poor and the Gini index, which implies that in the Blau and Blau study ?the effect of poverty is absorbed by the inequality measure when the measure of poverty is excluded from the analysis? (Williams 1984:288). In support of Blau and Blau (1982), Williams?s findings reaffirm their contention that racial economic inequality is a major contributor to and source of violence in the 46 country. Williams?s research finds the effect of regional location on the homicide rate reported in Messner?s study to be statistically insignificant with the new data. However, Williams (1984) presents evidence that shows that the effect of poverty on the homicide rate does vary by region, highlighting a marked difference between Southern and non- Southern Standard Metropolitan Statistical Areas (SMSAs). Consistent with Messner?s (1982) results, Williams (1984) also finds a strong association between a large black population and higher homicide rates that are not explained by economic variables alone. Williams (1984:283) concludes that in both the Blau and Blau?s study and Messner?s the researchers ?failed to detect a nonlinear pattern in the relationship between poverty and the homicide rate,? and therefore, mistakenly claimed that there was no association. Much like Williams (1984), Harer and Steffensmeier (1992) are concerned with evaluating the relative deprivation theory proposed by Blau and Blau (1982) and the adjunct supposition that relative deprivation is higher among blacks, resulting in the criminogenic consequences of economic inequality to be greater for blacks than whites (Harer and Steffensmeier, 1992). They examine the relationship between economic inequality and violent crime rates of both blacks and whites. Harer and Steffensmeier (1992) replicate the model used by Blau and Blau (1982) in an effort to discern whether or not the findings remain constant across time periods. Harer and Steffensmeier (1992) disaggregate their data by race and use alternative measures of poverty and inequality in an attempt to provide more theoretically appropriate indicators of overall inequality. Their indicators of poverty and inequality include measures of within-race inequality, alongside measures of between-race inequality and overall inequality. Harer and 47 Steffensmeier (1992) utilize race-specific arrest rates for homicide, aggravated assault, robbery, and rape as the dependent variables in their analysis. The results of their multiple regressions show that societal, or total inequality is strongly correlated with each of the violent crimes tested and has a significant, positive effect on the violent offense rates. Both percent black and the overall poverty measure are strongly correlated with all of the violent crimes and are strong, positive predictors of violent offense rates. All of these results coincide with and confirm the earlier work of Blau and Blau (1982) with the exception of one: between race or black/white inequality. Blau and Blau (1982) report that racial inequality has strong, positive effects on the violent offense rates, particularly, aggravated assault and homicide, whereas Harer and Steffensmeier (1992) find weak, negative effects. Contrary to Blau and Blau (1982), Harer and Steffensmeier (1992) insist that racial inequality is not a good predictor of aggregated violent crime rates, but offer that the variance might be due to the use of different measures of inequality. Within race inequality proves to be a strong, positive predictor for all violent offenses for whites; however, is not significantly related to any of the violent crimes for blacks. Other notable findings of their research include that the percent black has a significant, negative effect on all black violence rates except homicide and the poverty variable only has a significant and positive effect on white rates of violence (Harer and Steffensmeier, 1992). Harer and Steffensmeier conclude that inequality is a powerful predictor of high rates of white violence, yet a poor predictor of high rates of black crime. Harer and Steffensmeier (1992:1048) also stress the importance of disaggregating both 48 economic indicators and crime rates by race because the use of aggregate data would have ?masked racial differences in the economic sources of violent crime.? Lee, Maume and Ousey (2003) extend research on the southern subculture of violence by investigating the links between poverty concentration, socioeconomic inequality, and homicide in both rural and urban counties in the U.S. The goal of their research is to expand knowledge about possible structural influences on homicide in non- metropolitan areas. Although these rural areas remain somewhat overlooked in homicide research, they tend to be plagued by socioeconomic disadvantage, severe poverty, and high homicide rates as much as (and in some cases more) than their urban counterparts. Lee and his colleagues (2003:108) point to an ?urban bias? in the literature that largely ignores structural theories that were originally created to explain crime in both rural and urban areas. Lee, Maume and Ousey (2003) suggest that certain factors or characteristics, such as social isolation, that were initially only associated with urban areas have crept across the metro/non-metro divide. After dividing U.S. counties into metropolitan and non-metropolitan samples, Lee, Maume, and Ousey (2003) use negative binomial regression in their analysis. Their descriptive statistics indicate lower homicide rates for non-metropolitan areas, yet higher mean rates of the indicators of socioeconomic disadvantage (poverty concentration, unemployment, and high school dropouts) than urban areas. Lee and his colleagues find that regardless of location (metro or non-metro) there is a significant, positive relationship between homicide and the level of socioeconomic disadvantage. Poverty concentration emerges as a significant, positive predictor of homicide rates in metropolitan counties, yet has a negative, non-significant association in non-metropolitan 49 counties. The difference in the poverty coefficient between urban and rural counties is significant at the .01 level, suggesting that the effects of poverty concentration on homicide rates do vary by location (Lee, Maume, and Ousey, 2003). The results of the control variables show that in metropolitan counties, rates of homicide are higher where there are more divorced persons, greater population size and density, instability, and in counties in the Southern U.S. (Lee, Maume and Ousey, 2003). The pattern of findings for non-metropolitan counties is substantively similar, although the both the percentage of young people and residential segregation have significant, negative effects on homicide rates within these rural counties. Their analysis ?suggests that prominent structural covariates of urban crime? are also useful in explaining variations in crime and lethal violence in rural locations (Lee, Maume and Ousey, 2003:125). Lee, Maume and Ousey (2003:108) argue that although prior theory and research has focused heavily on urban settings, their results and substantive findings indicate that there is little justification for this type of ?urban bias.? Ball-Rokeach (1973:737) asserts that any empirical test of the subculture of violence theory must heavily rely ?on evidence of value differences between persons engaged in violent behavior.? She takes into consideration two types of violence in her research: interpersonal and deviant violence. Ball-Rokeach measures both attitudes and values toward violence and participation in violent crime in one independent survey of male prison inmates and another independent survey of male subjects conducted by the National Opinion Research Center. The prison subjects are Ball-Rokeach?s sample of a subculture, while the national sample most closely resembles the dominant or larger culture. 50 Ball-Rokeach finds weak support for the first hypothesis that would support a subculture of violence thesis: favorable attitudes toward violence are positively associated with more frequent participation in violent behaviors. The second testable hypothesis that follows from the subculture of violence theory is that ?persons who vary in participation in violence should also vary in underlying value patterns ? patterns that should logically be related to violent behavior? (Ball-Rokeach, 1973:737). According to Ball-Rokeach, this hypothesis also fails to gain empirical support. She next discusses the substantive significance of Wolfgang and Ferracuti?s (1967) value characteristics of a subculture of violence. Ball-Rokeach (1973) suggests that although she does find value differences between the inmate (subculture) group and the national sample (dominant culture), the differences are not large enough to confirm that values play a role as determinants of interpersonal violence. Ball-Rokeach concludes that the subculture of violence thesis is flawed and invalid as an explanatory tool for both violent crime and interpersonal violence. She offers an alternative explanation as to why attitudes and values are generally unrelated to violent behavior, suggesting that ?violence is primarily interpersonal rather than intrapersonal? (Ball-Rokeach 1973:748). The decision to participate in violence becomes a product of all of the attitudes and values of those involved (Ball-Rokeach, 1973). Violent behavior may also be better determined by other factors, such as demographic and situational characteristics. In yet another challenge to a cultural interpretation of southern violence, Loftin and McDowall (2003) refute the work of Cohen and Nisbett (1994) and their associates pertaining to the culture of honor theory. In a replication of Cohen?s earlier study, Loftin 51 and McDowall (2003:354) find Cohen?s conclusions that social stability has a ?positive effect on argument-related homicide in the South and West, but a negative effect on felony-related homicides in the South and West as well as on both types of homicide in the North? to be quite misleading. They critique Cohen?s data and his model, arguing that the relationships he presents are heavily influenced by a few counties with ?unreliably high estimates of homicide rates? and a greater number of counties with zero levels of homicide (Loftin and McDowall, 2003:354). Loftin and McDowall?s (2003) results serve to negate all of Cohen?s predictions. Loftin and McDowall argue that Cohen?s regressions were affected by a few extreme cases because he dropped 60 percent of homicide cases that contained no information on age, gender, race, or ethnicity. Loftin and McDowall (2003) do not attempt to disprove the culture of honor theory, in fact, they offer that simply because Cohen?s research did not provide empirical support for the theory, it does not invalidate it. However, the authors do claim that ?because all of the research on homicide that is cited in support of the Nisbett-Cohen (1994) theory is based on methods that share these problems, our findings raise doubt about the whole line of research? (Loftin and McDowall, 2003:365). Another study of Loftin?s, undertaken with Chu and Rivera (2000), reconfirms his earlier work in refuting the culture of honor theory. Chu and her colleagues re-examining one test of the Nisbett-Reaves theory; that the areas where topography or precipitation limited agriculture to herding would continue to have the highest homicide rates (Chu, et al. 2000). Chu, Rivera, and Loftin (2000) replicate the Nisbett-Reaves test exactly in an attempt to provide a more complete description of the statistical distributions that they examined, and find no support for the Nisbett-Reaves hypothesis. Chu (2000) and her 52 colleagues control for differences in poverty among whites and find that there is no correlation between white male homicide rates and environmental regions. The researchers report that the differences in Nisbett?s findings are, ?artifacts of unreliable estimates of homicide rates, skewed distributions, and the failure to control for differences in the distribution of white poverty? (Chu, et al. 2000:982). It is important to note that although Chu, Rivera, and Loftin (2000) found no supporting evidence for the Nisbett-Reaves hypothesis, they are not necessarily implying that the herding-culture of honor theory is false. Dixon and Lizotte (1987) also support a structural interpretation of Southern violence, calling into question the empirical evidence surrounding the Southern subculture of violence thesis. Dixon and Lizotte (1987) examine the relationships among three phenomena: region, subcultures of violence, and gun ownership, all of which they predict to be related. The researchers discuss two explanations that have been proposed for the high rates of gun ownership and homicide in the South. The first concerns the subcultural paradigm found in sociology, which claims that regional differences in firearms ownership and homicide can be attributed to unique cultural patterns in the South (Dixon and Lizotte 1987). The second explanation of these regional differences is consistent with the structural paradigm, and argues that variations in homicide and gun ownership can be traced to other demographic and economic factors (Dixon and Lizotte 1987). The researchers set out to determine whether there is evidence of a ?Southern subculture of violence characterized by a set of violent values that operate across a variety of structural dimensions? (Dixon and Lizotte 1987:397). They also investigate 53 whether or not firearms play a central role in subcultures of violence. Dixon and Lizotte (1987) look at the direct effects of education, race, income, religion, and city size per region in relation to homicide, gun ownership, and violent attitudes. They create a more refined measure of the region variable that they speculate does not present the problems with multicollinearity that many past studies have experienced (Dixon and Lizotte 1987:400). The researchers introduced a direct measure of the subculture of violence that is independent of measures of region and gun ownership; consequently, they were able to ?unconfound? the effects that region, specific structural variables, and the subculture of violence have on gun ownership (Dixon and Lizotte 1987:401). Dixon and Lizotte?s (1987) findings are not consistent with the findings of previous, similar studies and they attribute this to their development and enhancement of the region variable. Their results cast doubt on the Southern subculture of violence thesis because they find that when structural factors are controlled for, the violent values indicative of membership in a subculture of violence do not correlate with region (Dixon and Lizotte 1987). Their findings also question the premise that gun ownership is a definitive characteristic of violent subcultures (Dixon and Lizotte 1987). Dixon and Lizotte (1987) find that the Southern subculture of violence thesis is inadequate for explaining high rates of gun ownership and homicide in the South. Finally, in support of a structural position, Jacobson (1975) examines the Southern subculture of violence thesis in the context of regional variations in serious criminal offenses other than homicide. Jacobson (1975) examines the extent to which reported crime offenses between the South and non-South continue to manifest regional 54 differences. His data generally support a structuralist interpretation of converging crime trends. Jacobson (1975:239) argues that the traditionally strong impact of a Southern subculture of violence is now negligible, and that, ?the historically observed pattern of higher crime rates in the South remains today only in vestigial form.? Jacobson (1975) predicts that the rapid socioeconomic development of the South in recent years will continue to diminish the level of significance in regional differences in homicide rates. The Integrated Approach There is more than a sufficient empirical basis in the literature to justify support and merit the future investigation of both the structural and cultural perspectives of Southern violence. The findings have been very inconsistent and variable from study to study concerning the Southern subculture of violence; consequently, some researchers have adopted a more integrated perspective. In more recent research on the Southern subculture of violence, investigators argue for an integration of cultural and structural variables in explaining regional differences in homicide and violent crime rates. Researchers have expanded the Southern subculture of violence thesis to include structural characteristics that they initially found no value in. There are complex theoretical issues involved in the integrated approach, which has become a sort of causal debate. The difficulty is that both sides have a logical argument; some suggest that poverty directly influences the emergence and development of a violent subculture, while others speculate that maintaining violent values or adhering to a violent subculture keeps individuals in poverty (Parker, 1989). Gastil (1975:108) expands this theoretical link by suggesting that ?even if it appeared that the apparent influence of Southerness on homicide rates could be explained by those common 55 sociological variables such as low education and income that are also associated with Southerness, one could imagine that Southern culture ?caused? them all.? So the question remains, which factor (Southern culture or Southern structural characteristics) has a causal influence upon the other and to what degree, or do both factors exerting a continuous influence upon each other? Proponents of the integrated perspective would agree with the latter statement, offering that the demographic characteristics of an area and the area?s culture are inextricably linked and therefore; do no warrant a separate investigation attempting to isolate which of the factors are to blame for higher levels of lethal violence. The integrated perspective demonstrates the need for further investigation into the logical link and potential interplay between structural and cultural factors. Parker (1989:986) advocates a more integrated approach that combines both structural and cultural variables, suggesting that ?these two theories are complimentary rather than contradictory.? Parker (1989) offers several logical assumptions that support a close connection between cultural and structural variables. First, Parker (1989) argues that in a simultaneous analysis of the two types of variables, poverty will always have the dominant effect; however, like many other researchers, he suspects that violent subcultural norms may actually cause one to be poor. The result is subculture as ?an indirect cause of homicide, operating via the direct cause of poverty? (Parker, 1989:986). Land, McCall, Cohen (1990:955) conceptualize culture as the result of poor or sub-par structural conditions, asserting that culture can be viewed as a ?response to social structural constraints and opportunities.? Another possibility that appears to be highly likely may be that poverty and subculture are highly correlated predictors of homicide, with overlapping but somewhat independent effects on rates of homicide (Parker, 1989). 56 Other researchers demonstrate the significance of an integrated theory by insisting that the concepts of honor and respect associated with the Southern subculture of violence become much more viably dangerous when combined with structural factors such as poverty and inequality that plague the South (Parker, 1989). Land, McCall and Cohen (1990) assume a more integrated approach by utilizing both structural and cultural variables in an attempt to resolve some of the variance and empirical inconsistencies within the research and literature on the Southern subculture of violence thesis in regards to structural covariates of homicide rates that change over time periods and levels of analysis. They attribute the inconsistencies to differences in research procedures alongside problems of data analysis and statistical inference (Land, McCall, Cohen, 1990). Land and his colleagues (1990:927) evaluate contradictory empirical findings that share the same covariates but yield vastly differing results. The covariates that are typically examined and most commonly used in research on southern violence include: population size; population density; percentage of the population that is black; percentage of the population ages 15-29; percentage of the population of males ages 15 and over that is divorced; percentage of children 18 years old or younger not living with both parents; median family income; percentage of families living below the official poverty line; the Gini index of family income inequality; the unemployment rate; and some type of regional variable indicating location in the South. By examining the empirical results in 21 studies that employ the above variables, Land, McCall and Cohen (1990) find that most of the covariates are either significantly positive or null, making it exceptionally difficult to distinguish any sort of pattern or support for either the structural (poverty) perspective, or the cultural (regional) perspective. 57 Land, McCall and Cohen (1990) offer a baseline regression model using the list of common explanatory variables mentioned above in order to facilitate comparisons of all levels of analysis and across different time periods. Land and his colleagues utilize three levels of analysis (cities, states and SMSAs), and three time periods (1960, 1970, and 1980). Their statistical results provide more consistency over time and space, but still retain some variance. One of the major differences cited between the summarized study and the Land, McCall, Cohen (1990) study is that unit population size and the regional (South) variable have statistically significant (p < 0.05 level), positive effects across all three time periods and for one level of analysis (cities); these variables are also positive for the other two levels of analysis, but were not statistically significant. Percent black and percentage of the male population that is divorced also exhibit consistent, positive, and statistically significant (p < 0.05) results across all three time periods and for two levels of analysis (cities and SMSAs), and positive, yet not statistically significant results for states across all time periods (Land, McCall, and Cohen 1990). Six of the structural covariates examined are relatively inconsistent over the specified time periods and units of analysis. Land, McCall and Cohen conclude that most prior research on the Southern subculture of violence thesis is riddled with statistical estimation problems stemming from multicollinearity. In other words, the structural covariates or predictors of homicide are highly correlated. In some cases, they are more intercorrelated with each other than with the homicide rate; this is a major problem that is omnipresent in research across all time periods and units of analysis. Land, McCall and Cohen?s (1990:951) innovative model has great implications for future research; ?perhaps the most important of these is 58 that the theoretical presumption of relatively invariant relationships across time periods and social space appears to be corroborated by use of large samples, standard definitions, and the reduction of collinearity among structural covariates.? The work of Williams and Flewelling (1988) also uses a combination of structural and cultural variables in describing the social production of criminal homicide. Williams and Flewelling (1988) focus their integrated analysis on disaggregating the total homicide rate into subtypes with conceptual meaning. Williams and Flewelling outline a theoretical model which they believe describes the social production of criminal homicide. They compile what are viewed as the dominant sources of homicide (social disintegration, resource deprivation, and violent cultural orientation) in previous studies for analysis as exogenous variables in their own research Williams and Flewelling, 1988). They advance a theory that presumes social disintegration, resource deprivation, and violent cultural orientation are not only related to one another, but also either positively or negatively related to the availability of effective social control and the intensity of interpersonal conflict, both of which affect the rate of criminal homicide. Williams and Flewelling (1988) hypothesize that there is a strong link between violent cultural orientations and resource deprivation to a greater intensity of interpersonal conflicts (conflict homicides) that are more likely to result in acts of homicide. In addition, they assume that the more social disintegration and resource deprivation there is in an area, the less available effective social controls become. Williams and Flewelling?s (1988) results suggest that stranger homicides occur more often in large cities, a finding which may account for some of the variability that exits across different levels of analysis in past studies. As expected, a higher level of 59 intimacy between victims and offenders equates to a greater likelihood that homicides are the result of conflict incidents. Robbery and other types of felony homicides occur more often when the victim and offender have no prior relationship. Williams and Flewelling also provide evidence that resource deprivation has the strongest impact on different types of criminal homicide that do other cultural and structural factors. The justifiable homicide ratio, their measure of culture, has a significant and positive effect on all types of conflict homicide; whereas percent black and location in the Confederate South did not. Their study offers an alternative to the highly debated regional variable for future research to develop more comprehensive measures of culture. Williams and Flewelling (1988:428) conclude that their results ?suggest that resource deprivation, violent cultural orientation, and social disintegration play an important role in producing city-to city variation of criminal homicide.? Similar to Williams and Flewelling (1988), Rice and Goldman (1994) attempt to discover if a subculture of violence does exist in the South by examining the type of homicide, the circumstances surrounding homicide, and the victim-offender relationship. They presume that both argument-related homicide and acquaintance homicide will be higher in the South than the non-South. Literature on regional violence has suggested that most Southern violence is violence for cause and that Southerners are more likely to commit murder in circumstances that involve arguments and close relatives than non- Southerners (see Smith and Parker, 1980; Reed, 1982; Ellison, 1991). Using Supplementary Homicide Report (SHR) data from the FBI?s Uniform Crime Reports (UCR), Rice and Goldman (1994:378) find support for their hypothesis that argument related homicide is higher in the South than any other region, with 58 60 percent of murders in the South in 1991 categorized as argument related and 43 percent in the non-South. Rice and Goldman (1994:379) then employ logistic regression, finding that acquaintance homicide is also significantly higher in the South than elsewhere, with 81.3 percent of Southern murders involving family, friends or acquaintances compared with 77.3 percent in the North Central region, 71.8 percent in the Northeast, and 69.0 percent in the West. Rice and Goldman (1994:379) suggest that the data they present provides ?alluring evidence that a subculture of violence may actually exist? in the South; however, they admit that their results do not offer conclusive proof. Messner?s 1983 study deals with an integration of structural and cultural variables which examine the relationships among region, racial composition, and the homicide rate for a sample of 204 SMSAs. The significance of Messner?s study is that while it is a replication of previous studies by Loftin and Hill (1974), Hackney (1969), and Gastil (1971), Messner (1983) reanalyzes the effects of racial composition and region on the homicide rate by refining the unit of analysis to examine SMSA?s rather than states, and also adds to his study the structural poverty index introduced by Loftin and Hill. Using two regional variables and five additional control variables, Messner (1983) finds that for SMSAs outside the South, racial composition is strongly related to the homicide rate, whereas there is no significant relationship between these variables in Southern SMSAs. The strong associations between the homicide rate for percent black and for the two regional variables are consistent with the hypothesis that blacks and Southerners adhere to a subculture of violence, which increases the likelihood of lethal behavior. An important conclusion to be drawn from this research is that the high levels of poverty characteristic of the South cannot alone account for the high levels of homicide 61 for Southern SMSA?a with high percentages of blacks; therefore, Messner?s research provides evidence in support of the thesis of a Southern subculture of violence. Messner (1983) demonstrates that both the SPI, used by Loftin and Hill, and the Southernness index used by Hackney and Gastil are powerful predictors of homicide rates in SMSA?s. Another study conducted by Messner in 1983, involves a macro-structural theory that argues that economic discrimination against social groups is a significant predictor and determinant of national homicide levels. Messner?s (1983) results indicate that the effects of economic discrimination on national rates of homicide withstand controls for a variety of national demographics and characteristics such as political structure, standard of living, population size, and age-sex structure. However, his research and its findings are self-admittingly full of severe problems of multicollinearity, crude measures of economic discrimination and a small sample size. Huff-Corzine, Corzine and Moore (1986) attempt to resolve some of the questions that arise from the debate between the two competing theories on the Southern subculture of violence by analyzing state homicide rates for total populations in order to determine the effect of cultural and structural factors on Southern violence. Huff-Corzine and her colleagues identify the major shortcomings of past studies and attempt to rectify them by replicating Loftin and Hill?s (1974) research using updated data and incorporating variables and measures from other studies as well. The researchers extend Gastil?s (1971) work on developing an appropriate Southernness index by using the proportion of a state?s population born in the South as a measurement of Southerness. Their results confirm Gastil?s (1971) Southern index as a more consistent predictor of homicide rates than the percentage of those born in the South. 62 One finding that distinguishes Huff-Corzine and her colleagues? research from others in the field is that most previous studies have found poverty to be positively associated with the homicide rate, whereas their analysis discovers that the effect of poverty is restricted to the white population (Huff-Corzine, et al. 1986:917). Huff- Corzine (1986) and her colleagues insist that the regional effect on homicide levels cannot be ?explained away? by socioeconomic and demographic characteristics of the south alone. Their findings lend support to arguments that the high homicide rates of the South are tied to cultural differences and the existence of a violent subculture (Huff- Corzine, et al., 1986). Huff-Corzine, Corzine and Moore?s findings imply that the fluctuation of the influence of poverty on lethal violence rates in past studies may be a product of racial composition rather than a regional effect. Huff-Corzine, Corzine, and Moore (1991) continue their line of inquiry in a more recent study that investigates the relationship among suicide, homicide, and region, finding that both high rates of poverty and cultural differences influence levels of lethal violence in the South. Consistent with past literature, their study reaffirms the importance of disaggregating total homicide rates by race and using the SI and percent born in the South as a more accurate measure of Southerness. Their results indicate that the variables which affect levels of homicide and suicide differ by race. Huff-Corzine, Corzine, and Moore (1991) use a structural poverty index similar to that of Loftin and Hill (1974), and find that it has a significant effect on white homicide rates, but no such effect on black homicide rates. Both of their measures of Southerness produce an effect on levels of white homicide; however only the SI has an impact on levels of black homicide. 63 Huff-Corzine, Corzine, and Moore (1991:718) express ?the need to expand research on regional differences in lethal violence beyond a narrow focus on homicide rates,? making their own such contribution to research on Southern violence by examining data on suicide and homicide, the lethal violence rate, and the suicide- homicide ratio. Similar to previous research, Huff-Corzine, Corzine and Moore find that the South has a pattern of high homicide rates and low suicide rates. Huff-Corzine (1991) and her colleagues argue that for both races, structural characteristics have a more powerful influence on the lethal violence rate than do regional measures. In one particularly important finding, the lethal violence rate is not significantly related to the percentage of the population who are Southern born. Similar to Loftin and Hill (1974) among many other structuralists, Huff-Corzine, Corzine and Moore (1991) find severe poverty to be the strongest indicator of total rates of lethal violence for whites. Huff-Corzine, Corzine, and Moore (1991:723) insist that their research does not negate the importance of cultural factors in predicting variation in regional homicide rates suggesting that, ?it may be that cultural differences and regional socialization patterns exert an indirect effect on lethal violence rates.? Much like Hackney (1969), they rely on attribution theory to provide an explanation of southern violence offering that ?the most fruitful avenue will probably be one that seeks to identify and trace the development of a Southern world view that defines the social, political, and physical environment as hostile and that casts the white Southerner in the role of the passive victim of malevolent forces? (Huff-Corzine, et. al., 1991:725). Huff-Corzine, Corzine, and Moore (1991) conclude that future research should focus on Southern attitudes toward violence in an effort to determine if attributional processes differ by region. 64 Unnithan and his colleagues examine the relationship between homicide, suicide and region. They construct and propose a new model that includes an analysis of statewide Lethal Violence Rates (LVRs) and Suicide-Homicide Ratios (SHRs) for both blacks and whites. Their results indicate that both cultural differences and high levels of poverty are to blame for higher levels of lethal violence in the South. They make an important contribution to the field of Southern violence by extending their research beyond the frequent, isolated focus of homicide rates to include an investigation of both cultural and structural factors as potential predictors of the LVR and SHR for both races. Past research has typically focused on the influences that the Southern region has on the sheer volume or amount of homicide, but these researchers advocate a profound shift in this approach, suggesting that an examination of the ?Southern influence on the direction of violence? will prove more valuable (Unnithan, N. Prabha, Lin-Huff Corzine, Jay Corzine, and Hugh P. Whitt, 1994:147). Prior research has consistently provided evidence that violence in the South is characterized by high rates of homicide and low rates of suicide (see Porterfield, 1949; Hackney 1969; Unnithan, et. al, 1994). Unnithan (1994:148) and his colleagues find it plausible that Southerness, rather than producing high rates of homicide, may instead change ?the mix of lethal violence in favor of homicide over suicide.? Consistent with prior literature on Southern violence, they hypothesize that attribution theory, specifically, the attribution of blame, is a mitigating factor that has an affect on the relationship between Southern region and violence. They link the attributional processes and patterns specific to Protestant Fundamentalism, more common in the South than 65 elsewhere, to the higher rates of homicide and lower rates of suicide that characterize the region. They also view economic deprivation as a driving force behind increases in both homicide and suicide. Using OLS and ridge regression analysis, they find that the level and type of lethal violence differ by race, with the LVR for blacks more than doubling the white LVR and the SHR much higher among whites (Unnithan, et. al., 1994). Their results also show Loftin and Hill?s SPI to be the strongest predictor of variance in levels of lethal violence for whites. Their Southern regional variable, the percent of the population born in the South, is not significantly associated with the lethal violence rate for whites or blacks. Although their data provides evidence that structural variables exert a more powerful influence on lethal violence rates than regional measure do, they suggest ?that cultural differences and regional socialization patterns exert an indirect effect on lethal violence rates.? (Unnithan, et. al., 1994:154). Their results show a significant, negative relationship between their regional measure and the SHR for both races, indicating that Southerners are more likely than non-Southerners ?to express lethal violence as homicide.? (Unnithan, et. al., 1994:156). The SPI for whites shows a significant, positive association to the SHR among whites; however lacks a significant relationship to the SHR among blacks. One important theoretical finding in their research is that a Southern regional influence ?channels violence toward other persons rather than the self for both races? (Unnithan, et. al., 1994:158). They suggest that Southerners are more ?inclined toward external attributions,? which results in a ?propensity toward homicide,? reflected in the 66 higher levels of homicide and lower levels of suicide found in the region (Unnithan, et. al., 1994:158). The above implications must be interpreted cautiously because there are no studies that examine regional differences in attributions of blame, and the researchers treated attributional tendencies ?as an unmeasured, intervening variable? in their study (Unnithan, et. al., 1994:149). Their integrated model does however; provide a better explanation of Southern violence by incorporating both cultural and structural factors and illuminating the substantial impact that Southernness has on the direction, rather than the volume of lethal violence. The key to a full understanding of regional disparities in homicide rates lies in a thorough examination of both structural and cultural variables and their potential interplay in effecting rates of criminal violence. Regional structural characteristics as well as regional culture as predictors or influences on violence can be better understood from an integrated perspective. An integrated theory is a causal and circular theory; without one, the other could not exist. The two main theories concerning regional variations in homicide rates are often viewed as dichotomous (cultural or structural); rather, they should be seen as interrelated and relatively valid (Corcoran, 1995). It makes more sense that the causes of poverty would involve many interconnected variables that make up the structural environment, cultural processes and behavioral outcomes. An integrated theory of the causal factors involved in violence is an interesting research topic that could one day allow the exploration of the connection between micro and macro processes (Corcoran, 1995). An integrated perspective will also help to generate a more comprehensive understanding of the influence that both cultural and structural factors exert upon regional variations in homicide rates. 67 Present Study The present study seeks to evaluate the Southern subculture of violence thesis by explaining regional variations in homicide and aggravated assault rates from a more integrated approach which assumes that both cultural and structural variables will exert a divergent influence on these rates from the South to the non-South. This study will employ OLS regression in an effort to determine which cultural and structural variables are the strongest predictors of variance in regional rates of lethal and non-lethal violence. The inclusion of aggravated assault, a measure of non-lethal violence, as a dependent variable makes the present study particularly significant. The majority of research on the Southern subculture of violence relies heavily on homicide rates as the sole dependent variable. Past research on the southern subculture of violence has been too narrowly defined by focusing only on the most extreme form of violence: homicide. This study extends past research by using a more inclusive construct/measure of violence. Aggravated assault is included in the analysis as a dependent variable with the underlying contention that homicides are aggravated assaults with only a different outcome: the victim?s death. The present study views homicide and aggravated assault as analogous crimes on a continuum of violence. The present study is also significant because it extends research that links a subculture of violence with Southern religion, by using a religious denomination (conservative Protestant) as a variable to measure Southern culture. Consistent with the literature, the present study also hopes to be able to reconfirm the value in the use of both Gastil?s Southerness Index (SI) and the pioneering work of Land, McCall and Cohen 68 (1990) in developing a widely used resource deprivation/affluence component that reduces multicollinearity. Hypotheses 1. Rates of homicide and aggravated assault will increase as the level of poverty (RDAC component variable) increases. 2. Rates of homicide and aggravated assault will increase as the level of Conservative Protestant affiliation increases. 3. Rates of homicide and aggravated assault will increase with location in the South. 4. Homicide rates and aggravated assault rates will increase as the percentage of Blacks/African Americans increases. 5. Rates of homicide and aggravated assault will share the same predictors and relative pattern of commission, irrespective of region. 69 METHODOLOGY Data and Methods The dataset is available through statistics provided by the U.S. Bureau of the Census and consists of all counties (3109) in the lower 48 United States. Data on Alaska and Hawaii have not been included in the present study. County level data is not available on Alaska because the state is not separated into counties on a consistent basis and the counties and boundaries are constantly changing. Hawaii is culturally distinct from the rest of the U.S. and may therefore comprise its own region; also, any exclusion of the state will not severely affect the results because Hawaii only contains five counties. Research on Southern violence boasts the use of many different units or levels of analysis. Conjecture and ambiguity remain in the literature over which unit of analysis may be the most appropriate or preferred; however, empirical support exists for all levels of analysis, as well as against all levels. County level data has a clear advantage over state level data in the present study and other research that attempts to measure aspects of culture. County level data is essential because it does not presume that culture abruptly or conveniently ends at state boundaries. County level data is the only unit of analysis that can penetrate state lines in order to more accurately reflect regional culture. The use of county level data also allows for a larger sample size, covering almost the entire population of the United States. Another benefit of choosing counties as the unit of 70 analysis is that it permits both rural and urban areas to be examined, whereas a study that employs SMSAs as the unit of analysis does not. County level data for the homicide rates of each state was obtained from the FBI?s annual Uniform Crime Reports for the years 1999, 2000 and 2001. County level data for rates of aggravated assault is available in the serious offenses section of the Uniform Crime Reports. Next, a three-year average for both homicide and aggravated assault rates was computed. Dependent Variables One of the dependent variables for the analysis is the official homicide rate, or incidence of murder in a county, collected from the FBI's annual Uniform Crime Reports. The homicide rate is data that refers to the number of offenses per 100,000 populations known to the police. Since there is random fluctuation in the annual homicide totals, it is best to establish a rate over a multi-year period to provide a better representation of the true level of lethal violence (Huff-Corzine, et al. 1991). A three year average of U.S. county homicide rates was obtained using the years 1999, 2000, and 2001. Both dependent variables were transformed into natural logarithm form to reduce the effects of counties with extreme values either high or low. Aggravated assault is the other dependent variable; it has been measured and expressed in a manner similar to that of the homicide rate discussed above. According to the Bureau of Justice Statistics (2000), assault is defined as ?an unlawful physical attack or threat of attack.? An aggravated assault is classified as such by the presence of a weapon and how severe the injuries are; aggravated assaults include attempted murders according to the Uniform Crime Reports (FBI, 2000). Although homicide data can be 71 disaggregated by race, data on aggravated assault is not available by race or broken down into categories according to race. In an effort to remain consistent, neither of the dependent variables has been disaggregated by race. Independent Variables The present study utilizes One Least Squares Regression to analyze the effects of the following five predictor variables on rates of both homicide and aggravated assault. Population Structure Component The present study uses a ?population structure component? variable, initially introduced by Land, McCall and Cohen (1990:943). This component is comprised of the unit (county) population size (land per square mile) and population density (number of people per county) variables (Land, et. al., 1990). There is clear support for the use of such a component based on the findings of Land and his colleagues which indicates that unit population size and population density are one of ?two clusters of variables that consistently hang together over all levels of analysis and time periods? (Land, et. al., 1990:942). The present study includes the percentage black as a component of the population index, whereas Land and his colleagues did not. Resource Deprivation/Affluence Component (RDAC index) The other cluster of variables is referred to as a ?resource deprivation/affluence component? (see Williams and Flewelling, 1988; Land, et. al., 1990). This component is comprised of three variables that measure income (median family income, the percentage of families living below the official poverty line, and the Gini index of family income inequality) and one more social indicator (the percentage of children age 18 or under not living with both parents) (Land, et. al., 1990). These variables all pertain to a given 72 population?s composition of human resources, where the percentage of families living below the official poverty line is indicative of absolute deprivation and the Gini index is representative of relative deprivation (Land, et. al., 1990). Prior research has indicated that these variables are highly correlated; although they may be conceptually and operationally different, they remain empirically similar (Land et. al., 1990). The use of Land, McCall and Cohen?s resource deprivation/affluence component will aid in reducing inconsistencies that result from this type of collinearity. The amount of poverty per county is the average of the standardized scores for the four indicators. Percent Black Race is a complicated and controversial topic when it comes to research on homicide rates, particularly because it is a ?strong sociodemographic correlate of homicide? (Smith and Zahn, 1999:31). Some researchers attempt to correct the problems race poses as a variable by investigating race-specific homicide rates. Even with this improvement in methodology, the literature remains unclear and research continues to produce contradictory results. Other researchers choose to disaggregate data on homicide by race (see Huff-Corzine, Corzine and Moore, 1986; Corzine and Corzine, 1994). Past research has shown that the predictors and influences on homicide rates differ for blacks and whites. Although homicide data is available by race, aggravated assault data is not and in an effort to remain consistent, none of the data in the present study has been disaggregated by race. The present study diverges from the Land, McCall and Cohen (1990) study, by adding the percent black per county into the population index, rather than including it with the resource deprivation/affluence component. This research hopes to avoid the risk 73 of the percentage black variable being highly correlated with the indicators of poverty in the component variable and therefore, places percent black per county in the composite population index variable. When included in the population index, percent black is utilized as a statistical control for race in this study. Data for racial composition was obtained through the 2000 Census and includes the percentage of Blacks/African Americans in each county. Southerness Index As well as race, defining the South is an inherently complicated and controversial task, and many ambiguities remain in the literature. For many years, scholars from various genres have engaged in a rancorous debate when it comes to identifying the borders of the cultural regions of the U.S. There is frequent use of the Confederate South and the census South throughout prior research; however neither of these methods matches what most people imagine when they think of Southern culture, or delineates where Southern culture begins or ends. Smith and Zahn (1999:49) illuminate the difficulties: ?Some states that are generally considered culturally Southern (Kentucky, Oklahoma, and perhaps West Virginia) lie outside the Confederate South; conversely, the census South includes Delaware and Maryland, states that arguably have more in common with the Northeast than with Mississippi and Alabama. Florida, which is in but perhaps not of the South, presents a problem for both operalizations.? In order to refine these methodological problems, some researchers use the percentage of the population born in the South or Gastil?s Southerness index. However, Smith and Zahn (1999) insist that none of the operationalizations of the South that have been used in past research are ideal. It is impossible to precisely outline or create a 74 standard South because culture permeates city, county and state lines. Many researchers, including Smith and Zahn (1999) prefer the use of Gastil?s model, which allows investigators ?to draw cultural lines along the county lines where they are ultimately most useful? (Gastil, 1971:425). Gastil?s approach extends Southern culture into parts of the Border States and is more appropriate for county-level data. Gastil?s division of the country into 13 cultural regions is particularly useful for research on regional disparities in levels of violence, among many other topics. Gastil concentrates on the homogeneity of areas in drawing distinct regional boundaries, which reflects the differentiation of people into groups whom have identifiable cultural differences (Gastil, 1975). Regional borders are delineated ?where there are significant discontinuities? in cultural factors (Gastil, 1975:26). Gastil?s (1975:26) hypothesis supports a separation of the U.S. into definitive cultural regions ?as defined primarily by variations in the cultures of the peoples that dominated the first settlement and the cultural traits developed by these people in the formative period (where these are significant), and secondarily by variations in the cultures of peoples that dominated later settlements, as well as cultural traits developed subsequently.? This hypothesis has significant implications because it extends the idea of culture beyond its origins into contemporary cultural developments and emergent regional differences. Southernness is defined as the degree to which a person is socialized in the south (Hackney, 1969). The present study will utilize the Southerness Index developed by Gastil (1971) which is based on the degree to which a state was initially settled by Southerners. This index is used more frequently than any other regional indicator in research on the Southern subculture of violence and researchers have provided evidence 75 that Gastil?s SI and the percent of the current population born in the South are correlated at or above the .90 level (see Huff-Corzine, et. al., 1986). Gastil?s index remains a reliable measure of Southern boundaries and other cultural regions because its values, which rely on the initial settlement of states, do not change and can be confirmed through the Bureau of the Census; ?it would be difficult to reverse the relationship of a state to its neighbors or change greatly the extremes? (Gastil, 1975:110). The present study separates the U.S. into 8 regions using Gastil?s Southerness index which is in an integer format. The present study?s categorization departs only minimally from Gastil?s cultural regions: Gastil uses 11 regions (with the exclusion of Alaska and Hawaii taken into consideration), and this research combines his Pacific Southwest and Pacific Northwest into one region labeled the West. The present study also combines the New York Metropolitan region with the Pennsylvanian Region and this merged region is labeled the Mid-Atlantic region. Lastly, Gastil?s Rocky Mountain region is combined with his Mormon region to form one region. Gastil?s original regional border divisions were followed as closely as possible in combining regions. Percent Conservative Protestant Data on the conservative Protestant variable is available in the data-book for Religious Congregations and Membership in the United States, 2000 (Jones, Doty, Grammich, Horsch, Houseal, Lynn, Marcum, Sanchagrin, and Taylor, 2002). Religion is used in the present study as a measure of Southern regional culture in order to test the thesis of a Southern subculture of violence. The denominational bodies included in the data that comprise the conservative Protestant variable are a replication of original work by Roof and McKinney (1987). Roof and Mckinney?s scheme groups together 76 denominational bodies that adhere to the basic principles, morals and theology of Conservative Protestantism. Specifically, Roof and McKinney (1987) outline two requirements that individual denominations must possess in order to be considered adherents or members of the Conservative Protestant faith: a conversion experience and literacy of the bible. Denominations that do not meet these two basic requirements are not included as components of the conservative Protestant variable in the present study. There is precedent in the literature on Southern violence and religion for the use of Roof and McKinney?s religious grouping format (see Ellison, 1991; Ellison, Burr and Mcall, 2003). The variable ?conservative Protestant? in this study includes several denominations that comprise the following general religions: 1) Southern Baptists; 2) Churches of Christ; 3) Evangelicals/Fundamentalists; 4) Nazarenes, 5) Pentecostals/Holiness; 6) Assemblies of God; 7) Churches of God; and 8) Adventists (Roof and McKinney, 1987:135). Data for the conservative Protestant variable is the percentage of those affiliated with the denominational bodies mentioned above. The present study uses Ellison?s (1991:1226) definition of conservative Protestant as ?a member of a Western Christian church whose faith and practice are founded on the principles of the Reformation, especially in the acceptance of the Bible as the sole source of revelation, in justification by faith alone, and in the universal priesthood of all the believers.? Percent Lethal This variable is obtained by dividing the homicide rate by the sum of the homicide rate and aggravated assault rate for each county and then multiplying the result by 100 to give a percentage. The percentage is the percent of people who died in violent 77 confrontations either by homicide or aggravated assault. The lethality percentage shows the percent of violent events involving homicide and aggravated assault that result in the victim?s death. 78 RESULTS Descriptive data for the two dependent variables (three-year averages of county murder and aggravated assault rates), as well as the eight predictor variables (percent black, percent conservative protestant, percent living below the poverty line, the gini index, percent female head of household, median family income, total population and population density) from the 2000 U.S. census are presented in Table 1. Descriptive data are also examined in terms of the eight regions as constructed by Gastil (1975) and closely reproduced in this study. (Insert Table 1 here) A 10 X 10 correlation matrix was computed to examine the degree of association between the predictor variables, regional variables and the two dependent variables. As shown in Table 6, the murder rate has a strong, positive correlation with the aggravated assault rate (r = 0.75, p < 0.01) at the county level. Further, population density (r = 0.21, p < 0.01); percent Black (r = 0.31, p < 0.01); percent conservative protestant (r = -0.06, p < 0.01); the Gini index (r = 0.075, p < 0.01); percent female head of household (r = 0.36, p < 0.01); the percent of lethal events (r = 0.21, p < 0.01); and median family income (r = 0.27, p < 0.01) are significantly related to county level murder rates. Contrary to predictions, the percent living below the poverty line (r = -0.025, p = 0.16) is not significantly correlated with county level homicide rates. 79 A similar pattern of relations is observed between the predictor variables and county level aggravated assault rates. All predictor variables are significantly related to the aggravated assault rates with the exception of the Gini index of income inequality (r = 0.015, p = .399). For the regional variables, four regions show a significant, positive correlation with county level murder rates: the New England region (r = 0.059, p < 0.01); the Mid-Atlantic region (r = 0.185, p < 0.01); the West (r = 0.084, p < 0.01); and the South (r = 0.187, p < 0.01). The four remaining regions exhibit negative relationships with county level murder rates, with only three regions showing significant correlations: the central Mid-West (r = -0.20, p < 0.01); the upper Mid-West (r = -0.119, p < 0.01); the Rocky mountain region (r = -0.19, p < .0.01); while the interior Southwest is not significant (r = -0.005, p < 0.01). All of the regions are significantly correlated with county level aggravated assault rates. The direction of the relationships between each region and aggravated assault remain the same as they were for murder rates with the exception of the interior Southwest, which becomes positive (r = 0.039, p < 0.01). As demonstrated in Table 6, significant intercorrelations are observed between several of the predictor variables: the percent living in poverty; the Gini index of income inequality; the percent of female headed households; and median family income. To account for the significant intercorrelations between these variables and to reduce multicollinearity among the predictors a composite variable was created based on the recommendations of Land, McCall and Cohen (1990). Principal components analysis ?produces substantial invariance or greater stability among the structural covariates offered in homicide studies? (Parker and Pruitt, 2000: 1489). 80 Specifically, a principal components analysis was conducted on these four variables and a one factor solution which accounted for 66.08% of the total variance was obtained. A component score was calculated using the variable weights obtained from the principal components analysis. This component is referred to as the Resource Deprivation/Affluence Component (RDAC). Table 2 displays significant intercorrelations between other variables as well, which include the total population, the population density, and the percent black. Principle components analysis was also conducted on these variables, obtaining 64.21% of the total variance when the percent black is included in the analysis and 92.74% of the total variance when percent black is not included. This study incorporates four One Least Squares (OLS) regression models. An examination of the collinearity diagnostics for each model indicated that the degree of multicollinearity among the set of independent variables was within acceptable limits. Two of the OLS regression analyses were conducted to examine predictors of county level murder rates as well as the influence of the regional variable Southerness. The remaining two regression analyses examine predictors of county level aggravated assault rates in addition to the effects of Southerness on these same rates. In all four models, the eight regions have been dummy coded and the South left out of the regressions for the purposes of comparing the South to all other regions. In the first analysis (model 1), four predictor variables (the population index with percent black included, the RDAC index, the percent conservative Protestant, and the percent of lethal events), and all regional variables except the South were simultaneously entered with murder rate as the dependent variable. The only change in the second analysis (model 2), is that the percent black is extracted from the population index and 81 not included as a variable anywhere in the regression. The results of these two models can be compared and contrasted as to the influence of the variable percent black as a statistical control. Model three utilizes aggravated assault as the dependent variable, all four predictor variables with percent of the population that is black included and all regional variables except the South. The only change in the final OLS regression (model 4) is that percent black is not controlled for anywhere in the analysis. Regression Analyses Predicting County Level Murder Rates. As displayed in Table 2, results of the first analysis suggest that the set of four predictor variables and seven regional variables significantly predicted 66.4% of the total variance in county level murder rates (R 2 = 0.664, F= 503.88, p < 0.05). As shown in Table 2, each of the four predictor variables are positively associated with increased murder rates. Further, the coefficient matrix reveals that the population index with percent black included (? = 0.808, p < 0.05) is the strongest predictor of county level murder rates relative to all other predictor variables, followed by the percent of violent events that are lethal (? = 0.265, p < 0.05), the RDAC index (? = 0.071, p < 0.05), and the percent conservative Protestant (? = 0.014, p = ns.). Interestingly, in Model 1, the percent conservative Protestant was not a statistically significant predictor of homicide rates when entered simultaneously with the other predictor variables. (Insert Table 2 here) The regional variables indicate that four of the regions have significantly lower murder rates than the South: the New England region (? = -0.035, p < 0.05); the Mid- Atlantic (? = -0.033, p < 0.05); the central Mid-West (? = -0.118, p = < 0.05); and the upper Mid-West (? = -0.048, p < 0.05). The three remaining regions, the Rocky Mountain 82 region (? = 0.070, p < 0.05), the interior Southwest (? = 0.080, p < 0.05), and the West (? = 0.090, p < 0.05), all have significantly higher homicide rates than that of the South. Table 3 demonstrates the results of the second OLS regression model which indicate that the model is statistically significant and that without controlling for the percent of the population that is black, the set of predictor variables and regional variables significantly account for 66.8% of the variance in county level murder rates in the model (R 2 = 0.668, F = 512.12, p < 0.05). Each of the four predictor variables has a positive association with county level homicide rates. The population index without percent black included, significantly accounts for the most variance among the independent variables in this model (? = 0.812, p < 0.05), followed by the percent of lethal events (? = 0.277, p < 0.05), the RDAC index (? = 0.158, p < 0.05) and finally, the percent conservative Protestant (? = 0.015, p = ns.). (Insert Table 3 here) In the second analysis, there are four regions that display significantly lower homicide rates at the county level than that of the south: New England (? = -0.052, p < 0.05), the Mid-Atlantic region (? = -0.035, p < 0.05), the Central Midwest (? = -0.142, p < 0.05), and the upper Midwest (? = -0.071, p < 0.05). Table 4 shows that the Rocky Mountain region (? = 0.411, p < 0.05), the interior Southwest (? = 0.463, p < 0.05), and the West (? = 0.653, p < 0.05) all have significantly higher murder rates than the Southern region. Regression Analyses Predicting County Level Aggravated Assault Rates The third model, as displayed in Table 4, indicates that the four predictor variables and seven regional variables significantly account for 80.1 percent of the 83 variance in rates of aggravated assault at the county level when race is not controlled for (R 2 = 0.801, F = 771.969, p < 0.05).) Two of the independent variables, the percent conservative Protestant and the percent of violent events that are lethal, share a negative relationship with aggravated assault rates, while the other two predictor variables, the RDAC index and the population index without percent black included, have a positive association to county level aggravated assault rates. The strongest significant predictor of aggravated assault rates in the model is the composite population index (? = 0.904, p < 0.05), followed next by the RDAC index (? = 0.167, p < 0.05), the percent lethal (? = - 0.140, p < 0.05). The percent conservative Protestant displays the weakest relationship with the dependent variable (? = -0.017, p = ns.), and the association is not statistically significant. (Insert Table 4 here) Four regions show significantly lower rates of aggravated assault at the county level than those of the South: New England (? = -0.035, p < 0.05); the Mid-Atlantic (? = - 0.047, p < 0.05); the central Mid-West (? = -0.121, p < 0.05); and the upper Mid-West (? = -0.075, p < 0.05). Results of the regression suggest that aggravated assault rates in the Rocky mountain region (? = 0.075, p < 0.05), the interior Southwest (? = 0.069, p < 0.05), and the West (? = 0.083, p < 0.05) are significantly higher than aggravated assault rates in the South for the years 1999 through 2001. In the final regression, shown in table 5, the percent black was included in the population index composite variable. Model 4 was significant and accounted for 78.7% of the variance in county level aggravated assault rates, (R 2 = 0.787, F = 941.524, p < 0.05). The percent conservative Protestant and the percent of violent events that are lethal 84 are negatively associated with aggravated assault, while both the RDAC and population indices are positively associated with the dependent variable. The population index with percent black included (? = 0.892, p < 0.05) emerges as the most powerful predictor of county level aggravated assault rates, followed by the percent of lethal events (? = -0.155, p < 0.05) and the RDAC index (? = 0.068, p < 0.05). The regional cultural variable for the South, conservative Protestant, has a weak and non-significant relationship with aggravated assault rates at the county level (? = - 0.020, p = ns.). (Insert Table 5 here) The results of the final regression analysis indicate that New England (? = -0.017, p < 0.05), the Mid-Atlantic (? = -0.044, p < 0.05), the central Midwest (? = -0.097, p < 0.05), and the upper Midwest (? = -0.052, p < 0.05), all have significantly lower aggravated assault rates per county than the South. County level rates of aggravated assault in the Rocky mountain region (? = 0.097, p < 0.05), the interior Southwest (? = 0.105, p < 0.05) and the West (? = 0.112, p < 0.05), are significantly higher than those of the South. Discussion The results of the regression analyses indicate that the independent variables associated with this study explain a good portion of the variance in county level rates of both homicide and aggravated assault. The findings are also supportive of a more structural interpretation of southern violence, suggesting that structural characteristics such as population structure and various forms of poverty bear a heavier weight upon rates of violence than do cultural factors. These results using 2000 Census data serve to confirm the earlier work of Land, McCall and Cohen (1990), whose findings indicated 85 that the influences of Southern culture on rates of homicide were consistently diminishing over time. The results of all four regression models indicate that structural variables are far more influential on rates of southern violence than is the measure of southern culture utilized in this study. The religious cultural variable (percent conservative Protestant) emerges in the direction originally predicted, yet is not statistically significant in any of the models. This shift from prior research that supports a cultural interpretation may be a product of using the most recent 2000, Census data that was not previously available. It supports the notion of religious growth and change in the South and the fact that some conservative Protestant denominations are growing faster outside of the South rather than within it (Roof and McKinney, 1987). It also supports Ellison?s (1991) work, which is consistent with suggestions in the literature that regional distinctiveness in attitudes toward violence and differences in regional rates of violence are declining as the cultural lag between the South and other regions disappears. Although the directional changes in the conservative Protestant variable between the homicide models and the aggravated assault models are not statistically significant, they still retain substantive value. In both homicide models, conservative Protestantism positively influenced the murder rate. In the aggravated assault models, the direction changes and the higher the percentage of conservative Protestants that reside in a county, the lower the aggravated assault rate. This could indicate that Southern religious culture exerts some influence upon the lethality of a violent event, changing the direction and type of violence as suggested by Unnithan (1994) and his colleagues. This may be a 86 fruitful avenue for future research to explore, however; this study could not test such a hypothesis due to the non-significance of the variable. The regression models in this study show that culture does not appear to have much effect on rates of violence once structural factors are controlled for. In support of Loftin and Hill?s (1974) work, the Southerness variable does not emerge as a significant predictor of rates of homicide or aggravated assault when structural variables are taken into account. One notable finding in this study that differs from prior research is that regardless of whether percent black is included or excluded, the population index appears to be the strongest predictor of rates of violence relative to all other variables used in each model. Past researchers such as Blau and Blau (1982), Parker (1989), among many other structuralists have consistently found poverty to be the strongest predictor of homicide rates. This study finds that in all models, population composition is the most influential. In models two and three, there are no controls for race and the strength of the relationships between the predictor variables and county level homicide rates and aggravated assault rates increase. Although poverty is not the most influential variable, the results of all models confirm earlier predictions that both rates of aggravated assault and homicide will increase as the level of poverty increases. In agreement with hypothesis five, aggravated assault and homicide appear to share the same predictors in relatively the same order in each model. The strength of the relationships between poverty (RDAC index) and rates of violence become weaker in the two models that control for the percent of the population that is black. This confirms the need to control for race in research concerning southern violence. Although including percent black does not dramatically change the amount of 87 variance explained rates of homicide or aggravated assault across the models, it does however, weaken the effects or influences of other predictor variables. In model 4, the percent of violent events that are lethal is the second most influential of the variables on aggravated assault rates and shares a logically negative association with the dependent variable; as aggravated assault rates increases, the percent of events that are lethal decrease. The converse is true of the two homicide models; as county level murder rates increase, the percent of events that are lethal increase also. In each of the models, three of the more western U.S. regions display significantly higher county level rates of aggravated assault and homicide than those of the South. This finding, which uses 2000 Census data, lends support to more recent researchers such as Parker and Pruitt (2000), who suggest that violent crime rates in the West are converging with, if not surpassing those of the South. When percent black is not controlled for in this study, the regional differences in rates of violence in these three regions move closer to rates in the South. These results also reinforce Parker and Pruitt?s (2000) insistence on the importance of disaggregating data by race and the supposition that black migration patterns may have displaced the Southern subculture of violence into other regions as apparent here: into the interior Southwest, the Rocky mountain region, and the far West. The results here further serve to confirm Jacobson?s (1975) prediction that in years to come, the level of significance in regional differentials in homicide rates between the South and the rest of the country will continue to diminish. 88 Limitations and Future Research The problem with the research and literature on the southern subculture of violence thesis is that there is much confusion over how to ?construct an appropriate empirical test of its core arguments? (Ellison, 1991). It is probable that the use of percent conservative Protestant as a measure of culture is not a broad enough conceptualization of culture; however, this research finds it important to evaluate the role, if any, southern religion may play in influencing rates of violence. Future research should focus on compiling a variety of more inclusive measures of southern culture to more accurately test the thesis of a subculture of violence in this region. Perhaps combining several researchers? methods would be more appropriate, for instance, Reed?s use of the General Social Survey to ascertain attitudes and opinions toward different forms of violence coupled with Ellison?s use of the conservative Protestant factor and possibly Ball- Rokeach?s measure of violent values by surveying prison inmates versus civilians. Several measures of culture in a future study, such as those mentioned above, would allow for a wider range of cultural and situational variables and ultimately enhance the validity of the results. Another difficult task is in linking region, socialization processes, individual values and violence because research on crime rates typically requires the use of aggregate data (Ellison, 1991). This is a substantial problem in this study: the use of aggregate data which was not disaggregated by race. By developing four regression models, this research attempts to show the differences and changes that are affected in the models when percent black is included as a control variable. In any case, this does not solve the problem of using aggregate data. Not disaggregating the original data by race 89 can mask regional and racial differences and stifle the effects of a cultural variable. Future research should focus efforts on obtaining crime data that can be disaggregated by race. In this study, this problem is also a product of using 2000 Census data as it was in the process of being released to the public; therefore some of the data was not yet available in a disaggregated format. In any examination of Southern violence, the configuration of the South and other regions becomes quite problematic, particularly because there are no standardized or agreed upon methods or measures of ?Southerness.? Gastil?s (1975) Southerness index proved to be a rather difficult tool to use in constructing regions for this research. His index attaches values to each county that indicate the degree of Southerness; however in his separation of the country into regions, he does not specify which counties comprise which regions. His failure to precisely indicate his methods for U.S. regional divisions force the researcher to make judgment calls and decisions which could ultimately result in a biased or skewed division of U.S. counties into cultural regions. A future avenue to explore may be the combination of methodologies in developing a Southerness index, or research which examines separate indicators of Southerness with the same set of predictors in order to measure reliability. Another possible limitation of this research lies in the failure to control for cities and large urban areas that lie within various county boundaries. Not taking metropolitan or urban areas into account can increase homicide rates in regions where there are a larger number of cities, for example; the Baltimore-Washington area, New York City, Philadelphia, and many other large urban areas are all contained within the Mid-Atlantic region in the present study. A promising suggestion for future research that utilizes 90 county level data would be the inclusion of an urban/rural variable, which would denote which counties are of a metropolitan status. Finally, this study has a constraint in reporting comparisons across its regression models because no test of significance between the models has been calculated. 91 CONCLUSION This research can neither confirm nor reject the existence of a southern subculture of violence. Although the results of the regression models indicate that structural factors are more influential on rates of violence than are cultural factors, this study advocates an integrated approach concerning research on southern violence. It is reasonable to suggest that culture and structure are mutually reinforcing factors. A combination of structural and cultural variables provides a better opportunity for researchers to explain regional differences in homicide and other violent crime rates. Culture can still be useful in explaining continuities in action, such as the consistently higher rates of homicide and violence that characterize the South yet remain stable despite any structural changes that have taken place. As Parker (1989) suggests, in a simultaneous analysis of the two types of variables poverty will always have the dominant effect, but it is possible that violent subcultural norms may actually cause one to be poor. The result is subculture as ?an indirect cause of homicide, operating via the direct cause of poverty? (Parker, 1989:986). The significance of the cultural position remains undeniable because it is impossible to understand human behavior ?outside of the cultural and historical contexts within which it occurs? (Clarke, 1998:289). An integrated approach is the best possible path for future research to explore because an explanation of violence based solely on structural factors ignores various aspects of Southern history and culture that have functioned to shape the region. 92 The waning influence of culture on variations in regional crime trends and regional distinctiveness appears to be a product of the U.S. growing more culturally homogeneous over time. The vast expanse of advanced technology allows for culture to spread and be shared much more easily than in the past. Popular culture is readily dispersed through cell phones, television, the media and countless other facets. Another recent trend that allows culture to permeate regional boundaries is the transient nature of contemporary society. Families are spread out across the country and people move much more often than in the past, with hardly an opportunity to grow roots in any one place. Migration patterns seem to suggest that over time, the regions of the U.S. will simply reflect geographical boundaries rather than culture ones. Future research should continue to investigate homicide and aggravated assault as analogous crimes that differ only in outcome on a continuum of violence. This study has shown that these two crimes share the same predictors and regional indicators, and an effort should be made to research the two simultaneously in discussions of extreme types of violence. This study has served to broaden the scope of the concept of violence beyond the narrow focus of homicide and future research should continue to do so in any examination of a subculture that is presumed to be violent. This research also contains a unique variable, the percent of violent events that are lethal involving homicides and aggravated assaults. Following the research of Unnithan (1994:147) and his colleagues, this variable could aid future research in examining the ?Southern influence on the direction of violence,? if utilized as a dependent variable. The most difficult aspect in examining the thesis of a southern subculture of violence is precisely discerning the interplay between cultural and structural factors. 93 Structural conditions and culture are inextricably linked in a circular debate involving causality. It is possible that Southern culture created the structural characteristics of the region and vice versa. Rather than attempting to isolate whether it is culture or structure that is responsible for higher rates of violence, focus should shift to developing more appropriate constructs and measures of both culture and violence. This research hopes to have aided the study of southern violence by expanding the scope and definition of violence where subcultures are concerned. 94 REFERENCES Ball-Rokeach, Sandra J. (1973). Values and Violence: A Test of the Subculture of Violence Thesis. American Sociological Review, 38, 736-749. Blau, Judith R. and Peter M. Blau. (1982). The Cost of Inequality: Metropolitan Structure and Violent Crime. American Sociological Review, 47, 114-129. Brearley, H.C. (1932). Homicide in the United States. Chapel Hill: University of North Carolina Press. Bureau of Justice Statistics. (2000). Criminal Victimization in the United States, 1998: National Crime Victimization Survey. Retrieved March 11, 2006, from http://www.ojp.usdoj.gov/bjs/pub/cvus9802.pdf. Chilton, Roland. (2004). Regional Variations in Lethal and Nonlethal Assaults. Homicide Studies, 8, 40-56. Chu, Rebekah, Craig Rivera, and Colin Loftin. (2000). Herding and Homicide: An Examination of the Nisbett-Reaves Hypothesis. Social Forces, 78, 971-987. Clarke, James W. (1998). Without Fear or Shame: Lynching, Capital Punishment and the Subculture of Violence in the American South. British Journal of Political Science, 28, 269-289. Clinard, Marshall B. (1963). Sociology of Deviant Behavior. New York: Holt, Rinehart and Winston, Inc. Cohen, Dov. (1998). Culture, Social Organization, and Patterns of Violence. Journal of Personality and Social Psychology, 70, 961-978. Cohen, Dov and Richard E. Nisbett. 1994. Self-Protection and the Culture of Honor: Explaining Southern Violence. Personality and Social Psychology Bulletin, 20, 551-567. Corcoran, M. (1995). Poverty and Mobility in the United States. Annual Review of Sociology, 21: 237-267. 95 Curtis, Lynn A. (1975). Violence, Race and Culture. Lexington, MA: D.C. Heath. Dickinson, Bruce D. Jr. (1979). Violence and Culture in the Antebellum South. Texas: University of Texas Press. Douglas, Jack D. (1982). The Sociology of Deviance. Canada: Little, Brown and Company. Dixon, Jo and Alan J. Lizotte. (1987). Gun Ownership and the Southern Subculture of Violence. The American Journal of Sociology, 93, 383-405. Ellison, Christopher G. (1991). An Eye for an Eye? A Note on the Southern Subculture of Violence Thesis. Social Forces, 69, 1223-1239. Ellison, Christopher G. and Marc A. Musick. (1993). Southern Intolerance: A Fundamentalist Effect? Social Forces, 72, 379-398. Ellison, Christopher G., Jeffrey A Burr, and Patricia L. McCall. (2003). The Enduring Puzzle of Southern Homicide: Is Regional Religious Culture the Missing Piece? Homicide Studies, 7, 326-352. Farrell, Ronald A. and Victoria Lynn Swigert. (1975). Social Deviance. Philadelphia: J.B. Lippincott Company. Felson, R. and Steven F. Messner. (1996). To Kill or not to Kill? Lethal Outcomes in Injurious Attacks. Criminology, 34, 519-545. Gastil, Raymond D. (1971). Homicide and a Regional Culture of Violence. American Sociological Review, 36, 412-427. Gastil, Raymond D. (1975). Cultural Regions of the United States. Seattle: University of Washington Press. Grasmick, Harold G., Elizabeth Davenport, Mitchell B. Chamblin, and Robert J. Bursick, Jr. (1992). Criminology, 30, 21-42. Hackney, Sheldon. (1969). Southern Violence. The American Historical Review, 74, 906- 925. Harer, Miles D. and Darrell Steffensmeier. (1992). The Differing Effects of Economic Inequality on Black and White Rates of Violence. Social Forces, 70, 1035-1054. Hayes, Timothy C. and Matthew R. Lee. (2005). The Southern Culture of Honor and Violent Attitudes. Sociological Spectrum, 25, 593-617. 96 Huff-Corzine, Lin, Jay Corzine, and David C. Moore. (1986). Southern Exposure: Deciphering the South?s Influence on Homicide Rates. Social Forces, 64, 906- 924. Huff-Corzine, Lin, Jay Corzine, and David C. Moore. (1991). Deadly Connections: Culture, Poverty, and the Direction of Lethal Violence. Social Forces, 69, 715- 732. Jacobson, Alvin L. (1975). Crime Trends in Southern and Nonsouthern Cities: A Twenty-Year Perspective. Social Forces, 54, 226-242. Jones, Dale E., Sherri Doty, Clifford Grammich, James E. Horsch, Richard Houseal, Mac Lynn, John P. Marcum, Kenneth M. Sanchagrin, and Richard H. Taylor. (2002). Religious Congregations & Membership in the United States 2000: An Enumeration by Region, State and County Based on Data Reported by 149 Religious Bodies. Nashville, TN: Glenmary Research Center. Land, Kenneth C., Patrica L. McCall, and Lawrence E. Cohen. (1990). Structural Covariates of Homicide Rates: Are There any Invariances Across Time and Social Space? The American Journal of Sociology, 95, 922-963. Lee, Matthew R., Michael O. Maume, and Graham C. Ousey. (2003). Social Isolation and Lethal Violence Across the Metro/Nonmetro Divide: The Effects of Socioeconomic Disadvantage and Poverty Concentration on Homicide. Rural Sociology, 68, 107-131. Loftin, Colin, and Robert H. Hill. (1974). Regional Subcultures and Homicide: An Examination of the Gastil-Hackney Thesis. American Sociological Review, 39, 714-724. Loftin, Colin and David McDowall. (2003). Regional Culture and Patterns of Homicide. Homicide Studies, 7, 353-367. Luckenbill, D.F. (1977). Criminal Homicide as a Situated Transaction. Social Problems, 25, 176-186. Messner, Steven F. (1983). Regional and Racial Effects on the Urban Homicide Rate: The Subculture of Violence Revisited. The American Journal of Sociology, 88, 997-1007. Messner, Steven F. (1989). Economic Discrimination and Societal Homicide Rates: Further Evidence on the Cost of Inequality. American Sociological Review, 54, 567-611. 97 Nisbett, Richard E. (1993). Violence and U.S. Regional Culture. American Psychologist, 48, 441-449. Nisbett, Richard E., Gregory Polly, and Sylvia Lang. (1995). ?Homicide and U.S. Regional Culture.? Pp. 135-51 in Social and Cultural Aspects of Interpersonal Violent Behavior, edited by R. Barry Ruback and Neil Alan Weiner. New York: Springer Verlag. Parker, Karen F. and Matthew V. Pruitt. (2000). How the West Was One: Explaining the Similarities in Race-Specific Homicide Rates in the West and South. Social Forces, 78, 1483-1508. Parker, Robert Nash. (1989). Poverty, Subculture of Violence, and Type of Homicide. Social Forces, 67, 983-1007. Porterfield, Austin L. and Robert H. Talbert. (1952). A Decade of Differentials and Trends in Serious Crimes in 86 American Cities by Southern and Nonsouthern Pairs. Social Forces, 31, 60-68. Redfield, H.V. (1880). Homicide, North and South. Philadelphia: J.B. Lippincott. Reed, John Shelton. (1971). To Live and Die in Dixie: a Contribution to the Study of Southern Violence. Political Science Quarterly, 86, 429-443. Reed, John Shelton. (1972). The Enduring South: Subcultural Persistence in Mass Society. Massachusetts: D.C. Heath and Company. Reed, John Shelton. (1993). Surveying the South: Studies in Regional Sociology. Missouri: University of Missouri Press. Rice, Tom W. and Carolyn R. Goldman. (1994). Another Look at the Subculture of Violence Thesis: Who Murders Whom and Under What Circumstances. Sociological Spectrum, 14:371-384. Roach, Jack L. and Orville R. Gursslin. (1967). An Evaluation of the Concept ?Culture of Poverty.? Social Forces, 45: 383-392. Roland, Charles P. (1982). The Ever-Vanishing South. The Journal of Southern History, 48, 3-20. Roof, Wade Clark and William McKinney. (1987) American Mainline Religion: Its Changing Shape and Future. New Brunswick: Rutgers University Press. Shaw, Clifford R. and Henry McKay. (1942). Juvenille Delinquency and Urban Areas. Chicago: University of Chicago Press. 98 Smith, M. Dwayne and Margaret A. Zahn (eds). Homicide: A Sourcebook of Social Research. Thousand Oaks: Sage Publications, Inc., 1999. Smith, M. Dwayne and Robert Nash Parker. (1980). Type of Homicide and Variation in Regional Rates. Social Forces, 59, 136-147. Swidler, Ann. (1986). Culture in Action: Symbols and Strategies. American Sociological Review, 51, 273-286. Unnithan, N. Prabha, Lin-Huff Corzine, Jay Corzine, and Hugh P. Whitt. The Currents of Lethal Violence: An Integrated Model of Suicide and Homicide. Albany: State University of New York Press, 1994. Vold, George B., Thomas J. Bernard, and Jeffrey B. Snipes. (2002). Theoretical Criminology. Oxford: Oxford University Press, Inc. Weaver Greg S., Janice E. Clifford Wittekind, Lin Huff-Corzine, Jay Corzine, Thomas A. Petee, and John P. Jarvis. (2004). Violent Encounters: A Criminal Event Analysis of Lethal and Nonlethal Outcomes. Journal of Contemporary Criminal Justice, 20, 348-368. Williams, Kirk R. (1984). Economic Sources of Homicide: Reestimating the Effects of Poverty and Inequality. American Sociological Review, 49, 283-289. Williams, Kirk R. and Robert L. Flewelling. (1988). The Social Production of Criminal Homicide: A Comparative Study of Disaggregated Rates in American Cities. American Sociological Review, 53, 421-431. Wirth, Louis. (1938). Urbanism as a Way of Life. American Journal of Sociology, 44, 1- 24. Wolfgang, Marvin E. and Franco Ferracuti. (1967). The Subculture of Violence: Towards an Integrated Theory in Criminology. New York: Tavistock Publications. 99 APPENDICES 100 TABLE 1. Descriptive Statistics: 2000 Census Data, U.S. Counties. Variables Mean Std Deviation Population density 244.9102 1675.8322 Murder rate 5.0730 31.22968 Aggravated assault rate 276.0955 1474.9046 Percent conservative Protestant 21.305 16.19408 Percent poverty 14.1774 6.54528 Gini index 0.4241521 0.03802208 Percent female headed households 6.1006 2.40784 Median family income 42034.694 9817.5780 Total population 89927.127 293514.75 Percent lethal 2.3349 4.95461 Percent black 8.7961 14.56663 New England 0.0293 0.16859 Mid Atlantic 0.0647 0.24595 Central Midwest 0.1470 0.35416 Upper Midwest 0.1290 0.33523 South 0.4043 0.49084 Rocky Mountain 0.0991 0.29880 Interior Southwest 0.0630 0.24308 West 0.0608 0.23899 101 TABLE 2. Model 1: OLS Regression predicting County level murder rates with percent black included. Variables ? b Population index (with percent black included) 2.087* (0.034) 0.808* Percent conservative Protestant 0.002 (0.002) 0.014 RDAC index 0.186* (0.034) 0.071* Percent lethal events 0.137* (0.006) 0.265* New England -0.519* (0.178) -0.035* Mid Atlantic -0.329* (0.127) -0.033* Central Midwest -0.899* (0.105) -0.118* Upper Midwest -0.366* (0.103) -0.048* Rocky Mountain 0.611* (0.123) 0.070* Interior Southwest 0.808* (0.120) 0.080* West 0.929* (0.132) 0.090* Note. * = p < 0.01; ** = p < 0.05. 102 TABLE 3. Model 2: OLS Regression predicting County level murder rates without percent black included. Variable ? b Percent Conservative Protestant 0.002 (0.002) 0.015 RDAC index 0.417* (0.034) 0.158* Percent lethal events 0.144* (0.006) 0.277* New England -0.754* (0.176) -0.052* Mid Atlantic -0.348* (0.126) -0.035* Central Midwest -1.076* (0.104) -0.142* Upper Midwest -0.539* (0.101) -0.071* Rocky Mountain 0.411* (0.121) 0.047* Interior Southwest 0.463* (0.118) 0.046* West 0.653* (0.130) 0.063* Population index (without percent black) 2.098* (0.034) 0.812* Note. * = p < 0.01; ** = p < 0.05. 103 TABLE 4. Model 3: OLS Regression predicting County level aggravated assault rates without percent black included. Variable ? b Percent conservative Protestant -0.002 (0.001) -0.017 RDAC index 0.331* (0.020) 0.167* Percent lethal events -0.055* (0.003) 0.140* New England -0.385* (0.103) -0.035* Mid Atlantic -0.352* (0.074) -0.047* Central Midwest -0.695* (0.061) -0.121* Upper Midwest -0.434* (0.059) -0.075* Rocky Mountain 0.496* (0.071) 0.075* Interior Southwest 0.530* (0.069) 0.069* West 0.647* (0.076) 0.083* Population Index (without percent black) 1.764* (0.020) 0.904* Note. * = p < 0.01; ** = p < 0.05. 104 105 TABLE 5. Model 4: OLS Regression predicting county level aggravated assault rates with percent black included. Variable ? b Percent conservative Protestant -0.002 (0.001) -0.020 RDAC index 0.135* (0.020) 0.068* Percent lethal events -0.061* (0.003) -0.155* New England -0.190 (0.107) -0.017 Mid Atlantic -0.332* (0.076) -0.044* Central Midwest -0.557* (0.063) -0.097* Upper Midwest -0.300* (0.062) -0.052* Rocky Mountain 0.641* (0.074) 0.097* Interior Southwest 0.808* (0.072) 0.105* West 0.867* (0.079) 0.112* Population index (with percent black included) 1.740* (0.020) 0.892* Note. * = p < 0.01; ** = p < 0.05.