CJT5

PLEASE SCROLL DOWN FOR ARTICLE

This article was downloaded by: [UNC University of North Carolina at Pembroke] On: 21 August 2010 Access details: Access Details: [subscription number 731587264] Publisher Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37- 41 Mortimer Street, London W1T 3JH, UK

Sociological Spectrum Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t713666965

THE SOUTHERN CULTURE OF HONOR AND VIOLENT ATTITUDES Timothy C. Hayesa; Matthew R. Leea

a Department of Sociology, Louisiana State University, Baton Rouge, Louisiana, USA

To cite this Article Hayes, Timothy C. and Lee, Matthew R.(2005) 'THE SOUTHERN CULTURE OF HONOR AND VIOLENT ATTITUDES', Sociological Spectrum, 25: 5, 593 — 617 To link to this Article: DOI: 10.1080/02732170500174877 URL: http://dx.doi.org/10.1080/02732170500174877

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

THE SOUTHERN CULTURE OF HONOR AND VIOLENT ATTITUDES

Timothy C. Hayes Matthew R. Lee

Department of Sociology, Louisiana State University, Baton Rouge, Louisiana, USA

Research on subcultures of violence has mostly focused on the existence of a Southern subculture of violence or a black subculture of violence. Recent theorizing on this topic has addressed many of the deficiencies in this literature and has reframed the issue by developing the notion of a Culture of Honor. Specifically, this perspective argues that if there is a subculture of violence in the U.S., it is most prevalent among South- ern white males from rural areas. This study is designed to partially explore the veracity of these claims. Our analysis of GSS data reveals that Southern white males from rural areas are more approving of viol- ence only under certain conditions, some of which may be construed as defensive in nature, and all of which have moderate to low approval rat- ings to begin with. In other words, when given scenarios in which the overall approval rate from the American public is high, this group is indistinguishable. But when given scenarios where overall approval is moderate to low, this group is more inclined to approve of the use of violence—a finding which we interpret as partially supportive of a subcultural explanation.

INTRODUCTION

Subcultural explanations of violent behavior and attitudes draw upon one main assumption: something aside from social structural

A previous draft of this manuscript was presented at the 2002 Annual Meeting of the Mid

South Sociological Association, Memphis, TN. The authors thank R. Gregory Dunaway and

Peter B. Wood for helpful comments on the development of this project and for extensive

comments on previous drafts of the manuscript.

Address correspondence to Timothy C. Hayes, Department of Sociology, 126 Stubbs Hall,

Louisiana State University, Baton Rouge, LA 70803, USA. Email: [email protected]

Sociological Spectrum, 25: 593–617, 2005

Copyright # Taylor & Francis Inc.

ISSN: 0273-2173 print/1521-0707 online

DOI: 10.1080/02732170500174877

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

factors influences groups exhibiting high levels of violence. These theories stipulate that a particular social group or geographic area may be characterized by values either promoting, or at least toler- ating, violence. Historically, subcultural explanations have focused on race and region. Research in this tradition cites high rates of violence among urban blacks and Southerners as evidence for vio- lent subcultures. (see Curtis 1975; Gastil 1971; Hackney 1969; 1972; Huff-Corzine et al. 1986; Wolfgang and Ferracuti 1967). Although much of this research is compelling, there are at least two serious problems plaguing most of these studies. The first has to do with the measurement of subcultural values. For example, many macrolevel studies use ‘‘dummy variables’’ to assess the effect of Southern region on the violent crime rate and assume this taps into specific cultural values supporting viol- ence in the South (see Hackney 1969; Messner 1983a; and Parker 1989). Other studies have used percent black in a given area to predict violent crime rates and argue that this measure captures an urban black subculture of violence (see Messner 1983a; Messner 1983b; and Parker 1989).

It is difficult to see how either of these measures actually oper- ationalize differences in values supporting violence. For example, a dummy variable coded ‘‘1’’ for macrolevel units in Southern areas and coded ‘‘0’’ for all others, when placed in a multiple regression model predicting the rate of violent crime, simply gives the mean difference in the violent crime rate between Southern areas and all others when the influence of the other variables in the model are held constant. This not only does not provide evidence of sub- cultural values supportive of violence, but uses the actual outcome of interest (violence) as evidence of the supposed cause (different values). Similarly, the coefficient for a variable measuring percent black within a macrolevel unit simply gives the effect that a one unit change in the proportional representation of the black popu- lation has on the violent crime rate when the effect of the other variables in the model are held constant. Again, it is difficult to see how, in this case a purely demographic measure, provides evi- dence of a subculture of violence. At the very best, indicators such as these should be taken as indirect measures of a subculture of violence.

A second limitation of most existing subcultural explanations is that they ignore the complexity of crime and violence in American society. The prevalence of violence varies considerably not just by region or race, but by both region and race, as well as by gender and contextual location (e.g., rural versus urban

594 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

areas).1 This important point was recently recognized by Nisbett and Cohen (1996), whose important book offers a new subcul- tural model for explaining violence in the U.S. Their ‘‘culture of honor’’ model argues that if a subculture of violence does exist, it is most likely confined to white males residing in rural areas—primarily in the South. Nisbett and Cohen (1996) locate the source of this subculture of violence in the historical settlement of some areas of the rural South by herdsmen. These herdsmen were forced to use violence to protect their herds due to the constant threat of theft and the near-absence of law enforcement in remote rural areas. Thus a ‘‘culture of honor’’ was established in which the herdsman kept up the appearance of being strong and willing to engage in violence in order to protect their possessions, especially in the face of limited protection against theft or threats to the family from local authorities. The reference to herding in the argument, merely points to the fact that the past structural conditions of the South (remote rural areas with a lack of formal law enforcement) were conducive to the acceptance of violence in some situations. Nisbett and Cohen (1996) assert that this vigilante-type tradition, condoning violence in situations for protection of family members and possessions may still be prevalent in the South today due to a continuing transmission of these values through the socialization process.

This culture of honor thesis is different from most subcultural the- ories of violence in two important ways. First, the theory specifies that values condoning the use of violence will be observable in a very specific segment of the population: Southern white rural males. Second, the authors suggest that the violent values pertain to specific situations involving defense of self, personal honor, family, and per- sonal property. Nisbett and Cohen (1996) argue that those posing a threat in these areas may quickly become victims of violence by those who are most invested in this culture: Southern white rural males. This serves an immediate protective function by eliminating the threat, but it also has a deterrent effect. Individuals who are quick to act violently are less likely to be taken advantage of or threatened by others.

1A third potential problem with this literature is that some studies focus on lethal violence

and others only on non-lethal physical violence. It is difficult to discern how problematic this

is. Many times a homicide is just an assault where someone has died, and as Harris et al. (2002)

demonstrated, assault rates in the U.S. have remained high but this form of violence has

become less lethal. We thank a reviewer for bringing this important point to our attention.

Southern Culture of Honor 595

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

This study is designed to provide an empirical test of the test the emerging culture of honor model. We begin by reviewing the litera- ture on the traditional subcultural models of region and race followed by more recent literature focusing on the ‘‘culture of honor’’ expla- nation offered by Nisbett, Cohen and others. We then provide a test of this alternative model using General Social Survey (GSS) questions regarding the acceptance of violence in a variety of scenarios. The study concludes with a discussion of the theoretical importance of the research, limitations of the study and suggested directions for future research.

LITERATURE REVIEW

Although anecdotal and theoretical accounts of a violent Southern subculture go back more than 100 years, Wolfgang and Ferracuti (1967) are widely considered to have pioneered the scientific study of violent subcultures (see Messner 1983a). Their research stimulated writings dealing primarily with two types of subcultures: those based on race and those based on region. First, we briefly review the litera- ture focusing on regional subcultures of violence, which focuses almost exclusively on the South. We follow this with a brief review of the literature on racial subcultures of violence, before moving on to the most recent articulation of the culture of honor that is the central concern of this article.

Regional Subcultures

A great deal of debate surrounds the issue of regional subcultures of violence in the U.S. Nearly all available research on this topic cites the South as having the prevailing subculture of violence (see Corzine et al. 1999 for a recent review). Writings on southern culture and behavior by Cash (1941) consistently mention the South as a violent place both before and after the Civil War (refuting a popular argu- ment that violence in the South is the result of frustration over the loss to the North in the Civil War). Cash attributes the historically higher rates of violence in the South to its unique culture. His research, along with later studies by Hackney (1969) and Gastil (1971) helped start a long tradition of both macro- and microlevel research into the Southern subculture of violence.

At the macrolevel, the debate is most clearly found among the writings of Gastil (1971), Hackney (1969) and Loftin and Hill (1974). The former two analysts reported higher homicide rates among Southern states in the U.S., and concluded that cultural

596 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

factors were primarily responsible for the observed regional variation. In an important re-evaluation, Loftin and Hill (1974) re-examined these issues and reached a different conclusion. Although they did not completely discount a cultural influence, they did report that their structural poverty index was a strong, positive, and statistically sig- nificant correlate of the state level homicide rate, and that when it was included in a multiple regression context, the effect of measures of Southern culture were rendered insignificant.

Smith and Parker (1980) and Messner (1983a) continued the regional subcultural debate. While Smith and Parker (1980) report no significant regional variations, Messner (1983a) reported that location in the Southern region of the United States significantly affects (Standard Metropolitan Statistical Area’s (SMSA) homicide rate when poverty and other demographic variables are controlled. Messner used a ‘‘Confederate dummy’’ (as suggested by Hackney) as well as Gastil’s ‘‘Southerness index’’ to measure the effect of the Southern region.2 In both cases, the variables indicating Southern region (the dummy variable and the Southerness index) were found to be significant positive predictors of homicide rates in SMSA’s.

Messner (1983b) also found that poverty levels and income inequality have much stronger effects on homicide in areas other than the South. Though it was a significant predictor of homicide rates for non-Southern cities, the percentage of people below the official pov- erty line (or percent poor) was not a significant predictor for South- ern cities. The effect of percent black was also diminished in the Southern model, though it remained significant. These types of find- ings (see also Huff-Corzine et al. 1986) hint that there may indeed be a strong cultural difference between the South and the rest of the United States.

At the macrolevel then, there is little consensus as to the existence or the nature of a regional subculture of violence. Part of the reason for this is that it is difficult to determine cultural influences at the macrolevel of analysis. Hence, some analysts have attempted to sort out these issues by turning to individual level data sources. For example, Dixon and Lixotte (1987) used GSS data from 1973, 1976, 1980, and 1984, to examine attitudes toward violence. They divided the questions regarding violence into two categories: those suggesting defensive attitudes and those suggesting purely violent

2Gastil (1971, p. 425) developed his well-known Southerness Index, which has been fairly

widely used by those working in the subcultural tradition. As Loftin and Hill (1974, p. 715)

note, it appears that ‘‘proximity to the former Confederate States is clearly the major dimen-

sion being measured.’’

Southern Culture of Honor 597

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

attitudes. They interpreted their findings to indicate no significant difference in attitudes toward violence between ‘‘those socialized in the south and currently residing there’’ (Dixon and Lixotte 1987, p. 397) and those either socialized or currently residing in another region. They further conclude that violent attitudes are ‘‘more likely to be found among rural dwellers and among whites’’ (Dixon and Lixotte 1987, p. 398–399). In a notable set of commentaries however, Ellison and McCall (1987) and Corzine and Huff-Corzine (1989) both heavily criticized Dixon and Lixotte for incorrectly interpreting their results, and concurred that in fact their results supported the notion that Southerners are more supportive of violence in defensive situa- tions. Additional microlevel support for the Southern subculture of violence thesis is reported in Ellison (1991), who also reported that older Southern natives were more supportive of defensive or retaliat- ory violence than the younger Southern natives, suggesting that the subculture of violence is disappearing.3

In light of the good deal of research at both the macro- and microlevels on regional subcultures of violence, it is tempting to judge the debate resolved. However, many inconsistencies are apparent in this literature, and it is probably premature to develop extremely firm conclusions. Indeed, an entirely complementary body of research has evolved targeting the existence of another type subculture based on race. It is to this literature that we now turn.

Racial Subcultures

Speculation on the existence of a race-specific subculture of violence goes back many decades, and typically focuses on blacks (see dis- cussion in Hackney 1969, p. 920). Early empirical sociologists, such as Clifford Shaw and his colleagues at the University of Chicago (see Shaw et al. 1929) found that neighborhoods with high crime rates also had a racial make-up that was unrepresentative of the city. Often there was an overrepresentation of minority groups in these neighborhoods. This finding was consistent with the more anecdotal

3Several authors cite gun ownership as a predicting variable for the southern subculture of

violence. They reason that higher rates of gun ownership for the use of protection as well as

recreational activities points toward the violent nature of the South (Gastil 1971; Kleck 1979;

Dixon and Lixotte 1987) and may be somehow linked to the existence of a subculture of viol-

ence in the region (Reed 1971). Some argue that high levels of gun ownership translates into a

self-protection phenomenon in which individuals arm themselves for protection against the

perceived likelihood of being involved in an act of violence. Thus, gun ownership is not neces-

sarily a cause of high levels of violence, but one of its effects (Kleck 1979; Lizotte and Bordura

1980; Dixon and Lixotte 1987).

598 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

observations of others that Southern blacks had high homicide rates (see Hackney 1969, p. 920).

Several studies attempting to examine the black subculture of viol- ence thesis try to measure it at the aggregate level by including a mea- sure of the percentage of the population that is black in multiple regression models. For instance in his analysis of Southern and Non-Southern SMSA’s, Messner (1983a) showed that the relative size of the African American population (percent black) in non-Southern areas is a strong predictor of that area’s homicide rate when struc- tural factors such as poverty and unemployment are controlled. While replicating Messner’s work and introducing stricter controls for measurement error, Parker (1989) also found ‘‘percent black’’ to be a significant indicator of crime rates in a given area but failed to find any significant positive effect among other subcultural indica- tors. Parker concluded that new approaches must be utilized to find evidence for a true subcultural effect. Similarly, Blau and Blau (1982) analyzed SMSA’s and found that with controls for inequality, pov- erty and Southern region ceased to be significant predictors of violent crime. The effect of percent black, though still significant, was sub- stantially reduced. They conclude that a subculture of violence is likely to be linked to economic inequality rather than race or region (Blau and Blau 1982). Indeed, as Anderson (1999) details in his book ‘‘Code of the Street’’, some segments of the urban black population that are subjected to extreme deprivation and inequality may develop violent dispositions as a way of seeking status. As Sampson and Wilson (1995) argue however, it is likely that such adaptations are transient responses to unusually severe structural conditions, and will not be intergenerationally transmitted. This suggests that it would not be a race-based subculture of violence, but one more tied to economic situations.

Two of the most important studies to date find no support for a black subculture of violence. At the microlevel of analysis, Cao, Adams and Jensen (1997) criticize previous research dealing with the subculture of violence as neglecting race and focusing too much on violence in the South. They further state that previous measures of race, such as white and nonwhite, are inadequate. Instead, they suggest contrasting white with black and eliminating other races from the analysis. Using GSS data from 1983 to 1991 (omitting 1985 because necessary items were not included that year) they compare whites and blacks on defensive and offensive violent values. They find little support for the black subculture of violence, stating that white males report a significantly higher support of violence in defensive situations than black males. They find no significant difference

Southern Culture of Honor 599

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

between the two groups in offensive situations. At the macrolevel, Shihadeh and Shrum (2004) examine neighborhood level data and posit that high crime rates in urban neighborhoods often lead to a race-based explanation of crime before examining the structural con- ditions present. They find that structural conditions alone account for neighborhood variations in crime rates, and that in fact racial composition is not a significant predictor of crime.

Like the research on regional variations in support for violence, the research on racial variations produces mixed findings. The most recent evidence indicates that if anything, whites may be more toler- ant of violence and more supportive of defensive violence than other groups.4 Below we explore the most recent reasoning with regard to the intersection of regional and racial subcultures and how they may be reflected in attitudes toward violence.

The Culture of Honor

It is generally not disputed that the South has higher homicide rates than other regions of the nation. Reed (1982) indicates that part of the reason for such a high homicide rate is that the South has dispro- portionately high rates of homicide resulting from some type of argu- ment or dispute, be it some type of family dispute, or a spousal or lover’s dispute. He attributes this to a culture of violence in the South, meaning that violence is an acceptable form of resolving some disputes, rather than involving legitimate authorities, such as the police (Reed 1982). This type of a description of culture is not the familiar, broad definition often given by sociologists. Culture here is internalized in the individual members of a group to some extent and becomes something that they draw upon to negotiate everyday situations. Using Swidler’s (1986) metaphor, it is likened to a tool- kit that one can draw upon to negotiate everyday life (see also the dis- cussion in Corzine et al. 1999, pp. 45–46).

Other writings on Southern culture mention that the South has long been a violent region, one in which personal disputes are

4With regards to attitudes toward violence among blacks, Sampson and Bartusch (1998)

examined tolerance of deviance and violence using data from the Project on Human Develop-

ment in Chicago Neighborhoods (PHDCN). They found that blacks were generally less toler-

ant of deviance and violence than white respondents, which would effectively refute any

subcultural explanation assuming blacks are more violent than whites. However, African

Americans scored higher on a scale measuring legal cynicism than whites and much lower

on a scale measuring satisfaction with police. These data then suggest that while African

Americans are less tolerant of deviance (including violence) they are also generally less satisfied

with institutional responses to it from formal authorities.

600 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

considered matters of ‘‘honor’’ and are settled without the intervention of law enforcement (Cash 1941; Wyatt-Brown 1981). But why is this the case? In our view, a discussion of the culture of honor may help to answer this question.

As an emerging subcultural model, the culture of honor combines previous notions of race and region (Nisbett and Cohen 1996). This model states that white males reared in the rural South are more likely to approve of violence in some situations, specifically those involving protection of self, personal honor, family, or personal property. Recall, Swidler’s (1986) ‘‘tool-kit.’’ Southerners simply are more apt to draw a violent response from their tool-kit than others, given cer- tain situations. To view it another way, Southerners have violence near the top of their ‘‘cultural repertoire’’ when others may not. This is a different approach than previous subcultural explanations of viol- ence in the South because it focuses on a very limited and well-defined demographic group, and couples this with a focus on violence having protective motives. The culture of honor was most likely established when the first settlers moved into the Southern regions of the United States. As stated before, these settlers relied on a culture of honor as a self-policing tool due to the lack of law enforcement in rural areas. The high regard for personal honor was passed down through genera- tions, and is still somewhat present today (Nisbett and Cohen 1996).

Nisbett and Cohen (1996) state that they borrow the term ‘‘culture of honor’’ from anthropology, but insist that they confine their theorizing to only part of its complete meaning. For example, Wyatt-Brown (1982), who also discussed honor in the South indicates that honor is often considered multidimensional and deals with more than response to threats and handling personal disputes without involving the authorities. Rather, honor is (1) ‘‘an inner conviction of self-worth. . .,’’ (2) the ability to show that self-worth in public, and (3) the assessment by the public of the self-worth of the individ- ual (Wyatt-Brown 1982, p.14.) Thus to Wyatt-Brown, an individual must first believe that he or she is honorable, they must not hesitate to act honorably in public and finally, for the public or community to consider an individual honorable, their public actions must be interpreted as such.

There are clearly internal and external components to Wyatt- Brown’s definition of honor, and this definition of honor is very gen- eral and applies to virtually all public displays or overt behavior. So, in general a culture of honor refers to a group or society that relies on public displays to evaluate individual behavior and in a sense individ- ual worth. Specifically, if someone were to be threatened in a culture of honor, and especially if that threat were in public, the individual

Southern Culture of Honor 601

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

would feel the need to quickly respond to that threat to show that he or she is still an honorable person, both to himself and to the group. This is the main aspect that Nisbett and Cohen (1996) focus on in their theory.

Again, Nisbett and Cohen (1996) discuss the South as having a cul- ture of honor, but focus on the role of violence in this culture. Thus, from this point forward when we refer to the culture of honor, we are referring to Nisbett and Cohen’s (1996) discussion of violence and its use in the Southern culture of honor. The culture of honor primarily calls for violence in instances of personal threats or insults to an indi- vidual. Violence is regarded as an acceptable mode of defense when one’s honor or character is threatened. If a particular head of the household was viewed as unable to protect his possessions and fam- ily, he was viewed as more susceptible to attack from thieves and criminals. Therefore, when threatened, these males were usually quick to resort to violence as a proof that they were capable of defending themselves. Nisbett and Cohen suggest that:

To maintain credible power of deterrence, the individual must project

a stance of willingness to commit mayhem and to risk wounds or

death for himself. Thus he must constantly be on guard against

affronts that could be construed by others as disrespect. When some-

one allows himself to be insulted, he risks giving the impression that

he lacks the strength to protect what is his. Thus the individual must

respond with violence or the threat of violence to any affront (Nisbett

and Cohen 1996, p. xv).

This type of defense was needed in part due to the rural and remote nature of the South as well as the scarcity of law enforcement (Nisbett and Cohen 1996).

Nisbett and Cohen (1996) cited evidence for the subculture of vio- lence in the South based on several different research frameworks. First, they state that the South has a high rate of homicide. This proposition held regardless of the population size of the unit of analy- sis. It also held regionally for whites and not for blacks. Second, Nisbett and Cohen examined attitudes toward violence from the Northern and Southern perspectives. They found that Southerners were more likely to support violent responses to a variety of protec- tive situations than Northerners. Finally, they used controlled experi- ments to see if these attitudes toward violence among Southerners were accompanied by a measurable physiological response. They found that Southerners responded to a potentially insulting situation

602 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

with significantly higher levels of aggression than those from other regions (Nisbett and Cohen 1996; also Nisbet 1993; Cohen 1996; Cohen et al. 1996; and Cohen 1998).

It is notable that one assertion in the theory which has recently been challenged is that homicide rates are higher where herding was more prevalent. Chu, Rivera, and Loftin (2000) take issue with this assertion and report that the relationship between herding areas and homicide is disappears when controlling for levels of poverty among non-hispanic whites. In our view this does not undermine the efficacy of the entire theory, in large part because of substantial population redistribution that has occurred over time.

Summary and Expectations

The main contribution of the work by Nisbett and Cohen is that they specify that if there is a subculture of violence, it is adhered to by a very specific segment of the population. In a sense, their perspective clarifies a good deal of the mixed findings of prior research on both regional and racial subcultures of violence. The culture of honor framework is clearly advantageous in that it combines region and race with gender and rural residence to create a well-defined group of people expected to have a high propensity towards prescribing to values supportive of violence in defensive situations in particular. Unfortunately, their framework has not been subjected to the same empirical scrutiny as the more traditional regional and racial subcul- tural studies. With this in mind, we derive two main hypotheses.

Hypothesis 1: Southern white rural males will be more approving of violence than members of other groups.

Hypothesis 2: White rural males socialized in the South (but not neces-

sarily still residing there) will be more approving of violence than

members of other groups.

These hypotheses are not meant to test the actual level of violence that Southern white rural males commit. Instead, they are meant to test the core assumption of the culture of honor thesis: that Southern white rural males are more supportive of violence.

Data and Measures

To test these hypotheses we use data from the GSS. The GSS uses a three-stage full probability sampling design based on the most recent

Southern Culture of Honor 603

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

U.S. Census data beginning with Standard Metropolitan Statistical Areas and nonmetropolitan counties in the first stage, block groups and enumeration districts in the second stage, and if necessary the segments are further divided and sampled for a third stage. The final stage of sampling includes households selected from the block groups and enumeration districts or segments. The end result is a national sample of about 2500þ U.S. households; the actual sample size var- ies depending on the year (Davis et al. 2001).

Dependent Variables

The main dependent variables are measures of approval of inter- personal violence used in some part by Dixon and Lixotte (1987) and Cao, Adams and Jensen (1997), among others. The variables are constructed from six GSS items regarding approval of violence in a variety of situations. While several past studies distinguish between offensive and defensive violence, the present study does not. Rather, all the measures of interpersonal violence in the GSS are analyzed separately in order to form a more complete under- standing of variations in violent attitudes among several demo- graphic groups, regardless of differing perceptions on what constitutes offensive and defensive violence. Although we depart from prior research on this important point, we feel we are justified in doing so for at least two reasons.

First, it is simply impossible to determine how respondents clas- sify a particular scenario without asking them. This is apparent in the case of Dixon and Lixotte (1987, p. 394) for example, where they classify the following question as capturing a purely violent attitude as opposed to a defensively violent attitude: ‘‘Would you approve of a man punching an adult male stranger if the adult male stranger was drunk and bumped into the man and his wife on the street?’’ On its face, an affirmative answer to this question would seem to imply a defensive orientation because it captures an apparent physical threat, in part towards a female companion, occurring in public space. Yet based on a factor analysis, the authors place this item in their Purely Violent Attitude scale. This approach seems to place an inordinate degree of confidence in the ability of a factor analytical model to discriminate among concep- tual categories.

The second justification we use for departing from prior research in measuring our outcome of interest focuses on the homogeneity of effects assumption implicit in this approach. The essence of the problem is that creating a scale out of multiple items and using it

604 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

as the dependent variable assumes that the effects of the relevant explanatory variables are constant or at least similar across a series of outcomes. This is an extraordinarily strong and dangerous assumption to make. We argue that this is especially true when the distributions of the responses vary considerably from outcome to outcome—which is the case with these data—as we demonstrate below. In light of these issues then, we opt to examine each outcome separately.

The GSS includes a six-item series of questions regarding approval of interpersonal violence in a variety of situations. The first item (HITOK) asks ‘‘Are there any situations that you can imagine in which you would approve of a man punching an adult male stranger?’’ The second item (HITBEATR) asks, ‘‘Would you approve of a man punching an adult male stranger if the adult male stranger was beating up on a woman and the man saw it?’’ The third item (HITCHILD) asks the respondent ‘‘Would you approve of a man punching an adult male stranger if the adult male stranger had hit the man’s child after the child accidentally damaged the stranger’s car?’’ The fourth item (HITDRUNK) asks ‘‘Would you approve of a man punching an adult male stranger if the adult male stranger was drunk and bumped into the man and his wife on the street?’’ The fifth item (HITMARCH) asks ‘‘Would you approve of a man punching an adult male stranger if the adult male stranger was in a protest march showing opposition to the other man’s views?’’ Finally, the sixth item (HITROBBR) asks, ‘‘Would you approve of a man punching an adult male stranger if the adult male stranger had broken into the man’s house?’’ The valid response categories for each of these measures are yes and no. Each of the variables is recoded into a binary measure, coded 1 ¼ Yes and 0 ¼ No.

Explanatory Variables

The main explanatory variables employed in the following analyses are constructed from a series of dummy variables that tap into region, race, sex and rural residence. The REGION variable from the GSS is recoded in such a way that the existing categories in the original data- set are combined into their major Census division, thus leaving four major divisions of the U.S. as defined by the U.S. Census (Northeast, Midwest, West, and South). For example, the categories indicating residence in the South Atlantic, East South Central, and West South Central regions are combined into one category indicating Southern residence. In addition, as an alternative indicator of regional

Southern Culture of Honor 605

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

influence, a second measure indicating regional residence at age 16 (REG16) is coded into the four major Census divisions plus a cate- gory for foreign residence at 16. These new regional variables are then used to construct dummy variables for each region at the time of interview and at 16 years of age. The variables for race (RACE) and sex (SEX) are also drawn from the GSS, and coded so that a sep- arate dummy variable exists for each racial category (black, white and other) and each sex (male and female).

Finally, two variables indicating rural or nonrural residence are also constructed. The variable indicating current rural residence is constructed from the XNORCSIZ variable in the GSS dataset. For the purposes of the present study, respondents are coded as rural if they indicated they lived in a town of 2,500 people or less (this includes those who do not officially live in a town). From this, a dummy variable is developed, coded 1 ¼ rural and 0 ¼ nonrural. A variable for rural residence at age 16, using PLACE16 (indicating the type of place where the respondent lived at age 16) is recoded so that those indicating a rural residence (on a farm or open country) at 16 are coded 1 and others coded 0. From this series of dummy vari- ables we derive our two main variables of interest: A) an indicator variable coded 1 ¼ Southern white males residing in rural areas, and 0 ¼ all others, and B) an indicator variable coded 1 ¼ white males residing in rural areas that were living in the rural South when they were 16, and 0 ¼ all others. We use these measures as our two key explanatory variables of interest.

Control Variables

In the multivariate analysis, it is important to control for the poten- tial influence of other factors that may condition respondent’s atti- tudes toward violence. One potentially important factor is prior exposure to violence. Respondents who indicated that they had been hit by another person, whether as a child or as an adult, or indi- cated that they had been threatened or shot at with a gun are coded as having some history of exposure to violence. Again, separate dummy variables for each category are constructed for use in the regression stage of the data analysis, with those having prior exposure to violence coded as 1, and all others 0. Several other control variables are also employed in the multivariate analyses, including respondent’s age in years (AGE), educational attain- ment in years (EDUC), income in actual dollars (REALRINC), and the respondents marital status, coded 1 ¼ never married and 0 ¼ otherwise.

606 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

ANALYTICAL METHODS AND RESULTS

Descriptive Statistics

Complete data are available for the dependent variables for the years 1976, 1978, 1980, 1983, 1984, 1986–1991, 1993 and 1994. Each year listed is included in the analysis, regardless of differences in number of responses, due to differences in sample size or item rotation. With the respondents having missing data on one or more of the variables used in the analysis removed, the final sample yields an N of 8,940 respondents from the years 1976, 1978, 1980, 1983, 1984, 1986– 1991, 1993, and 1994. The descriptive statistics for selected variables are presented in Table 1. With respect to the dependent variables, the majority of respondents approved of a person hitting a stranger in most of the situations. For example, 86.5 percent approved of a man hitting a stranger if the stranger was beating up a woman (HITBEATR), and a nearly identical proportion approved of a

Table 1. Coding and distribution for selected

variables

Dependent variables

HITOK

1 ¼ Yes 70.1% (6264)

0 ¼ No 29.9% (2676)

HITBEATR

1 ¼ Yes 86.5% (7730)

0 ¼ No 13.5% (1210)

HITCHILD

1 ¼ Yes 61.6% (5511)

0 ¼ No 38.2% (3429)

HITDRUNK

1 ¼ Yes 6.9% (614)

0 ¼ No 93.1% (8326)

HITMARCH

1 ¼ Yes 2.7% (241)

0 ¼ No 97.3% (8699)

HITROBBR

1 ¼ Yes 85.3% (7625)

0 ¼ No 14.7% (1315)

Key explanatory variables

Southern rural white male 3.1% (279)

Rural white male residing in the

rural south at age 16

1.9% (171)

Total N for Sample ¼ 8940.

Southern Culture of Honor 607

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

man punching an adult male stranger that had broken into the man’s home. It is notable that for two of the scenarios, the overwhelming majority did not approve of hitting a stranger (HITDRUNK and HITMARCH). These scenarios clearly do not call for a violent response by most respondents. However, we would predict a strong Southern component within the small percentage that do approve of acting violently in these cases.

Descriptive statistics for our two key explanatory variables are also presented in Table 1. We can see here that 3.1 percent of the sample is comprised of Southern white males living in rural areas, and 1.9 percent of the sample is white males living in rural areas that resided in the rural South when they were 16 years old.5 It is both of these groups which we expect to exhibit higher levels of approval for viol- ent actions, particularly those that in some way appear defensive in that they are designed to protect family, property, or personal honor.

Logistic Regression

Because the outcomes of interest are binary in nature, logistic regression is an appropriate analytical strategy to evaluate differences in the level of approval between our key demographic group of inter- est and all others while controlling for the possible influence of a series of confounding factors (Menard 1995; Pampel 2000). We begin testing our hypotheses in Table 2, which presents a series of six logistic regression models.6

5We do not report other descriptions of the sample in tabular form, but the sample is 51.6

percent male, 85.6 percent white, 11.8 percent are black, and 2.6 percent other. The sample is

fairly equally distributed between regional categories with 20.1 percent reporting from the

Northeast, 27.6 percent from the Midwest, 19.5 percent from the West and 32.8 percent from

the South. In addition, 14.7 percent of the respondents reside in areas categorized as rural.

The age of our respondents ranges from 18 to 89 years old with a mean of 38.86. Average edu-

cational attainment is 13.16 years, and the median income is $16,563, with only 21 percent of

the sample indicating they have never married. Finally, 43.1 percent claim to have been

punched or beaten by another person, and 23.2 percent have been threatened or shot at with

a gun. 6One of the main threats faced when performing multivariate regression analysis is multi-

collinearity. Multicollinearity refers to the problems encountered when two or more inde-

pendent variables are highly correlated with one another and with the dependent variable.

When this is the case, tests of statistical significance can be biased, among other things. To

probe for this potential problem, these models were re-estimated using Ordinary Least Squares

Regression to obtain Variance Inflation Factors, a standard tool for diagnosing multicollinear-

ity. The informal standard in sociology for diagnosing multicollinearity is that variance

inflation factors higher than 4.0 may indicate a problem. None of the variance inflation factors

in the following models were above 2.5, suggesting that multicollinearity is not a problem in

these models (Fox 1991).

608 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

T a b le

2 .

L o g is ti c re g re ss io n m o d el s p re d ic ti n g vi o le n t a tt it u d es

w it h so u th er n ru ra l w h it e m a le s a n d co n tr o ls

D ep en d en t va ri a b le s

H IT

O K

H IT

B E A T E R

H IT

C H IL

D H IT

D R U N K

H IT

M A R C H

H IT

R O B B E R

K ey

ex p la n a to ry

v a ri a b le :

S o u th er n ru ra l w h it e m a le s

1 .3 0 7 (. 1 4 0 )

1 .1 0 2 (. 1 7 9 )

1 .7 6 8 ��

� (. 1 4 0 )

2 .3 3 4 ��

� (. 1 7 4 )

1 .8 2 1 � (. 2 6 3 )

1 .2 1 8 (. 1 9 0 )

C o n tr o l v a ri a b le s:

N ev er

m a rr ie d

.7 4 1 ��

� (. 0 6 5 )

.8 7 8 (. 0 8 5 )

.8 0 7 (. 0 6 0 )

1 .1 5 9 (. 1 1 7 )

1 .4 1 1 (. 1 7 6 )

.9 4 7 (. 0 8 1 )

T h re a te n ed

o r sh o t a t

1 .2 1 1 ��

(. 0 6 2 )

.8 9 5 (. 0 7 8 )

1 .4 0 5 ��

� (. 0 5 8 )

1 .1 0 0 (. 1 0 4 )

1 .5 2 8 ��

(. 1 5 5 )

1 .1 6 6 (. 0 8 1 )

H it b y a n o th er

p er so n

1 .3 7 2 ��

� (0 .0 5 2 )

1 .1 1 0 (. 0 6 9 )

1 .1 4 3 ��

(. 0 4 9 )

.9 3 6 (. 0 9 3 )

.9 2 4 (. 1 4 7 )

1 .3 2 6 ��

� (. 0 6 8 )

A g e (i n ye a rs )

.9 9 3 ��

� (. 0 0 2 )

.9 8 8 ��

� (. 0 0 3 )

.9 7 8 ��

� (. 0 0 2 )

1 .0 1 3 ��

� (. 0 0 3 )

1 .0 0 7 (. 0 0 5 )

0 .9 8 9 ��

� (. 0 0 2 )

E d u c a ti o n (i n ye a rs )

1 .1 3 5 ��

� (. 0 0 9 )

1 .0 8 3 ��

� (. 0 1 2 )

1 .0 1 4 (. 0 0 9 )

.9 2 6 ��

� (. 0 1 6 )

.8 2 9 ��

� (. 0 2 5 )

0 .9 8 1 (. 0 1 2 )

In c o m e (i n d o ll a rs )

1 .0 0 0 (. 0 0 0 )

1 .0 0 0 (0 .0 0 0 )

1 .0 0 0 (. 0 0 0 )

1 .0 0 0 (. 0 0 0 )

1 .0 0 0 (. 0 0 0 )

1 .0 0 0 (. 0 0 0 )

C o n st a n t

.6 9 0 � (. 1 7 2 )

3 .2 6 7 ��

� (. 2 2 1 )

2 .0 9 4 ��

� (. 1 5 8 )

.1 0 0 ��

� (. 3 0 0 )

0 .2 2 6 ��

(. 4 6 4 )

8 .3 9 8 ��

� (. 2 1 5 )

M o d el

ch i- sq u a re

3 4 9 .2 3 7 ��

� 1 1 8 .7 7 0 ��

� 3 0 2 .1 8 1 ��

� 8 5 .1 7 0 ��

� 1 3 5 .8 7 0 ��

� 1 0 1 .1 3 0 ��

�

N a g el k er k e R -s q u a re

.0 5 4

.0 2 4

.0 4 5

.0 2 4

.0 6 9

.0 2 0

C o ef fi c ie n ts

a re

o d d s ra ti o s w it h st a n d a rd

er ro rs

in p a re n th es es .

� p <

0 .0 5

�� p <

0 .0 1

�� � p

< 0 .0 0 1 .

609

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

In this table each model predicts a dichotomous outcome with the main explanatory variable—the indicator for Southern white rural males, and a series of control variables. Overall, these models indicate that Southern white rural males are more supportive of violence in three of the six models. Specifically, the odds ratio for the HITCHILD model (coefficient ¼ 1.768) indicates that being a South- ern white rural male increases the odds of responding yes to this item by 77 percent. Another way to interpret this effect is to convert the odds ratio to a probability. This is done simply by multiplying the odds ratio by the original odds of saying yes (which is the percentage of all respondents indicating yes on this item divided by the percent- age which said no), and dividing by this new figure plus one. To illus- trate, the odds ratio is 1.768. The original odds can be expressed as 61.6=(100 – 61.6), which equals 1.604. This figure is then multiplied by 1.768, which gives a figure of 2.836 (1.768� 1.604 ¼ 2.836). This number is then divided by itself plus 1 (2.836=3.836), which results in 0.739, or 73.9 percent. Thus, net of controls, being a Southern white rural male increases the probability of responding yes to this item by about 12 percent (73.9 – 61.6 ¼ 12.3).

The coefficients for the HITDRUNK and HITMARCH models indicate that being a Southern white rural male increases the prob- ability of responding yes to these items by 7.8 percent and 2.2 percent respectively holding all other variables in the model constant, and the proportional change in the probability is 113 percent and 81 percent respectively. While the increase in these probabilities is relatively modest, it is important to point out that overall support for these two scenarios is very low to begin with (6.9 percent and 2.7 percent respectively, see Table 1). Overall, these models indicate that being a Southern white rural male is associated with a greater probability of supporting violence for the HITCHILD, HITDRUNK, and HITMARCH items. In contrast, Southern white rural males are no more likely than other segments of the population to indicate approval for the HITOK, HITBEATER, or HITROBBER items. It is notable that had these items all been collapsed into indices, these variations in the effect of the key explanatory variable may have been concealed.

In terms of the control variables, the odds ratios indicate that those who are never married have lower odds of responding yes to the HITOK item, but are no different than their contrast group on any of the other items. Interestingly, having a history of exposure to viol- ence increases the odds of supporting violence under several of the scenarios. The odds ratios for the age measure indicate that as people get older, they generally become less supportive of the use of violence,

610 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

T a b le

3 .

L o g is ti c re g re ss io n m o d el s p re d ic ti n g vi o le n t a tt it u d es

w it h ru ra l w h it e m a le s in

th e so u th

a t a g e 1 6 a n d co n tr o ls

D ep en d en t va ri a b le s

H IT

O K

H IT

B E A T E R

H IT

C H IL

D H IT

D R U N K

H IT

M A R C H

H IT

R O B B E R

K ey

ex p la n a to ry

v a ri a b le :

R u ra l w h it e m a le s in

so u th

a t 1 6

1 .1 4 4 (. 2 0 8 )

1 .0 1 9 (. 2 1 6 )

1 .6 9 9 ��

(. 1 7 5 )

2 .4 5 5 ��

� (. 2 1 1 )

1 .9 6 6 � (. 3 0 9 )

1 .3 0 3 (0 .2 0 1 )

C o n tr o l v a ri a b le s:

N ev er

m a rr ie d

.7 3 9 ��

� (. 0 6 3 )

.8 7 7 (. 0 8 5 )

0 .8 0 5 ��

� (. 0 6 0 )

1 .1 5 2 (. 1 1 7 )

1 .4 0 6 (. 1 7 6 )

0 .9 3 4 (0 .0 8 4 )

T h re a te n ed

o r sh o t a t

1 .2 1 3 ��

(. 0 6 2 )

.8 9 6 (. 0 7 8 )

1 .4 0 6 ��

� (. 0 5 8 )

1 .0 9 9 (. 1 0 4 )

1 .5 2 2 ��

(. 1 5 5 )

1 .1 3 6 (0 .0 8 3 )

H it b y a n o th er

p er so n

1 .3 7 4 ��

� (. 0 5 2 )

1 .1 1 0 (. 0 6 9 )

1 .1 4 8 � (. 0 4 9 )

.9 4 6 (. 0 9 3 )

.9 3 0 (. 1 4 7 )

1 .3 5 8 ��

� (0 .0 7 0 )

A g e (i n ye a rs )

.9 9 3 ��

� (0 .0 0 2 )

.9 8 8 ��

� (. 0 0 3 )

0 .9 7 7 ��

� (. 0 0 2 )

1 .0 1 3 ��

� (. 0 0 3 )

1 .0 0 7 (. 0 0 5 )

0 .9 8 8 ��

� (0 .0 0 3 )

E d u c a ti o n (i n ye a rs )

1 .1 3 3 ��

� (. 0 0 9 )

1 .0 8 2 ��

� (. 0 1 2 )

1 .0 1 4 (. 0 0 9 )

.9 2 6 ��

� (. 0 1 6 )

.8 2 9 ��

� (. 0 2 5 )

0 .9 7 9 (0 .0 1 2 )

In co

m e (i n d o ll a rs )

1 .0 0 0 (. 0 0 0 )

1 .0 0 0 (0 .0 0 0 )

1 .0 0 0 (0 .0 0 0 )

1 .0 0 0 (0 .0 0 0 )

1 .0 0 0 (0 .0 0 0 )

1 .0 0 0 (0 .0 0 0 )

C o n st a n t

.7 0 7 � (. 1 7 2 )

3 .3 0 9 ��

� (. 2 2 1 )

2 .1 4 5 ��

� (. 1 5 8 )

.1 0 3 ��

� (. 3 0 0 )

.2 3 0 ��

� (. 4 6 3 )

1 9 .0 1 1 ��

� (0 .2 5 6 )

M o d el

ch i- sq u a re

3 4 6 .1 0 8 ��

� 1 1 8 .4 7 7 ��

� 2 9 4 .2 3 9 ��

� 8 0 .2 9 4 ��

� 1 3 5 .4 2 4 ��

� 1 3 3 .8 9 0 ��

�

N a g el k er k e R -s q u a re

0 .0 5 4

0 .0 2 4

0 .0 4 4

0 .0 2 3

0 .0 6 8

0 .0 2 7

C o ef fi c ie n ts

a re

o d d s ra ti o s w it h st a n d a rd

er ro rs

in p a re n th es es .

� p <

0 .0 5

�� p <

0 .0 1

�� � p

< 0 .0 0 1 .

611

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

with the exceptions of the HITDRUNK and HITMARCH items. The opposite pattern holds for years of education, where increases in educational attainment are associated with elevated odds of sup- porting violence for two of the indicators (HITOK, HITBEATER), and depressed odds of supporting violence for the HITDRUNK and HITMARCH items. Finally, the measure of respondent’s income is generally not associated with support for violence.

Table 3 provides the same series of logistic regression models with the alternative operationalization of the main independent variable, White rural males who lived in the rural South at 16 years old. The intent of this model is then to try and get at potential differences between those who were socialized in the rural South and all others. The results from these models closely parallel the results reported in the previous table. For example, like the previous table, rural white males in the South at 16 have higher odds of supporting violence for the HITCHILD, HITDRUNK, and HITMARCH items. The probabilities for these models indicate that being a rural white male who lived in the rural South at age 16 is associated with an increase in the probability of responding yes to these items of 11.6 percent, 8.5 percent, and 2.5 percent respectively when all other variables in the model are held constant. These probabilities correspond relatively closely to the estimates derived from the previous series of logistic regression models, and provide another layer of support for the idea that Southern rural white males are more supportive of violence in some situations.

In terms of the performance of the other variables in the models, individuals who were never married were less likely to indicate support for the HITOK and HITCHILD items. In contrast, those who had been exposed to some sort of gun related violence in the past had a higher odds of indicating support for the HITOK, HITDRUNK and HITMARCH items, while those previously exposed to physical violence had a higher odds of indicating support for the HITOK, HITCHILD, and HITROBBER items. The effects of the other variables in the models are almost identical to the coeffi- cients observed in the previous table, meaning that changing the operationalization of the main independent variable simply does not affect their performance overall.

DISCUSSION AND CONCLUSIONS

This analysis was designed to test the emerging Culture of Honor model set forth by Nisbett and Cohen (1996). In contrast to prior research which focused almost exclusively either on a regional

612 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

(Southern) subculture of violence or a racial (black) subculture of violence, the Culture of Honor model takes into account the intersec- tion of region, race, gender and locational context. Drawing on his- torical migration and settlement patterns, this model suggests that Southern white rural males in particular will be more approving of violence than other demographic groups when used in defense of honor, family, or personal property.

Prior research on this issue has provided some support for the model, but the conclusions of most studies should be treated with some caution. Macrolevel studies that look at ecological variations in homicide do not capture the essential component of the theory— cultural attitudes. In contrast, some available microlevel studies do examine attitudes but conflate numerous outcomes into a single (sus- pect) summary measure, or do not actually measure how the main demographic group of interest differs from other segments of society. Our analysis was therefore designed to extend prior research on this topic. Our results indicate that both Southern white rural males and white rural males that lived in the rural South at 16 years of age are more likely to support the use of violence under three of the six sce- narios we considered: (1) if an adult male stranger had hit the man’s child after the child accidentally damaged the stranger’s car, (2) if an adult male stranger was drunk and bumped into the man and his wife on the street, and (3) if an adult male stranger was in a protest march showing opposition to the other man’s views. But our target group did not differ significantly from the rest of the sample on any of the other three items we examined.

In interpreting our results, one thing that stands out most clearly is that Southern white rural males are more likely than the rest of the sample to support the use of violence under the three scenarios that have the lowest levels of support overall according to the descriptive statistics. For example, large majorities of the respondents answered in the affirmative to the HITOK, HITBEATR, and HITROBBR items. In contrast, less than two-thirds of the respondents indicated an affirmative response to the HITCHILD item, and only roughly seven percent and two and half percent of respondents support the use of violence for the HITDRUNK and HITMARCH items.

While we do not consider this overwhelming support for our hypotheses, the results do present an intriguing interpretational issue. On the one hand, the fact that our target group does not differ signifi- cantly from the rest of the sample on half of the outcome items may be viewed as a refutation of our hypotheses. On the other hand, the evidence that Southern white rural males are more supportive of viol- ence under scenarios where overall support is moderate to low may

Southern Culture of Honor 613

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

be construed as at least partially supportive of our expectations. While much extant theorizing on the subculture of violence issue emphasizes the defensive nature of Southern violence in particular, it is also possible that among those reared in this subcultural orien- tation, a broader range of situations may come to be encompassed under the umbrella of defensive violence. If this is the case, then we would expect that adherents to the culture of honor would be more inclined to approve of violence under scenarios where overall support in the population at large is lacking, because they are going to be more likely to view the context as a threat to honor, status, family, and so forth. When framed in this manner, we suggest that our results indicate at least partial support for our expectations.

Perhaps the most severe limitation of this study is the problematic nature of trying to operationalize the concept of defensive violence. As we noted above, previous research such as that by Dixon and Lixotte (1987) has relied on a factor analytic approach, which even on its face seems wholly inadequate. It seems to be the case that defensive violence is truly in the eye of the beholder. A good example of this may be the HITMARCH item contained in the GSS that we employ. While the most tolerant segments of society would probably see an affirmative response to this item as an offensively violent atti- tude, who is to say that the individuals responding in the affirmative are not doing so because they view the other person’s actions as a direct threat to their way of life, their family, their honor, etc? Unless the respondents are actually queried for whether they think their response taps into defensive versus offensive attitudes, it is nearly impossible to disentangle this issue.

A second limitation to our data which may have undermined our ability to fully detect the culture of honor has to do with a decay in norms and values over time. As most regional subcultural theorists note, the roots of Southern violence appear to predate the civil war. While there is certainly merit to the intergenerational trans- mission thesis, it is unclear through how many generations the values supporting the culture of honor can or will survive. There is almost certainly a process of decay, as new generations that are supposedly steeped in this cultural tradition are also bombarded with counter- vailing influences as they are reared in modern times. It is unlikely that the degree of adherence to the culture of honor is as strong among current generations as it was, say, 100 years ago. This may be one reason why our data did not reveal a strong effect for all of the outcomes we analyzed. In other words, it may be that the grip the culture of honor has on its adherents has weakened over time. Evidence buttressing this argument may also be contained in the fact

614 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

that rates of Southern violence, while still eclipsing that of other regions, are only marginally higher than the Western region of the country. That is, the distinctiveness of the South may be declining.

Future research could greatly expand on this line of thinking by undertaking several primary tasks. First, it would greatly enhance our understanding of the culture of honor if a larger variety of sce- narios were made available to respondents. Few of the questions available in the GSS probe potential responses to direct threats to the respondent. In light of the emphasis of the culture of honor on defensive violence, such items appear warranted. Additional items probing responses to direct threats to, say the sexuality of a man’s wife, the chastity of his daughters, or his own masculinity may further illuminate the nature and extent of the culture of honor.

Second, the concept of honor itself is typically presumed to be cap- tured in some of the available questions, but it may be useful to actu- ally ask questions to respondents about whether they view a particular action as a threat to their personal honor. Those who are heavily invested in the culture of honor may be more sensitive to slights, and as noted above, may have a broader range of issues which they feel entail issues of honor. Evidence to this end is mainly speculative at this juncture, but future researchers who are undertaking primary data collection would do well to consider this alternative.

REFERENCES

Anderson, Elijah. 1999. Code of the Street: Decency, Violence and the Moral Life of the

Inner City. W.W. Norton and Company: New York, NY.

Blau, Judith R. and Peter M. Blau. 1982. ‘‘The Cost of Inequality: Metropolitan

Structure and Violent Crime.’’ American Sociological Review 47:114–129.

Cao, Liquin, Anthony Adams, and Vickie J. Jensen. 1997. ‘‘A Test of the Black Sub-

culture of Violence Thesis: A Research Note.’’ Criminology 35:367–379.

Cash, Wilbur Joseph. 1941. The Mind of the South. Alfred A. Knopf: New York, NY.

Chu, Rebekah, Craig Rivera, and Colin Loftin. 2000. ‘‘Herding and Homicide: An

Examination of the Nisbett-Reaves Hypothesis.’’ Social Forces 78:971–987.

Cohen, Dov. 1996. ‘‘Law, Social Policy, and Violence: The Impact of Regional Cul-

tures.’’ Journal of Personality and Social Psychology 70:961–978.

——— 1998. ‘‘Culture, Social Organization and Patterns of Violence.’’ Journal of

Personality and Social Psychology 75:408–419.

Cohen, Dov, Richard E. Nisbett, Brian J. Bowdle, and Norbert Scwarz. 1996.

‘‘Insult, Aggression and the Southern Culture of Honor: An Experimental Eth-

nography.’’ Journal of Personality and Social Psychology 70:945–960.

Corzine, Jay and Lynn Huff-Corzine. 1989. ‘‘On Cultural Explanations of Southern

Homicide: Comment on Dixon and Lizotte.’’ The American Journal of Sociology

95:178–182.

Southern Culture of Honor 615

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

Corzine, Jay, Lynn Huff-Corzine, and Hugh P. Witt. 1999. ‘‘4. Cultural and Subcul-

tural Theories of Homicide.’’ Pp. 42–57 in Homicide: A Sourcebook of Social

Research, edited by M. Dwayne Smith and Margaret A. Zahn, Sage Publications:

Thousand Oaks, CA.

Curtis, Lynn A. 1975. Violence, Race and Culture. Health Publishing.

Davis, James, A. Tom W. Smith, and Peter V. Marsden. 2001. ‘‘General Social Sur-

veys, 1972–2000: (Cumulative File)’’ Computer file. 3rd version. Chicago, IL:

National Opinion Research Center (producer), 2001. Storrs, CT: Roper Center

for Public Opinion Research, University of Connecticut=Ann Arbor, MI: Inter-

university Consortium for Political and Social Research (distributors).

Dixon, Jo and Alan J. Lixotte. 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 Subculture of Viol-

ence Thesis.’’ Social Forces 69:1223–1239.

Ellison, Christopher G., and Patricia McCall. 1987. ‘‘Region and Violent Attitudes

Reconsidered: Comment on Dixon and Lizotte.’’ The American Journal of Soci-

ology 95:174–178.

Fox, John. 1991. Regression Diagnostics: An Introduction. (Sage University Paper Ser-

ies on Quantitative Applications in the Social Sciences, series no. 07–079). Sage

University Press: Thousand Oaks, CA.

Gastil, Raymond D. 1971. ‘‘Homicide and a Regional Culture of Violence.’’

American Sociological Review 36:412–427.

Hackney, Sheldon 1969. ‘‘Southern Violence.’’ The American Historical Review

74:906–925.

———. 1972. ‘‘Origins of the New South in Retrospect.’’ The Journal of Southern

History 38:191–216.

Harris, Anthony R., Stephen H. Thomas, Gene A. Fisher, and David J. Hirsch.

2002. ‘‘Murder and Medicine: The Lethality of Criminal Assault 1960–1999.’’

Homicide Studies: An Interdisciplinary and International Journal 6:128–166.

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.

Kleck, Gary. 1979. ‘‘Capital Punishment, Gun Ownership, and Homicide.’’

American Journal of Sociology 84:882–918.

Lixotte, Alan J. and David J. Bordura. 1980. ‘‘Firearms Ownership For Sport and

Protection: Two Divergent Models.’’ American Sociological Review 45:229–244.

Loftin, Colin and Robert H. Hill. 1974. ‘‘Regional Subculture and Homicide: An

Examination of the Gastil-Hackney Thesis’’. American Sociological Review

39:714–724.

Menard, Scott. 1995. Applied Logistic Regression Analysis. Sage Series: Quantitative

Applications in the Social Sciences. Sage University Press: Thousand Oaks, CA.

Messner, Steven F. 1983a. ‘‘Regional and Racial Effects on the Urban Homicide

Rate: The Subculture of Violence Revisited.’’ American Journal of Sociology

88:997–1007.

———. 1983b. ‘‘Regional Differences in the Economic Correlates of the Urban

Homicide Rate.’’ Criminology 21:477–488.

616 T. C. Hayes and M. R. Lee

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0

Nisbett, Richard E. 1993. ‘‘Violence and U.S. Regional Culture.’’ American Psycho-

logist 48:441–449.

Nisbett, Richard E. and Dov Cohen. 1996. Culture of Honor: The Psychology of Vio-

lence in the South. Westview Press: Boulder, CO.

Pampel, Fred C. 2000. Logistic Regression: A Primer. Sage Series: Quantitative Appli-

cations in the Social Sciences. Sage University Press.

Parker, Robert N. 1989. ‘‘Poverty, Subculture of Violence and Type of Homicide.’’

Social Forces 67:983–1007.

Reed, John S. 1971. ‘‘To Live—and Die—in Dixie: A Contribution to the Study of

Southern Violence.’’ Political Science Quarterly 86:429–443.

———. 1982. ‘‘Below the Smith and Wesson Line: Southern Violence.’’ Pp. 139–

153, From One South: An Ethnic Approach to Regional Culture. Louisiana State

University Press: Baton Rouge, LA.

Sampson, Robert J. and Dawn J. Bartusch. 1998. ‘‘Legal Cynicism and (Subcul-

tural?) Tolerance of Deviance: The Neighborhood Context of Racial Differ-

ences.’’ Law & Society Review 32:777–805.

Shaw, Clifford R., Henry McKay, Frederick Zorbaugh, and Leonard S. Cottrell.

1929. Delinquency Areas: A Study of the Geographic Distribution of School Truants,

Juvenile Delinquents, and Adult offenders in Chicago. University of Chicago Press:

Chicago, IL.

Shihadeh, Edward S. and Wesley Shrum. 2004. ‘‘Serious Crime in Urban Neighbor-

hoods: Is There a Race Effect?’’ Sociological Spectrum 24:507–533.

Smith, M. Dwayne and Robert Nash Parker. 1980. ‘‘Type of Homicide and Vari-

ation in Regional Rates.’’ Social Forces 59:136–147.

Swidler, Ann. 1986. ‘‘Culture in Action: Symbols and Strategies.’’ American Socio-

logical Review 51:273–286.

Wolfgang, Marvin and Franco Ferracuti. 1967. The Subculture of Violence. Tavistock

Publication: London, UK.

Wyatt-Brown, Bertram. 1982. Southern Honor: Ethics and Behavior in the Old South.

Oxford University Press: Oxford, UK.

Southern Culture of Honor 617

D o w n l o a d e d B y : [ U N C U n i v e r s i t y o f N o r t h C a r o l i n a a t P e m b r o k e ] A t : 1 6 : 4 4 2 1 A u g u s t 2 0 1 0