| *** PRELIMINARY DRAFT - PLEASE DO NOT CITE OR QUOTE *** (Forthcoming, Homicide Studies, 1999) Effects of an Execution on Homicides in California ERNIE THOMSON University of Houston Clear Lake Abstract Contemporary research on the effects of executions on homicides has turned up two different results. Research examining short term effects of executions has sometimes found decreases in homicides after executions (a deterrent effect) while research examining longer term effects has sometimes found that homicides increase following executions (a brutalization effect). This study employed a quasi-experimental before-and-after analysis coupled with a disaggregation strategy to examine the short and long term effects of executions on different types of homicides in Los Angeles, California in the aftermath of the 1992 execution of Robert Alton Harris. Findings indicated that there were both short term deterrent and long term brutalization effects of the execution on different types of homicides, but the net effects included increases in overall homicides and most disaggregated types of homicides consistent with brutalization theory. In recent years, the longstanding debate over the question of whether or not capital punishment is a more effective general deterrent than alternative forms of punishment has taken a significant new turn.[FOOTNOTE 1 HERE] Numerous studies over a period of several decades have produced little evidence that the use of capital punishment leads to lower homicide rates, while mounting evidence from recent research indicates that executing accused killers might actually inspire some murders and lead to increased homicide rates, as brutalization theorists have argued. One of the basic rationales for capital punishment is the idea that executions serve as a stronger deterrent against murder than long prison terms and thus provide greater protection for society against this especially abhorrent crime (Bailey & Peterson, in press). Underlying this rationale is the assumption that in the wake of executions potential murderers will identify with the executed offenders, fear the possibility of being executed themselves, and therefore be deterred from subjecting themselves to possible execution by committing a capital crime. In short, they will �think twice� before committing a murder if one consequence of the murder might be their own death. Further, this deterrent effect is expected to be reinforced by the message that society, in imposing such a severe punishment for murder, highly values the sanctity of human life (Bailey & Peterson, in press). Capital punishment would thus be expected to strengthen general social inhibitions against lethal violence and lead to lower homicide rates. Brutalization theorists have challenged virtually all of the assumptions and arguments of deterrence theory. Proponents of the brutalization view point out that most homicides are more or less spontaneous acts that occur in the midst of highly charged social situations involving arguments and other disputes and thus are not as �rational� as proponents of deterrence imply. Instead, most homicides are spur of the moment acts that apparently don�t involve much thought in the first place and thus are unlikely to be influenced by second thoughts (Thomson, 1997). Brutalization theorists also argue that potential killers are probably as likely to identify with the executioner as with the person executed. As Bowers (1988, p. 53) points out, even potential killers would not likely identify with those who are executed since they tend to be very unattractive people accused of �... cruel or cowardly acts without provocation or remorse. They may have strangled small children, killed whole families, dismembered their victims, and the like. Will calculating potential murderers identify with such persons, or will they not infer that the death penalty is reserved as punishment only for people unlike themselves?� This is consistent with the oft quoted phrase from Beccaria that the state sets a �savage example� for society to follow when it executes its own citizens, and in the process devalues the sanctity of human life rather than reinforcing it (see Bailey & Peterson, in press). According to brutalization theorists, then, executions would be expected to weaken general social inhibitions against lethal violence and ultimately lead to higher homicide rates. THE RESEARCH EVIDENCE As Berns (1979, pp. 87-88) has pointed out, it is very difficult for researchers to isolate and measure the effects of capital punishment on homicide because of the many different variables that influence homicide rates. Early studies in the United States and several other countries, designed to overcome this problem, used approaches comparing homicide rates in abolitionist and retentionist jurisdictions, comparing homicides in a single jurisdiction that had either abolished or reinstated capital punishment, or examining the short term effects of executions on homicides (see Bowers, 1988 for a more detailed review of these studies). As Bailey & Peterson (1997) and Bowers (1988) have pointed out, the findings from these studies have uniformly failed to support deterrence theory. Some more recent researchers have employed sophisticated techniques like time series designs or multivariate analysis, and the findings from these studies have been mixed. Ehrlich (1975), Phillips (1980), Layson (1985), and Stack (1987) are among the few studies that have reported finding deterrent effects of capital punishment, but these studies have been effectively discredited on methodological grounds (see Bowers, 1988 for a critique of Ehrlich, Phillips, and Layson, and Bailey & Peterson, 1988 for a critique of Stack). Meanwhile, a number of other studies using these techniques have either reported evidence of brutalization effects (e.g., King, 1978, Bowers & Pierce, 1980, and Decker & Kohfeld, 1990), or have found little support for either view (e.g., Peterson & Bailey, 1991. Unlike earlier reviewers of the long tradition of deterrence studies, most of whom concluded that the deterrence issue remained intractably unresolved (e.g., Baldus & Cole, 1975; Barnett, 1981) contemporary reviewers have mostly concluded that the evidence is now �overwhelmingly contrary to deterrence theory� (Bailey & Peterson, 1997, p. 153; also see Harries & Cheatwood, 1997). These reviewers, though, have stopped short of endorsing Bowers� (1988) conclusion that the balance of the evidence now clearly supports brutalization theory. As Bailey & Peterson (1997, p. 153) have pointed out, the recent studies reporting brutalization effects of executions have not yet been subjected to close scrutiny and thus it is difficult to draw any real conclusions from these studies. They also point out that a number of basic issues have not yet been adequately addressed by proponents of brutalization theory, including the questions of what kinds of people are likely to be inspired to kill by an execution and in what circumstances this response is likely to occur (1997, p. 153-154). Bowers (1988) indirectly pointed to the same problem when he acknowledged that the increases in homicides following executions are typically quite small and thus hard to interpret. Similarly, in their recent comparative study of the deterrence issue using county level data, geographers Harries & Cheatwood (1997, p. 107) found no support in their data for deterrence theory. Pointing out that their capital punishment variables accounted for only very small changes in homicide levels, they went on to report that the correlations that did appear between capital punishment and violent crime were positive, a finding which is consistent with brutalization theory (p. 108). The Disaggregation Strategy In their study of the aftermath of an execution in Oklahoma, Cochran, Chamlin, & Seth (1994) proposed an explanation for both the difficulty that researchers have had in isolating the effects of executions on homicides and the small increases in homicides found in so many previous studies. They argued that much of the deterrence research suffers from aggregation bias. The idea is that executions are likely to have different effects on different kinds of homicides, and aggregating these different kinds of homicides into a single measure used as the dependent variable (�the homicide rate�) leads researchers to overlook these different effects (1994, p. 116). To address this problem they formulated a research strategy that involves disaggregating homicides into types that can be theoretically related to potential deterrent and brutalization effects of executions. Cochran et al. (1994) took advantage of a naturally occurring quasi-experimental set of circumstances in Oklahoma to employ their disaggregation strategy for the first time. Like a number of other states, Oklahoma stopped executions in the 1960's, and the execution of Charles Troy Coleman in 1990, after a twenty-five year hiatus, provided a suitable situation for a before and after quasi-experiment. Using the disaggregation strategy, they examined three hypotheses. First, they predicted that the Coleman execution would have a deterrent effect on felony-related homicides, those that they thought most closely approximate �rational� killings in which a potential offender might be likely to consider the personal consequences of committing a homicide and �think twice� before killing. Second, they predicted that the execution would have a brutalization effect on �stranger� killings, which they argued are more likely to involve spur of the moment responses to social disputes in which a potential offender�s inhibitions against violence are already somewhat relaxed. Third, they predicted that the execution would have no significant impact on the overall (aggregated) homicide rates because deterrent and brutalization effects would nullify each other in the overall rates. They found that felony homicides did decrease slightly following the execution, but the decrease was not statistically significant and thus their deterrence hypothesis was rejected. A further analysis of several subtypes of felony homicides (death-eligible felony homicides, robbery homicides, etc.) also led to null results. They found that overall homicide rates increased slightly after the execution, but again the increase was not statistically significant, which supported their hypothesis regarding overall changes. Finally, a large, statistically significant, increase was found in stranger homicides and also in several subcategories of stranger homicides (e.g., stranger-nonfelony and stranger-argument), each consistent with the predictions of brutalization theory. In their interpretation of these findings, they suggested that executions have a demoralizing and dehumanizing effect on society and weaken social and cultural inhibitions against the use of violence to settle social disputes, thus producing �... a brutalization effect in situations where prohibitions against killing are weakest and where the offender perceives having been wronged� (1994, pp. 129-30). In an examination of the effects of a 1992 execution in Arizona, Thomson (1997) used the disaggregation strategy to both replicate and expand the approach of the Oklahoma study. Like Oklahoma, Arizona stopped executions in the 1960's, and the 1992 execution of Donald Eugene Harding was the first in Arizona in almost thirty years. Thus Arizona was an appropriate setting for a replication of the Oklahoma study. The findings from this study were mostly consistent with those of the Oklahoma study. In each area where deterrence theory would predict decreases in homicides following an execution (felony homicides, robbery homicides, etc.) small to moderate increases in homicides were found, along with a moderate increase in overall homicides. As in the Oklahoma study, hypotheses based on brutalization theory were supported by the Arizona data - there were large increases in homicides following the Harding execution in each category where brutalization theory would predict such increases (stranger homicides, argument homicides, nonfelony homicides, etc.). Large increases in nonstranger homicides were also found, which raised some questions about the part played by the relationship of offender and victim in the effect of executions on homicides. The expansion of the disaggregation strategy in this study involved examining a number of homicide categories that were not considered in the Oklahoma study. Thomson (1997) argued that if there is a brutalization effect, homicides by teen offenders should increase following an execution because teens tend to be more impulsive and imitative than adults and are probably less likely than adults to consider the personal consequences of their actions. Deterrence theory, on the other hand, would predict a decrease in homicides by adults following an execution because they are probably more likely to identify with the person executed and are also more likely to consider the personal consequences of their actions. He also argued, based indirectly on the assumptions of brutalization theory, that gun-related homicides would increase in the aftermath of an execution since it is apparently the ready availability of guns that converts many disputes into spur of the moment homicides. Findings from the Arizona study were mostly consistent with the predictions of brutalization theory while failing to support the predictions of deterrence theory. Large increases in teen offender homicides and gun-related homicides were found, along with a much smaller increase in adult offender homicides and a small decrease in homicides involving weapons other than guns. Thomson (1997, p. 124) concluded with the suggestion that the brutalization effect occurs primarily in nonfelonious social conflict situations, especially among teenagers, and where firearms are readily accessible. In summary, the Oklahoma and Arizona studies, like much of the previous research in this area, consistently failed to find empirical support for deterrence hypotheses while consistently finding increases consistent with brutalization theory in various types of homicides and with various types of offenders following executions. Evidence of a �Death Dip� Following Executions As the evidence in support of brutalization theory has increased in recent years, some researchers have continued to explore what Stack (1995) has termed the �death dip,� the notion that executions deter homicides primarily in their immediate aftermath (Phillips, 1980; Phillips & Hensley, 1984; Stack, 1987, 1990, 1995). The death dip research typically examines the effects of executions on overall (aggregated) homicide rates within a few days, or at most a few weeks, following executions, and one proponent of this approach has summarily dismissed the issue of longer term effects by arguing that examining the effects of executions over a longer period of time involves a �questionable methodological technique� (Stack, 1995, p. 174). Death dip researchers have consistently reported decreases in homicides in the near aftermath of an execution (or an �execution story� appearing in the media). Phillips (1980) examined the effects of executions on homicide rates in London between 1858 and 1921. Although he reported sizable decreases in homicides during the week of and the week following executions (a death dip effect), homicides increased again over the next few weeks leaving a net effect of slight increases in homicides. In a follow-up study, Phillips & Hensley (1984) examined the effects of several publicized life sentences, death sentences, and executions on homicides in the United States, and again reported a statistically significant short-term drop in homicides. As Bowers (1988) has pointed out, though, an attempt to replicate Phillips� (1980) London study with data from the United States failed to support the death dip hypothesis, while the Phillips & Hensley (1984) study failed to separate the effects of executions and death sentences from those of life sentences, thus raising questions about whether the death dip was related to deterrent effects of capital punishment at all. Stack (1987) examined the effects of publicized executions on homicides in the United States between 1950 and 1980 and reported statistically significant decreases in homicides during months that execution stories appeared in the media. In other studies, he found death dips following publicized executions in South Carolina (Stack, 1990) and a drop in homicides nationwide by Caucasian offenders, but not African American offenders, in the four days following publicized executions in the United States from 1977 through 1984 (Stack, 1995). Stack�s studies have been criticized for a variety of methodological flaws, and a replication of his 1987 study found that �... merely correcting [Stack�s] coding errors for the execution publicity variables eliminated the reported significant association between execution publicity and homicide rates.� (see Bailey & Peterson, 1997). Despite the convincing critiques of the death dip research, the fact remains that there seems to be at least some suggestive evidence that there is a death dip following executions. For example, a careful examination of data presented in the Oklahoma study to illustrate overall homicide trends (see Cochran et al. 1994, p. 113) clearly shows a decrease in homicides in the three or four weeks following the highly publicized Coleman execution before homicide levels in Oklahoma increased in the following months. Similarly, a re-examination of the Arizona SHR data used by Thomson (1997) shows a decrease in homicides during the month of the Harding execution before overall homicide rates in Arizona increased over the next few months. The death dip research, along with the evidence cited above, suggests that the effect of executions is complicated. The death dip research, which has focused on aggregated homicide rates and short term effects of executions, appears to suffer from both offense and temporal aggregation bias, while research based on the disaggregation strategy suffers from temporal aggregation bias in not more closely examining the short term effects of executions on disaggregated homicides. The study presented here used homicide data from California to examine some of the issues raised by this recent research. METHOD This study used a quasi-experimental before-and-after analysis of average monthly criminal homicides in the city of Los Angeles (FOOTNOTE 2 HERE) to examine the impact of the 1992 execution of Robert Alton Harris on homicide rates in that city. (FOOTNOTE 3 HERE) Data came from the CD-ROM version of FBI Supplemental Homicide Reports (SHRs) for California for 1992 and included the population of cases for that year rather than a sample. (Fox, 1994. See Appendix A for a further description of the data set and definitions of variables). Since SHR data include both criminal and noncriminal homicides (homicides committed in the line of duty by police, citizen homicides involving self-protection, and a few other categories) noncriminal homicides were purged from the data set. Comparison periods were the four months immediately preceding and the eight months immediately following the Harris execution. Using the disaggregation strategy developed by Cochran et al. (1994) and Thomson (1997), the first part of this study involved a simple before-and-after analysis of the effects of the Harris execution on different types of homicides which can be logically related to deterrence and brutalization theories. Based on the assumptions of deterrence theory, decreases in felony and adult offender homicides were expected in the aftermath of the execution. Based on the assumptions of brutalization theory, increases in nonfelony, gun/knife, and teenage offender homicides were expected. Based on the assumption that deterrence and brutalization effects tend to nullify each other in overall homicide levels, no changes in overall homicides were expected. In the second part of the study, changes in homicide levels in the immediate aftermath of the execution (the month following the execution) were examined for evidence relating to the death dip hypothesis. This was followed by use of the disaggregation strategy to examine and compare short and long term changes in the kinds of homicides that might be subject to this particular kind of deterrent effect. Based on the findings of previous death dip research and the apparent death dip in the Oklahoma and Arizona studies cited above, a death dip in homicides following the Harris execution was expected. Further, since the death dip hypothesis is based on the assumptions of deterrence theory, short term decreases were expected to be larger for felony and adult offender homicides than for nonfelony and teen homicides. FINDINGS Cochran et al. (1994) used weekly data collected by a state agency in their groundbreaking study of the return to executions in Oklahoma, and the large number of time periods yielded by weekly data allowed them to use ARIMA (autoregressive integrated moving average) techniques to control for trend, drift, and autocorrelation. Unfortunately, weekly SHR data for California are not available, and the much smaller number of time periods based on monthly data precludes the use of ARIMA techniques. The key issue of homicide trends in Los Angeles was examined by looking at raw unadjusted SHR homicide levels from 1981 through 1992. Homicide levels in Los Angeles remained remarkably stable from 1981 through 1988, ranging from a low of 1000 to a high of 1114. In 1989, about the time the national crack epidemic began, there was a large increase in homicides (1431) followed by another large increase in 1990 (1639). In 1991, homicides began to decrease (1573) and a larger decrease followed in 1992 (1455), the year of the Harris execution. Thus there was an overall downward trend in homicides in Los Angeles in the year preceding the Harris execution. Table 1 shows the changes in average monthly homicides following the Harris execution for eleven different categories of homicides. Previous research using the disaggregation strategy found small increases in overall homicides following executions and the pattern in California is similar - overall homicides increased by 22 percent in the eight months following the Harris execution. ------------------------------------------------------------------------------------------------------------------------------------- Table 1. Changes in average monthly homicide levels in the city of Los Angeles following the 1992 Harris execution. ------------------------------------------------------------------------------------------------------------------------------------- Homicide Type % Change Average Monthly Homicides Before/After Execution Overall + 22 106/129 Adult offender + 13 52/59 Felony + 4 27/28 Robbery + 6 16/17 Other weapon - 4 26/25 Stranger + 13 30/34 Teen offender + 59 22/35 Nonstranger + 26 61/77 Nonfelony + 26 68/86 Social Conflict + 31 62/81 Handgun/Knife + 30 79/103 SOURCE: Data from CD ROM version of FBI Supplementary Homicide Reports (Fox, 1994). See Appendix for description of variables. -------------------------------------------------------------------------------------------------------------------------------------- Previous research also found little empirical support for hypotheses based on deterrence theory and the California data are also consistent with this pattern. Small increases in felony homicides (4 percent increase), robbery homicides (6 percent increase), and adult offender homicides (13 percent increase), along with a small decrease in other weapon homicides (4 percent decrease) were found following the Harris execution. Previous research has found support for hypotheses based on brutalization theory and the results of this study are consistent with these findings. Moderate to large increases were found in social conflict homicides (31 percent), handgun/knife homicides (30 percent), nonfelony homicides (26 percent), and teen offender homicides(59 percent). Each of these findings is consistent with the view that the brutalization effect occurs mainly in homicides involving teenagers and/or social disputes where a lethal weapon is readily available. Unlike some previous research, though, a moderate increase was found in nonstranger homicides (26 percent) along with a smaller increase in stranger homicides (13 percent). The following discussions compare in more detail the findings of this study with those from previous research. Circumstances of the Homicide Cochran et al. based their deterrence hypothesis on the view that a deterrent effect would be most likely in situations where potential offenders might consciously �... anticipate the need for lethal force for the successful completion of another crime.� (1994, p. 129). In the near aftermath of an execution, this anticipated need for lethal force might be counterbalanced by a conscious calculation which includes the risk of execution if lethal force is actually used. As a result, some potential offenders might �think twice� about committing a homicide, and the aggregate result would be a decrease in felony homicides. Accordingly, Cochran et al. suggested that felony homicides, especially those involving robberies (the largest category of felony murders) should be deterred by an execution. However, they found only small and statistically insignificant decreases in felony and robbery homicides, and concluded that their Oklahoma data did not support the deterrence hypothesis. Thomson (1997) reported a small increase in felony homicides along with a moderate increase in robbery homicides in Arizona following the Harding execution, and concluded that his data also failed to support the deterrence hypothesis. Findings of small increases in felony and robbery homicides in California following the Harris execution were generally consistent with those in the Oklahoma and Arizona studies and also provided no support for the deterrence hypothesis. These findings also seemed to indicate that these types of homicides are probably not influenced by a brutalization effect of executions either. Cochran et al. reported null findings for both nonfelony and argument homicides in Oklahoma, while Thomson found fairly large increases in each of these categories in Arizona. The moderate increases in nonfelony and argument homicides found in California were consistent with the Arizona study and were also generally consistent with the brutalization hypothesis - executions tend to increase homicides involving noncriminal social conflict. Relationship Between Offender and Victim Cochran et al. based their brutalization hypotheses on the notion that stranger homicides often involve social disputes in which �affronts to honor� lead to spur of the moment homicides, and thus they argued that stranger homicides would be the most likely type to be influenced by the brutalization effect of executions. Their data appeared to support this view when they found a large increase in stranger homicides following the Coleman execution, although they did not report findings for nonstranger homicides to compare with this increase in stranger homicides. Thomson also reported a large increase in stranger homicides following the Harding execution in Arizona, but he also reported a large increase in nonstranger homicides, a finding that raised doubts about whether relationship between offender and victim is really a relevant variable. The finding in the present study of a moderate increase in nonstranger homicides coupled with a smaller increase in stranger homicides leads to the conclusion that the relationship between offender and victim is not salient in explaining the effects of executions on homicide rates, a conclusion consistent with Thomson�s (1997, 119) view that �... where Cochran et al. find that an execution appears to inspire homicides in conflicts between strangers, our findings suggest that an execution is likely to inspire homicides in conflicts between both strangers and nonstrangers.� Age of Offenders In his expanded use of the disaggregation strategy, Thomson proposed that executions would likely have different effects on adults and teenagers. Based on assumptions of deterrence theory, he proposed that potential adult offenders should be deterred by an execution. Based on assumptions of brutalization theory, he proposed that executions should lead to increased teen offender homicides. He found a small increase in adult homicides following the Harding execution in Arizona along with a very large increase in teen homicides, and concluded that the Harding execution had no discernible deterrent effect on adult homicides but had a brutalization effect on teen homicides. The findings of a large increase in teen offender homicides and a small increase in adult offender homicides following the Harris execution in California are consistent with Thomson�s Arizona findings. The fact that juvenile homicide levels in California varied greatly between younger and older juveniles and the fact that the adult homicide category included such a wide range of ages (adult offenders in the California data ranged from 20 to 71 years of age) led to a closer examination of the age variable. It quickly became clear from the data that homicide offending increased sharply from 15 to 16 year olds and dropped sharply among those 30 and older. Based on these observations, the analysis was expanded from two age categories (teen and adult) to five: young teens (15 and under), older teens (16-19), adults 20-29, adults 30-39, and adults 40 and over. Of the 1,455 homicides in the data set, the age of offenders was known in 1,052 (72 percent). Of these 1,052 offenders, 48 were young teens, 326 were older teens, 476 were in their 20�s, 141 were in their 30�s, and 61 were 40 years of age or older. Following the execution in California, young teen homicides increased by 90 percent, homicides by older teens increased by 52 percent, homicides by offenders in their 20�s increased by 20 percent, homicides by those in their 30�s did not change, and homicides by those 40 and over decreased by 5 percent. This seemed to indicate that the brutalization effect was strongest among the youngest potential offenders (young teenagers) and decreased with age, finally disappearing completely among those 30 years of age and older. It was unclear whether the small decrease in homicides among offenders 40 and over represented a deterrent effect or simply the absence of a brutalization effect. Weapons Used in Homicides After pointing out that most homicides appear to be spur of the moment events and that the ready availability of lethal weapons plays a key part in escalating some social conflicts into homicides, Thomson (1997) proposed that gun-related homicides would increase following an execution due to the brutalization effect. He found a large increase in gun-related homicides after the Harding execution, which is consistent with the brutalization hypothesis, and a small decrease in homicides involving other weapons. There was a 20 percent increase in gun-related homicides, and a 34 percent increase in other-weapon homicides following the Harris execution in California, which raised some doubts about the part played by type of weapon. In examining this issue more closely, the weapons that seemed to fit the spur of the moment argument of brutalization theory, handguns and knives, were analyzed separately and then compared with �other firearms� and �other weapons� (weapons other than firearms and knives). This analysis indicated that in the aftermath of the Harris execution, handgun homicides increased by 28 percent, knife homicides increased 44 percent, other firearm homicides decreased by 10 percent, and other weapon homicides increased by 6 percent. This finding led to a reconceptualization of the weapon variable in which handguns and knives were combined into one category while other firearms and other weapons were combined into a second category. Analysis of these new categories found that handgun/knife homicides increased by about 30 percent following the execution while other weapon homicides decreased by 6 percent (see Table 1). This seemed to suggest that part of the brutalization effect of executions might be an increase in those homicides where the ready availability of a lethal weapon plays a part in escalating the violence of a social conflict. The Death Dip Hypothesis In the second part of the analysis, the question of whether there was a death dip following the Harris execution was examined by comparing average monthly homicides for four months preceding the execution with homicides during the month following the execution (the death dip month) and average monthly homicides for seven months following the death dip month. The disaggregation strategy was used to look for patterns consistent with deterrence and brutalization effects. The results are reported in Table 2. -------------------------------------------------------------------------------------------------------------------------------------- Table 2. Changes in average monthly homicide levels in the city of Los Angeles following the 1992 Harris execution. -------------------------------------------------------------------------------------------------------------------------------------- Homicide Type T-1 T-2 T-3 Death dip, followed by return to previous level: Felony 27 24 31 Adult offender 52 50 60 Robbery 16 9 18 Stranger 33 18 36 Death dip, followed by increase over previous level: Overall 106 95 134 Nonfelony 68 64 88 Soc. Conflict 59 53 83 Handgun/Knife 79 70 108 No death dip, followed by increase over previous level: Teen offender 22 23 37 Nonstranger 61 70 78 T-1 = average monthly homicides for 4 months before execution. T-2 = number of homicides for month following execution. T-3 = average monthly homicides for 7 months following T2. SOURCE: Data from CD ROM version of FBI Supplementary Homicide Reports (Fox, 1994). See Appendix for description of variables. -------------------------------------------------------------------------------------------------------------------------------------- As Table 2 shows, each type of homicide fit one of three patterns. For adult offender homicides, felony homicides, robbery homicides, and stranger homicides, there were decreases in homicides during the death dip month followed by a return to about the previous levels (actually small increases) in the seven months following the death dip month, resulting in net effects of small increases in each category. Three of these four (adult, felony, and robbery) were types of homicides that were expected to be most subject to the deterrent effects of an execution and in each case there was a death dip following the execution. For adult and felony homicides, though, the decreases were so small that they might very well be a result of the fluctuation normally found in monthly homicides. It is interesting that the largest death dips were found in robbery and stranger homicides. Harris was executed for the murder of two strangers in the course of preparing for a bank robbery. For overall, nonfelony, argument and handgun/knife homicides there were decreases in the death dip month followed by increases over previous levels in the following seven months. Here, too, the short term decreases were very small and it was not clear whether the death dips related to the execution or resulted from normal fluctuation in monthly homicides. With this caveat, this pattern could be interpreted as involving a combination of small short term deterrent effects and larger long term brutalization effects. In each case, the long term increase in homicides more than made up for the short term decrease, so that the net effect was a sizable increase in each homicide type. Three of these four categories (nonfelony, argument, and handgun/knife) involved types of homicides for which a brutalization effect was expected, and in the fourth category (overall homicides) small increases have been found in each of the studies using the disaggregation approach. In the remaining two cases, teen and nonstranger homicides, the pattern involved no death dip, followed by large increases after the death dip month. Teen offender homicides closely fit the logic of brutalization theory and the finding was consistent with a strong brutalization effect. The findings for stranger and nonstranger homicides seemed to reinforce the ambiguity of relationship between offender and victim as an explanatory variable, as noted above. SUMMARY AND CONCLUSIONS There are two key findings from this examination of the effects of the highly publicized execution of Robert Alton Harris on homicides in Los Angeles, California. First, in the month following the execution there were small decreases in most of the types of homicides examined consistent with the predictions of the death dip hypothesis. Further, the largest decreases were in the two types of homicides most similar to those for which Harris was executed, robbery and stranger homicides. While the small size of most of the short term decreases makes any interpretation questionable, these findings suggest that there might be a small short term deterrent effect of executions. Second, in the seven months following the death dip month there were increases in each type of homicide examined consistent with a brutalization effect. The smallest increases were in types of homicides theoretically least likely to be influenced by a brutalization effect (felony, adult, robbery) and the largest increases were in those categories most likely to reflect this influence (social conflict and teen offender homicides). These long term increases were much larger than the short term death dip decreases and the net effect was a sizable increase in most types of homicides along with a sizable increase in overall homicides in the months following the execution. These findings suggest that there is a long term brutalization effect of executions. While there is clearly a need for great caution in drawing conclusions from before and after studies of single executions, quasi-experimental studies employing the disaggregation strategy have begun to provide a clearer picture of the effects of executions on homicides. Bailey & Peterson (1997, p. 153-154) have suggested that a key weakness of brutalization theory is its failure thus far to address the basic issues of which potential offenders might be subject to a brutalization effect and in what kinds of circumstances this effect operates. While these studies do not provide conclusive answers to these questions, they do suggest specific directions that future research might take in looking for such answers. This approach has a number of other limitations which can also be addressed in the context of suggestions for future research. First, the executions examined in these studies were very highly publicized because they were the first executions to occur in these states for a very long time. The question of the differential effects of highly publicized versus less publicized executions, which has been a key concern of some of the death dip research (see Stack, 1995), has not yet been adequately addressed in studies using the disaggregation approach. At issue is the question of whether there is a change in the balance of short and long term effects following less publicized executions. Second, each of these studies looked at one execution with an aftermath uncomplicated by subsequent executions. Future research needs to examine the effects of multiple executions (for example, several executions during a single week or month or year) as well as the effects of different intervals between multiple executions over a period of time. For example, would an execution every month or every two months extend the short term death dip effect or increase the long term brutalization effect or both? Third, there are several problems that derive from the scope of these studies. No attempt has yet been made to employ the disaggregation strategy to address gender issues (are deterrent, brutalization, and death dip effects similar or different for male and female offenders?). Also, Thomson (1997) found that long term brutalization effects appear to be different for black and white offenders following the execution of a white prisoner in Arizona, and Stack (1995) found differences in the nationwide death dip effect for black and white offender homicides following highly publicized executions in the United States. The effect of race/ethnicity on these processes also needs to be examined in future research. Finally, the basic mechanisms through which deterrent and brutalization effects marginally influence individual decisions about whether or not to commit a homicide need clarification. In this respect, the logic of deterrence theory seems to be fairly clear. The idea is that in the aftermath of an execution at least some potential offenders will identify with the executed person, fear the threat of being executed themselves, and engage in a more or less conscious cost-benefit analysis in which the threat of execution outweighs any possible benefits from carrying out a homicide. This sketch not only makes good common sense, but also provides a framework for understanding likely different impacts of an execution based on variations in the component parts of the theory - those who are similar in terms of gender, age, and race or ethnicity to a person recently executed would arguably be more likely than those dissimilar to be deterred by an execution. The basic mechanism through which the brutalization effect influences homicide decisions has not been so clearly or so logically elaborated. Bowers & Pierce (1980) suggested that the brutalization effect acts on those already �primed to kill� while Cochran et al. (1994) proposed that the effect occurs mainly in emotionally-charged situations involving social conflicts between strangers. The Arizona study and this study located the effect more precisely among young potential offenders who are involved in social conflicts where lethal weapons are handy. Modeling and imitation of state executions, the lowering of social inhibitions against violence, and the devaluation of human life have all been suggested as possible effects of executions, and all of these would seem likely to have a stronger effect on the very young than on adults. But which if any of these factors underlie the brutalization effects of executions, and the specific mechanisms through which they operate remain to be clarified. APPENDIX: DESCRIPTION OF VARIABLES Overall homicides: all SHR homicides, with these exceptions: (1) Homicides involving a single offender and multiple victims were coded as one homicide, (2) Homicides involving multiple offenders were coded as one homicide per offender, (3) Homicides involving unknown offender(s) were coded as one homicide, (4) Homicides by police were omitted. Felony homicides: homicides involving concurrent felonies or suspected felonies. Nonfelony homicides: homicides not involving concurrent felonies. Stranger homicides: homicides involving strangers. Nonstranger homicides: homicides involving family members or acquaintances. Robbery homicides: homicides involving robberies. Social Conflict homicides: homicides involving arguments or youth gangs. Teen offender homicides: homicides involving offenders 19 years of age and under. Adult offender homicides: homicides involving offenders 20 years of age and older. Handgun/Knife homicides: homicides involving handguns or knives. Other weapon homicides: homicides involving weapons other than handguns or knives. NOTES 1. At issue in the deterrence debate has been the question of whether executions deter homicides by potential offenders more effectively than alternative sanctions like lengthy imprisonment. Proponents of brutalization theory argue that there is a marginal general brutalization effect of executions or, in effect, that executions tend to inspire homicides that would not have occurred in the absence of the executions. 2. Data for the city of Los Angeles instead of statewide data for California were used because careful examination of the 1992 statewide SHRs indicated that at least some jurisdictions failed to submit reports for at least part of 1992, while the data from the city of Los Angeles appeared to include complete reports for the whole year. The sample bias resulting from the use of data from a large urban area suggests special caution in generalizing findings to dissimilar areas. 3. The 1992 execution of Robert Harris, the first California execution since the 1960�s, was a controversial and highly publicized event which drew national and international interest and media coverage (Richards & Easter, 1992). Harris, who with his brother had murdered two teenagers in order to steal their car for use in a bank robbery, was held out by proponents of the death penalty as a prime example of the kind of offender that justified capital punishment (see Baker, 1992, and Lungren & Krotoski, 1992). The Harris case was regarded by critics of the death penalty, though, as emblematic of serious defects in the capital punishment adjudication process. Harris�s appeals lawyers pointed out that the case had been rushed to trial in the midst of hostile publicity and that Harris�s court-appointed defense attorney had been refused the time and resources necessary to adequately prepare for trial (see Sevilla & Laurence, 1992). As a result, Harris received a blatantly unfair trial which subsequent appeals courts refused to correct. REFERENCES Bailey, W.C. & Peterson, R.D. (1997). Murder, capital punishment, and deterrence: A review of the literature. In H.A. Bedau (Ed.), The death penalty in America (pp. 135-161). New York: Oxford University Press. Bailey, W.C. & Peterson, R.D. (in press). Capital punishment, homicide, and deterrence: An assessment of the evidence and extension to female homicide. In M.D. Smith & M.A. Zahn (Eds.), Homicide studies: A sourcebook of social research (pp.257-276). Thousand Oaks, CA:Sage. Baker, Steve. (1992). Justice not revenge: A crime victim�s perspective on capital punishment. UCLA Law Review, 40, 339-344. Baldus, D. & Cole, J. (1975). A comparison of the work ofThorsten Sellin and Isaac Ehrlich on the deterrent effect of capital punishment. Yale Law Journal, 18, 170-186. Barnett, A. (1981). The deterrent effect of capital punishment: A test of some recent studies. Operations Research, 29, 346-370. Berns, W. (1979). For capital punishment. New York:Basic Books. Bowers, W.J. & Pierce, G. (1980). Deterrence or brutalization: What is the effect of executions? Crime and Delinquency, 26, 453-484. Bowers, W.J. (1988). The effect of executions is brutalization, not deterrence. In K.C. Haas & J.A. Inciardi (Eds.), Challenging capital punishment (pp. 49-59). Newbury Park, CA:Sage. Cochran, J.K., Chamlin, M.B., & Seth, M. (1994). Deterrence or brutalization? An assessment of Oklahoma's return to capital punishment. Criminology, 32, 107-134. Decker, S.H., & Kohfeld, C.W. (1990). The deterrent effect of capital punishment in the five most active execution states: A time series analysis. Criminal Justice Review, 15, 173-191. Ehrlich, I. (1975). The deterrent effect of capital punishment: A question of life and death. American Economic Review, 65, 397-417. Fox, J.A. (1994). Uniform Crime Reports: Supplementary Homicide Reports, 1976-1992 [computer file]. ICPSR version. Ann Arbor:ICPSR [distributor]. Goetting, A. (1995). Commentary on Stack. Criminal Justice and Behavior, 22, 186-189. Harries, K. & Cheatwood, D. (1997). The geography of executions. London:Rowman & Littlefield. King, D.R. (1978). The brutalization effect: Execution publicity and the incidence of homicide in South Carolina. Social Forces, 57, 683-687. Layson, S.K. (1985). Homicide and deterrence: A reexamination of the United States time-series evidence. Southern Economic Journal, 52,68-89. Lungren, D.E., & Krotoski, M.L. (1992). Public policy lessons from the Robert Alton Harris case. UCLA Law Review, 40, 295-326. Peterson, R.D., & Bailey, W.C. (1991). Felony murder and punishment: An examination of the deterrence question. Criminology, 29, 367-395. Phillips, D.P. (1980). The deterrent effect of capital punishment: Evidence on an old controversy. American Journal of Sociology, 86, 139-148. Phillips, D.P. & Hensley, J. (1984). When violence is rewarded or punished: The impact of mass media stories on homicide. Journal of Communications, 34, 101-116. Richards, J.I., & Easter, R.B. (1992). Televising executions: The high-tech alternative to public hangings. UCLA Law Review, 40, 381-422. Sevilla, C.M., & Laurence, M. (1992). Thoughts on the cause of the present discontents: The case of Robert Alton Harris. UCLA Law Review, 40, 345-380. Stack, S. (1987). Publicized executions and homicide: 1950-1980. American Sociological Review, 52, 532-540. Stack, S. 1990). Execution publicity and homicide in South Carolina. Sociological Quarterly, 31, 599-611. Stack, S. (1995). The impact of publicized executions on homicide. Criminal Justice and Behavior, 22, 172-186. Thomson, E. (1997). Deterrence vs. brutalization: The case of Arizona. Homicide Studies, 1, 110-128. |