RKBA: Who are the killers - criminals or "law abiding citizens"? by Robobagpiper Abstract (aka, the TL;DR version): The intentional homicide rate in the United States in 2012 was 5.2 per 100,000 - lower than during the peak crime wave of the 1990s, but still unacceptably high compared to other industrialized nations. It has become axiomatic to state, then, that Americans are an intrinsically violent people. This sentiment is especially strong among those proposing to strongly restrict the instrument used in most homicides; specifically firearms, on one of two behavioral models - either that possession of firearms is itself the cause of the higher rates of intentional homicide; or that the populace is so intrinsically violent that the means must be removed from easy access, to protect Americans from themselves. These models rely on the assumption that high homicide rates (as well as ancillary violence) are endemic to the population as a whole. A review of the data, however, shows this to be contrafactual. Combining Bureau of Justice Statistics data on murder defendants with information about juvenile homicide defendants, it can be demonstrated that the overwhelming majority of homicide is confined to those with prior felony convictions, and that no more than 18.5% of homicides are committed by those with no prior convictions, putting the ceiling for the criminal homicide rate of "law-abiding citizens", that is, persons without a prior adult or juvenile criminal conviction, in the 1.1 per 100,000 range, a number that compares favorably with countries considered to have low homicide rates, such as most of Europe and Canada, and in which rates of access to defensive firearms is far lower than in the United States. A separate calculation based on homicides by CPL permittees, who have had to pass criminal background checks, produces a similar result of 0.62 per 100,000. By contrast, individuals with a prior felony conviction commit homicides at the rate of approximately 62.7 per 100,000, a factor of at least 57 times higher than persons with no prior criminal convictions, and a factor of 100 times that of CPL permittees. With this in mind, it is possible to conclude that American culture is not particularly prone to homicide per se, even with nearly half of households possessing the main instrumentality of violence, but rather, that its small subset of career criminal sub-culture is unusually violent. It is indeed true that (http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2010/crime-in-the-u.s.-2010/offenses-known-to-law-enforcement/expanded/expandhomicidemain) 42% of homicides were precipitated by arguments, while only 23% took place in the course of another felony; and that 53% of victims were killed by someone they knew, and 25% by family members. These facts are, however, a red herring. The assertion that the majority of homicides are committed by people with a criminal record should not be taken that they necessarily did so in the course of a criminal act against an innocent victim. Criminals have friends, family, and acquaintances too - and among those acquaintances are their partners, customers, suppliers, and rivals in criminal enterprise, and the friends and families of the same. The question is whether people unconnected with criminal enterprise are likely to engage in homicide on par with individuals with criminal records. Gun control advocacy groups routinely insist that they are. The facts show this claim, however, to be not only false, but prejudiciously so. The principal data used were from the Bureau of Justice Statistics "Felony Defendants in Large Urban Counties" reports for 1998, 2000, 2002, and 2006 (http://www.bjs.gov/index.cfm?ty=pbse&sid=27). The data for 2006 was the most recent available for download; 2004's report was just a summary, with limited tabulated data, so was unusable. The reports contain a wide variety of statistics about felony defendants in the largest urban counties in the US. Because these statistics are collected for only the 75 most populous urban counties, there is the possibility that extrapolating these statistics nation-wide may bias the results; however urban areas represent a disproportionate fraction of homicides, so this will dampen any effects the peculiarities rural homicides might have when extrapolating the urban results nation-wide. It should be noted that the BJS is receiving these statistics from state courts, who are in the business of adjudicating cases, and not tracking the prior criminal records of defendants. Criminal records are reported haphazardly, leading up to 50% of arrest records lacking important information such as final disposition (http://www.bjs.gov/content/pub/pdf/rporchr.pdf). Similarly, when counties note whether a defendant has prior criminal offenses in a given case, generally such a note applies only to prior offenses in that county. Because of this, it should be taken as a given that each category of prior conviction among homicide defendants will suffer from underreporting. What we are seeing is a minimum fraction of the homicide defendants that have prior criminal records, and by extension, the maximum fraction of defendants without. Also, as we shall see, because prior *juvenile* records are generally sealed, they are not counted in the statistics. Because of this, we will be forced to estimate their contribution, shown below. I have chosen to focus on convictions, rather than arrest records, to make the estimate more conservative. But it is important to note, in this context, that it is generally understood that there as many as 100 crimes committed for every 2 reported to police (source this!); and that someone can have a long criminal career with only a few arrests and fewer convictions; and that some career criminals will escape a record entirely. The first statistic of relevance to the question of who are committing homicides is the frequency of prior conviction. Averaged between 1998 and 2006, approximately 43% of murder defendants had at least one prior felony conviction as an adult (those that did averaged around two), with another 12% with only at least one misdemeanor conviction on their adult record (again, those that did averaging several). A larger number turns up in a tabulation of prior arrests. 75% of murder defendants have a prior adult arrest record; and of these a median of 5-9 prior adult arrests. Approximately 80% of those that had any prior arrests had an adult felony arrest, and averaged several adult felony arrests. An additional 8.5% of murder defendants were under 18 at the time of arrest, and could not have an adult criminal record. Juvenile records are generally sealed, but some cities have provided data on the subject. Boston, in particular, is revealing - 77% of youth knife and gun homicide offenders between 1990 and 1994 had prior juvenile arraignments, 54% of whom had been sentenced to probation. (Kennedy, David M. et al. 1996. "Youth Violence in Boston: Gun Markets, Serious Youth Offenders, and a Use-reduction Strategy." Law and Contemporary Problems. Vol. 59, no. 1. P. 148). By comparison, averaging the results from the four FDiLUC reports, 75% of adult murder defendants had prior arrests. Though not all arrests lead to arraignment, and only a fraction of convictions lead to probation, the youth offenders appear to have similar rates of prior criminal records to adult defendants (77% juvenile arraignments vs. 75% adult arrests, 54% juvenile probation sentences vs. 55% adult convictions). If anything, since not all arrests lead to arraignment, and not all sentences include probation, youth defendants appear to have slightly *higher* rates of prior offense than adult defendants, possibly a result of adult undercounting (from both exclusion of their own juvenile records, and from counties reporting mainly prior offenses of adults, who are more mobile, in the jurisdiction where the homicide took place). Acting conservatively, for the purposes of this analysis, youth homicide offenders' rates of prior offense, and severity, will be treated as essentially the same as the adults'. This leaves 36.5% of adult murder defendants with no adult conviction record. Again, this does not mean that a third of murder defendants were "law-abiding citizens until they pulled the trigger". Criminologist Don Kates notes: "The fact that only 75% of murderers have adult crime records should not be misunderstood as implying that the remaining 25% of murderers are non-criminals. The reason over half of those 25% of murderers don't have adult records is that they are juveniles. Thus, by definition they cannot have an adult criminal record." (Kates, Don B., et. al, Guns and Public Health: Epidemic of Violence or Pandemic of Propaganda? 61 Tenn. L. Rev. 513-596, 1994) Thus, the 36.5% of adult murder defendants with no adult conviction record includes three distinct populations: 1) persons engaged in no history of criminal activity prior to their adult arrest for murder; 2) persons engaged in criminal activity as adults prior to their adult arrest for murder, but who had escaped conviction (or conviction was pending - see below), and 3) persons with a juvenile criminal conviction, who escaped further conviction before being arrested after 18 for murder. Another statistic bears on this question - the percentage of persons, at the time of their arrest, that were in the justice system already. 33% of all murder arrestees were either on probation, in custody, incarcerated, or on parole at the time of their murder arrest. Because of the comparatively long times between arrest and adjudication, an offender can rack up multiple additional offenses before his first conviction. The 36.5% of adult murder defendants without an adult conviction record should then be considered "conviction record unknown" rather than "no conviction record", because any potential juvenile records are sealed. The only available approach is to estimate, based on the known records of adult and juvenile offenders; assuming that nothing magical occurs on the 18th birthday. One should expect continuity between the typical prior criminal history of murder defendants arrested six months before their age of majority as those arrested six months after. As such, it is reasonable to suspect that those adult murder defendants without a prior adult conviction record will have prior juvenile records at the same rate as juvenile murder defendants, which is itself very similar to the rates of adult defendants with known prior adult records. Using this extrapolation of prior juvenile offense rates iteratively estimate the juvenile conviction rates of the adult murder defendants, one arrives at the following estimate of the prior criminal records of murder defendants, averaged from 1998-2006. 43% adults with adult felony convictions (and possible juvenile convictions) 12% adults with adult misdemeanor convictions only (and possible juvenile convictions) 20.5% adults with juvenile convictions (felony/misdemeanor ratio unknown) EST. 16% adults with no prior convictions EST. 5% juveniles with juvenile convictions (felony/misdemeanor ratio unknown) 3.5% juveniles with no prior record If the ratio of prior felony to prior misdemeanor only conviction rates for adult murder suspects holds for juveniles, and that adults with misdemeanor-only records have typical juvenile records, this can be broken down further via a second iteration. 43% adults with adult felony convictions (and possible juvenile convictions) 5% adults with adult misdemeanor convictions only, and juvenile felony convictions EST.^2 15% of adults with juvenile felony convictions. EST.^2 4% of juveniles with juvenile felony convictions. EST. ____ Min 67% with prior felony record, juvenile or adult. 7% adults with adult and juvenile misdemeanor convictions. EST^2 5.5% adults with juvenile misdemeanor convictions. EST^2 1% juveniles with misdemeanor juvenile convictions only EST. ____ Min 13.5% with prior misdemeanor record only, juvenile or adult. 16% adults with no prior convictions EST. 3.5% juveniles with no prior convictions ____ Max 18.5% with no prior convictions, juvenile or adult. I have not tried to estimate the effect of undercounting in these statistics from counties reporting, primarily, prior offense rates of their defendants in the county, leaving offenses out-of-county unreported, or other potential sources of undercounting. I will simply note that the fraction of murder defendants with some sort of felony conviction is a minimum possible value, and that the fraction of murder defendants with no prior convictions at all is a maximum possible value. So, the summary: A *minimum* of 67% of murder defendants are estimated to have at least one adult or juvenile felony conviction. A *minimum* 13.5% of murder defendants are estimated to have no felony convictions, but some sort of misdemeanor conviction, adult or juvenile. A *maximum* of 18.5% of murder defendants are estimated to have no prior convictions, adult or juvenile. At this point, we have to note that we've so far been talking about *defendants*, not convicts. On average, 24% of murder defendants are not convicted; the overwhelming majority of those have their cases dismissed, with only a small fraction actually being acquitted. Many, if not most, of these will be justifiable or excusable homicides. One method of approaching this subset would be to subtract off justifiable and excusable homicides from the raw homicide total to produce the number of criminal homicides. In 2010, 665 justifiable homicides were reported by law enforcement to the FBI - 387 by police in the line of duty, and 278 by private citizens. (http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2010/crime-in-the-u.s.-2010/offenses-known-to-law-enforcement/expanded/expandhomicidemain). This does not include the related legal category of "excusable" homicides, nor does it include those homicides that led to arrest and prosecution but were dismissed by *courts* as justifiable or excusable, and does not include data from jurisdictions where police are not permitted to determine whether a homicide is justified or not in their report. Studies that compared detailed local homicide data in Dade County, Detroit, and Seattle to that reported as "justifiable" to the FBI suggests that the actual number of justifiable/excusable homicides, as determined by final court disposition, is more than four times higher than that reported as such to the FBI; equivalent to 10% or more of all total homicides. This suggests over sixteen hundred justifiable and excusable homicides a year by civilians (Kleck, Point Blank, pp. 111-116). Unfortunately, there are no statistics given as to the relationship between prior convictions of murder defendants and their conviction rate. Given the close correlation between criminal history and murder charges, it would not seem unreasonable to suspect that a disproportionate number of dismissals and acquittals would be among the population least likely to be charged with murder in the first place - those with no criminal history; however, as persons with a criminal history are also themselves at the highest risk of victimization by criminal violence (a topic for a future essay), and thus more likely to engage in uses of force in acts of self-defense, civilian legal homicides (justifiable and excusable) will applied to the raw homicide number, and only police line-of-duty homicides to the "no criminal record" group exclusively. Needless to say, with at least 82% of homicide defendants having at least one prior criminal conviction, and on average, multiple convictions, with only a maximum of 18% lacking any prior criminal conviction, adult or juvenile, the notion - promulgated almost universally in gun control circles - that the majority of people who commit homicide would be "considered law-abiding before they pulled the trigger" is revealed to not only be wrong, but a dangerous and offensive misrepresentation of the facts. Indeed, among that 18% of homicide defendants with no prior conviction will be many with a long *arrest* record, and a significant number will either be awaiting trial for a prior offense for which they have not been convicted yet. This disproportionate representation of felony convicts among murder defendants has strong implications for America's unfortunately high homicide rate. It is instructive, then, to break down the national homicide rate by criminal history of the perpetrator. If we are to calculate rates of homicide, by prior conviction record type, we need to tease out the fraction of the population with felony records, and the number with non-felony conviction records. Unfortunately, these statistics are not available to the BJS, who only report the number of Americans who have served time in prison (5.6 million in 2001 - http://www.bjs.gov/index.cfm?ty=qa&iid=404). It is possible to back-calculate this number from the FDiLUC reports, which break down the sentencing of felony defendants into severity; the number who received prison time, jail but no prison, and probation but no jail or prison. Typically only 38% of felony convicts are given prison sentences - suggesting that the total number of people with felony convictions is around 2.6 times the number who have served time in prison. This suggests that as of 2001, some 14.7 million people, or 5.2% of the population, had an adult felony conviction record in that year. The number cited for total number of Americans with a criminal record is typically 30% (http://www.nelp.org/page/-/65_Million_Need_Not_Apply.pdf?nocdn=1), after possible redundancies are accounted for. It's difficult to estimate the number of these with convictions, but about a third of felony arrests never lead to conviction, suggesting that the percentage with a conviction is closer to 21% (cited in the above source). Subtracting off the estimate fraction with a felony record, this suggests that 16% of the population has an adult misdemeanor conviction, but no felony convictions, on their record. 79% of Americans, then, would have no criminal conviction record at all. In 2012, there were 16,275 homicides (http://www.cdc.gov/nchs/fastats/homicide.htm); reducing this by the estimate of 10% that are determined by police report, court, or jury to be excusable or justifiable, there are were an estimated 14648 criminal homicides (plus police line-of-duty homicides) that year. The US population in 2012 was 313,000,000. Using these, and the above estimated breakdowns of murder defendants by estimated total prior number of homicides committed, we get: At least 9814 criminal homicides are estimated to have been committed in 2012 by persons with a prior felony conviction as an adult or juvenile; namely 5% of the population. The homicide commission rate by persons with a prior felony conviction is at least 62.7 per 100,000. At least 1977 criminal homicides are estimated to have been committed in 2012 by persons with a prior misdemeanor conviction, but no felony, as an adult or juvenile; namely 16% of the population. The homicide commission rate by persons with a prior misdemeanor conviction is at least 3.9 per 100,000. At most 2710 non-civilian legal homicides are estimated to have been committed in 2012 by persons with no prior convictions as juvenile or adult; namely 79% of the population. 387 of these were line-of-duty police homicides, leaving 2323 criminal homicides committed by this group. The homicide commission rate by persons with no criminal conviction record is *at most* 1.1 per 100,000. In other words, the homicide rate among Americans with no record of involvement in criminal enterprise is in a range similar to that for the Canada and most of Europe, but with a much, much higher rate of access to "weapons 'designed' solely to kill people", overwhelmingly pistols. It should be noted that all of these overstate the number of *murderers* in each subpopulation; as they are calculating the number of homicide victims produced by each group, and as some murderers kill several people in their lifetimes, the killer-to-victim ratio is greater than one. The comparative low number of homicides attributable to "law abiding citizens" might seem a startling result to some, especially in the context of long-standing gun control doctrine, and if possible, it would be desirable to compare it to a similar number using a different methodology. One presents itself in the study of homicide rates among CPL (Concealed Pistol License) permit holders. CPL permittees invariably must pass a criminal background check prior to an issuance of their license, ensuring that they will, at least in that area, be similar to the larger general population without prior conviction. CPL permittees would, however, differ in another major area - they are much more likely at any given moment to be armed than the general law-abiding population. The Violence Policy Center, no friend to gun owners, maintains a list, culled from media reports, of the number of persons killed by individuals "legally allowed to carry a concealed handgun" (http://www.vpc.org/CPLkillers.htm), with the inflammatory name of "Concealed Carry Killers". Leaving aside what the organization is attempting to broadly imply about CPL permittees by way of its title, it should be possible to calculate a rough estimate of the homicide rate commission among CPL permittees. Possibly in response to criticisms of the results presented below, they assure us that their numbers are an "unknown fraction of the unreported number of similar incidents that routinely occur across the nation", but culled the media reports sufficiently thoroughly to manage to include multiple incidents of death by negligent discharge among the intentional homicide numbers they claim. Furthermore, VPC's count includes incidents involving people carrying in states where no permit or background check is required to carry, where the perpetrator could just as easily have a criminal record as not, biasing their number instead significantly upwards. Additionally, the list deceptively counts the suicides in a murder-suicide, and often the offender when shot by police, as "victims" of the concealed carrier's homicide, and in several other cases, counting those killed by a stolen or borrowed gun as victims of the concealed carrier! Most strikingly, and with no justification whatsoever, a full quarter of VPC's counts are simple suicides among CPL permittees from Michigan, reported in bulk by the state. Subtracting out all suicides by the CPL permittees, deaths by persons other than the CPL permittee (including police), unintentional shootings, and those killed in states not requiring a permit, the VPC count yields 315 criminal, or potentially criminal (as many cases are pending) homicides in the period from May 2007 through August 2013 (6.33 years). A July 2012 GAO report estimated, based on responses from states, that in 2011 over 8 million Americans possess CPL permits, and conceded their value likely suffers from significant undercounting. A calculation of these numbers puts the approximate criminal homicide rate among CPL permittees at about 0.62 per 100,000 per year - a close match to the above upper-bound estimate of 1.1 per 100,000 per year for all persons lacking a prior felony conviction. The VPC's assurance that their data set is a lower bound should be countered with the fact that the GAO's estimate of the number of CPL permittees is *also* a lower bound due to state under-reporting; and that the VPC's inclusion of 62 pending criminal cases also biases their number upwards, considering that homicide cases have non-conviction rates of around 24%. These biases will act against the bias towards undercount in the corrected VPC's statistic. At the same time, it should be remembered that virtually all of the CPL permittees are armed, and significant number of them are armed in public on a regular basis. The fact that this sub-population commits homicide at rates comparable to, or (cautiously) lower than, the broader segment of the population with no prior criminal conviction record, strongly indicates that easy access to a firearm, in itself, is a very poor predictor of the rates of commission of criminal homicide. < Full conclusion paragraph, especially discussing implications of how this knowledge should affect gun control policy. I plan to point out stats that show handgun ownership among criminals is higher than among the general public, and they do it for protection, mainly, themselves. The big recommendation will be the de-escalation of the drug war, and concentrated urban jobs efforts, to hit the homicide/violence problem hardest where it's worst. > Raw numbers from BLJ FDiLUC reports averaged above: Year: 1998 2000 2002 2006 Murder defendants: % Min. 1 conviction: 70 47 56 56 % Min. 1 felony conviction: 54 34 42 42 % Min. 1 prior arrest: 81 67 72 81 Under 18 at arrest: 9 10 8 7 % On probation/parole/in custody/incarcerated at arrest: 45 30 27 31 % Not Guilty/Case Dismissed: 32 30 17 18 % Felony Convictions: % of felony convictions to prison: 34 40 38 40 VPC numbers VPC claim: 516 people killed by "concealed carry killers" as of 9/2013. After performing a cursory examination of their report, the following was found: 11 were in Arizona, killed by non-permittees. 36 are the suicide component of murder-suicides. 12 were ruled accidental. 8 were perpetrators shot by police or other intervening parties. And 2 were shot not by the CPL holder, but by someone who stole his gun. An astounding 129 of simple suicides from Michigan, and 3 from other states (one of whom was included because the gun was borrowed from a CPL permittee at a gun range), were included in the count. This leaves 315.