index Crime in the Nation's Capital

Trying to Define Both the Problem and the Solution


In September of 2003, the FBI published its annual Uniform Crime Report detailing all types of crimes committed across the US. And after a several year lapse, DC moved back almost to the top of the list of murders per capita amongst all American cities. It is a reason for substantial embarrassment in the nation's capital, as well as a serious concern for its residents. It also raises a variety of questions about how to mitigate the problem.

As is NARPAC's wont, we have rummaged about in the readily available statistics in the hopes of finding some cogent quantitative explanation for the problem, as well as a corollary path towards a long-term solution. Such hopes have not been fully rewarded. Nevertheless, it is possible to try to put the problem in some perspective without intending to appear dismissive.


For understandable reasons, the impact of homicides on daily life is accorded more anguish and concern than it deserves relative to others means of becoming a mortality statistic. Murder rates have in fact been declining in the US and DC, if for no other reason than some of their root causes have also been declining (primarily poverty among young males). Murder rates are not a good indicator of overall crime levels. They are however statistically related to the circumstances suffered by US minorities, particularly in an urban environment. The root problem will be better mitigated by addressing the plight of single moms in poverty, the over- representation of minorities in urban public schools, and the associated excessive depressing drop- out rates, than by adding cops to the beat.

The following sections essentially use annotated charts to grope with the issue:

1) homicides are not a major cause of deaths in the US

Despite the understandable effrontery of being killed, often unexpectedly and sometimes by a stranger, one's chances of being murdered are relatively small compared to being taken by any of the major diseases or accidental causes. In fact, there are still more people killing themselves than each other. And the combination of the two in automobile accidents still deserves far more attention. Nationwide in 1999 (according the latest Statistical Abstract of the US), drivers killed almost three times as many people with their cars as did criminals with weapons. However, it should be noted that in DC, homicides killed four times as many as cars. In most of these major categories, DC (a city) ranks worse than the national averages consolidated at the state level. By these comparisons, DC is also very close to being the cancer capital, the diabetes capital, and the HIV capital of the US.

2) homicides are generally not random things

With the exception of kids who are mostly (82%) killed by adults (top line), and females who are mostly (90%) killed by males, most murders are between two persons of the same race and sex. And whites are somewhat more likely to kill females, and blacks are somewhat more likely to kill males.

Equally interesting, the lower chart also shows that 75% of the victims in 2002 (where any relationship was determined) knew their killers. 35% knew them well, and 40% knew them more casually. Nevertheless, being killed by a stranger, and one of a different sex, age category, or race remains really quitesmall indeed.

3) homicides are not always carried out with guns

This chart, below, also drawn from data in the Statistical Abstract) indicates that there has been virtually no change in the fraction (67%) of homicides carried out with firearms. Though one is most likely to die of a bullet wound, knives, blunt objects (including fists), and many other available items also contribute to the death toll.

4) most murderers are in the prime of life

It is also clear from FBI/UCR statistics that those who commit homicides tend to be in their late teens and early twenties, whether they live in cities or the suburbs. Half of all such crimes are committed by age 23 in urban areas, by age 25 in more ruralareas, as shown below. (But be careful of FBI designations which place most Americans in "cities", many of which hardly earn the title of village.)

5) murder has been declining significantly in the US

The murder rate has been declining significantly over the past decade from an unusual peak in the early '90s. This is true for both DC and the country as a whole, as shown below. This is based on city data by police district from DC's Metropolitan Police Dept, and national data from the Statistical Abstract. While murders spiked in DC in '96 (for reasons unknown to NARPAC), they have consistently remained somewhere near 2% of all US homicides(!), even though DC is home to less than 0.2% of the US population. Without question, US cities are crime-prone.

6) jail/prison populations have special characteristics

Not surprisingly, those incarcerated do not fit the average American mold, as shown below, based on sketchy data in the latest available (1998!) Annual Dept of Justice report on Criminal Populations. The left column applies to the broad category of all those convicted of serious crimes, which the middle column applies only to those of death row. Hence it is clear that the criminal population is: disproportionately young, male, and black; four times more likely to have dropped out of high school; and slightly more likely to have come from a one-parent home. These clearly are not independent variables, but begin to show the obvious demographic tilt of those locked up, and in some cases awaiting execution.

7) murder rates are very sensitive to jurisdictional boundaries

It should also be clearly understood that crime rates are particularly sensitive to the boundaries included when there is a mix of high and low murder rates in adjacent jurisdictions. This chart shows how DC's statistics would be changed by two jurisdictional moves. Move #1 transfers the high poverty/high crime part of the city East of the Anacostia (EoA) to the State of Maryland which can absorb it with only minor changes in statistics. Conversely, Move #2 brings Alexandria and Arlington back into DC (i.e. the missing part of the diamond West of the Potomac (WoP) with similar effects. The enormous difference in crime rates between the two jurisdictions WoP and EoA deserves further study in itself. In either move, the DC murder rate is magically reduced by about 30%, while the effected neighboring states register very small changes in their very favorable crime rates.

8) there's a lot of serious crime to contend with besides murder

While it would be inappropriate to suggest that homicide is simply a quirk amongst serious crimes, one should surely be more concerned about the likelihood of being victimized in other, albeit less lethal, ways. Murder and non-negligent manslaughter are only a bit more than 2% of all Violent Crimes (which also include forcible rape, robbery, and aggravated assault), and violent crimes ("V-crimes" to NARPAC) are less than 40% of all Property Crimes (including burglary, larceny-theft, motor vehicle theft and arson). And these categories, which make up the FBI's "Crime Index" ("I-crimes" to NARPAC), are just about one-quarter of total arrests made by law enforcement officers in 2002. The click-up graphic to the left is drawn from NARPAC's recent challenge to the recent flaky GAO study trying to relate DC's needed police force levels to only its murders, not its overall crime requirements. (Police force sizing is discussed again below)

9) there are large variations in relative murder, violent crime, and index crime rates

Contrary to the speculation by the GAO, the number of murders in a particular jurisdiction has very little to do with the overall number of violent crimes or index crimes or the number of full time police. In fact, there are probably other first order terms which enter the proportionality equation. The long click-up chart to the right here ranks the 41 cities in the US that in 1996 had either more than 500,000 persons in residence, or experienced at least 100 murders. They are ranked here by murders per 100,000 residents, violent crimes (including murder) per 10,000 residents, and index crimes (which include violent crimes) per 1000 residents. Note that while DC is very near the top in murder rate (red bars), it is perhaps sixth in violent crime rate (green bars), and tenth or lower in index crimes (blue bars).

An even more stark example of this variation in serious crime elements is shown by comparing the ratio of crime rates for DC to those of the entire US. DC's murder rate hovers around 8x the national average, while DC maintains about 3-3.5x the national violent crime rate, and at best a rather stable 2x the index crime rate. While NARPAC has no firm idea of why this should be, it may very well be related to crime opportunities: there are many parts of DC where it may be easier to shoot someone in an out-of-control dispute than commit profitable larceny or auto theft. But it is also interesting to see how little these ratios have changed, even though overall crime rates have dropped substantially.

10) there is little correlation between crimes and law enforcement personnel

If there is little correlation between crime types (above), there is even less in the setting of police force levels. This click-up chart shows how many residents (red bars), violent crimes (blue bars), and minority residents (green bars) there are per full time law enforcement personnel (uniformed and civilian). On this chart, for variety (!) small numbers (short bars) are "good", and long bars indicate excessive police forces for the job at hand. This issue has been raised in several different contexts by NARPAC in the past. The cities are listed in the same order of decreasing murder rates.

11) there is little correlation between murder rates and city density, but.....

The left hand scatter chart below plots murders per thousand I-crimes against the black share of the population in the same 41 big US cities. The right hand twin plots I-crimes per thousand population against those same black population share. Although there is more than enough scatter to disappoint the analyst, one trend is clear: as the black population share increases, so does the murder rate, and so does the I-crime rate. The "spread" in the data indicates there are almost certainly many other forces at work as well. On the other hand, the red dots represent higher density populations while the green dots show the lower densities. There is no evident correlation indicating that urban density impacts consistently on murder or I-crime rates.

12) ...crime rates clearly relate to racial demographics....

The impact of race on crime categories can be shown on this oversimplified comparison. Black and white (including Hispanic) contributions to crime (foreground) are compared to their respective population shares (background). The first conclusion is that the "other" demographic categories (Asian, Native Americans, etc.) are seriously under-represented in the crime domain. The second is that the black population takes up their slack in the general categories, and is, for whatever reason (see ahead), over-represented in violent crimes including particularly homicides.

13) ....and perhaps more directly to other socioeconomic considerations

As is often the case, however, it is also possible to find informative trend lines that do not stress black population as the cause. In fact, the "tightest" scatter plot NARPAC has found to date shows a strong correlation between the overall murder rate and the prevalence of families with single moms whose kids are in public schools. Because single moms with public school kids tend to be black (and poor), there is of course understandable ambiguity as to whether skin color or economic plight is driving the crime rate. NARPAC opts for the latter. To be strictly correct, the crime data should lag the demographic data by about a decade to give the kids time to morph into full-blown, arrestable criminals.

14) and school demographics tend to amplify minority populations

It is also clear from this data base that, as has been illustrated somewhere else on this web site, school demographics tend to exaggerate racial minority trends. While NARPAC uses straight lines to show the "knee in the curve" (which analysts so admire), it is clear from this 41-city sample that as the minority share grows in the population, it grows substantially faster in the public schools. By the time the population has become 40% minority, the schools have become 80% minority and rise asymptotically towards 100%. This over-representation of minorities has, NARPAC believes, important implications for a rapid decline in school performance, including almost certainly a rapid rise in drop-out rates: the shortest path to incarceration.

15) big city crimes and crime types are clearly related to minority prevalence.

To suppress the scatter problem, NARPAC divided the 41 cities into five groups of continuously declining murder rates (same as the rankings on the prior charts) and aggregated the crime data within each cluster (see chart below). Minority shares (tan) and murder rates (red) decline together, while violent crimes (blue) decline more slowly, and the total index crimes (yellow) decline even less. Even across these five clusters totaling 41 cities, then, murder rates drop fourfold, violent crime rates drop to less than half, while index crimes fall less than 30%. Almost 40% of the nation's total murders occur in these cities whose population is only about 15% of the national total. These cities also harbor almost 40% of all the nation's families with kids that live below the poverty line (both two- and single-parent families).

16) but race is more a condition than a cause

The final chart in this series attempts to explore whether the shockingly high murder rate in the US is caused by race itself, or other socioeconomic circumstances. The chart can become intelligible with a little explanation. The purpose is to demonstrate that the number of murders in a given year is well documented for the numerator of the murder rate factor, but the denominator is subject to substantial refinement, which can impact quite differently on the white and black shares of total murders. As shown below, the number of murders are well known, and so are the gross population samples. In 2002, blacks committed 1.05 times as many murders as whites with 0.16x the population. This yields a 0.030 murder rate for whites and a 0.196 murder rate for blacks (per 100K persons). As shown in the central pink column, the black murder rate is 6.49 times as high as for whites.

However, in subsequent rows across the table, other considerations are introduced to refine the base population most likely to contribute to the crimes. These are done individually for each race in proportion to known circumstances. For instance, there are slightly more blacks in the 18-44 age group which commits most of the crimes. This reduces the ratio to 6.19. And so on, cumulatively down the table. There are more black males in poverty, more black single-mom households in poverty, disproportionately more blacks in public schools (see above), and those blacks have a higher drop-out rate, and may in fact be inclined to commit more crimes per perpetrator. While rigorous statisticians will object to simply factoring all of these considerations sequentially on top of one another, the fact remains that real-world considerations can probably effect the black/white murder rate ratio by a factor of 10. On this basis, there is no evident reason to introduce skin color as a fundamental source of violent behavior. Fix the socioeconomic conditions, fix the murder rates for everyone.

17) PS: crime involving school kids

This postscript takes a direct look at the latest available trend data on crimes against school kids from 1992 to 2000. The 8-part chart below considers total crimes on the left, and violent crimes on the right, differentiating between crimes committed in or very close to schools (red bars), and those further away and presumably not directly related to school (blue bars). Comparative trends are shown for national vs urban settings, male vs female victims, younger and older teenagers, and between white and black kids. In virtually every category, crime levels have been dropping both near and away from the schools, frequently by as much as 50%. Note that the scale for violent crimes is about half of that for total crimes, a much smaller spread than in the adult world.

We note the following trends or lack thereof:

o Crimes are somewhat higher in urban areas than across the nation as a whole, but most of the urban increase takes place away from schools, presumably on the urban streets;

o Females are victimized somewhat less frequently than males, but males are more likely to get into trouble, and violent trouble, away from school than females:

o The younger kids are more likely to become victims in and around their schools, while the older kids get into more crimes, and more violent crimes, away from the school environment; and

o White kids get victimized somewhat more than black kids (per capita) in or away from the school environment although the differences had become minimal by 2000.

The graphs below show how total arrests in the US for total (upper) and index (lower) crimes have peaked and then declined over the past ten years:

The fact remains, however, that even in 2000, 15 out of 100 school kids are involved in a crime of some sort each year in or outside of American urban areas, and better than one third of those will be subject to a violent crime, more likely outside of the school zone than within. Their odds of remaining safe will get considerably better as they get older and further from school! It is not a pretty picture.


For understandable reasons, the impact of homicides on daily life is accorded more anguish and concern than it deserves relative to others means of becoming a mortality statistic. Murder rates have in fact been declining in the US and DC, if for no other reason than some of their root causes have also been declining (primarily poverty among young males). Murder rates are not a good indicator of overall crime levels. They are however statistically related to the circumstances suffered by US minorities, particularly in an urban environment. The root problem will be better mitigated by addressing the plight of single moms in poverty, the over- representation of minorities in urban public schools, and the associated excessive depressing drop- out rates, than by adding cops to the beat.

Did you find this of interest?
Please give us your FEEDBACK

This page was updated on Jan 5, 2004



© copyright 2007 NARPAC, Inc. All rights reserved