States, Race, Politics, & Murder
When homicide rates spiked in 2020, Republicans and Democrats pointed fingers at each other. Republicans accused Democrats of being soft on crime, and Democrats pointed to high homicide rates in Republican-leaning states. I want to revisit the homicide statistics because they are complicated by the enormous disparity between victimization among whites and blacks. The horrifyingly high murder rate of African Americans means that homicide rates may be elevated in a state because of a large African American population. How do comparisons among states look when we take this into account?
The Rates
First let’s review the total (race-independent) homicide rates (see figure), which come from the CDC (reference at end of this post). I display them with a size/intensity plot, representing each state with a box whose height and width shows the state’s population and murder rate, respectively. To save space I made 2 stacks of states rather than one long one. States are color-coded based on how they voted in the 2016 presidential election, blue if Clinton got a firm majority, red if Trump, and grey if the margin was relatively small.
Homicide rate is given in deaths per million, which I think is more useful than the usual deaths per 100,000. Populations of states, nations, and large cities or counties are usually given in millions, so this unit makes it easier to go from rates to total number of deaths.
The figure confirms the observation, made by many others before me, that the states with the highest murder rates are predominantly Republican-leaning, and most of the ones with the lowest rates tend to vote Democratic. There are a handful of low-homicide “red” states with rates of about 40/million or less, but these are low-population states that collectively have far fewer people than the red states with 130/million or more. Two “blue” states – MD and IL – are among the high-homicide states. More about that below.
The US average homicide rate in 2020 was about 80/million, and about half the population of the country lives in states with rates between 60 and 100 /million. However, the total range is much broader than that, a factor of ten difference between LA and MS (around 200/million) and ME, VT, and NH (around 20/million or less).
The Spike
Any discussion of recent changes in homicide rates should start with the reminder of the tremendous drop in murder that took place a generation ago. National homicide rates dropped from almost 100/million in 1990 to less than 60/million by 2000 and a low of under 50/million during the Obama administration.
Since then, rates crept up starting in 2015, and then spiked in 2020 from about 60/million to almost 80/million (see figure above). The cause of the spike is uncertain, but speculation has linked it to the two big changes that occurred in 2020, the Covid-19 pandemic and deteriorated community-police relations in the wake of the George Floyd protests and riots. Homicide rates increased further in 2021, but by a much smaller amount.
Whatever the cause of the spike, it must be something affecting virtually the entire nation, because the jump was ubiquitous. The figure above displays averages for 2017-2019 and 2014-2016. Homicide increased by a large amount (10-40%) in 29 states regardless of the partisan lean of the state or the prior homicide rate. Only 2 states (AK and NH) had decreases from the 2017-2019 average to 2020, and two (LA and MS) had increases of over 50/million.
Race
An unfortunate feature of this country is the very high murder rate for African Americans. For 2015-2016, homicide rates were around 30/million for non-Hispanic Whites, 50/million for Hispanics, and over 200/million for non-Hispanic Blacks. Most victims are of the same race as their murderer. We only know the race of the offender when someone is convicted, but among such cases, about 82% of White victims and 91% of Black victims were killed by someone from their own race according to FBI statistics for 2018.
Since the high Black victimization rate is tied to a high Black offender rate, it is important to not let the numbers feed negative stereotypes. Nevertheless, this is a key statistical feature which can not be ignored when trying to understand the problems of violence in America.
The large disparity in homicide rates means that the state homicide rate can be sensitive to the relative size of the African American population. Consider 2 states, each with a homicide rate of 30/million for Whites, and 200/million for Blacks. For each demographic group, the homicide rate is exactly the same in either state. But if state A has a 5% Black population, state B has a 30% Black population, and in both states the rest of the population is White, then the homicide rate will be 38.5/million in A and 81/million in B. The statistic might motivate people to look for a policy or economic conditions that makes state B have the worse rate, but really social/economic conditions are driving the same racial disparity in both. It’s just that the demographics are different.
To remove this factor, I plot separate White and Black homicide rates for each state. I use 2008-2016 data from Riddell et al., 2018. This source only includes men; since men have higher homicide rates than women, the rates here will be higher than for the whole population. As in the previous figure, individual states are colored according to political lean based on the 2016 election. The size of the square for a state represents the state population, and each state is marked to indicate the percentage of the Black population.
The figure is also shaded to divide each axis into terciles: about 1/3 of the nation’s White population lives in states in the horizontal dark band, which covers the range of White homicide rates from 20/million to 34/million; 1/3 live in states with lower rates and 1/3 live in states with higher rates. Similarly, the vertical dark band (230/million to 340/million) divides the Black population into equal segments with low, medium, and high rates.
The data shows that, even controlling for racial demographics, there is a tremendous range of homicide rates. For Whites, rates range from 10/million or less for RI and MN, to 50/million or more for 7 states. Black rates range from about 100/million (RI again) to over 600/million for MO.
Generally, the separation between high-homicide “red” states and low-homicide “blue” states (those with margins >15% in the 2016 election) persists when we separate White and Black rates. A few small Republican-leaning states (the largest of which is UT), with very small Black populations, also have low rates, but more states are in the high-homicide part of the graph. A square also represents the average for blue and red states. Homicide for each race is weighted by the state populations of that race. The average homicide rate for the red states is much higher than for the blue states for White men (about 38/million versus 22/million) and slightly higher for Black men (400/million versus 330/million).
The “Gulf Southern” states of MS, AL, LA, and AR have large Black populations, but their overall homicide rate is high because their White homicide rate is among the highest in the nation. Similarly, WV, KY, and TN all have small or moderate-sized Black populations, but the high White homicide rate produces high total rates. In MO, the White homicide rate is somewhat above the national average, and the Black rate is the highest of any state.
In contrast, all the most Democratic-leaning states have low-to-average homicide rates for both demographic groups, except for IL (high Black rate). RI, MA, and NY all have the low rates for both races.
Looking at the less partisan states, there is less separation. Moderately Republican states such as SC, TX, and GA all have moderately low Black rates and moderate-to-high White rates, as does moderately Democratic NM. Moderately Democratic OR, VA, CT, and moderately Republican IA, have low-to-moderate rates for both races. States with election margins within 5% are scattered randomly across the graph. MN, now infamous for racial tension in the wake of the George Floyd murder, has one of the lowest homicide rates for both Blacks and Whites.
There is an interesting geographic cluster of states centered on the Midwest: the block of states including OH, IN, PA, MI, IL, and WI are all in the lower right corner of the graph, with Black rates higher than average and White rates average or lower.
I’m happy to see that the high murder rate in MD, where I live, is a product of demographics. The homicide rate for Whites is well below the US average, and the Black rate is only slightly above it. The other high-homicide “blue” state, IL, has mixed statistics, with one of the highest rates for Blacks and a low rate for Whites.
One narrative of Black migration goes something like this: Blacks left the oppressive and dangerous South to make a better life in Northern states. They found fewer official restrictions, but unofficial restrictions and frightening levels of crime made their new homes a mixed blessing. Some are even “returning” to the states their grandparents or great-grandparents fled. Did moving North involve a trade-off between oppression and murder? The homicide data shows that the answer is: it depends. Both Southern and Northern states have a wide range of Black homicide rates. Someone moving from MO to MN would go from almost 650/million down to less than 200/million, but Arkansas to IL goes the other way, 120/million to 520/million. Along the East Coast, Southern states and Northern states are mostly in the moderate/low range for black homicides.
What to Make of All This
Raw comparison of homicide rates show that they tend to be higher in Republican-leaning states than in Democratic-leaning ones. Since this finding could be an artifact of demographics, I looked separately at homicide rates for White and Black men. This produced a similar result: “red” states tended to have much higher rates for Whites and somewhat higher rates for Blacks. If anything, separating the races increases the disparity between red and blue, because Maryland, a strongly Democratic-leaning state with a large Black population, has a high homicide rate but a low rate for Whites and a moderate one for Blacks.
A surprising finding was a region of high Black rates stretching from PA to WI. White rates are low-to-moderate in this region.
The race-based data used here was from before the big jump in murders in 2020. However, since the jump affected virtually all the states, I suspect that the results here would not be very different if updated with more recent data.
While higher homicide rates are associated with a tendency to vote Republican, this does not mean that it is caused by Republican governance. Social scientists have trouble explaining differences in homicide between different regions or times, so I do not attempt to do so. Still, it is worth knowing that the party that claims to represent “law and order” in the country tends to run states where people are murdered at a higher rate.
For anyone tempted to draw racist conclusions about the high African American homicide rate, its worth noting the vast difference between the White homicide rate in WV (50/million) and MA (10/million), even though the ethnic background for both of them is predominantly English and North and West European. For comparison, homicide rates are about 10/million in the United Kingdom, Germany, and Ireland, the origin countries for many in WV. These vast differences suggest that if West Virginians’ relatives in Europe can have a low homicide rate, economic and social conditions can be altered in West Virginia to achieve the same degree of peace. Similarly, the high African American homicide rate is a problem that is, in principle, solvable.
Doing a crude extrapolation, if the average male Black American lives to 78, then the chance of being murdered is something like (80 years) x (320 /million/year) = 2.5%. More elaborate calculations based on 1998 data also found that the probability of being murdered by age 45 was over 2% for Black males and .3% for White.
The homicide rate is smaller than the Covid death rate, which was almost 1200/million for the period 3/1/2021 to 2/5/2022. However, there is good reason to feel the same sense of urgency as we did for Covid. Though the Covid numbers are much higher, Covid deaths mostly affect people over 50, whereas the homicide rate is highest for 15-49 years old. A high homicide rate impede progress in multiple ways. Murders have a toxic effect on how people in high-crime areas see the world and how the world sees them.
Now that people are – justifiably – spooked by the rise in homicide since 2019, this might be an especially good time to concentrate effort getting the rate down to 2015 values – and much lower.
Data Sources
State Data: CDC 2020 and previous homicide mortality data from https://www.cdc.gov/nchs/pressroom/sosmap/homicide_mortality/homicide.htm.
National Data: QuickStats: Age-Adjusted Homicide Rates, by Race/Ethnicity — National Vital Statistics System, United States, 2015–2016. MMWR Morb Mortal Wkly Rep 2018;67:462. DOI: http://dx.doi.org/10.15585/mmwr.mm6715a8external icon.
Race Data: Riddell CA, Harper S, Cerda M, Kaufman JS, 2018: Comparison of rates of firearm and nonfirearm homicide and suicide in black and white non-hispanic men, by US state, Annals of Internal Medicine, doi:10.7326/M17-2976.
Photo: Homicide victims, Indianapolis IN, 2017: https://www.wrtv.com/news/local-news/crime/indianapolis-sets-all-time-homicide-record-for-third-year-in-a-row.