According to the count kept by the Baltimore Sun, there have been 319 homicides in 2017 up until November 30.^{1} That is already more than the 318 seen in all of 2016, but it’s well below the record 344 from 2015. Although Baltimore experienced 353 homicides in 1993, it had about 110,000 more residents, so the homicide rate then was 4.8 homicides per 10,000 people. In 2015, it was 5.5 homicides per 10,000 people.^{2}

Using the number of homicides per day for the year adds an arbitrary time horizon that we probably should not be using because things that cause all of this violence and death do not reset on January 1st. If anything, they continue on without worrying much about the calendar. Things that do seem to influence the rate at which homicides are occurring are many, and they go from the natural (e.g. the weather^{3}) to the societal (e.g. unemployment^{4}) to the individual (e.g. drug addiction^{5}). (Admittedly, drug addiction has social and natural components to its causation as well.)

A better measure is a rolling average of the homicide rate per day for the city of Baltimore. (I learned this technique from a colleague.) What we do is take the number of homicides in a 365-day period and divide them by 365. Then we take the number of homicides in the 365-day period one day after the previous period and divide those by 365. Slowly, this builds an average per day of homicides. For example, on from January 1, 2015, to January 1, 2016, there were 345 homicides, for a per day rate of 0.945 homicides per day on that day. There would not be a homicide again until January 6. By that day, the rate had dropped slightly to 0.931 as 5 days had gone by and in the 365 days before January 6, there were 340 homicides.

This way of looking at the per day rate allows us to not lose sight of the significant figure, and, as epidemiologists do, it allows us to detect an outbreak. Here, then, is what that looks like since mid-2005:

This measurement tells us that the rate of homicides was pretty high from 2005 to 2007, then it began a very well-sustained decline in 2008. That means that there were lots of days without homicides in that year, so much so that the rate dropped below 0.600 homicides per day, or about one every other day. The rate would climb again and remain steady-ish from 2009 to 2011. Then there is another severe dip in late 2011 and early 2012. In 2011, Baltimore saw the lowest number of homicides (196) since 1977 (171). It was all over the news, and a lot of politicians congratulated themselves for it.

It wasn’t to last, however. The rate increased again between 2012 and early 2014. Then there is a short-term dip below 0.600 up until April of 2015. In April 2015, according to one epidemiologist, a severe outbreak of homicides began in Baltimore. While the rate per day dipped in 2016 before it reached one per day, it steadily went back up this year. So, by November 20, 2017, Baltimore is at 345 homicides in the previous 365 days for a rate of 0.945.

With that rate of 0.945 in mind, one could calculate that the 31 days in December will see about 29 more homicides (0.945 multiplied by 31). Add those to the current number of 319, and Baltimore smashes through the 2015 record by hitting 348. But that is not the only number possible here.

Since 2005, December in Baltimore has averaged about 21 homicides. Add those to 319, and Baltimore doesn’t break the record but comes close by hitting 340 homicides for 2017. This is a more probable number since it accounts for weather and social factors that come with December, like the holidays and the activities surrounding the holidays.

Finally, a third number to look at is 8%. Since 2005, Decembers have had about 8% of the homicides for the whole year. So, if 319 represents about 92% of the homicides Baltimore will see this year, then 346 represents the full 100% we will probably see. Again, this number surpases the 344 homicides of 2015 and would represent a rate of 5.6 homicides per 10,000 people in Baltimore for the year.

So will Baltimore see a record number of homicides this year? The answer is, sadly, that it probably will. Two of the three models say so, with 346 and 348 as the possible number of homicides. One model that looks at past Decembers has it just slightly down at 340. Because these are not fully fledged mathematical models done in a supercomputer that accounts for thousands of variables, I hesitate to give the odds of the record-breaking number (345) happening… But it’s going to come close. (Only a multi-day snowstorm may keep that number from happening, or some big event that disrupts everyone’s lives, including the lives of the would-be murderers and/or their victims.)

The policy implications for this are many, of course. The Baltimore Police Department, the Mayor, and the City Council all need to take a hard look at the interventions for violence in general and homicide in particular that are in place in the city. They need to assess interventions that have worked in other cities of similar size and social composition and see if those interventions can be brought to Baltimore. And you cannot forget civic organizations trying to bring peace to their neighborhoods, sometimes against all odds and with hardly any kind of funding.

Stay tuned.

Sources:

- Baltimore Sun Data: http://data.baltimoresun.com/news/police/homicides/
- FBI Uniform Crime Reporting: https://ucr.fbi.gov/ucr
- Cheatwood, D. (1995). The Effects of Weather on Homicide.
*Journal of Quantitative Criminology,**11*(1), 51-70. Retrieved from http://www.jstor.org/stable/41954121 - Raphael, S., & Winter‐Ebmer, R. (2001). Identifying the Effect of Unemployment on Crime.
*The Journal of Law & Economics,**44*(1), 259-283. doi:10.1086/320275 - Lammers, S.M.M., Soe-Agnie, S.E., et al. (2014) Substance Abuse and Criminality: A Review.
*Tijdschrift voor Psychiatrie*. Volume 56, Issue 1, Pages 32-39.