Murder is the only official statistics used for measuring gun violence nationally, but as I’ve previously argued, tracking shootings does a better job of gauging gun violence levels. But even counting every time a person is shot misses a fair amount of gun violence in New Orleans. A research paper by Jennifer Doleac and Jillian Carr proved this by using ShotSpotter data in Washington, DC and Oakland to show that only 2-7 percent of gunfire incidents result in a shooting.

New Orleans does not have ShotSpotter, but the city’s Calls for Service database does provide a mechanism for evaluating how much gun violence is missed by only counting shootings and murders. Turns out it’s quite a bit.

Signal 94 in NOPD’s Calls for Service data represents a firearm discharge report. These are calls that rarely result in reports being written as evidenced by about 87 percent of these calls so far in 2016 ultimately receiving a disposition of ‘Gone on Arrival’, ‘Unfounded’ or ‘Duplicate’ (e.g. multiple neighbors calling to report the same thing).

Yet these calls are important because they expose the degree to which we are missing gun violence.

Since the start of 2010 there have been nearly 20,000 firearm discharge reports and over 2,800 shooting incidents in New Orleans. That works out to a ratio of 6.95 firearm discharge reports per shooting incident through August 26, 2016. On a day-to-day basis there’s virtually no relationship between firearm discharge reports and shooting incidents, but over a six month period the correlation is nearly perfect (R-squared = .91).

As firearm discharge reports go up, so do shootings, and vice versa. This can be seen clearly in the below graph which shows firearm discharge reports over 90 days in blue and shooting incidents over 90 days in red. Note that I have artificially smoothed December 31 to January 1 and July 3 to 5 for each year by replacing the reported value with 9 firearm discharge reports (the daily average since 2010). This helps avoid spikes in these reports when fireworks are shot off over New Years and the 4th of July.

The same sentiment is captured in the below table which provides the ratio for each six month period starting in 2010. The last column on the right shows how the ratio of firearm discharge reports to shootings tends to vacillate above and below 6.95 over six month stretches.

2014 was slightly above the average for the full year suggesting a regression to the mean was inevitable. In other words, the same number of firearm discharges in 2015 would be expected to result in fewer shooting incidents, and that’s essentially what we saw.

There were 2,908 firearm discharge reports in 2014 and a virtually identical 2,881 in 2015, but there were 432 shooting incidents in 2014 and 392 in 2015. Firearm discharge reports have risen as seen in the blue line above, but they’ve been outpaced by the rise in shooting incidents.

Just as randomness/luck appears to be a driver in murder increases/decreases, randomness/luck appears to be a driver in the number of shootings that occur over time. Consider the case this past weekend of a carjacking perpetrators who shot at three people (they missed all three). The research and available evidence suggests there are several more incidents we don’t hear about for every one incident we do hear about.

For 2016 this means that the present rate of firearm discharge reports should lead to fewer shootings in the not too distant future. Another way to look at it is provided in the graph below. This graph compares the expected number of shootings over 180 days — found by dividing discharge firearm reports over 180 days by 6.95 — to the number of shooting incidents over 180 days. If the difference is positive then there have been more shootings than expected based on discharge firearm reports.

The current spike over the last six months matches the biggest on record going back to 2010. The two biggest spikes came in August 2010 and September 2015. There were 115 and 111 shootings respectively in the 90 days before those spikes. There were 92 and 75 shootings in the 90 days that followed.

This doesn’t mean murder will slow down if the percentage of shooting incidents ending in a fatality continues to slowly rise. But it does suggest that sooner rather than later there will be a slowdown in shootings in New Orleans.