And we are less than one-third of the way through November.
Cumulative casualties seem to be rising at a faster and faster pace.
Figure 1: Cumulative mass shooting fatalities (dark red), non-fatal injured (pink), from 1982M08, through November 8, 2018.. Orange denotes 2017M01-2018M11, light orange 2016M11-2017M01. Source: Mother Jones, accessed 11/8/2018, and author’s calculations.
The frequency of mass shooting events seems to be increasing. A negative binomial count regression (estimated using QML) indicates 1.6 more events per month under Trump (estimate statistically significant a 1% msl).
Figure 2: Mass shooting event count, from 1982M08, through November 8, 2018.. Orange denotes 2017M01-2018M11, light orange 2016M11-2017M01. Source: Mother Jones, accessed 11/8/2018, and author’s calculations.
Fatalities are increasing over the Trump era.
Figure 3: Mass shooting fatalities per quarter, from 1982Q4, through November 8, 2018. Orange denotes 2017Q1-2018Q4, light orange 2016Q4-2017Q1. Source: Mother Jones, accessed 11/8/2018, and author’s calculations. [graph modified 11/9 to put into quarterly frequency]
f = -6.19 + 0.028pop + 5.590trump
Adj.R2 = 0.21, N = 145, DW = 2.09, bold denotes significance at 10% msl using HAC robust standard errors.
Where f denotes mass shooting fatalities, pop is population in millions, trump is a dummy variable for Trump administration (all on monthly basis, estimated using quarterly frequency data)
On a quarterly basis, the regression is:
f = -18.51 + 0.084pop + 15.344trump
Adj.R2 = 0.19, N = 145, DW = 2.08, bold denotes significance at 10% msl using HAC robust standard errors.
[edits in italics 11/9, noon Pacific]
One can interpret this as follows: a Trump administration
quarter month is associated with 5.6 greater fatalities from mass shootings, or 22.4 67.2 on an annualized basis. (Over 1982q4-16q4, the average fatalities per quarter is 4.876 14.6). Inclusion of a deterministic time trend yields a negative coefficient on population, and a trump coefficient (12.03) significant at 1% msl.
Data: [xlsx] as of 11/8/2018.
Update, 11/10/2018, 1:30PM Pacific: Reader Sam thinks I should conduct some out of sample validation. He also conjectures I am misleading readers by reporting only 10% significance levels. I address his argument that I don’t understand out of sample cross validation in this post.
Here is a bar chart of Trump dummy coefficients, with HAC robust standard errors. I show *** for 1% msl, ** for 5% msl.