Gun violence risk isn't equal across the country
The chance of a school shooting happening is higher in Kentucky, Tennessee, and the Carolinas than California. Two social theories can help explain why.
One of the most common misconceptions I hear about school shootings is they are rare and random which means every school has equal risk. While an attack can happen anywhere, the risk (quantifiable chance of experiencing something) is not equal. In this map, it’s easy to see that shootings at k-12 schools are not equally distributed across the country. This map is a logarithmic scale (measures values by orders of magnitude) meaning the difference between 0.1 and 2 represents a 20x increase.
A newly published study uses the K-12 School Shooting Database incidents from 2000-2024 with a covariate-adjusted kernel regression framework that embeds an orthogonalized residual risk surface within a semiparametric binary model. This spatio-temporal analysis accounts for both time and space to identify stable school-level associations with markedly higher risk for a shooting. (Translation: this study looked at both location and time to find schools that consistently have a higher risk of a shooting while controlling other factors that might influence that risk.)
Something that immediately jumps out at me is the paper has four authors with completely different academic backgrounds:
Tilman M. Davies (Department of Mathematics and Statistics, University of Otago)
Michael R. Desjardins (Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health)
Alexander Hohl (Department of Geography, University of Utah)
Guangzhen Wu (Department of Sociology & Criminology, University of Utah)
Multidisciplinary academic work is really powerful because it blends crunching numbers with broader cultural and sociological frameworks for understanding the world. The study finds that the risk of a shooting in the Appalachia region of Virginia, Kentucky, southern Indiana and Illinois, Tennessee, and the Carolinas is systematically higher than would be expected from the background distribution of schools alone.
Smaller but still coherent localized pockets of elevated residual risk are also present in the far Northwest, the northern Rockies, and around parts of the Great Lakes. The spatial fit shows that the risk of school gun violence incidents is not diffuse, risk is organized into several geographically coherent zones. This pattern should not be read simply as a consequence of sparse school coverage because the controls are drawn from the background school population. Across much of the West, the fitted residual risk is generally near or below the covariate-adjusted background level.
This rigorous statistical study that aligns with the locations of high profile attacks including Marshall County High (KY), Westside Middle (AR), Apalachee High (GA), Oxford High (MI), Springwater Trail High (OR), Pearl High (MS), Heath High (KY), Andrews High (SC), Chardon High (OH), and Heritage High (GA).
Full paper --- Orthogonalized Kernel Regression for Spatial and Spatio-Temporal Residual Risk: Application to School Shootings in the Contiguous United States by authors:
What’s explains this geographic pattern?
Here are two different explanations. First, the political left and right both get the primary drivers of gun violence wrong. These are not regions of the country full of evil people (right wing theory) or desperately poor people (left wing theory). Most people who shoot someone in public are teens and young adults who carry guns and then make really bad decisions during an argument.
The Jens Ludwig’s ‘Unforgiving Places’ theory disproves the popular narrative that shootings are the calculated acts of malicious or desperate people by showing how most shootings actually stem from interpersonal conflicts, especially arguments that escalate into shootings.
This theory applies to the Appalachia and Great Lakes regions because both political parties create narratives for these former industrial and railroad towns are violent because of economic devastation, drug use, bad people (e.g., immigrants are eating the dogs and cats in rural Ohio towns), and communities with little opportunity/education. In ‘Ungiving Places’ like Appalachia, easy access to firearms from minimal state-level regulation combined with simple disputes results in higher risk of shootings. For example, Dr. Ludwig’s analysis of crime data shows that 50-80% of all homicides each year in Chicago since 1965 resulted from disputes and interpersonal conflicts.
A second theory is looking at the cultural maps of the United States. American Nations: A History of the Eleven Rival Regional Cultures of North America is a book written by Colin Woodard in 2011 that proposes a framework for examining American history and current events based on a view of the country as a federation of eleven nations. Each region is defined by a shared culture established by each nation’s founding population.
Findings from the geospatial study I described above match the highest school shooting risk area with the cultural region of Greater Appalachia.
This is the area populated by waves of immigrants from Ireland, England, and Scotland. Greater Appalachia covers the highlands of the American South including Kentucky, most of Tennessee, West Virginia, and the western parts of Virginia and North Carolina, the southern parts of Ohio, Indiana, and Illinois, the Ozarks of Arkansas and Missouri, most of Oklahoma, and Texas Hill Country. The Appalachia region is culturally defined by the fighting spirit by figures like Davy Crockett, Andrew Jackson, and Douglas MacArthur…all of whom handled disputes with firearms. The phrase “shoot first, ask questions later” originated in George Forrester Williams’ fictional 1882 book about Civil War soldiers fighting battles across Appalachia and the upper Deep South.
The 95% upper risk area (dotted line) includes the Midlands Region founded by English Quakers and Pennsylvania Dutch. The border city of Chicago is shared with Yankeedom and St. Louis with Greater Appalachia (both Chicago and St. Louis are unforgiving places for gun violence). According to Woodard it is culturally the most “American” of the nations.
Looking at all forms for gun violence (not just school shootings), the American Nations map shows the same patterns of significantly increased risk in Greater Appalachia, the Deep South, and New France. The map on the right is national level data and the left is county level.
Read more from Colin Woodward and Neighborhood Lab: The Geography of U.S. Gun Violence
Ask “where” not “who”?
The convergence of spatio-temporal risk modeling, Ludwig’s unforgiving places theory, and Woodard’s cultural framework combine to suggest that maybe we’ve been asking the wrong question about how to stop school shootings and how to explain the risk of gun violence.
Instead of who commits these attacks, the more relevant question may be where does the cultural permission to respond to conflict with lethal force run deepest. Furthermore, how does firearm availability transform that permission into tragedies at k-12 schools? I would hypothesize that risk is not randomly distributed because culture is not randomly distributed. The good news for violence prevention is that unlike individual psychology, culture has a geography that we can describe, quantify, and measure.
David Riedman, PhD is the creator of the K-12 School Shooting Database, Chief Data Officer at a global risk management firm, and a tenure-track professor. Listen to my podcast—Riedman Report: Risk, AI, Education & Security—or my interviews on Freakonomics Radio and the New England Journal of Medicine.