On 25 March 2026, the International Academy on the Science and Impact of Cannabis (IASIC) hosted a presentation by Dr Ragy R. Girgis, M.D., M.S., Professor of Clinical Psychiatry at the Columbia University Department of Psychiatry and the New York State Psychiatric Institute. His subject was cannabis and mass shootings, and the findings come from what is now the most comprehensive mass murder repository ever assembled: the Columbia Mass Murder Database.
The database contains more than 2,300 cases of personal-cause mass murder documented worldwide since 1900. It was not built to prove a point. It was built because Dr Girgis and his colleagues decided they wanted to understand, as definitively as possible, what was actually driving mass shootings rather than relying on assumption, media coverage, or politically convenient narratives.
What it is beginning to reveal about cannabis and mass shootings is not comfortable reading.
Building the Database
Before getting to the findings, it is worth understanding what the Columbia Mass Murder Database actually is, because the rigour of the methodology is what gives the results their weight.
Dr Girgis and his team began by reviewing popular databases of mass murder including Wikipedia, the Stanford database, Mother Jones, Everytown, and others. From those sources they identified potential cases, then excluded anything that did not meet their strict criteria: events had to involve three or more fatalities not including the perpetrator, had to be perpetrated for personal reasons rather than war, terrorism, gang activity or organised crime, and had to be supported by primary sources in English, meaning court and police records or reliable news media rather than secondary websites.
What remained was a dataset of around 1,700 mass murders in the original iteration, now grown to more than 2,300 cases through to 2023 or 2024. It is the largest such dataset in the world.
One of the key design decisions was the use of comparison groups. Most prior research on mass shootings had not included them, which introduced significant bias into the findings. Dr Girgis’s team divided the sample into perpetrators who used firearms and perpetrators who used other methods, on the basis that weapon choice was the most meaningful differentiator between types of mass murder. That comparison structure is what makes the cannabis findings meaningful rather than merely suggestive.
What Most People Get Wrong About Mass Shootings
Before arriving at the cannabis data, Dr Girgis walked through some findings that challenge widely held assumptions.
The first concerns mental illness. Most violence, and most mass shootings, is committed by people without psychotic illness. This is consistent with what the research has long suggested, but it continues to be misrepresented in public debate. Mental illness, specifically psychosis, accounts for roughly 5% of the contributing factors to mass shootings. It is a slice of the pie, not the whole thing.
The second concerns the types of mass shooting people tend to think about. School shootings, which dominate news coverage and public imagination, represent only about 13% of all mass shootings. The most common type, accounting for around 45% of cases, is familicide: mass murder involving a spouse, children, or other family members. The second most common is felony-related mass murder. The public-facing, stranger-targeted mass shooting is actually the least common category, though it attracts the most attention.
The third concerns alcohol. Despite a widespread cultural assumption that alcohol and violence go hand in hand, alcohol misuse showed no meaningful relationship with mass murder in the data. The signal in the data is not about alcohol. It is almost entirely about cannabis.
The Number That Should Prompt Questions
Here is what the data on cannabis and mass shootings actually shows.
In the United States, the prevalence of cannabis involvement among mass shooters before 1996 was around 5%. After 1996, the year medical cannabis was first legalised in California, that figure more than doubled to over 11%. The result was statistically highly significant.
Dr Girgis chose 1996 deliberately, and not arbitrarily. He and his colleagues were aware that using a single date as a policy marker has limitations. So they tested the result using 1990 and 2000 as alternative cutoff years and found the same pattern held. They also noted the well-documented diffusion effect of cannabis laws, whereby legalisation in one state influences use and availability across many others, which further supports the use of a single national date as a meaningful marker.
When the same analysis was applied to perpetrators of mass murder who did not use firearms, there was no increase. No relationship at all. The doubling of cannabis involvement was specific to mass shootings.
That specificity is the most important feature of the finding. It is not simply that cannabis use has increased across the general population since legalisation began, which it has. The relationship shows up specifically in people who used firearms to commit mass murder, and not in a closely comparable group of people who committed mass murder by other means. That is the kind of pattern that, in research terms, begins to suggest something more than coincidence.
Motive, Suicide, and What Cannabis Involvement Tells Us
One of the more revealing findings from the data concerns what cannabis involvement predicts about a perpetrator’s behaviour at the time of the event.
Mass shooters with cannabis involvement were significantly less likely to take their own life at the time of the shooting compared to those without cannabis involvement. To understand why that matters, Dr Girgis explained how researchers think about the barriers that prevent someone from perpetrating a mass murder.
There are three of them. The first is the person’s own moral or value system, whether that is religious, ethical, or psychological in origin. The second is the moral framework they have internalised from the people around them: family, partners, community, society. The third is the rational deterrent of getting caught, prosecuted, and imprisoned.
When a perpetrator plans to die at the scene, that third barrier is removed entirely. It becomes significantly easier, psychologically, to carry out the act. Mass shootings, more than other forms of mass murder, are associated with perpetrators who intend to take their own life, and it is partly for this reason that mass shootings are so much more common than other forms of mass murder involving comparable planning and motivation.
What the cannabis data suggests is that perpetrators with cannabis involvement are operating with a different set of motivations. They are more likely to intend to survive. That changes the profile of the event and potentially the profile of the intervention that might prevent it.
Cannabis involvement was also associated with significantly younger perpetrators. That finding is perhaps less surprising given what is known about cannabis use demographics, but it reinforces the picture of a distinct subgroup of mass shooters for whom cannabis is a meaningful variable.
Correlation, Causation, and What Comes Next
Dr Girgis is precise about the limits of what the data shows. These findings are correlational. Establishing causation would require a randomised clinical trial, which is not feasible in this context. The data cannot tell us that cannabis caused these individuals to commit mass shootings.
What the data can do is establish a specific, statistically significant pattern that is consistent with the broader and well-established body of evidence linking cannabis use and violence more generally. Dr Girgis described that broader evidence as very clear. The relationship between cannabis and violence in general is not contested in the literature. What the Columbia Mass Murder Database adds is a specific examination of that relationship in the context of mass shootings in the United States, with a comparison group designed to account for confounding.
The honest summary is this: cannabis involvement in mass shooters more than doubled after cannabis legalisation began, the relationship is specific to mass shootings rather than mass murder more broadly, and it is consistent with what we already know about cannabis and violence. That is not a finding to dismiss because it is inconvenient.
A Piece of the Pie
Dr Girgis frames the contributing factors to mass shootings as pieces of a pie rather than a single cause. Mental illness accounts for roughly 5%. Recreational drug use, and in this dataset that means almost exclusively cannabis, accounts for another portion. Social, cultural, and situational factors make up the rest.
That framing is important. It resists the temptation to reduce a complex phenomenon to a single explanation, which is exactly what public debate on mass shootings tends to do. After every high-profile shooting, attention immediately focuses on one cause: mental illness, gun laws, social media, school culture. The data suggests the reality is more distributed than that.
But distributed does not mean equal. And a doubling of cannabis involvement among mass shooters following legalisation is not a small signal in a noisy dataset. It is a finding that deserves to be part of the policy conversation.
The Conversation That Keeps Getting Avoided
Cannabis and mass shootings is not a pairing that survives long in public debate. Legalisation is politically popular. The cannabis industry is large and commercially motivated to keep the public narrative positive. And findings like these are easy to dismiss as correlational, preliminary, or simply too uncomfortable to act on.
But the data on cannabis and mass shootings is now emerging from the most rigorous dataset ever applied to this question. Governments expanding cannabis access, and many are, are making long-term decisions about public health without seriously engaging with this research. That is a failure of due diligence dressed up as progress.
Dr Girgis and his colleagues are not campaigners. They are researchers who built a database, applied a comparison structure to control for bias, and reported what they found. That finding is specific: cannabis involvement among mass shooters in the United States more than doubled after legalisation began. It is statistically significant. Moreover, it is consistent with what the broader literature on cannabis and violence already tells us.
The Columbia Mass Murder Database now contains more than 2,300 cases. The pattern around cannabis and mass shootings is clear enough that the question is no longer whether it exists. The question is whether anyone in a position to act on it is paying attention.
Dr Ragy R. Girgis, M.D., M.S., Professor of Clinical Psychiatry, Columbia University Department of Psychiatry and New York State Psychiatric Institute, presented at the IASIC Speaker Series on 25 March 2026. (Source: WRD News)