Advancing Understanding, and Informing Prevention of Public Mass Shootings: Findings from NIJ Funded Studies, Part 2

MARY JO GIOVACCHINI: Good afternoon, everybody, and welcome to today's webinar, Advancing, Understanding, and Informing Prevention of Public Mass Shootings: Findings from NIJ Funded Studies part two, hosted by the National Institute of Justice. At this time, I am going to turn it over to Barbara Lopez.

BASIA E. LOPEZ: Good afternoon, everyone. Welcome again today. For those of you who joined us in addition to our event we held yesterday, as well as all the new audience, my name is Basia Lopez. I'm a Social Science Research Analyst at the National Institute of Justice. Here at NIJ, I oversee federally-funded research, evaluation, and data collection projects related mainly to firearms violence, including mass shootings. I will turn this webinar during which we will from a team of our renowned experts, followed by a discussion on the implications of their findings. I'm also accompanied by NIJ Graduate Research Assistant, Danielle Crimmins. Danielle, would you please introduce yourself?

DANIELLE CRIMMINS: Yes. Thank you, Basia. My name is Danielle Crimmins. I'm a Research Assistant here at NIJ, and also a graduate student at Purdue University. I'll be leading the Q&A session following the discussion on the implications and the findings we have on the criminal justice system and prevention. Thanks.

BASIA E. LOPEZ: Thank you, Danielle. Very briefly, I already talked to this yesterday, but I would like to repeat it for the new audience, that our institute has sponsored research in an effort to assist our nation in reducing firearms violence by supporting rigorous studies and evaluation of programs since the 1980s. We focus mainly on building knowledge and advancing evidence-based practices that aim to reduce the firearms violence and mass shootings. But we built on various previous projects that were carried out in NIJ's Comprehensive School Safety Initiative. In 2018, we prioritized research on mass shootings under firearms violence portfolio, and today our panelists are recipients of these awards. One from the Comprehensive School Safety Initiative and two others are from the firearms violence program of research.

We will hear first from a couple of members. From CSSI, Dr. Steven Chermak and Joshua Freilich. Dr. Steven Chermak is a Professor of Criminal Justice at Michigan State University, and his colleague, Dr. Joshua Freilich, is a member of the Criminal Justice Department at John Jay College, CUNY. They will present on “Understanding School Shootings in the USA, from 1990 to 2016.” Second, Dr. Jillian Turanovic, she's an Associate Professor in the College of Criminology and Criminal Justice at Florida State University. And along with her colleague, Dr. Travis Pratt, a fellow at the University of Cincinnati Corrections Institute, she will present on their projects funded in 2018 under NIJ's firearms violence portfolio. And she will present on “Deadly Mass Shootings in America: Features and Trends across Four Decades.” Finally, but not lastly, Jillian Peterson. She's an Associate Professor of Criminal Justice and Director of Forensic Psychology at Hamline University in Minnesota. She will present along with her colleague, Dr. James Densley, who is a Professor of Criminal Justice and university scholar at Metropolitan State University in Minnesota. They will present on the “Reflections on researching the lives and crimes of mass shooters.” So, with no further ado, we will start with Dr. Chermak and the presentation on understanding school violence. So, I'm going to pass the ball to you, Steve.

STEVEN CHERMAK: Good afternoon. Thank you for joining us this afternoon. We're happy to present some preliminary results from our 2016 NIJ-funded project on understand school shootings in the United States. The outline for today's presentation will talk about the goals and objectives of the project. We will also talk about the importance of the study. And then I'll do a brief overview of the methodology we used, and then we'll present some of the key findings.

In terms of our goals and objectives, the general goal was to expand overall research knowledge about school violence and school shootings specifically and address gaps in the research to better understand the cause of the school shootings. By doing this, we also wanted to create an open source database which includes all publicly known shootings with at least one injury that occurred between 1990 and 2016. The two theoretical lenses that we focused on, in terms of thinking about the school shootings--in order to better understand offenders, we applied criminology, developmental social control theory, and then, to better understand events, we applied criminology situational crime prevention. And our hope, of course, is that our research findings are of great value to law enforcement, to school officials, policy makers, as well as the social sciences.

Just briefly in terms of the importance of the study, I mean, it's obvious that the victim and the victim's family after school shootings are significantly impacted by the event, but we also know that such events impact the entire school, would impact the entire school district, teachers, other students, administrators, as well as the entire community. In addition, although there's a growing body of research in this area, there are still significant limitations regarding the existing research. I mean, school violence is difficult to study systematically. There are only a few national-level databases where the inclusion criteria as well as definitional criteria may differ somewhat. There's a wide range of different reporting practices that are used. And because of the rarity of these events, conventional methods that we apply in criminology may not be particularly useful. The major obstacle to date has been the lack of reliable empirical data. Some other limitations in the research is sort of a general lack of understanding of the school environment. Also, because of the rarity of having a large-scale data project, the inability to study changes over time and differences in how school violence has changed over time. There's been--not enough research understanding the attributes of victims of school violence. And in survey here, we used an open source database, which is an area that has grown significantly to study rare events. You'll hear two other presentations that use open source methodologies on mass shootings. But in the areas of terrorism, police use of force, and other things, open source have grown, and so it was important for us to talk and think a little bit about how such approaches can really use cutting-edge measurement, as well as address reliability in other research concerns. And so, our thinking is that, with better data, we'll get a better understand of the problem, and thus give us an opportunity to think about policy solutions, as well as evaluate policy initiatives that are put in place.

Our methodology was both use quantitative and qualitative methods. We built an open source database, which we'll talk a little bit today about the inclusion criteria. We'll talk briefly about the process of putting together that database, and then also how we assessed the quality of the information we put into the database. And then our qualitative analysis focused on building out 30 detailed case studies of a random sample of the school shooters from the database, where we focused on building a better understanding of the life course of the offenders but also looking at situational crime prevention as it relates to the schools.

So, first, in terms of thinking about our database, the inclusion criteria, first, we used open sources to identify the events that are included in the database. As the title of our slide noted, the dates of interest for this particular project were January 1st, 1990 through 2016. And our key inclusion criteria was to include shootings with injuries or fatalities on K to 12 school grounds, including outdoor crimes that occurred outdoors--outside the school, but excluding university-related shootings. These are all shootings that happened in the United States and U.S. territories. And, as I mentioned, there had to be a gun that was fired on school grounds that caused at least an injury or a death. So, this, of course, includes rampage, mass workplace shootings, but also drug and gangs. We also included information about suicides, as well as accidental shootings that went off in school. We even exclude plots where either a shot was fired but no injury occurred and/or the plot was stopped before the shooting took place.

This may be a little bit difficult to see, so I apologize for that. But this is just an example of the decision tree we used to sort of apply our inclusion criteria to make sure we consistently identified those cases that fits a criteria of whether or not a shooting happened, whether or not there was an injury. And really the critical inclusion criteria and really the most difficult for us to apply was deciding whether or not the event occurred on school grounds. We went to considerable length to identify the location of the shooting when it happened outside the school using a variety of different sources to ensure that the events that got into the database met that particular criteria. In terms of identifying cases, again, we used open sources. We reviewed over 35 different sources, including other existing databases, looking at scholarly materials including books, official documents, and certainly using the media as well, looking for a listing of school shootings to try to capture all known school shooting events that were reported in open sources. This slide gives you a little bit of sense of sort of the process we went through in terms of looking at particular events and having to exclude some because they didn't meet our inclusion criteria. So we looked at almost 1,400 different events, and about half of them, for a variety of different reasons, fell out because they didn't meet all of our selection criteria. And of that group, we have 652 events within the database, including intentional shootings--intentional school shootings but also suicides, accidental shootings, and legally-justified shootings. In terms of searching cases, once we identified events, we collected all publicly available information about each of the cases. We reviewed and used--we went through 60 different web engines to--and applied using both an offender, an event-level search protocol that was followed. And, independently, we took the name of the school and ran them through an independent protocol to collect as much information about the school as possible, including school policies, demographic characteristics of the school, and the community. All of our searchers received extensive training on sort of going and using these search strategies, and we identified a variety and a large range of variety--of different open source materials which we'll talk about more in a second. This slide just briefly just sort of shows you the table that we used and we asked our researchers to follow when collecting the open source information. And so, they would go source by source. So they'd start, say--Lexus Nexus. And then after they did a full search using a variety of different search words, they would note in the file that they completed the search, and then they would move to the next source, and then the next source, the next source. And after they got down the list of all 60 then the case would be considered quoted. This often led to some duplication of effort but we also felt that it was better to be overly inclusive of the open source search material rather than to be under inclusive.

This particular slide here shows you what I think is a real success in terms of the effort and how thorough we were in being able to collect information about each of these events. On average, we were able to identify about 85 documents for each case, and a total of--we collected over 30,000 documents. As you can see, it's broken down by the type of document. Not surprisingly, the main source of information about school shootings come from news sources simply because of the newsworthiness of such events. But we were also able to obtain court records on cases, police reports, and police information, including other government documents, as well as website and scholarly material per event.

Once we created a search file with all of those documents, then of course we code the information. We have an offender-level codebook where we attempted to operationalize a variety of developmental social control indicators, as well as a large number of control and other variables about the offender and characteristics of the offender. We also had an event-level codebook, as well as a school codebook. So, again, we're here, we're trying to operationalize elements of situational crime prevention and other types of indicators related to the event and the school, extending some of the other work we've done in the area of terrorism using our extremist crime database. For the case studies, we created a template that for every case we followed, again, by attempting to dig a little bit deeper into the areas of understanding the offenders and their life course, as well as situational crime prevention.

I'm just going to go through a few of the findings. This first slide just takes the 652 events in the database and presents it over time from 1990 through 2016. As you can see, the vast majority of school shootings involve intentional offenders. Sort of the--what you think about when you think school shooting is offender but then the other--about 15% of the events in the database are self-harm or suicides, 11% are accidental, and less than 1% are justified. This particular slide shows considerable variation over time but, on average, there's about 24 intentional suicide and accidental school shootings that happen in a particular year. It's not static, in that there's sort of peaks and valleys. You can have as low as eight shootings that occurred in 2002, 12 shootings that occurred in 1990, whereas 36 different shootings happened in 1994. There appears to be a slight uptick in the number of events when you compare the last three years in the database, 2014 through 2016 compared to 2010 and 2012. This particular slide just takes those 75% of the whole where they're intentional U.S. school shootings, and it breaks it down by where we are able to identify the shooter and also those events where the shooter was unidentified. There's a couple reasons why we weren't able to identify the shooter. One is that they just simply--the crime remained unsolved. That there was a school shooting, somebody was injured, but the event--offender was never identified. And the other issue has to do with having juvenile shooters that oftentimes for privacy concerns, their names weren't provided in any of the sources that we were able to access.

Again, this slide shows there's about 18 intentional shootings on average within the U.S. And, again, just like the previous slide, the numbers vary considerably by year, whereas there's only seven in 2002 and nine in 2012 but 30 in 1993 and 29 in 2016. It's interesting. Early on, from '91 to '94, account for nearly--over 20% of the total number of intentional U.S. school shootings, and then it's bookended by small numbers in both '90 and 1995. Again, there appears to be a slight increase in the number of intentional shootings from 2012 to 2016. Most of the shootings in the database involve known offenders, and, on average, there's about 13 of these shootings each year. Again, there's some variation by year. This particular slide compares nonfatal to fatal events. Not surprisingly that the number--there's somewhat more nonfatal school shootings than fatal school shootings. This makes sense in that, you know, the number of attempted homicides is far greater than the number of homicides every year within the United States. When we examine only the fatal events that are in the database, then we find, on average, there are eight deadly shootings each year. And, again, we're finding variation by year. There's only two fatal shootings that occurred in 2002 and four in 2011, four in 2012, but 11 in ‘92 and 13 in '94. Interestingly, I think the number of nonfatal intentional shootings has increased every year starting 2012 through 2016.

This next slide, just as a crude geographic representation of where the shootings occurred within the United States. The vast majority, 43%, occurred in the South. About 23% occurred out West. Twenty-one percent occurred in the Midwest. And only about 12% occurred in the Northeast. This next slide simply compares the known intentional shooting events, and we break this out comparing adolescent to adult shootings. So, there was 352 shooting events where we knew the offender. And then, of those, about 250 are adolescent school shootings, and 102 are adult shootings. Again, I'll just talk about a few of the variables here. As you can see, the number--comparing the number of fatal shootings to--from adolescent to adult, a higher percentage of adult school shootings resulted in fatality, about 64%, compared to only 46% of the adolescent school shootings that ended up in a fatality. You can see that the number of mass shootings, regardless of what definition you use to sort of operationalize the notion of mass shootings, whether or not you use three fatalities or higher, or four fatalities, the number of mass shootings occurring within schools are relatively rare events. That is the number of adolescent mass shootings with three or more victims was eight over the 27-year time period from our study, and the number of adults with three fatalities or more was three during that time period. A couple of interesting findings regarding where the shootings occurred. The majority of school shootings in our database occur outside the school, someplace on school grounds. Fifty-eight percent of the adolescent school shootings happen outside the school. And close to 72% of the adult school shootings happen outside the school, in a parking lot, in a field that's right next to the school. Not surprisingly then the timing of the event is consistent with the events occurring outside the school. About 57% of the adolescent school shootings happen before or after school hours and 70--about 73% of the adult school--adult shootings happen before or after school hours. And then almost 57% of the adolescent school shootings events in our data were committed by a current student of that particular school.

The last part--bit of results--research results we want to talk about today. First, we want to present just a little taste of the situational crime prevention results. We collected a variety of different indicators, and here's four that we wanted to report out today. About 5% of the schools in our database had metal detectors at the time of the shooting. About 45% either had school guards or police officers within the school when the shooting occurred. About 27% had some sort of barrier, usually a fence at the shooting. And about seven out of 10 schools in the database had limited access, meaning that they control access: locked particular doors, allowed only a single door or single access into the school. We did want to talk a little bit about the case study results that's really intriguing. We wrote about 15 to 20 single-spaced pages for each case study and really provided some detailed insights into really the dynamics of what's going on leading to school shootings. So, we just want to talk about two issues that were pretty interesting to us. One is the notion of criminal opportunities. Really, the reality is most of the shooters were able to get easy access to schools despite having whatever barriers were in place. That even when there were metal detectors, which were rare, students were able to sort of get around and get access to the schools by planning either going through a side door, using an accomplice to open up a door, bringing the gun in after school hours when nobody was manning a metal detector or access was wider, and they would hide the gun and then use it the next day. The other issue was how most shooters gained access to guns. Most of these guns were legally bought but they were taken from a parent, from a grandfather, grandmother or another--a sibling. So, it was a gun in the house that was easily accessible. The shooter was able to take the gun from the house and then use it in the school shooting.

The last issue I wanted to talk about today is just the notion of leakage or warning behaviors that occur. In terms of the 30 case studies that we did, 17 of the school shooters warned, or gave some warning, that they were about to do a school shooting. And most of them did it the day of the shooting, but several did it days before as well. They did it through--in a variety of different ways as you'd expect. Some of them did it through social media and making announcements that way, but the vast majority simply announced to the people that they knew, either their friends, their acquaintances. In some cases, they even announced it to the victim. Out of the--so 17 out of 30 leaked the information. Of those 17 cases where somebody leaked information, only three was there any action taken. That is--in 14 out of the 17, the people who were warned that a school shooting was about to occur just didn't believe that it was going to happen. They thought the person was joking. They didn't take the threat serious, and so they decided not to do anything and report that information to any school officials, law enforcement, or anybody for that matter. With that, I'm going to conclude my presentation, and I'm going to pass the ball to Jillian Turanovic who's going to speak more on mass shootings.

JILLIAN J. TURANOVIC: I'm Jillian Turanovic. I am an Associate Professor in the College of Criminology and Criminal Justice at Florida State University, and it's an honor to be presenting in front of all of today. Thank you to those of you who have tuned in, and also to NIJ for organizing this event. So today I'm going to be giving you an overview of NIJ-funded research that seeks to build a comprehensive database of deadly mass shootings in the U.S. since 1980. And this data includes those mass shootings that occurred in both public and private spaces. So, before we begin though, I'd like to formally acknowledge Kristen Neville who's a PhD student at FSU who works as our Project Manager, my Co-PI, Dr. Travis Pratt, as well as the many outstanding undergraduate and graduate students at FSU who have contributed to the creation of this data.

So, as we know, deadly mass shootings have emerged as one of the most prominent social problems in contemporary America. And tragedies such as those that have occurred in recent years, such as in Orlando, in Parkland, and in Las Vegas, have understandably traumatized and shocked the nation. And it is clear from multiple polling sources that there is a public perception that deadly mass shootings are becoming alarmingly frequent in recent years. It's not uncommon, for example, to see news media headlines, such as these, that state that mass shootings are on the rise, that they are spreading like a disease, that they are increasing significantly, or that they have more than doubled since Columbine. Now, this is a concern, but from a data and research standpoint, there's questions whether there may be a disconnect between the reality of these deadly incidents, and how often they occur, and the public's perceptions of them. I say this because violent crime, in general, tends to be lower in recent years than it was in prior decades such as in the 1980s and 1990s. In Patrick Sharkey's book, Uneasy Peace, 2014 was regarded as the safest year in modern history based on violent crime rates. Now, this fact tends to be lost on most of the American public who still report that crime has been constantly on the rise for years. So, this chart here shows the disparity between crime rates and the public's perceptions of crime. So, the top darker blue line reflects the percentage--the percentage of Americans who think crime increased over the past year and the bottom line reflects the actual violent crime rate from the National Crime Victimization Survey. So, you can see that the--people's perception don't always match up with reality. And this gap may be due, in part, to the publicity afforded to high profile attacks in public spaces, as well as the media's increasing sensationalization of crimes.

But a more serious issue is that we currently have little in the way of a shared definition with respect to what constitutes a mass shooting. And contributing to this confusion is that existing databases, especially those that are publicly available and routinely drawn from by the media, paint very different pictures of the mass shooting problem. So, for instance, the Mother Jones data, which has heavily been relied on by the public and the media, currently defines a mass shooting as three or more people shot and killed in a public space, not including the shooter. And excluded from these data are mass shootings that occur in private spaces, those where the shooter is unknown, or any shootings that may be motivated by other felonies such as armed robberies or gang violence. And so, these data that mass shootings are more common today than they were in earlier years and show 10 incidents that occurred in 2019. Alternatively, the Gun Violence Archive, which is another popular data source, defines a mass shooting as four or more shot, not including the shooter, where nonfatal incidents are also included. There are no restrictions based on what is included in these data, you know, namely where the shootings happened or why they happened. And these data show that over 400 mass shootings happened last year. So, the disparity between 10 and 400 is vast, and it creates mass confusion over the nature and frequency of mass shootings.

Now, our sources from the FBI also present a different picture. The FBI Active Shooter data include incidents where individuals actively engaged in killing or attempting to kill people in a populated area with one or more firearms. Excluded from these data are gang and drug-related shootings, and those that didn't intentionally put people's lives in danger. These data show incidents to be increasing where one, for instance, was recorded in 2000 and 28 were recorded in 2019. The FBI Supplemental Homicide Reports, too, are not necessarily complete or accurate in the reporting of mass murders by firearm. There are several incidents recorded in these data that are mischaracterized as mass killings, and there are also other issues in that not all jurisdictions report and cases are missing even in jurisdiction that do.

So, altogether, when the body of work kind of looks like this, we are motivated to collect data on mass shootings and together, you know, when we first started this project, it's clear that there really wasn't any comprehensive, publicly available, or centralized database on mass shootings overtime, especially during those years when crime and violence tended to be higher in general, such as in the 1980s and 1990s. And without accurate and complete data going back in time, it's very difficult to determine the extent and nature of the current mass shooting problem. And this is consequential because any successful policy or even a fruitful discussion of what that policy should be about enhancing public safety or curbing mass violence first requires something really important: that those at the policy table have a shared understanding of the nature and extent of the problem itself. And we didn't necessarily have that clarity with respect to mass shootings.

So, to attempt to fix this issue, with the support of NIJ, we're currently compiling a mass shooting database to assess all deadly mass shootings in the US dating back to 1980. And we aim to identify all mass shootings over time, again, examine their trends and features, and also to assess the key individual, situational, and contextual features of these shootings in order to hopefully to develop better informed and data-driven policies. So, we began by defining a deadly mass shooting as one in which four or more victims were killed by a firearm in one event in one location, not including the shooter, consistent with Congressional Research Service and the FBI's definition of mass murder. Now, in our data, we include mass shootings occurring in both public and private spaces, including those that maybe tied to other crimes or gang or drug violence. So any homicide where four or more victims were shot and killed in a single incident. We are compiling our data from open source records, including news reports, FBI data, police files, and court records if available. And we plan to make these data publicly available on completion of our project.

So this effort has proceeded in three primary phases. So, first, we had to identify all deadly mass shootings that we could dating back to 1980. The second is to gather extensive information on all of these incidents. And, third, to code each incident according to a host of characteristics. Now, we're pretty much finished with phase one and phase two, and phase three is currently underway. So, we still have lot more information to code and to analyze, but hopefully I can provide you with at least some understanding today of what the basics of our data will look like.

So, to locate matching incidents, we used three primary sources. First, existing reports and data on mass shootings, knowing of course that most of these are incomplete. Second, local, state, and national homicide report, and court records. And, third, news media reports. So, given the issues with existing data sources that I described previously, nothing was automatically entered into our data if it appeared in an existing database. So instead we validated it through multiple sources and news media reports. And this is kind of what our verification process looked like. So anytime we came across an incident that looked like could possibly be a mass shooting, we verify it through additional news reports, case files, and court records if available. So, after extensive searching, we have included and verified in our data 761 deadly mass shootings between 1980 and 2019. Recall that these are not only incidents that occurred in public spaces. So any event where four or more people were shot and killed were included.

So given that, the first question we can ask is what is the overall trend in mass shootings look like? So here's what our data look like so far. Now, each bar represents the raw number of deadly mass shootings that occurred in a given year between 1980 and 2019. And so, as you can see, the trend over time is relatively flat, indicating that the raw number of deadly mass shootings that occur per year has been somewhat stable over the past four decades. So based on this, it appears that mass murder by firearm does not necessarily appear to be a new social problem but rather one that has been an issue in society for some time now. So during any given year, the number of mass shootings ranges from 11 to 27. And in our data, the years with the most mass shootings of any kind are 1991, 1993, and 2008. Several years in our data contain more than 20 deadly mass shootings. And so, as you can see, these--you know, the years that have higher frequency, the mass shootings tend to be more common in recent decades. Now, when I address the number of mass shootings by population increases that have occurred in the US over time, that is when I create a rate of mass shootings. You can see that the trend actually tilts slightly downwards in that deadly mass shooting--the deadly mass shooting rate was highest in the early 1980s and 1990s. Although, in general, there doesn't seem to be many dramatic fluctuations in the rate over time. So--and, again, in the '80s and '90s, you see these rates were the highest.

So, in seeing these trends, the next question you can ask is, well, what are the features of these incidents? Well, the majority of these are occurring in private settings with only 31% representing the mass shootings in public spaces. Over half involve victims who are known to their shooter, that is they were non-strangers. And of these non-strangers, most were family members, where family victims are involved in 42% of a deadly mass shooting in our data. Most of those family killings also target a romantic partner. So in 27% of incidents, a current or former romantic partner was killed. Additionally, 33% of all incidents were also felony-related, meaning that they were carried out in the context of another crime, many of which were gang and drug-related activities or armed robberies. Incidents also primarily involve male shooters, although there are some female mass shooters in our--shooters in our data in 4% of incidents. In the majority of incidents, victims were also killed by a single shooter, although 22% involves two or more shooter. Now, in terms of location, not surprisingly, most of these incidents were occurring in private spaces, meaning a home or a residence, and 34% took place in other public spaces. And note that only 2% of the deadly mass shootings in our data reflect school shootings.

So given that we included a variety of deadly mass shootings including those that were family and felony-related, it might be helpful to look at some of the trends in these incidents separately. So, first, I'll show you those mass shootings that involve family members as victims. So these are raw counts, and you can see that they tend to be slightly more common in recent years with the most occurring in 2011. Note, however, that because these are raw numbers, these are not a rate and that they do not reflect or adjust per population increases over time. In terms of felony-related mass shootings or those that were carried out in the context of other crimes, you can see that these were slightly more common in the early to mid '90s than today, and many of these incidents actually reflected urban violence. The highest of those incidents occurred in 1992. Now, here is what the public mass shootings in our data look like, those that do not involve family members and which were not carried out in the context of another crime. You can see that the trend is increasing slightly, and that the trend increase reflects on average though of three incidents from 1980 to 2019. And you can see, in recent years, that these tend to be more common, slightly.

So in addition to whether mass shootings have become more frequent in recent years, I mean, there's also questions raised concerning the number of victims killed in these incidents each year. So in our data, the range of victims killed is between 46 to 164, and the two deadliest mass shootings have occurred in recent years. So the next question is then: "Are mass shootings becoming deadly over time?" So here is the raw count of victims killed in mass shootings per year in our data. And you can see again a slight increase. And the years with the highest number of victims killed are 1991, 2016, and 2017. Notably with 2016 and 2017 representing Pulse and Las Vegas shootings. Looking at the average number of kill per incident, however, the trend is relatively stable over time up until more recent years in which multiple high-casualty incidents have occurred. So this reflects the average number of victims killed per mass shooting incident. So our data are still a work in progress. And we're currently coding each incident according to all of the information that we've gathered thus far on the shooters, the situation, and the precursors of these shootings. And we plan to also merge our data with additional state-level information on gun legislation, as well as various characteristics from the US Census. So, once we finished coding, we'll be able to conduct more informative analysis to hopefully advance knowledge on why these instances happen and how to prevent them. And the data will be made publicly available when we are finished. So that's all that I have to present to you today. Thank you. And I am going to pass the ball over to the other Jillian.

JAMES DENSLEY: And although Jillian Peterson is going to be controlling the slides, I'm going to be talking first for the first few slides. So hi, everybody. My name is James Densley. I'm a professor at Metropolitan State University. I want to thank NIJ for funding this work, and they're putting their faith in us, and for hosting this event today. I also want to thank the other presenters and panelists for their excellent work. And then also, although it's Jill and I's name on this presentation, I'd be remised if I did not thank the students at Hamline University who helped us build the database that we're about to talk about and also supported many aspects of this project. They had been fantastic, and we would certainly be lost without them. So I just want to acknowledge them that. So very much like the previous presentation, there was a lot of work on the front end of this project establishing exactly the phenomena that we were focused on. And our work is slightly different to what was described by Jillian in the last presentation, which is that we are focused on public mass shootings, not private mass shootings. And also we are focused on shootings that are separate from felony offences, gang-related shootings, urban violence. And we also have a kind of threshold for whether or not the number of victims, if it was 50% or less, were family members for inclusion within this particular database. So slightly different definition that we're working with. We're much focused on public shootings. And, to that point, we also focused this project in terms of public criminology. And the reason we say that is because for the reasons outlined in the previous presentations, mass shootings are a sort of public health issue and they also capture the imagination of the public and our policy members—makers. And we were very dissatisfied with the level of public discourse about this phenomena: the mixed definitions, the way in which we often just sort of chalk this phenomena up to--something to do with gun control versus mental health. And we felt that we needed better data to be able to drive decision-making.

So, we've been very deliberate and intentional in the way we've approached this project as a public criminology project being sort of quite vocal in the media to some extent, and making sure that the information was publicly available, which brings me to the next slide, which is that we built a database that we released to the public on a website that we also built to host it. And the database you can interact with right now. So, if you go to the violenceproject.org, you can download the full database. And we intentionally wanted to put it in the hands of the public. We know, to this date, that we've probably had about 3,000 downloads of the database. It's been downloaded by professors, policymakers, journalists. It's been downloaded by graduate students. And we're really excited about that, and we encourage it because we know that the super smart people out there, they're going to do really cool things with the data, and we want to encourage them to do so. And so here I am again, in public, saying please download it, please play with the database. And if you find interesting things with it, publish with it, just acknowledge us when you do. But we want to make sure that we're getting the information out there so we can influence decision-making.

Now, the database actually has 106 different variables looking at the life histories of mass shooters. And that was really a major focus of this work, which is not just to address the events themselves, but the characteristics of the perpetrators. And with regard to that, we have looked at everything, from their high school records through to their mental health histories, to how they got access to guns. We also have, within the database, a tab that focuses on the victims of the mass shootings because we want to be very intentional about making sure that their lives are also documented in this project. And much like Jillian outlined in her presentation previously, we went through the same sort of coding process. In fact, it's very almost identical the process that we went through in terms of identifying the shooting cases, and then using publicly available sources and records to try and get as much information as possible about those events and about those shooters, including primary sources and secondary sources.

Now, what is different about our project to the others, however, is it includes a pretty extensive qualitative component. So back when we were first at the beginning of this research project, we were using what was the existing publicly available databases out there, for instance The Washington Post had a database which is built on Grant Duwe's work. Grant Duwe was one of the present--one of the teams that presented yesterday in part one. And at that time, there were 150 mass shooters listed from 1966 to 2017. Well, in conversation one day, Jill and I had a crazy idea about what would happen if we could identify those who are currently incarcerated and alive, and write to them, and see whether or not they'd be willing to talk to us? So, we did just that. By the time we identified who was actually alive and able to respond to us, that list of 150 went down to 31 because so many mass shooters die at the scene. They either take their own lives or they are killed by law enforcements. We then also realized that some of these mass shooters were not fit for a research study because of their pleading not guilty by reason of insanity. But we wrote to these individuals, and when we discovered who was not pending litigation or going through an appeals process, et cetera, et cetera, we actually got nine responses back. And of those nine responses, we had two who were declined and said thanks very much but we don't trust psychologists and sociologists, and we don't want to talk to you. And we also got two who were denied become of the prison authorities and other--and other things. In the end, we were able to correspond with five incarcerated mass shooters that fit our definition. These were individuals that we have written a series of letters back and forth with. There were issues about getting into the prisons to actually interview these people face to face. And then obviously when COVID-19 kicked in, that really upended that sort of work. But we have written back and forth with these individuals multiple times as these letters show on the screen. And we, in some cases, have 20 or 30 letters where we've asked a series of deliberate questions about the life histories of these mass shooters.

Now finally, in addition to the mass shooters themselves, we have conducted a series of what we call community stakeholder interviews. So beyond the perpetrators, we have been interacting with the people in the sort of orbit of those individuals. This includes their parents, their family members, their spouses, high school friends, childhood acquaintances, old teachers, and social workers, anyone who is sort of willing to talk to us about the lives of mass shooters we've been trying to interview. We also, through this project, which started to get quite a lot of media attention in the last year, ended up getting sort of unsolicited emails and request from individuals who said that they had a story to tell and wanted to tell it. Now, of course, we had to do a little bit of vetting with this, but through our kind snowball sampling method, we have been introduced to the parents of perpetrators of mass shooters, victims of mass shooters and their families, FBI investigators, and other stakeholders, and also individuals who had planned mass shootings but never actually participated in them. And so, we've looked at this both from a mixed method standpoint. We have a sort of quantitative work with the database and then have the qualitative work with the interviews. And so, with that, I'm going to pass it over to Jillian who is going to actually dive into the data.

JILLIAN PETERSON. Hi, yes. Thanks, James, and thanks everyone for being here. So, I'm Jillian Peterson. I am an Associate Professor of Criminal Justice at Hamline University. And I am a Psychologist by training, so that's kind of the perspective I bring to this work. And prior to being an academic, I was actually an investigator. I worked with New York City Capital Defenders Office and worked on developing life histories of people facing the death penalty. So, that was, I think, one of our main goals of this project is if we can actually really deeply understand the psychosocial history that gets perpetrators to the point of committing a shooting, then we are more able to design prevention and intervention strategies that may be more data-driven and helpful.

So, this is our train data, which looks slightly different because of our slightly different definition, so, four more people killed in a public mass shooting and you can see our data shows a fairly dramatic rise over time, especially looking at 2017, 2018, and 2019. Looking at the location of these shootings, the most common place was a workplace this was nearly always employees of that workplace who had recently been fired or reprimanded. The next most common being retail, a restaurant establishment, and then you see although church shootings, K-12 school shootings, and college and university shootings get a lot of attention, they're actually much lower on our list in terms of frequency. This is just some sort of quick higher-level frequency data. The majority of our mass shooters are male, 98%; I think we have four women in this sample. About almost a third had military background, which is much higher than the general population. You can see nearly two-thirds had a criminal record or a violent history, so this was not their first offense. Domestic violence background is 28%. We did look at things like violent video games and being bullied. And we took kind of an in-depth look at the variable we called "in crisis" which nearly--or over 80% of our sample met criteria. So, we defined being "in crisis" as a market change in behavior and the days or weeks leading up to the shooting which was communicated to other people. And we said it was when your current situation overwhelmed your ability to cope. So, it was a really broad definition. And if you look at the types of behaviors that were seen, so things like increased agitation and increase in abusive behavior, isolation, losing touch with reality, mood swings, depressed mood. And, for most perpetrators, they were having multiple of those crisis signs with nearly 40% of the sample showing five or more of those changes.

We also took a real in-depth look at mental illness. So, being a psychologist, this is something I was very interested in because it is so often part of our public policy conversation. And again, this is using publicly available records. Ideally, we would have access to full medical files. But using what we could find, we have 20% had a prior hospitalization, about 30%, prior counseling. About a quarter of the sample used a psychiatric medication, which is pretty consistent with the general population. Where we do see a significant difference from the general population is in thought disorders, so that was 26, 27%, where in the general population it's more like 1%. And if you cluster all those mental health variables together and collapse them into what we call any mental health background, it's about 70% of the sample has some sort of mental health history. But we are also interested in going more nuance into this because just because someone has a history of mental illness, it doesn't mean that symptoms of that mental illness actually influence or cause them to commit that crime.

So, we did a deep dive into psychosis specifically because symptoms of psychotic disorder, hallucinations and delusion are not generally found amongst individuals without mental illness whereas symptoms of depression and even bipolar things, like irritability, can be found amongst the general population. So, when looking specifically at psychosis, we ended up creating this continuum. We started out coding: did psychosis play a roll, yes, no; and quickly realized that it's so much more complicated than that and that perpetrators had these kind of complex motivations. So, we found that psychosis played no role in the crime 70% of the time. A small role, 11% of the time, a significant role but it was not the only motivation around 9% of the time and then psychotic symptoms completely motivated the crime 10% of the time. So, in terms of policy, I think this amplifies the idea that whether or not mental illness causes mass shootings is not a black and white conversation. But that the data show this is pretty nuance for the majority of mass shooters, it does not play a role; for 10% of them it completely caused the crime.

We found that 31% of mass shooters were suicidal prior to their attacks. Meaning they were telling other people they were thinking of suicide, or writing about suicide, or had previous attempts. Forty percent were suicidal during the shooting; they may kill themselves in the shooting. And 59% died on scene. So, this finding that mass shootings are oftentimes suicides instead of only homicide was a striking finding for us, and I'll talk more about that later. We looked at motivations of mass shooters and this is tricky to code. We used all of the available data we had. So, including what the perpetrators stated their own motivation was, interviews with other people who knew the perpetrator about their motivation, and then looking at what the precipitating incident was. And we look at these over time, so 1960 to 2000. We looked at 2000 to 2015, and then we looked up the last five years. So, you can see employment-based shooting. These are typically workplace shootings, shootings motivated by legal issues and motivated by misogyny have been decreasing over time, whereas shooting's motivated by interpersonal issues, psychosis, domestic violence, and relationship issues, have been fairly consistent over time. And where we've seen the real increase in the last five years is mass shootings motivated by racism, by fame seeking. So, kind of that quest for notoriety and religious hatred.  

We also ended up building an additional database that is our gun data base. So, in that database, it has every gun that was used in the mass shootings since 1966. So, it has 378 guns in that database, including the exact type of gun it was, when it was obtained, how it was obtained. And you could see handguns are used most commonly, 56% of the time. Assault weapons are used about a fifth of the time, which is significantly different than other forms of homicide where assault weapons are only used around 1% of the time. And, like James said, we also did all of these interviews. And for us I think that was a really--I don't know, it was a really powerful project. I think it made us realize the importance of these types of case studies and those subjective narratives. And it served both kind of as a validity check for our publicly available data so we could look and say, "Here's what the publicly available data says. What do our interviews and in-depth case studies say, and do those match?" It also, sort of, let us discover new themes within the interview data and then be able to look at those within our quantitative data. So, in some cases, we interviewed people on both sides of the same mass shooting, so, a perpetrator's parents and the victim's parents. And you just really got these powerful stories of trauma, and of coping, and of resilience. And we are--currently, we just finished writing a book really heavily based on the interview data that will be released next year. So, in terms of scenes that really emerge from the qualitative data, one was trauma. That sort of very significant adverse childhood experiences were really common amongst mass shooters. This is oftentimes sexual abuse, physical abuse, a parent committing suicide, really significant trauma that mass shooters reach the breaking point and oftentimes becomes suicidal that this is meant to be their final act. That was a common theme that emerged and these noticeable signs of crisis that people were noticing. ‘Three’ we're calling social proof, but kind of capturing the idea that mass shooters study other mass shooters, that they look for guidance and what people have done in the past that a lot of them get radicalized online and that search for kind of validation and feeling part of something bigger than they are was a really common theme. And then fourth is opportunity, so just perpetrators need both the means to shoot, access to guns, and access to the places and people they want to target, and the places and people they target were really representative of that grievance that they held.

So, just a touch on some implications, some early implications from our data, in terms of the firearm data, it would indicate that closing illegal loopholes is really important, and safe storage, which was talked about in the first presentation so that school shooters tend to almost exclusively get their guns from family members that they're stealing. Another big finding was that school shooters and workplace shooters, in particular, are insiders, not outsiders, so they're students of the school or employees of that workplace, which can lead to some questions around our current strategies around things like drills and hard security because the perpetrator would be going through that same security and through those same drills. The fact that the majority of our sample was suicidal and that there was noticeable signs of crisis in their--or signs of crisis made us think about suicide prevention and what we can learn from the world of suicide prevention and apply it to potentially prevent mass shootings. Because when someone is actively suicidal, our traditional deterrent strategies are not as effective if someone plans to die in the act, so what can we learn from the suicide prevention world? Same thing around things like punishing threats and criminally charging threats: we know in many cases that only exacerbates the grievance and intensifies the crisis. We know that leakage was really common, so what reporting systems do we have for leakage? We talked to plenty of family members even who knew. I had one interview with a mother where she said, "You know, I knew something was wrong, but what was I going to do? Call the police on my own kid for something they hadn't done yet?" And so where can people go when they have concerns on law enforcement and how we do we build those systems? The spike in fame-related mass shootings have implications for how we're covering mass shootings in the media and how we are sensationalizing those. And also the fact that many mass shooters get radicalized online has implications for the role of social media companies in sort of combating some of the hateful rhetoric that's occurring.

So, this is our email addresses. That is the website where you can download the database, if you'd like. And we've got some articles and books and things coming out over the next year. Thank you. I'm passing it back to you, Basia.

BASIA E. LOPEZ: Thank you very much to all the presenters for this very important information and summary of your findings. Before we move on, we will have a Q&A answers, questions from our panelists in a moment. But first I would like to ask all of you on the implications--or what are the implications that findings have for the criminal justice systems and on prevention. So Jill, you touched them on the prevention already and that--but please go ahead and start elaborating. We will start with Steven and Josh, then we will move to Jillian Turanovic, and then we will go back to our team, Jillian and James. Steven?

STEVEN CHERMAK: So I guess in terms of the implications based on our data, I mean, to me, it's really interesting, or to us it's really interesting, the multifaceted phenomenon, which are school shootings that is--I think any prevention model must recognize the multifaceted nature of school shootings. The violence that we saw really encapsulates many categories, including suicides, including accidents, and intentional shootings. In addition, we have some cases involving targeted students versus others that are domestic violence, that are workplace violence, et cetera, and we have a wide variety of offender characteristics. So, thus, you know, there's really no one profile or policy that adequately embodies the significant criminological problem. But intentional school shootings, that is, especially those that occur outside the building appear to be somewhat representative of youth violence more broadly and though they also share several unique qualities that merit some consideration and discussion, so [INDISTINCT] K to 12, firearms violence would also do well to factor these complexities. Again, a lot of the school shootings that we saw really take place in urban areas, and many also occur outside the building, but on school grounds. Incidentally, a minority of intentional school shootings occur in classrooms. Many shootings occur when school is not in session, at night, or on the weekend, or during a break, and some of the school shootings were gang-related, as well as drug-related. Thus, a lot of school shootings, while happening on school grounds, are not school related. It appears some of these shootings are wider community problems that spill into the school and preventing the opportunity to consider the complexities.

The only other thing I'd say, and I'm probably speaking too long, but the leakage variable that Jillian and James just mentioned, as well as with our data, it's really an important finding. You know, our deeper dive in the finding shows that there were interesting differences in leakage across modality, type of warning behavior, as well as a time lapse between behavior and a shooting occurring. Most of the shooters, when sharing their intentions, did so offline, primarily conveying their thoughts to their friends. And one of the key challenges we saw is that most warnings, most leakage, occurs very close to the time of the event. So, it's sort of crucial to think about policies that recognizes the really small window of opportunity to take advantage of, in order to prevent an attack.

BASIA E. LOPEZ: Thank you very much, Steven. Josh, do you want to add anything?

JOSHUA D. FREILICH: Sure. Can you hear me?

BASIA E. LOPEZ: Yes, we can hear you now.

JOSHUA D. FREILICH: Yeah. I wanted to thank you guys here for taking the time to organize. And also, on behalf of Nadine Connell who's also a PI in the project from Griffith University. You know, just to follow up on what Steve said, you know, when you think about it, even in terms of leakage, that is going to be very important for intentional shootings, when you're planning it beforehand. But you're not really going to have leakage for an accidental shooting. So, I think it really highlights that when--and Steven noted this already, but it's really a multifaceted problem, and in some sense, when you look at school shootings as a whole, it's really a series of problems that occur in school grounds. Right? So you have the accidental discharges, you have suicides, you have the intentional shootings, and it's the shootings where students are potential targets that get a lot of attention. But we also have intentional shootings that involve domestic violence or workplace violence, and they might require a different type of strategy. So, I think one takeaway is maybe not looking for a one size fits all, but rather possibly taking into account a series of targeted responses depending upon the particular problem at play.

BASIA E. LOPEZ: Thank you very much. Thank you, Josh and Steven. So, let's move with the same questions to Jillian Turanovic.

JILLIAN J. TURANOVIC: You know, similar to the school shootings data, our mass shooting data that we are compiling is very diverse. I mean, it captures very different types of mass shooting incidents. And so we have family killings, we have mass shootings that reflect urban violence, and then we have, you know, those public mass shootings, those more random, kind of targeted attacks that are occurring in public spaces. And so, although we couldn't quite finish compiling all of our data yet to really put forth policy implications, what I hope that we will be able to identify once we complete that is factors that may be unique to certain forms of mass shootings, and those that may actually be general across different types. And if we could target maybe some of those general factors, we could possibly see a diffusion of benefit to reduce various forms of mass violence in private spaces and in public spaces. And so I think by looking at these different types of mass shootings, and mass shootings as a whole, we'll be able to see ones that are--or factors that are particular to certain types of mass shootings, and then those that may be general to all forms. So that's what we hope to use this data for. And also, to look over time as well at, you know, the different forms of gun legislation that have been implemented over time, and how that may have affected different types and patterns of different forms of mass shootings. So, I don't have clear-cut policy responses yet, but that is the goal with our project.

BASIA E. LOPEZ: Okay. Thank you very much, Jillian. And now, I will give it back to James and Jillian Peterson.

JAMES DENSLEY: And, yeah, I would just echo what both groups have just said, which is, you know, we--the different types of issues need different types of responses, and I think that's really well-articulated in these presentations. The only thing I really want to stress, which I think is maybe a different approach or different thought is, you know, having spent a lot of time with families, victims, survivors, first responders, and others, the one thing that really strikes me in this work is that we have, in our media and in our culture, sort of painted mass shooters as these kind of scary monsters that cannot be explained, and, therefore, we should just kind of accept this as a sort of, you know, rare event reality that is part of American culture and society. But actually what we find with our work is that until we understand that these individuals aren't these outside monsters, they are the insiders, they are people that we know, they are our children, they are our classmates, they are our family and friends, and this issue around leakage and everything else, is really very clear in our work that we have to understand and appreciate mass shooters as human beings so that we can have a sort of human response to what's going on. We can't run, hide, and fight our way out of this problem. That is only a small part of the solution, and there's so much more that we can be doing, and we can only get to that stuff once we really understand who those individuals are.

JILLIAN PETERSON: Yeah, I'm not sure I have much to add. I would just emphasize that these findings that these are suicides as well as homicides, I think, has some really important implication both for what we're doing right now, because if you're suicidal, somebody with a gun isn't going to deter you on being punished for making a criminal threat, isn't going to deter you, it's a really sort of different place. And so, thinking about how do we give someone hope, how do we identify when someone's going down this pathway, and then how do we pull them out, I think that's where our prevention efforts are going to need to go moving forward.

JAMES DENSLEY: Yeah. It's really about identifying opportunities for intervention. The pathway to violence is very long, and we've got to build more off-ramps to get people off.

BASIA E. LOPEZ: All right. Thank you very much for these answers. For--the last question that I would like to ask is just very briefly if one of the panelists could answer it. Based on your research findings, what are the future directions for the research field at this time? Not practice, but for the research field. So Steven or Josh, if you would please answer this question first.

JOSHUA D. FREILICH: I can just jump in. So, you know, we talked about this quite a bit. Nadine, Steven, myself, and we think one issue really relates to greater use of comparison groups. In the study that we presented today, we had a pretty broad inclusion criteria in the sense that we only needed one injury to occur on school grounds, but that allows us to compare fatal to nonfatal, so we have a comparison group. But going forward, it would be very useful to also look at foiled plots. Steve talked about leakage, but when you have foiled plots and completed shootings, you can really dig deep into where leakage occurs and under what circumstances. Furthermore, you could also look at the types of policing strategies that might be successful at preventing these attacks. Obviously, some shootings are going to occur, and the police or others are going to have no idea. In others, there might have been a warning sign that was missed. But it would be useful to not only try and capture particular investigative strategies that police are using in a single case, but also what the police--what the police agency does on a more global level. So have they employed problem-oriented policing or community-oriented policing? These would be very useful factors to compare between the completed and the foiled.

And then another comparison that would be pretty interesting to look at, especially in terms of situational crime prevention and risk assessment, what the vulnerable targets are, it would be interesting to take a school where a shooting occurred and to match it under some criteria with a school where no shooting occurred. So then you could really begin to dig deep, once you do this comparison, to find out: are there particular attributes that make a particular type of school more vulnerable to attack? And then, finally, all of the projects that they have used open sources, as Steve and the others have noted today, open sources have become so much more popular to study rare events in criminology. But with the greater use of open sources, I also think--we also think comes greater responsibility. So I think as researchers, we should think about ways to make the raw data that is being used to code these variables more transparent. And there are a number of ways that this could be done, such as talking about the types of documents, which we did a little bit about today. But for key indicators and key variables, in addition to the substantive value, it might be useful in creating reliability indices as well, where you could look at the types of information that give you the values of that code, whether there was any type of contradiction among sources, the number of sources, and the level of conscience. And then share that with other academics as well. So moving beyond just here in our database, here are the values, giving researchers the opportunity to dig deep into the raw data behind the values.

BASIA E. LOPEZ: Yes. And the same question. Thank you very much, Josh, for this. The same question to Jillian Turanovic.

JILLIAN J. TURANOVIC: All right. From a research perspective, especially with data like ours, I'm interested in how the problems of mass shootings fit within the broader study of gun violence. So, to the extent that some of these mass shootings could follow similar patterns to homicide in general, follow similar trends in homicides, especially given that, you know, a lot of our mass shootings reflect problems with urban violence and domestic violence. So, to the extent that certain policies can be implemented to curb those forms violence, as well as some of the types of mass violence that occurs in our data, I'd be really interested in looking at that. And in addition, the extent to which different types of gun laws not only impact the incidents of mass shootings, but also the conditions under which these laws are most effective. So, are there certain forms of mass shootings that they are most likely to deter and how they impact? Yeah. There's different forms of gun violence in different situations and different population. So, I think in the future maybe merging some of the general research on gun violence with mass shootings to see if there's overlap there, and again, to hopefully curb violence in general. So, that's all I have.

BASIA E. LOPEZ: Thank you, Jillian, very much. And back to James and Jillian again, please.

JAMES DENSLEY: Yeah. I would say one question that springs to mind is how COVID-19 impacts public mass shootings. So, you know, you look at this year, for example, and we had a mass shooting in Milwaukee back in February. But we've not had another public mass shooting, four or more people killed by the sort of definition that we were looking at, since then. And I think that's because of two things; one, opportunity. When public places are shut down, you can't have public shootings. And two, I think it's the impacts of the contagion effect to some degree. What I mean by that is we're not talking about mass shootings anymore. The media is not talking about them. We are focused on other issues at the moment. Young people are focused on other issues. And maybe that's having an impact as well. So, that, in turn, though, gives us some sort of indication of, like, what possible intervention points might be in those ways. So, that's certainly a research question I think would be very interesting to look at if it were possible to do so. And also, that depends how long the pandemic lasts as well, I guess.

And then I think the other thing as well is, just to echo what Jillian T. said earlier, which is, you know, thinking about mass shootings in the context of what we already know about other forms of violence, particularly gun violence and how there are similarities but also differences with this type of phenomena, where does this fit in terms of urban violence and gang violence, where does this fit in terms of domestic violence, and so on. And so that, I think--which is why I'm very excited about her project and the ability to be able to sort of separate these things out. I think that's going to be really great.

BASIA E. LOPEZ: Thank you, James. Jillian, would you like to add anything?

JILLIAN J. TURANOVIC: I think James covered it. We can move on to other Q&A.

BASIA E. LOPEZ: Terrific. Thank you very much. Again, thank you to all my panelists for your excellent presentation and for sharing all this information with us. I will pass the ball now to my assistant, Danielle. And she will ask some questions from the Q&As that you posted during this webinar. For those of you who cannot stay online, I encourage you to look toward--forward to an email when we will send the recording of this presentation. Thank you. Danielle.

DANIELLE CRIMMINS: Thanks, Basia. The first question is for Joshua and Steven. "Forty-five percent of schools had a school guard or officer at the time of the shooting. Does that mean they were employees assigned to the campus for the school year or they were physically present at the time of the shooting since so many occur outside of school hours?"

STEVEN CHERMAK: Yeah. I'll answer that. It's a great question. Thanks for whoever sent that. We've looked a little bit at that because--and it actually applies to all those situational--one of the issues we're thinking about for all the situational crime prevention related measures in terms of making sense of whether or not they matter, whether or not they contribute to differences in fatal versus nonfatal. So, in general, what we presented there, the 45% represented whether or not the school had a school officer or a school guard. That's it. But looking at it and sort of breaking that out by, you know, looking at just the events where the shooting happened during school hours and whether or not they are present is an important element of that--understanding that phenomenon. Thanks for the question.

DANIELLE CRIMMINS: Thank you, Steven. This next question is for Jillian T. "What are the characteristics of female shooters? Why are men so dominant in mass shootings?"

JILLIAN J. TURANOVIC: That's a good question. The majority of female shooters in our data were women who killed their families or--and there was maybe one or two that were operating with men to carry out mass shootings. And so, I don't have the answer for why these tend to be so male-dominated. But, like other forms of violent crime, men are overrepresented here. But hopefully when we start coding more about our sources and triggers of these events, hopefully I can get some more information. But the female shooters are interesting. And, yeah, the majority of those are family-related shootings.

DANIELLE CRIMMINS: Great. Thank you, Jillian. This next question is for James and Jillian P. with regards to the database. "You mentioned having a tab for the victims of the shooting. Was this focused on victims who were killed or does it also include victims who were injured?"

JILLIAN PETERSON: I can take that. The victim's database is just those who were killed. We would have liked to do also injured, and we may add that in the future. But in terms of being able to gather enough information about them, like, their name and include, like, gender, race, and if they knew the perpetrator, we just started out with victims killed.

DANIELLE CRIMMINS: Great. Thank you. This next question goes back to Steven and James. "Can you explain how shooting events are determined to be justified in quotations with 0.061% in parenthesis?"

STEVEN CHERMAK: Sure. I can take it. So, I believe that's referring to--we had four cases that we classified as legally justified. Those were cases where charges were--it was investigated and no charges were ever brought. So, it was deemed to be a self-defense situation. I don't have the facts of all of them. But I believe at least one of them occurred outside, after school hours. And there was some type of altercation in the yard. And then after the investigation, no charges were ever brought.

DANIELLE CRIMMINS: Great. Thank you. This next question is for Jillian P. "How do events like the 2015 San Bernardino attacks fit in your definitions involving more than one location?"

JILLIAN PETERSON: That's a good question. So, we started out with one event in one location. And as we got into the data more, we realized that how do we define what one location is? There was many where--so, for instance, in addition to the San Bernardino killing. I mean, there was one, for instance, where a man shot people inside his house, went next door, shot people there. Then there was another where someone drove across town and shot more people. So, we actually modified this in practice. And especially after--last year, we had a research meeting at NIJ. And so, we changed our definition instead of one location to a 24 hour period. And so, if these are now shootings occurring in a 24 hour period and four people died or were shot, we're including it. So, that's how we got around that issue because it was too tricky to trying to determine what's a certain mile radius that someone may have travelled so that's what we're using now. And so many of our shootings occur in multiple locations.

DANIELLE CRIMMINS: Great. Thank you. This next question is for Jillian P. and James. "With respect to DV, domestic violence, as a variable in shooters, what was the personal history or exposure in their childhood home?"

JILLIAN PETERSON: So, we actually coded--so, we coded domestic violence. And then we also coded childhood trauma, which I didn't talk about, does the childhood trauma variable then also goes into where they physically victimize, emotionally victimize, other--some other form of trauma, who did it, who perpetrated it, close family member, and so, if you download the database and dig into that variable, you can sort of get more in-depth with it. The domestic violence one was: were they perpetrators of domestic violence in their adulthood?

DANIELLE CRIMMINS: Great. Thanks. So, we have one that is for the whole group, I believe. "With all the different databases being used to gauge mass shootings, how can we get a handle on the nature and extent of mass shootings and address them as information concerning the prevalence? Is it possible to have one comprehensive data source?"

STEVEN CHERMAK: I'll jump in. And then the mass shooting folks can, you know. I think it's a good question but I don't think it's necessary. I actually think there's extraordinary value in having a variety of different databases coming at this from different perspectives with different--somewhat different inclusion criteria and definitional issues and sort of bringing those out. All of us, I think, if we could just talk probably for hours in terms of some of the data-related challenges we face and decisions that we made, and they made sense and we sort of laid them out in terms of--but ultimately, it impacts, you know, sort of the end, as well as what we find. And I think that's okay. It's just, sort of, one lens that you can look at an issue like mass shootings. And then you can look at another dataset and it's another lens. But they're both important lenses to really get a good understanding of the problem.

JILLIAN PETERSON: I was just going to say that I really agree with Steve. I think the diversity is a great thing. And as long as you're very clear on what your definition that you're using is and so people can interpret that accordingly and recognize that, depending on how you define the problem, it has consequences and not--so, if you're defining a mass shooting as four or more people shot but not necessarily deadly, that doesn't mean that that--that there's four hundred incidents like Vegas or El Paso happening in a given year. I think it's just being mindful of the definitions that you use. But I agree that the diversity and data sources is a good thing. It's how you would get all different kinds of information on different features of the problem.

DANIELLE CRIMMINS: Great. Thanks. I think we are now at 15 minutes over. I think I'm going to hand it back to Mary Jo.

MARY JO GIOVACCHINI: No. I really appreciate that. And thank you to all the presenters that offered their presentation today. It was wonderful. So, on behalf of the National Institute of Justice and everybody here today, I would like to thank you very much for joining today's webinar. Have a wonderful day.