The Safety Of Trains v. Cars and Trucks

Understanding Rail Accident Risks With Appropriate Context

Honestly, it is surprising that so many railway deaths are not in the right of way (ROW). The defining characteristic of rail transportation is that it travels on fixed tracks which have a right of way.

There are extremely rare incidents of someone getting killed from a train while not being on the tracks - like when it falls off a bridge and kills someone below as one did not so long ago on I-25 near Pueblo, or when an accident releases a poisonous gas that kills people outside the ROW but nearby which happened to the parents of one of my first clients - but those are vanishingly rare.

Of course, what they are really doing in these statistics is distinguishing between deaths in the ROW at highway crossings and deaths in the ROW at something other than highway crossings, and basically ignoring the tiny percentage of deaths that fall in neither category.

The regulatory focus on highway crossings to the near exclusion of safety measures in railroad railways isn't unreasonable.  According to the Federal Railway Administration:

Highway-rail grade crossings are intersections where highways cross railroad tracks at-grade. Approximately 212,000 highway-rail grade crossings exist on the approximately 140,000 miles of track that make up the United States’ railroad system.

If we conservatively assume that the average highway rail grade crossing is 105.6 feet (i.e. 0.02 miles), then there are 4,240 miles of track in highway-rail grade crossings, which is about 3% of the total miles of track in the U.S. (and realistically the truth is probably closer to 1-2%). 

While only 30% of rail deaths nationally are at highway crossings, the number of deaths per mile of track are 10-30 times greater at highway crossings than they are away from them. But, the cost of safety measures to prevent rail deaths is roughly proportional to the length of the track when they are put in place.

Also, both passenger rail and freight rail cause far few deaths per passenger-mile than cars, to both passengers (who are safer by a factor of seventeen) and bystanders like pedestrians and bicyclists and people other cars, and per freight ton-mile than trucks carrying freight to both operators and bystanders.

In the U.S., in 2024, there were 954 railroad deaths and 6,542 nonfatal railroad injuries.

(Source)

There is about 1 fatal rail death away from highway crossings per 196.5 miles a track away from highway crossings, and there is about 1 fatal rail death per 806 highway crossings at highway crossings. Vanishing few fatal rail deaths are to people on the train itself 

People on railroad tracks who shouldn't be (i.e. trespassers) account for 69% of railroad deaths, but only about 10% of nonfatal railroad injuries, because 51% of rail accidents involving trespassers that make it into official statistics are deadly. 

The vast majority of rail traffic in the U.S. is freight. U.S. freight rail carried 1.71 trillion ton-miles per year with about 100 tons per train car. So, about 17,000 million train car-miles per year. 

Amtrak provides 6544 million passenger miles per year of transportation. About 35% of those are in the Northeast Corridor which run about 80% full on average and about 65% are outside the Northeast Corridor which run closer to 50% full on average. A coach car holds about 74 people which is about 59 people per car in the Northeast Corridor and about 37 people outside it. So, about 39 million train car miles per year in the Northeast Corridor and about 115 million train car miles per year outside the Northeast Corridor, for a total of about 152 million train car miles per year.

So, about 0.9% of train car miles are passenger train car miles and about 99.1% of train car miles are freight train car miles.

If rail deaths are no more likely, per train car mile, to be caused by freight train cars than by passenger train cars (although arguably the risk is per train and not per train car, in which cars passenger trains which have far fewer cars per train are much more dangerous, even though this is harder to estimate), we can reasonably estimate that passenger trains kill about 12 people per year on average (2 passengers and 10 other people), while freight trains kill about 942 people per year on average. Adjusting for train car length who shift this somewhat from freight trains to passenger trains, but even so, freight trains would be the dominant issue.

Do Existing Rules Reflect The True Nature Of The Problem

Of course, it only makes sense to take measures to reduce right of way deaths away from highway crossings if there is an effective (and cost effective) way to do so. 

People walk across and along train tracks all the time, and since trains are noisy, and you only need to move a few feet away from train tracks to avoid being hit by one, normally walking across or along train tracks is pretty safe, so long as you are able to hear them coming, you pay attention, you don't fall asleep on them, you don't get stuck on them, and you get out of the way when they come. And, it isn't really clear what the statistics use to define a trespasser, and we can't really understand the statistics without knowing this definition.

Is a trespasser merely someone who is in the right of way at all? Is a trespasser someone who tries to hitch a ride on a train that isn't using it like a passenger on a typical passenger train? Or what?

Common intuition wouldn't consider crossing over a train track when there is no indication that a train is coming, away from a highway crossing, to be trespassing, nor would common intuition consider walking along a train track in the right of way when there is no indication that a train is coming to be trespassing. I did that on my walk to and from school in junior high school almost every school day for two years without anyone even suggesting that I was trespassing.

Crossing a highway crossing when there are indications that a train is coming such as flashing lights or barriers that come down probably is trespassing, but that obviously doesn't apply when you aren't at a highway crossing.

The Federal Railway Administration, however, begs to differ, stating:

It is illegal to access private railroad property anywhere other than a designated pedestrian or roadway crossing. Trespassers are most often pedestrians who walk across or along railroad tracks as a shortcut to another destination. Some trespassers are loitering or are engaged in recreational activities such as taking photographs, jogging, bicycling, hunting, or operating recreational off-highway vehicles (ROVs). Riding ROVs along railroad tracks leads to the erosion of an important part of the track foundation known as ballast, or the rock and soil material that supports the ties and rail.

But, this definition is awfully harsh. From a pedestrian's perspective, it is often much safer to walk across or along railroad tracks away from a designated roadway crossing, than it is to cross at a highway crossing where there is much more frequent and less noisy car traffic, or across open ground away from the railroad right of way (indeed, railroads that are no longer in use are often converted to pedestrian trails for just this reason) which may constitute trespassing on private property (which is dangerous in much of rural America) and is sometimes uncleared brush. 

Sticking to official pedestrian routes is challenging. Many rural areas have no sidewalks, and walking along rural roads at night can be much more dangerous than walking along railroads. 

Designated pedestrian crossings are extremely very rare outside of dense urban areas and resort areas, because they aren't cost effective in places with little pedestrian traffic and only marginal safety risks to the pedestrians who do without them. 

But train tracks have to be crossed somewhere to get from point A to point B in many cases, because tracks often split whole states in two. 

In flat farmland, highway crossings are often a mile to six miles apart, and highway crossings can be even more sparse in mountains, forests, wetlands, and deserts. 

These detours can be tolerable in a car or a motorcycle, but represent immense delays relative to crossing a train track away from a highway crossing for a pedestrian who may have to cross that track to get to school or work or a friend's house on a regular basis.

So, simply writing off people who are simply meeting the strict FTA definition of rail right of way trespassing as "at fault" is really unreasonable, blames the victim, and unreasonably protects railroad operators from legal liability when the common law "Learned Hand" test for negligence liability (i.e. that the probability adjusted risk of harm is greater than the cost of a measure that would prevent that harm) would hold railroad operators liable for these injuries if they didn't take reasonable and cost effective measures to prevent them.

In the face of unreasonable trespassing definitions and regulations of pedestrians near train tracks, even extremely expensive preventative measures, like tall fences, will be routinely circumvented. And, authorities won't do much to prevent the harm, because the authorities seeing the situation up close will be aware that these work arounds that circumvent overkill prevention measures like tall fences, will mostly do more good than harm.

Also, my intuition is that ordinary pedestrians walking across or along railroad tracks, especially in rural and suburban areas, are not at all typical of people who are killed in rail accidents, because it is so easy to hear trains coming and because it is so easy in most cases to get out of the way. I suspect that these cases probably actually make up less than 10% of rail accidents involving pedestrians, even though the FTA itself admits on its own website that this is the predominant form of "trespassing" in rail right of ways.

Instead, I suspect that the typical cases that result in serious injury or death are very different. 

Maybe the pedestrian is a young child to little to appreciate the risk of being on train tracks when one can hear a train coming. Maybe the pedestrian is deaf or wearing headphones. 

Maybe the pedestrian falls asleep on the tracks while drunk, on drugs, or homeless, mistakenly thinking that this particular track which has low frequency traffic has been abandoned entirely. Maybe the pedestrian has been assaulted or knocked unconscious and left on the tracks. 

Some of these cases may be close to the boundary between reckless behavior indifferent to the grave risk that they will be killed and full fledged intentional suicide a bit like playing Russian roulette. 

Maybe the pedestrian is trying to hitch a ride on a freight train and stumbles. 

Maybe the pedestrian is on a rail bridge or other part of the right of way where there is no easy place to move and get out of the way of the train when it is coming. Maybe the pedestrian somehow gets stuck on the tracks with a foot wedged into a gap or is trying to rescue someone else.

Ideally, better data would add insight that the official statistics do not, and we wouldn't have to rely on intuition. But, relying in intuition grounded in common sense and lived experience until better data is available, is usually better than taking no action at all until better data is available, especially if the measures suggested by that intuition aren't terribly expensive or difficult to implement.

If my intuition is right and these are the predominant cases of result in rail deaths or injuries to "trespassers", then the broad FTA definition in addition to blaming the victim, also undermines its effectiveness by being so broad that it fails to distinguish between reasonable conduct that is only dangerous in the most freak circumstances and unreasonable conduct that really does pose an extraordinary risk of death which should be actively policed to prevent rail deaths and injuries. By not tolerating reasonable conduct, the overbroad definition undermines its own effectiveness at preventing the harms it is designed to discourage.

Distinguishing between "safe trespassing" within the FTA definition and "unsafe trespassing" within the FTA is also critical to determining what kind of preventative measures could reduce rail right of way deaths and injuries.

Possible Infrastructure Based Solutions To Reduce Harm

If the real problem is predominantly people who fall asleep or are otherwise unconscious or stuck or inert on train tracks away from highway crossings, then the solution might be to periodically put something like a LIDAR sensor along active train tracks that warn the train operator of the obstruction in time to stop the train (which could take more than a mile of warning about a minute before the train reaches that point), which would prevent not only deaths to trespassers, but also train collisions with fallen trees, large sleeping wildlife (for which "cow catchers" were designed in an earlier era), and debris carried there by a storm. These preventative measures weren't really technologically feasible or cost effective in the late 19th century and early 20th century when existing freight rail systems were invented. But now, these kinds of sensors and a way to send their signals to trains on the routes are cheap and easy to retrofit commercial off the shelf technologies that require only minor modifications for this application.

At crossings, something like a LIDAR system, or even just an "emergency stop" switch that someone could activate in the highway crossing infrastructure, could reduce the number of collisions arising from vehicles stuck at a highway crossing. Even if the warning was transmitted to the train when it was too late to come to a full stop, on oncoming train aware of an imminent collision could slow down as much as possible and activate maximum warning sirens to mitigate the magnitude of the harm from an unstoppable collision and could alert first responders so that they could on their way to the scene of the nearly inevitable collision before it even happened, in situation where every moment counts in saving someone's life. 

If a big part of  the problem is people who are deaf or wearing headphones, bright, flashing, motion sensor activated lights that only go off when trains are actually coming, located away from highway crossings, which are also cheap, easy to retrofit, commercial off the shelf technologies could do the trick, and might even reduce the need for loud train horns in the middle of the night.

The solutions above would be cheap, and could be implemented at a very manageable cost by railroad operators in response to newly adopted safety regulations (which would prevent a race to the bottom by profit minding companies competing with each other) over a period of perhaps five or ten years.

If the problem is people trying to hitchhike on train cars, hobo style, maybe a partial solution would be to design train cars so that they are difficult to ride or mount, without ladders or anything else to grip, and with the top of enclosed freight cars designs with pitched roofs that are too steep to rest upon for any length of time.

This would be very expensive to retrofit, but if the new design standards were imposed only prospectively to newly purchased freight cars as they were replaced over their useful lives, the new design would probably not be much more expensive than the old one, and as the percentage of new design cars increased over time, the entire strategy of hobo style hitchhiking on freight trains would grow increasingly less popular.

This gradual replacement would also dovetail with reduces consumption of coal, which is the single largest component of of rail freight, in favor of renewables, nuclear, and natural gas transported by pipeline, and declining consumption of liquid fossil fuels (another major component of rail freight) as electric vehicles gradually replace internal combustion engine vehicles, with freight rail instead being used to ship more containerized cargo.

Regular underpasses or overpasses for wild animals might not save many human lives but might prevent train collisions with wild (or domestic) animals that are unable to get out of  the way in time or don't realize the risk, which is surely something that both conservationists and train operators would appreciate, which is fundamentally, very low tech, like Roman Empire class civil engineering. But these measured could potentially be quite expensive to retrofit into existing rail lines. Still, good data on "rail kill" could identify the areas where these kinds of collisions are most common so that the investment in these solutions could be implemented first where it would do the most good, and these standards could be established for new rail lines (mostly proposed high speed rail lines).

Admittedly, none of these measures would be very effective against someone who is simply reckless to the point of being virtually suicidal, or someone who is actually suicidal. 

For these cases, a highly targeted campaign of public service announcements, that focuses only on the conduct that other preventative measures can't deal with, that actually is high risk, in channels that the people most likely to act this way are most likely to hear and take seriously may be the best possible solution - a bit like the campaigns in the 1980s about drunk driving and the importance of having a designated driver. A PSA campaign probably wouldn't be as effective as some of the new safety infrastructure suggested in this post, but it would almost surely make some difference if it was well done.

Civilian Gun Ownership Rates

The state by state breakdown may be using a different methodology than the international one.

Seemingly, five U.S. states have lower gun ownership rates than England and Wales, which I am skeptical of, although the relative rates of gun ownership should be solid. Utah, an ultra conservative red state in the "cowboy west", remarkably, has the third lowest gun ownership rate.

 (Source)

Civilian gun ownership rates are surprisingly low in Switzerland, which is used when looking at gun control measures internationally as a country with widespread gun ownership, yet low crime.

Canada has gun ownership rates less than 30% of those of the U.S. 

Russia has gun ownership rates just barely more than 10% of those of the U.S.

 (Source)

In Europe, civilian gun ownership rates are highest in Scandinavia, especially rural Scandinavia, although it is below U.S. rates everywhere.

(Found on Facebook without a citation for its source, probably E.U. statistics prior to Brexit)

Middle Class Crime

Middle Class Prison Inmates Are Rare

There is a lot of data on the characteristics of offenders and nature of the offenses committed overall, and there is significant data on sentencing for felonies (sentencing data for misdemeanors and ordinance violations is much harder to come by).

One of the things that we know from that data is that having any college education, even attending a single semester at community college and then dropping out, profoundly reduces you odds of being a prison inmate relative to have no college education at all. Your odds of being a prison inmate are 40-fold or more lower.

Clearly, this is mostly a sorting effect. 

Your likelihood of being a prison inmate is highest if you don't graduate from high school and also don't earn a GED. 

The next highest likelihood of being a prison inmate is earning a GED (even though it takes a higher IQ to do so than it does to graduate from high school in the ordinary course), rather than graduating from high school and then not receiving any further higher education. This reflects the fact that people, especially men, drop out of high school not just because they are low in IQ and not succeeding in school for that reason. They also drop out because they have trouble behaving appropriately in a high school setting, which is highly connected to their likelihood of committing crimes and going to prison.

Graduating from high school in the ordinary course rather than earning a GED, and then not receiving any further higher education is the next highest likelihood of being a prison inmate, and is still quite high.

But, some college or more dramatically reduces your likelihood of being a prison inmate, and the number of prison inmates who are in that category is so small that the statistics often aren't broken down further into people who have only some college, people who have associate's degrees, people who have four year degrees, and people who have graduate or professional degrees beyond a four year undergraduate degree.

By comparison, in the general population:

In 2022, the highest level of education of the population age 25 and older in the United States ranged from less than high school to advanced degrees beyond a bachelor’s degree.

9% had less than a high school diploma or equivalent.

28% had high school as their highest level of school completed.

15% had completed some college but not a degree.

10% had an associate degree as their highest level of school completed.

23% had a bachelor’s degree as their highest degree.

14% had completed advanced education such as a master’s degree, professional degree or doctorate. . . . 

In 2022, 30.1% of men age 25 and older had completed a high school diploma or GED as their highest level of educational attainment, compared with 27.0% of women age 25 and older.

In 2022, 39.0% of women age 25 and older, and 36.2% of men in the same age range, had completed a bachelor’s degree or more as their highest level of educational attainment. 

About 16% of adults have passed a GED exam. About half of people who pass the GED exam go on to have at least some college. So, about 8% of adults have a GED as their highest educational credential (roughly the same as the percentage of high school dropouts who haven't passed a GED exam). 

Both high school dropouts and people who have a GED exam as their highest educational credential are disproportionately men. In the case of the GED, this is, in part, because a large share of GED exams are taken high school dropouts who are in prison, and prison inmates are disproportionately men.

So, about 36% of men have at least a four year degree, 24% of men have some college but no degree or an associate's degree, about 20% of men graduated from high school in the ordinary course, 10% have a GED rather than a high school diploma and no college, and about 10% of men are high school dropouts who have not earned a GED.

The roughly 60% of men with at least some college have a dramatically reduced likelihood of going to prison. The 20% of men who are high school dropouts or only have a GED have a highly elevated likelihood of going to prison, and the 20% of men who graduate from high school but have no further high education have an intermediate likelihood of going to prison. On average, these men are in the bottom quarter of high school graduates academically (although a variety of other factors influence a decision not to pursue any higher education after graduating from high school).

One expects that among the 20% of men who graduate from high school but have no further education, men who are at the top of that group academically and behaviorally but didn't go to college because they had decent job prospects or went into the military or couldn't afford to go to college, probably have a likelihood of going to prison similar to that of people who drop out of college soon after starting college. But men who are the the bottom of that group academically and behaviorally probably have a likelihood of going to prison similar to but somewhat lower men who drop out of high school and then earn a GED but don't pursue further education. They are able to behave marginally well enough to meet the minimum requirements to graduate, and are not totally stupid.

So, the likelihood of going to prison is probably most elevated among men in the bottom 30% to 1/3rd of success in the formal education system, and are profoundly reduced among men who are more successful in the formal education system.

Operationally, however, since there is almost no data sorting high achieving and well-behaved high school graduates with no college, from low achieving and poorly behaved high school graduates with no college, we can define "middle class" inmates as inmates who have some college or more.

Questions That Are Hard To Answer

1. Some of the reason that middle class inmates are so rare is that middle class criminal defendants tend to have only minimal prior criminal records and are seen as having good prospects of rehabilitation. They also tend to have been legal representation in the court system, make good choices about when to accept plea bargains, take actions that judges view as mitigating circumstances, and behave in the way that sentencing judges want them to in the court process, relative to non-middle class criminal defendants. 

So, a middle class criminal defendant who commits the same crime as a non-middle class criminal defendant is more likely to receive probation, a fine, community service, time in a half-way house, or a jail sentence, than an ordinary prison sentence, especially for less severe crimes. Likewise, even if they receive a prison sentence, it is likely to be shorter than the sentence that a typical non-middle class criminal defendant would receive for the same crime.

2. Middle class criminal defendants probably tend to commit less serious crimes, when they are convicted of crimes. They tend not to commit "blue collar" property crimes that can send you to prison like car theft, burglary, or grand larceny of tangible personal property (as opposed to fraud or theft of intangible assets). They also probably tend to commit the same sorts of crimes that are disproportionately committed by women.

3. I suspect that middle class criminal defendant who actually go to prison have mostly either committed high dollar/drug volume non-violent crimes like fraud or drug dealing, or have committed serious violent crimes for which a prison sentence is really the only plausible sentence. But, I don't know that for a fact and I haven't seen good data on that point.

4. I don't know and would be interested to know how educational attainment beyond high school influences one's likelihood of being a prison inmate. I would suspect that the risk is reduced with greater educational attainment, but the data isn't there to show how significantly protective education beyond "some college" is in terms of likelihood of being a prison inmate.

4. The trio of serious substance abuse issues, serious mental health issues, and traumatic brain injury are pervasive in a huge share of all prison inmates. Middle class prison inmates are probably no exception. Indeed, these issues probably make up a bigger share of middle class prison inmates, since economic pressures on these inmates are usually more mild.

But notably, the protective effect of even some college is huge, despite the fact that the rates of serious substance abuse and serious mental health problems is very significant even among college educated people. Most likely, people with at least some college (overall, not just prison inmates) both have mental health and substance abuse problems that are sufficiently moderate that they can function well enough to finish high school and spend some time in college, and they are better at coming up with mental health treatment, substance abuse rehabilitation, and self-help programs to manage to live an at least moderately normal life despite these challenges, while the most severely impaired people have behavioral problems early enough that they can't finish high school and some college, and/or they can't figure out a self-help regime or access the mental health care system resources that they need.

Also, some mental health and substance abuse problems have a relatively late onset. 

Susceptibility to substance abuse has a very strong genetic component. But people who grew up in strict families, or avoided substances after seeing how it affected other family members, may not have had access to substances that they can abuse, or strictly avoided substances that they can abuse, until they are in or have graduated from college.

Most mental health conditions are genetic or congenital. And, most of them, like ADHD, autism spectrum disorders, psychopathy, anxiety disorders, and OCD typically manifest before you finish high school, so people with particularly severe cases of these conditions that aren't treated adequately either don't graduate from high school for behavioral reasons or just barely scrape by at that time. Most of them, except clinical levels of autism spectrum disorder and psychopathy are quite common. Substance abuse, ADHD (as a result of high levels of impulsivity and impairments to working and being a student at regular jobs), and psychopathy seem to have the strongest association with criminal conduct that can send you to prison. Other mental health conditions, while also common, seem much less likely to be associated with criminal activity.

But psychosis (i.e. manic-depression and schizophrenia) tends to manifest in late adolescence or in one's 20s as your neural connections thin out in the brain transition from childhood to adulthood. It tends to emerge a little later in men than in women because their brains mature a little bit later than women. Psychosis is rare but devastating in its effect on your ability to function. Psychoses are also among the most heavily genetic mental health conditions.  Even if early mild symptoms of psychosis manifest in late high school or college, these way be manageable until a few years later. (As an aside, one thing that is almost a litmus test for schizophrenia is heavy nicotine use, which despite its other harmful health effects, provides some limited relief from schizophrenia symptoms. A very large percentage of schizophrenics who have access to nicotine, legally or illicitly, use it.) 

M.S. typically manifests at ages similar to psychosis but has a viral cause, and its neurodegenerative symptoms can mirror major mental health issues and gets worse over time since it is degenerative. Further, traumatic brain injury, major depression (apart from manic depression) and PTSD are predominantly not genetic or congenital and can manifest at any age. Finally, some forms of dementia, whose symptoms are often only unmistakable in late life, and stokes (including TIAs) can also give rise to symptoms similar to major mental health issues.

These mental and cognitive health conditions that manifest after one typically starts college probably disproportionately impact middle class inmates, since these impairments do not interfere with behavior until these key educational landmarks are reached. 

Judicial Background And Standards Of Review

As judges, empirically, non-criminal law practitioners are less deferential to trial court decisions than criminal law practitioners, and judges with mixed backgrounds or backgrounds as non-practicing lawyers (e.g. law professors) are the most deferential to trial court decisions. 

Over the years much ink has been spilled defining, explaining, and critiquing standards of review. Countless lawyers, judges, and scholars have flyspecked distinctions among questions of law, fact, and discretion in an effort to derive a coherent theory explaining when and whether appellate judges should endeavor to correct trial court error. Most of these theories have been premised on the notion that standards of appellate review, although sometimes ill-defined, are applied based on consistent legal or rational standards. Our research, however, supports those scholars who posit that standards of review are often influenced by extraneous factors not anchored in a coherent legal conception of deference. 

We observe that across a broad spectrum of cases, different panels of jurists apply standards of review in a disparate manner, influenced by their personal backgrounds. Our research explores numerous aspects of personal background, including prior professional legal experience, length of time on the trial court, gender, and political affiliation. Among these categories, we discovered that only one exhibited a statistically significant impact on the selection and application of the standard of review: the type of prior professional legal experience of panelists. Specifically, we find that the criminal or civil practice background of jurists on a reviewing panel influences ultimate outcomes but also shapes the selection of the standard of review.  

Based on our findings, we hypothesize that the collective training and experience of a panel in civil or criminal law significantly shapes their analogic reasoning, i.e., their mental model. Consequently, this background factor exerts more influence than others in determining how and when jurists defer to the trial court.

Kira L. Klatchko & Quinn A. Keefer, "Judicial Backgrounds Influence the Standard of Review," 55 U. Pac. L. Rev. 1 (2023). 

The conclusion of the paper states: 

“Civil law only” background panels are most likely overall to apply the de novo standard of review, meaning they are least likely to afford deference to trial court decisions even where they would have the option to do so when considering “dynamic issues.” 

“Criminal law only” panels are least likely overall to review for abuse of discretion, and in reviewing criminal cases are most likely to review for substantial evidence.  

“Other background” panels are, overall, most likely to review for abuse of discretion and least likely to review issues de novo; they are also least likely of all panels to reverse issues.

Footnote 71 of the paper defines these categories:

We define “criminal law only” background to mean that while in law practice, and before taking the bench, a justice practiced criminal law only and did not report any experience practicing any form of civil law. 

We define “civil law only” background to mean that while in law practice, and before taking the bench, a justice practiced civil law only and did not report any experience practicing criminal law. Civil law, for this purpose, includes all non-criminal law, including but not limited to general civil law, probate, family, and transactional law. 

We define “other background” such that it includes a mixture of both civil and criminal law practice experience and also nontraditional practice experience that cannot be fairly classified as either civil or criminal law, e.g., law professor. 

A panel classified as having a “majority” of “criminal law only” panelists will consist of at least two members who have a criminal law only background. A panel classified as having a “majority” of “civil law only” panelists will consist of at least two members who have a civil law only background.

The Policy Status Quo Is Very Powerful

Democracies strongly preserve status quo policies even when they no longer make sense and are unpopular. The paper below theorizes that policies that were previously the subject to decisive, close political fights stay that way after they are decided because people stop caring about those issues enough to change them after they are resolved.

Policy choices sometimes appear stubbornly persistent, even when they become politically unpopular or economically damaging. This paper offers the first systematic empirical evidence of how persistent policy choices are, defined as whether an electorate’s or legislature’s decisions affect whether a policy is in place decades later. 

I create a new dataset that tracks the historical record of more than 800 state policies that were the subjects of close referendums in U.S. states since 1900. In a regression discontinuity design, I estimate that passing a referendum increases the chance a policy is operative 20, 40, or even 100 years later by over 40 percentage points. 

I collect additional data on U.S. Congressional legislation and international referendums and use existing data on state legislation to document similar policy persistence for a range of institutional environments, cultures, and topics. 

I develop a theoretical model to distinguish between possible causes of persistence and present evidence that persistence arises because policies’ salience declines in the aftermath of referendums. The results indicate that many policies are persistently in place—or not—for reasons unrelated to the electorate’s current preferences.

Zach Freitas-Groff, Persistence in Policy: Evidence from Close Votes (2023).

There is some irony in this result. One of the major reasons for opposition to democracy in the transition from monarchy, and for opposition to direct democracy, in particular, was the fear that this would led to frequent, radical policy changes as the public's mood shifted from election to election. It turns out that they had nothing to be afraid of on that score.

This empirical result also bodes well for the future of abortion rights in the United States now that Dobbs has overturned Roe v. Wade.

Red states like Kansas and Ohio have backed abortion rights in state referendums (Ohio passed a state constitutional right to abortion with almost 57% of the vote yesterday.) And, this academic study shows that those referendum decisions are likely to endure.

Image from the New York Times.

Image from the Washington Post.

Notably, and surprisingly, these massive regional abortion restrictions have not greatly impacted the number of abortions carried out nationally, as people have used abortion pills and travelled to other states to obtain abortions.

What Criminal Justice Policies Influence Murder Rates?

International data and other analyses tends to find that gun control does indeed influence murder rates. 

One difficulty may be that the effectiveness of U.S. gun control law is impaired by having a border free zone that makes buying guns where they are legal and transporting them to places where they are banned or restricted a trivial matter.

Overall, I'm skeptical of the statistical methods used, which make a lot of seemingly ad hoc methodological decisions. Moody's publication history shows him to be a conservative hack. But nonetheless, it is a study that deserves mention.

This study investigates the effects of most of the major firearm and crime control policies on murder. We use two-way fixed-effects models based on state-level panel data from 1970-2018. We include a comprehensive list of relevant policy variables to control for their influence in determining the effect of each. We do a specification search using four commonly used econometric methods to estimate three models of the crime equation. A Bonferroni correction is used to control for false rejections. A robustness check using new difference-in-differences estimators confirms the results.

We find that, with the possible exception of constitutional carry laws, no firearm policy can be shown to have a significant long-run effect on murder. However, we find that the traditional policies of prison incarceration and police presence significantly reduce murder in the long run. We also find that executions have no significant long-run effect on murder. Finally, there is considerable evidence that three-strikes laws increase murder in the long run.

Carlisle Moody, "A Comprehensive Analysis of the Effect of Crime-Control Policies on Murder" (September 14, 2023). Available at SSRN: https://ssrn.com/abstract=4574358 or http://dx.doi.org/10.2139/ssrn.4574358

Why Do Cops Do Bad Things?

A new study blames insufficient training for inappropriate law enforcement actions.

What causes adverse policing outcomes, such as excessive uses of force and unnecessary arrests? 

Prevailing explanations focus on problematic officers or deficient regulations and oversight. Here, we introduce a new, overlooked perspective. 

We suggest that the cognitive demands inherent in policing can undermine officer decision-making. Unless officers are prepared for these demands, they may jump to conclusions too quickly without fully considering alternative ways of seeing a situation. This can lead to adverse policing outcomes. 

To test this perspective, we created a new training that teaches officers to more deliberately consider different ways of interpreting the situations they encounter. We evaluated this training using a randomized controlled trial with 2,070 officers from the Chicago Police Department. In a series of lab assessments, we find that treated officers were significantly more likely to consider a wider range of evidence and develop more explanations for subjects' actions. 

Critically, we also find that training affected officer performance in the field, leading to reductions in uses of force, discretionary arrests, and arrests of Black civilians. Meanwhile, officer activity levels remained unchanged, and trained officers were less likely to be injured on duty. 

Our results highlight the value of considering the cognitive aspects of policing and demonstrate the power of using behaviorally informed approaches to improve officer decision-making and policing outcomes.

Oeindrila Dube, Sandy Jo MacArthur & Anuj K. Shah, "A Cognitive View of Policing" NBER WORKING PAPER 31651 (September 2023) DOI 10.3386/w31651

A Clever Research Design Tracks Corruption In Italy

It is possible to identify lies about birth dates in public records statistically, even if you can't tell which particular birth dates are lies. This trick was used to track honesty by region in Italy, and shifts in the honest of different regions over time.

Using census data, we study false birth-date registrations in Italy, a phenomenon well known to demographers, in a setting that allows us to separate honesty from the benefits of cheating and deterrence. By comparing migrants leaving a locality with those who remain in it, we illustrate the tendency of Italians to sort themselves across geographic areas according to their honesty levels. Over time, this tendency has modified the average honesty level in each locality, with relevant consequences for the distribution across geographic areas of outcomes like human capital, productivity, earnings growth, and the quality of local politicians and government.

Massimo Anelli, Tommaso Colussi, and Andrea Ichino, "Rule Breaking, Honesty, and Migration" 66(2) Journal of Law and Economics (2023) (Supplemental Data here). This closed access paper is analyzed in the comments at Marginal Revolution and at another blog that it links to discussing the paper.

Tort Law Wasn't Very Effective In Streetcar Accidents

The effectiveness of early 20th century tort law has been greatly exaggerated relative to the historical reality.

Streetcars were the greatest American tortfeasors of the early 20th century, injuring approximately one in 331 urban Americans in 1907. This empirical study presents never-before-assembled data concerning litigation involving streetcar companies in California during the early twentieth century.

This article demonstrates the methodological folly of relying upon appellate cases to describe the world of trial-court litigation. Few cases went to trial. Plaintiffs lost about half their lawsuits. When plaintiffs did win, they won very little money. Regarding the bite taken out of the street railway company, the Superior Court was a flea.

Professor Gary Schwartz and Judge Richard Posner have presented inaccurate empirical data concerning early twentieth-century personal injury litigation. Professor Gary Schwartz was wrong to characterize tort law as generous. Likewise, Judge Richard Posner has been wrong to call tort law efficient. Like Professors Lawrence M. Friedman and Morton Horwitz, I see the amount taken from the street railway companies as quite small. However, I see no deliberate efforts to subsidize the industry.

Thomas D. Russell, Blood on the Tracks (Seattle University Law Review, Vol. 47, 2023) on SSRN.

Guns, Homicides, Suicides, and Drugs

The Big Picture

Gun control and drug policy are two areas where overwhelming empirical evidence shows that liberal policies on these issues would profoundly increase public welfare, and that conservative policies on these issues are deeply misguided and do profound harm without providing meaningful benefits to society. 

The misguided status quo policies in the United States on these issues are also among the most important reasons that U.S. life expectancies are lower than in other developed countries. While the raw numbers of deaths caused by these policies is not extremely high relative to diseases that cause deaths, because they results in deaths of much younger people on average than other leading causes of death, their impact on U.S. life expectancies is outsized.

Weak gun control laws are a root cause of about 75-80% of the homicides and more than half of all suicides in the United States. 

Insufficiently strong gun control laws cause about 30,000 more deaths per year in the United States than it would have if strong gun control laws were in place. 

These preventable deaths take a particularly strong toll on people who are more than one year old and are not yet elderly, who otherwise tend not to die of "natural causes" and are the leading cause of death for children in the United States. These deaths disproportionately kill men and racial and ethnic minorities in the United States, especially black adolescent boys, and young black men. Deaths due to weak gun control laws also disproportionately kill young white men in the South and in rural America.

This is also the reason that the United States leads the world in its number of mass shootings. And, the evidence is overwhelming that widespread gun ownership does not meaningfully mitigate mass shootings that do occur, and does not prevent mass shootings. Instead, it makes them more common.

The criminal justice system does not deter mass shootings. They continue to occur, even though it is widely known that almost every mass shooter (98% of whom are men): (1) commits suicide, (2) dies in the act as law enforcement tries to shop him or arrest him, (3) is convicted of multiple murders and remains in prison for life (or in very rare cases is executed a decade or more later), or (4) is declared insane or incompetent and is involuntarily committed and never walks free again. Almost no mass shooters escape death shortly after or maximal criminal justice system punishments. The clearance rates for these cases is almost perfect. Because of this fact, the only way to reduce the number of mass shootings is to prevent them, and stricter gun control laws are well proven to greatly reduce mass shootings.

Consideration of homicides and suicides alone fails to consider the way that the pervasive threat of armed crime triggers excessive uses of force by law enforcement and the militarization of law enforcement, sometimes resulting in unjustified law enforcement killings, in legally unjustified killings by civilians claiming to be acting in self-defense, in justified law enforcement and self-defense killings that could have been prevented if guns were less widely available, and in riots causing mass property damage, injuries, and sometimes deaths. These circumstances claim hundreds of lives each year and also lead to hundreds of law enforcement deaths of each year.

Of course, this doesn't even begin to consider the harms associated with gun involved crimes such as non-fatal shootings, extortion, robberies, burglaries, and rapes that are committed with firearms. The rates at which these aggravated crimes would be committed would be significantly reduced if strict gun control were in place, although comparative crime rate studies suggest that the reductions would not be nearly as great as the reductions in the rates of homicides.

The evidence is also overwhelming that the widespread availability of armed self-defense does not significantly prevent crimes from taking place, or mitigate the harm associated with crimes. Instead, gun ownership increases the rate at which gun owners and non-gun owners alike are victims of crimes and commit suicide. Armed self-defense and armed defense of others does succeed, at least partially, in a tiny number of cases, but the benefits of armed self-defense in the rare cases where it is used are profoundly overwhelmed by the harms that widespread gun availability facilitates, even to people who are generally law abiding when they buy firearms who purchase them in good faith solely for the purpose of defending themselves, their homes, and their families.

Widely available firearms, by making crimes more serious, also drives mass incarceration in the United States, by turning people who otherwise would have committed less serious crimes into people who commit serious violent crimes. People serving long sentences for violent crimes make up a large share of all prison inmates and would make up a significantly smaller share of prison inmates in a world with strict gun control.

Weak gun control laws in the United States and the prohibition rather than regulation of controlled substances in the United States are also a leading cause of homicides and other gun crimes in Canada, the Caribbean, and Latin America. Strong gun control laws in the United States would save tens of thousands of lives each year elsewhere in the Americas.

The facts that countries with strict gun control laws like the U.K. and Japan have healthy democratic systems, that guns and threats of violence are increasingly being used to thwart the democratic process, and the results of comparative and historical studies of the impact of armed populations on tyranny and the democratic process, all soundly demonstrate that the political theory underlying the Second Amendment is profoundly incorrect as an empirical matter.

Unequivocal evidence clearly shows that the United States, and the Western Hemisphere more generally, would be profoundly better off if the Second Amendment to the United States Constitution were repealed, and the United States then adopted strict national gun control laws along the lines of those current in place in the United Kingdom and Japan.

In short, the Second Amendment is a suicide pact.

A policy towards illegal drugs and unauthorized use of prescription drugs that deals with drugs as a public health problem, rather than treating this as primarily a criminal justice problem would also greatly reduce ever rising number of drug overdose deaths in the United States. It would also profoundly reduce organized crime and gang activity and greatly reduce property crimes committed to fund illegal drug purchases. In the year 2021, 106,699 people in the United States died of drug overdoes (mostly of opiates), and this would be profoundly reduced if the United States instead took a public health approach to the problem. Perhaps 90% of these deaths could be prevented with better drug policies. This has been convincingly demonstrated in places that have shifted fully or partially from a war on drugs criminal justice approach to a public health approach to the problem of substance abuse such as France, Switzerland, and Portugal, and in places that have legalized recreational marijuana. The illegal drug trade that U.S. controlled substance laws facilitate also fuels organized crime worldwide, often making it a powerful rival to the civilian governments of the countries where it is present.

This post addresses many, but not all of the claims above, others of which have been addressed in previous posts at this blog.

Gun Control Laws Compared

The U.S. has the most lax gun control law in the world other than Ethiopia and Yemen (Switzerland is lax, but not as lax as the U.S., Yemen and Ethiopia are also more lax than they seems as Yemen is in the middle of a civil war and Ethiopia is the midst of a lower grade military insurgency).

The U.K. and Japan have the most strict gun control laws in the world, with Japan's regulation of guns being more strict and more effectively enforced, in part because the borders of the U.K. are more open to countries with less strict gun control laws. Japan also might tightly regulates bladed weapons than the U.K. does.

Homicide Rates Compared

How does this affect homicide rates in the respective countries?

The lion's share of the difference in homicide rates between the U.S., U.K., and Japan can be attributed to gun control.

Total Homicide Rates Per 100,000 people:

* U.S.    4.7 (74% involving guns)

* U.K.    1.17 (5% involving guns, i.e. 35 gun homicides per year).

* Japan   1.02 (less than 1% involving guns, i.e. 9 gun deaths including suicides and accidents per year).

Gun Homicide Rate Per 100,000 people:

* U.S.    3.48 (about 58 times as great as the U.K.)

* U.K.:  0.06 (more than 6 times more than Japan)

* Japan: less than 0.01 (more than 348 times less than the U.S.)

The U.S. would have about 11,300 fewer gun homicides per year if it had the U.K. gun homicide rate instead of its own.

More generally (involving slightly different rates due to age adjustments and data from different years):

Non-Gun Homicide Rates Per 100,000 people:

* U.S. 1.22 (21% more than Japan and 10% more than the U.K.)

* U.K. 1.11 (10% more than Japan)

* Japan 1.01

The U.S. would have about 363 fewer non-gun homicides per year (about one less homicide per day, nationwide) if it had the U.K. non-gun homicide rate rather than its own non-gun homicide rate.

Some of the difference in the non-gun homicide rate between the U.S. and the U.K. and Japan (but probably less than 10%), may reflect the inferior health care system of the U.S., which unlike the U.K. and Japan is not universal causing people who need emergency medical care to avoid hospitals.

Some of the difference in the non-gun homicide rate between the U.K. and Japan (but probably not more than 10%) may reflect stricter controls on bladed weapons in Japan than in the U.K.

The fact that the population of Japan is older than the populations of the U.K. and the U.S. may account for some of the difference.

Another factor may  be greater economic inequality and higher poverty rates in the U.K. Economic inequality and poverty are higher in the U.S. than either the U.K. or Japan.

But, all other factors explaining the differences in homicide rates between these countries pale in comparison to gun control.

Mass Shootings Compared

Unsurprisingly, the U.S. also (almost) leads the world in mass shootings:

Mass shootings are a fairly modest share of all murders committed with guns in the United States (about 2.6%):

The Gun Violence Archive, an online database of gun violence incidents in the U.S., defines mass shootings as incidents in which four or more people are shot, even if no one was killed (again excluding the shooters). Using this definition, 513 people died in these incidents in 2020.

But mass shootings have a disproportionate impact on our public sense of security, because they are comparatively random and unpredictable.

There are no mass shooting in Japan, which has a population of about 125 million people.

The U.S. has:

about twenty-five times as many mass shootings per capita as the U.K., 

about twelve times as many as Italy, 

about eight times as many as Australia, 

about five times as many as Germany, 

about three times as many as Canada, 

about two and three-quarters times as many as Austria, 

about two and two-thirds times as many as the Netherlands, 

about two and a half times as many as France, 

one and three-quarters times as many as Belgium, 

about one point six times as many as the Czech Republic, and 

33% more the Switzerland. 

Finland actually has 80% more mass shootings per capita than the U.S., in part due to random variation in a very small number over a twenty-one year period in a country with a small population (and probably involves fewer victims per capita than the U.S.).

Per capita rates are also problematic and not as statistically significant in countries with only one mass shooting during a twenty-one year period when the country has a small population, where random chance at a given rate and "rounding error" type issues come into play.

Gun Suicides

Stricter gun control would also greatly reduce firearm suicides without corresponding increases in suicides from other causes. 

According to the Pew Center:

In 2020, 54% of all gun-related deaths in the U.S. were suicides (24,292), while 43% were murders (19,384), according to the CDC. The remaining gun deaths that year were unintentional (535), involved law enforcement (611) or had undetermined circumstances (400). . . .

Nearly eight-in-ten (79%) U.S. murders in 2020 – 19,384 out of 24,576 – involved a firearm. That marked the highest percentage since at least 1968, the earliest year for which the CDC has online records. A little over half (53%) of all suicides in 2020 – 24,292 out of 45,979 – involved a gun, a percentage that has generally remained stable in recent years. 

As noted by the Kaiser Family Foundation:

Variation in state-level suicide rates is largely driven by rates of suicide by firearm.  

Suicides involving firearms vary from the lowest rate of 1.8 per 100,000 in New Jersey and Massachusetts to a high of 20.9 per 100,000 in Wyoming, representing an absolute difference of 19.1. 

In contrast, the rate of suicide by other means is more stable across states, ranging from a low of 4.6 in Mississippi to a high of 11.4 in South Dakota, representing an absolute difference of 6.8. . . .

More than twice as many suicides by firearm occur in states with the fewest gun laws, relative to states with the most laws. . . .

Taking a look at suicide deaths starting from the date of a handgun purchase and comparing them to people who did not purchase handguns, another study found that people who purchased handguns were more likely to die from suicide by firearm than those who did not–with men 8 times more likely and women 35 times more likely compared to non-owners.

Non-firearm suicides rates are relatively stable across states suggesting that other types of suicides are not more likely in areas where guns are harder to access.

Similarly:

A study by the Harvard School of Public Health of all 50 U.S. states reveals a powerful link between rates of firearm ownership and suicides. Based on a survey of American households conducted in 2002, HSPH Assistant Professor of Health Policy and Management Matthew Miller, Research Associate Deborah Azrael, and colleagues at the School’s Injury Control Research Center (ICRC), found that in states where guns were prevalent—as in Wyoming, where 63 percent of households reported owning guns—rates of suicide were higher. The inverse was also true: where gun ownership was less common, suicide rates were also lower. 

Spillover Effects

Stricter gun control laws in the U.S. would also significantly reduce homicide rates in Canada, the Caribbean, and Latin America, where a very significant share of all gun homicides are committed with guns smuggled into those countries illegally from the United States, where guns are easy to obtain. 

For example, according to a July 2022 article, in "Ontario, Canada's most populous province . . . when handguns involved in crimes were traced in 2021, they were overwhelmingly - 85% of the time - found to have come from the United States. . . . 70% of all traced guns used in crimes in Ontario came from the United States, while so far this year the U.S. share has risen to 73%, according to the data from the Ontario police's Firearms Analysis and Tracing Enforcement (FATE) program."

A significant share of homicides in Latin America are also attributable to the trade in illegal drugs involving drug cartels and other forms of organized crime from Latin American to meet U.S. demand, which would be greatly reduced if those drugs were legalized but regulated in the United States.

Taking a global view, the six countries with the highest age-adjusted rates of firearm homicides are:

1. El Salvador
2. Venezuela
3. Guatemala
4. Colombia
5. Honduras
6. Brazil

Research has found high levels of homicides in these countries are associated with drug cartels, the illegal trade in firearms from the US, and firearms flowing to civilians after conflicts end, as summarized in the Global Burden of Disease study.

From here.

Thus, stricter gun control in the United States and more enlightened controlled substances laws in the United States would greatly reduce homicides almost everywhere in the Americas.

As an aside, the linked Global Burden of Disease study concludes that the drug trade and smuggled firearms from the U.S. have a much smaller impact on suicide rates in Latin America than these factors do on homicide rates there. 

There is good reason to think that this is also true in Canada.