Is there really an "Epidemic" of "Traffic Violence?," Part I
Background and USA Data
Note: This post is long in length and contains several charts making it difficult to view in an email client or on a mobile device.
It’s best viewed on a large screen using a web browser.
Introduction: Huffery and Puffery
“Now you, too, can out-double-talk your accountant, confuse your political opponent, prove that your product has built-in goodness,” says the front cover of the first edition of the book, How to Lie with Statistics.
First published in 1954, Darrel Huff’s best selling book is not only accessible to the layperson but is a humorous guide to the misuse of statistics, demonstrating how easily statistical data can be manipulated to mislead the public. Its straightforward explanations and entertaining illustrations by Irving Geis have made it a staple in educational settings, particularly in courses focused on critical thinking and data literacy.
How to Lie with Statistics likely deserves credit for memeing the common phrase “correlation does not imply causation,” into existence but also contains a series of other lessons. These lessons in the book include highlighting issues with bias in sampling and the different types of averages and how they can be useless or misleading. Then there’s of course intentionally omitting data to distort the outcome and a chapter dedicated to common tricks used when presenting graphs, charts, and pictograms.
It would later be discovered that Huff was, let’s just say, getting a bit high on his own supply. Perhaps when he wrote in the book, “Statistics can be used to deceive. Data can be consciously or unconsciously biased — by the statistician or the source,” he was subtly telling his readers something else.
Decades after the original publishing of the book, researcher Alex Reinhart revealed something sinister: in the 1960s, Huff was hired by the tobacco industry executives and attorneys to defend against growing evidence linking smoking to serious health risks. Using the same techniques he detailed in How to Lie with Statistics, Huff attempted to cast doubt on the growing scientific consensus by highlighting ambiguities and inconsistencies in their data.
I. The Foundations for Semantic Fuckery
“Concept creep" was introduced by Australian psychologist Nick Haslam in his 2016 paper titled "Concept Creep: Psychology's Expanding Concepts of Harm and Pathology." Haslam observed that concepts related to harm, such as "abuse," "trauma," and "mental disorder," were broadening in scope, leading to changes in how these terms are understood and applied in psychological and social contexts.
Haslam in his paper identified both “vertical” and “horizontal” forms of concept creep.
Vertical concept creep occurs when the severity or intensity threshold of a concept is lowered. For example, the term "trauma" originally referred to severe, life-threatening events but has expanded to include less severe experiences, such as everyday stressors to include “microaggressions.” Trauma stretches "vertically" to include milder forms of the original phenomenon. Horizontal concept creep happens when a concept extends into new domains or contexts that were not previously associated with it. For instance, "bullying" was once used specifically for peer aggression among children but has expanded to include behaviors in workplaces and online interactions. Bullying thus stretches "horizontally" to cover different situations or types of behavior not included in the original definition.
Other terms to experience a change in meaning via concept creep are violence and epidemic. Violence has long meant the use of physical force, be that armed with a weapon, a tool, or one’s body parts against another. Violence in a post concept creep world refers to the use of language viewed as harsh or offensive by a so-called victim from a so-called oppressor. But now, at least in some circles, virtually anything and everything can be “violence.” Adjectives placed in front of the word violence further distort the original concept with “structural violence” (defined here in a circular definition as “a form of violence wherein social structures or social institutions harm people by preventing them from meeting their basic needs.”) and “traffic violence,” a term that’s not used to refer to motorists intentionally using their vehicles as weapons to physically hurt others but instead as the overall term by transportation, urban planning, and Cluster B(ike) activists for what non-activists and the rest of us typically call traffic crashes, collisions, or accidents.
Authors Jonathon Haidt and
popularized Haslam’s term outside academia in their September 2015 article in The Atlantic called The Coddling of the American Mind, later expanded into a best selling book, and now a full-length documentary here on Substack.A great deal of concepts out of The Coddling of the American Mind, in any of its forms, are arguably directly applicable to modern bicycling advocacy such as the demand for “feeling safe” while cycling even if that “protected” bicycle lane (arguably a form of a safe space) creates magnified risks of the bicyclist getting into collision traffic violence event with a motor vehicle. That’s all for another piece though.
Streetsblog, whose writers began using the term “traffic violence” over a decade ago describe it as, “the epidemic of death and serious injuries that has raged on our roadways since the advent of the automobile.”
They go further with what can be described as an utterly Critical approach, akin to the “Motornormativity” concept (mentioned by and Bret Weinstein on a recent Dark Horse Podcast) to the overall injuries and fatalities on the nation’s public roadway network.
They continue:
Violent systems do not come into existence by accident, and they certainly do not persist by accident. America has collectively chosen, every day, to perpetuate a violent transportation landscape that results in roughly 38,000 preventable deaths every single year.
Violence that occurs in the traffic realm is, by definition, traffic violence. A car crash is always a car crash, even if the driver is ultimately found not guilty of criminal negligence. And when we feel a sense of unease saying these words, it is because we have been raised in a culture that has systematically and deliberately worked to shift the blame for that violence onto victims, down to the subtlest workings of the language in the news reports we read.
…..
The ideas of "empathy" and "intention" are common to most criticisms of the phrase “traffic violence.” When we hear the story of the long-serving bus driver who unintentionally killed a jogger, virtually anyone who has ever piloted an automobile — and if you live in America, you probably have — does feel compelled, at least for a moment, to put themselves in that driver’s seat and imagine how it would feel to realize that a human being has sprinted into your blind spot, and in a single, sickening moment, you have ended his life.
Word salad aside, what is going to be discussed here is whether there is an “epidemic” of “traffic violence,” or not, at least in the United States.
Future pieces will attempt to address these and hopefully eventually, with good data, further analysis can be done on other countries.
I. It’s All Greek to Me
“There's an epidemic killing millions of people across the world — and no, we're not talking about the coronavirus. We're talking about the pandemic levels of traffic violence on our streets, which claim the lives of 3,698 people worldwide every day1.”
The "epi" in "epidemic" comes from the Greek prefix "ἐπί" (epi), which means "upon" or "above." Therefore, "epidemic" literally means (or meant) "upon the people," signifying a disease or condition that spreads widely and affects many individuals within a community, population, or region.
An disease outbreak is considered an "epidemic" when the number of cases increases above what is normally expected in a given population, geographical area, or season.
The key factors in defining an epidemic include:
Increase in Case Numbers: There is a significant rise in the number of cases compared to the baseline level of the disease in the area.
Population or Area Affected: The increase occurs within a specific population or geographical region.
Time Period: The increase happens over a particular period, usually a short one, indicating a sudden spike rather than a gradual rise.
Compare the etymological exercise above to a related term (pandemic), the "pan" in "pandemic" comes from the Greek prefix "πάν" (pan), which means "all" or "every." For the rest of this piece, only the term epidemic will be discussed since the main difference between the two terms relies on the population or area for a pandemic extending to the global level.
Unfortunately, epidemiology, despite being science, er often times The Science™, doesn’t maintain an adherence to harder sciences or strict definitions.
A Columbia School of Public Health informational page admits:
Epidemics can refer to a disease or other specific health-related behavior (e.g., smoking) with rates that are clearly above the expected occurrence in a community or region.
The paper entitled, “When is an Epidemic an Epidemic?” admits:
The word ‘epidemic’ is an emotionally charged term. It means different things to different people, and professionals using the term may have an intended meaning quite different from the public's perception of the word.
An NPR piece even admits referring to something as an epidemic, “creates a sense of emergency or crisis!”
This expansion into other specific heath-related behavior is the first iteration of the term’s horizontal concept creep. The second, and more severe case of horizontal concept creep is when someone, often with a political, social, or activism-driven agenda.
A number of phenomena are labeled epidemics and they span across the whole spectrum of the expansion of the term.
They include but are certainly not limited to:
Opioid overdoses (largely caused by Big Pharma & Public Health, oops.)
Obesity (in this example the freaking World Health Organization confuses epidemic and pandemic!)
Eating disorders (largely caused by Public Heath’s “Science” in support of Fiat Foods)
Systemic Racism (to the point where over 1,000 Public Health clowns called for an exception for COVID lockdowns so they could protest their pet cause)
Firearms fatalities (termed “gun violence” which often misleads by including suicides)
And, of course, “traffic violence.”
Public Health activists, having already done a great deal of misleading in the “gun violence space” by making it entirely political2, are also starting to analogize driving motor vehicles to firearms3 and to tobacco usage.
Anti-car activists and anti-gun activists both make identical claims invoking what Nassim Taleb calls pedophrasty about what demon item is the greatest killer of children. One says firearms, while another says “traffic violence.” The bullshit “Motonormativity” paper tried to argue morally that driving motor vehicles were akin to tobacco use, and of course Public Health are no angels in this domain either. Streetsblog’s Kea Wilson even wrote an article wondering, “Why Traffic Violence Victims Can’t Easily Sue Automakers That Advertise Cars That Kill,” citing lawsuits against firearm makers from victims of gun violence.
Well okay, let’s see some examples of people claiming “traffic violence” or less inflamatory terms are considered to be an “epidemic.”
II. Identifying the Contagion
Recall Walker, Tapp, and Davis’ infamous motonormativity paper. In the introduction to the paper, they insist “easy hypermobility” is the cause of at least three “public health epidemics” in the United Kingdom.
Here in the United Kingdom, like in many societies around the world, we are in the midst of environmental degradation and no fewer than three parallel health epidemics thanks to the easy hypermobility (Adams, 2001) afforded by private motor vehicles. We have an epidemic of collisions, with 1752 deaths and 25,945 serious injuries in 2019, the last year before the Covid pandemic (DfT, 2020); we have an epidemic of physical inactivity – responsible for 22-23% of coronary heart disease, 16-17% of colon cancer, 15% of diabetes, 12-13% of strokes and 11% of breast cancer (World Health Organisation, 2002) – despite 24% of car trips being under 2 miles and so mostly amenable to walking or cycling (DfT, 2022); and we have an epidemic of pollution, with vehicle exhaust fumes causing cancer (Raaschou-Nielsen et al., 2013), heart disease (Hoek 54 et al., 2013) and diabetes (Rao et al., 2015) at such levels that estimates have put the UK air pollution death toll at 40,000 per year (Royal College of Physicians, 2016). Even a future switch to electric vehicles would address only one of these three epidemics (Walker & Bösehans, 2016). It is clear we must acknowledge a simple fact: transport issues are not just environmental issues: they are also inherently public health issues.
With each of these so-called epidemics, notice the use of figures that stand on their own and are not compared to past figures to determine their change but are also not compared to the UK’s population? Also notice the vague use of percentages. With each of these so-called epidemics, Walker and his associates use figures that stand out on their own and are not compared across time or take into account UK population. Their use of percentages may properly reflect the data somewhere else but also may just be useless noise. Their figures may sound scary but these authors are not using anything remotely close to the original definition of an epidemic. Instead the closest they get are other specific heath-related behaviors but assert their activism-driven motives later in the paper.
Last fall, in a press release from LA City Council District 4’s Nithya Ramen not only used the term “traffic violence” but spit out a bunch of scary sounding numbers and predictably, calling the issue an epidemic:
“In 2023 alone, 134 people walking have been killed by people driving in the City of Los Angeles and 427 people have been severely injured,”
and
“We have seen an alarming increase in pedestrian fatalities not just across the city, but across the country, with over 7,500 pedestrians estimated to have been killed by traffic violence in 2022 – the highest number since 1981.”
and
“These numbers are nothing short of an epidemic, and it is critical that we employ whatever measures we can to prioritize pedestrian safety. Walking in Los Angeles should not have to come with a safety warning.”
The press release also quotes Damian Kevitt, the Executive Director of an activist group called Streets are for Everyone who invoked the use of the term public health crisis with:
“We have a worsening public health crisis on our roads due to traffic violence with pedestrians being hit and killed at far higher numbers than any other group.”
These sentences surely sound scary and crisis-invoking but they fall for many of the same statistical trucks used in the Motornormativity paper’s authors.
Then there’s the Los Angeles Times, Enemy of Regular Californians and no strangers to piss poor “reporting” on traffic issues wrote this in a January 2023 article.
According to Los Angeles Police Department data, 312 people were killed in traffic collisions last year, 5% more than in 2021 and a 29% increase over 2020.
L.A.’s streets remain particularly deadly for pedestrians and bicyclists, with 159 people killed in collisions involving pedestrians and motorists, a 19% rise compared with 2021, and 20 people killed in collisions involving bicyclists and motorists, an 11% rise.
They even provided a handy graph - with no adjustment to city population of course and then referred to what was going on in LA as “traffic carnage" that “outpaced national trends.”
Now, just what are these national trends?
This is all LA Times give us:
The National Highway Traffic Safety Administration reported that fatalities from traffic crashes were essentially flat nationwide in the first nine months of 2022 compared with 2021. Fatalities among cyclists and pedestrians rose nationwide last year but were lower than the percentage rise seen in L.A.
As it turns out, there’s an entire book dedicated to calling (at least some of) the traffic fatalities an “epidemic,” and that’s ex-Streetsblog writer Angie “How Our Transport System is Biased Against Women” Schmitt’s 2020 book, “Right of Way: Race, Class, and the Silent Epidemic of Pedestrian Deaths in America.”
Will we find the answer in there?
Despite the use of the term epidemic in the title of the book and as part of her introductory chapter ( called “Introduction: Outline of an Epidemic”), Schmitt spends virtually no time actually making the case for whether these deaths are considered an epidemic.
In fact, in all 220 pages of her book, the use of epidemic is mentioned a total of 13 times: one is the front cover, another the inside title page, then one mention in the table of contents, then on the first page of said introductory chapter, seven are located on the header of the pages of the introductory chapter, along with one in the book’s index.
The actual copy of her book contains just two usages of the term, and one is that’s her quoting a woman, Amy Cohen, a resident of Park Slope who lost her young son after he’d been struck by a car chasing his soccer ball by a motorist who admitted to speeding to make time.
Cohen said:
“I had no idea this was a preventable epidemic. That [it] wasn’t just a freak accident .. . that our family was marked for this tragedy. That it could have been prevented.”
What happened to Cohen’s son is, of course, horrible, and no parent deserves to have to bury their child, but this is an anecdote and provides no basis for labeling what happened as part of an epidemic.
The other case Schmitt covers is the death of Ignacio Duarte-Rodriguez, a senior citizen who was hit while trying to cross a busy arterial in Phoenix. Schmitt tells her readers his death, which she never mentioned was a hit and run, along with two pedestrians deaths on that same road and eleven others in “the greater Phoenix area,” were as she put it, “are part of a mounting epidemic.” Schmitt made a, predicable in this author’s eyes, statement wondering too, “had eleven people been shot by a mass shooter in Phoenix, it would have made the national news.”
Schmitt then gives some figures, and much like the LA Times article, a bit of context but not enough to paint the whole picture.
In 2018, more pedestrians were killed in the United States than at any point in a generation. That year—the most recent for which we have official data—6,283 pedestrians died, a number not seen since the mid-1990s. (An additional 1,500 were killed in driveways or parking lots and other locations that are considered private property and so are not counted as “traffic fatalities.”)
Activists often like to focus alone on number of fatalities, even though their definition of “traffic violence” (sometimes) includes serious injuries.
They’ll blabber, such as Wilson did above as even in terms of deaths per day (“Car domination is directly responsible for at least one in every 34 deaths around the world”) or such as below, that "drivers4” kill 100 others a day in the US, or in other words, 36,500 people.
For the purpose of this piece, let’s just run with that assertion and give these activists their due: it’s the “drivers” who are responsible for all the killings. (Never mind these killer drivers sometimes kill just themselves.)
Let’s try to figure out whether the claim this is an epidemic or not is true but we want to stick closely to the original definition of the term.
Later in the paper, “When is an Epidemic an Epidemic?”, the authors ask for more precision when using the term:
Furthermore, descriptive words should be used to better define the epidemic. The nature of the spread, number of cases, case-fatality, and area affected should be included. Thus, the disease should be described as contagious or non-contagious, with mild, moderate or severe clinical manifestations. The size of the epidemic could be described as small, medium or large, depending on the percentage of the population affected. Finally, the distribution of the epidemic should be classified as widespread or localized.
The analysis in the rest of this post will attempt to keep those points in mind.
III. Setting the Curve
In a developed country such as the United States, it’s extremely difficult to hide a death. That’s especially the case when the death occurs when someone is traveling from one place to another on the public roadway network. Emergency services are almost always eventually called to the scene including medical services to take the victim to hospital for treatment or to a morgue should the worst already happened. Law enforcement officers trained in accident investigation play real life CSI by trying to figure out what happened.
Ultimately, a crash investigation report is filed to local or state databases (see How To Lie With Statistics: Vulnerable Road User Edition which covers California’s database) and then to a national database called the Fatal Accident Reporting System, otherwise known more commonly as FARS. FARS is maintained by the National Highway Traffic Safety Administration (NHTSA), an agency of the US Department of Transportation (USDOT).
To say that FARS data are immensely detailed is an understatement. A good general summary of what someone can find in FARS can be found at Arizona Bike Law , and for bicycling fatalities specifically, Paul Schimek, a prominent critic of “protected bike lanes” and the flawed studies that claim to support the safety claims made to support them, digs deeply into FARs bicycle fatalities5.
FARS has limitations, for example, if someone dies from a traffic crash over a month after the event occurred, it will not be counted as a record in FARS. Bicycle dooring fatalities also do not meet the criteria due to a definition issue of the parked car not being considered “in transport.” FARS does not include traffic fatalities that occur on private property such as private subdivisions or non-highway areas such parking lots. That was something Angie Schmitt mentioned too. This increase in number of deaths is where the National Safety Council (NSC), a non-profit's data comes in to play.
The NSC try their best to add that FARS doesn’t count by looking at death certificates. As a result, NSC’s figures are typically higher than FARS data. A caveat is that NSC don’t have nearly as detailed data as FARS when it comes to crash and user specifics meaning researchers such as Paul Schimeck cannot drill into their data as he does with FARS.
Both FARS and NSC data are easily accessible to download as raw tables and often in the form of easy to read charts and we’ll use both here to try to answer the epidemic question.
Fatality data from both go back at least a century as does vehicle miles traveled estimates and US population. This allows for an honest look at what’s going on with traffic-related fatalities in the same way
looked at over a century of homicide data (in units of deaths per 100k people) when he debunked the claim that there is a “gun deaths epidemic.”FARS data includes traffic fatality counts going back to 1899 - that was when the streets looked like this:
The US Census was also well-established so we know how many people lived at the time, and how the population changed between then and now.
In 1921, as demand for both automobiles and roadway networks to accommodate them grew, so did the government and society’s interest in increasing roadway traffic safety. (Related reading: "My Bike? Or My 2-ton Land Missile? and Bob Shanteau's "The Marginalization of Bicyclists").
This was all generations before the Roadside Design Guide, headrests, tubeless tires, seat belts, child car sears, anti-lock brakes, air bags, cultural shaming of drink driving, and bans on protruding hood ornaments. Then, in the 21st Century came backup cameras, blind spot detectors on wing mirrors, adaptive cruise control, automatic braking, lane departure assist, and of course advancements with the goal of human-free autonomous driving.
It was also in 1921 when the first estimates for vehicle miles traveled were compiled grown steadily with a few dips possibly due to events such as WWII rationing, and energy crisis or two, economic downturns, and a certain event four years ago.
They provide all these data, absolute fatalities, fatalities per 100k, and million VMT in an Excel spreadsheet.
Feel free to download it and plot a chart with the years for the X-Axis and for the Y-Axis: Total Fatalities, Fatality Rate per 100k People, and Fatality Rate per 100 Million VMT.
Either plot these separately or on the same graph, if doing the latter enable a secondary Y-axis.
Set the Y-Axis to zero6!
The intent is to establish a set of baselines similar to what
did in “The Gun Homicide Epidemic Isn’t.”Availability bias, be damned, Streetsblog started in the late 00’s with an explicit anti-motoring agenda, and their claims of “traffic violence” being an “epidemic” began to pick up steam in the 2010s.
Rinsing and repeating using NSC data, which as a reminder fatalities only go back to 1913 but they typically record more deaths per year than FARS.
So okay, someone may say, “both absolute traffic fatalities and fatalities per 100k people are increasing.” That may be so and makes for some of the “scare” numbers presented by media, activists and the like, but fatalities for 100k people are almost at a record low which could be used to argue the opposite.
What gives? How does one make sense of all of this in answering whether traffic fatalities should be considered an epidemic or not?
Streetsblog apparently want to insist the epidemic has been occurring since the advent of the automobile (they also include “serious injuries” yet recall these data are more difficult to get ahold of the further one goes back) but that’s a bit ridiculous and gives into their anti-car bias but a century plus time period is way too long for an epidemic. If they wanted to argue the first several years, if not decades, of automobiles brought on an epidemic, that could slide, but trends change as more years come by. Recall that actual disease or disease adjacent epidemics require three components: a significant increase in case numbers over some baseline, an area or population affected, and a time period for this significant increase in cases to occur (often relatively short.)
Let’s look at some basic statistics with a basic review of why such statistics are useful.
The mean (often simply called average) gives us a central value so to speak of where all the data points tend to cluster. If the cluster is tight, then the standard deviation, the measure of the spread is also tight, and vice versa if the data is scattered. The mean number of traffic deaths for NSC and FARs respectively were 40,388 and 37,985 deaths with standard deviations of 8,473 and 8,895 deaths. Traffic deaths per 100k people were 20.30 and 19.22 with standard deviations of 5.14 and 19.22. Mean traffic deaths per 100 million VMT were 6.08 and 6.19 with large standard deviations of 5.32 and 5.56.
Now, data can be skewed or inconsistent across such a large span of time so it helps to find the median and some percentiles. Percentiles help in understanding the distribution of the data by showing a rank. In a basic statistics class or analysis, people tend to calculate the 50th Percentile all the time without ever knowing it, and that’s because it’s also known as the median value which indicates that 50% of values will fall above and the other 50% will fall below that value. We’ll go as far low as the 25th percentile and as far high as the 95th percentile. The rest are shown in the table below along with the statistics already typed out above.
Depending on the disease, the 90th or 95th percentile is used in part to define whether a spread’s rate is significant enough (again over typically a short time period and over a specified area or population) to be considered an epidemic.
The tables below show the last five years of both NSC and FARS data for all three metrics (total fatalities, fatality rate per 100k people, and fatality rate per 100M VMT). The other columns show whether these values exceeded the mean, median, 75th percentile and the two epidemic threshold percentiles, the 90th and 95th.
IV. In Closing
“The fact is that, despite its mathematical base, statistics is as much an art as it is a science. A great many manipulations and even distortions are possible within the bounds of propriety. Often the statistician must choose among methods, a subjective process, and find the one that he will use to represent the facts.”
-Darrel Huff
Loud activist Progressive organizations such as Streetsblog, Transportation Alternatives, Families for Safe Streets, bicycling organizations such as CalBike, politicians, and journalists scream the inflated terms “traffic violence,” and “epidemic.” Yet at the same time they demand drastic changes to the way people live and move around with programs such as Vision Zero, road diets, and “reconfiguring” the roads so “people on bicycles” can “feel safe” (even if that includes manufactured conflicts and door zones). Some even call this “mobility justice.”
While this is an ultra low level analysis, void of fancy statistical models, and likely other caveats. The argument for an epidemic given these constraints in terms of traffic fatalities in the US just doesn’t seem to be there.
Looking at things on a different time scale, zooming in on cities or states, adding serious injuries, and dividing the road users up in to motor vehicle passengers, motorcyclists, bicyclists, and pedestrians are all things that need to be done to further tease out this question and will be the subject of upcoming pieces. There’s also quantifying the medical expenses of those involved, job and productivity losses, and property damage costs that result from both fatal and non-fatal collisions.
Comments are open for critiques or suggestions on how to improve the analysis done here.
For anyone who didn’t do the math already, 3,698 people times 365 days equals 1,349,770 traffic deaths (or approx. one San Diego) worldwide per year.
Kea Wilson didn’t cite her source for this but assuming it’s accurate, that puts traffic deaths at “best” at number 11 on this WHO chart (also assuming it too is accurate) for leading causes of deaths globally. (She wrote the article in early 2020 and this WHO chart shows 2019 deaths, so they’re probably close enough for this rough comparison.)
“As Bordua, Cowan, and Southwick observed, a prejudice against gun ownership pervades the public health field. Deborah Prothrow-Stith, dean of the Harvard School of Public Health, nicely summarizes the typical attitude of her colleagues in a recent book. "My own view on gun control is simple," she writes. "I hate guns and cannot imagine why anybody would want to own one. If I had my way, guns for sport would be registered, and all other guns would be banned." Opposition to gun ownership is also the official position of the U.S. Public Health Service, the CDC's parent agency. Since 1979, its goal has been "to reduce the number of handguns in private ownership," starting with a 25 percent reduction by the turn of the century.”
In terms of bicyclist fatalities, there is a crash type that doesn’t make into FARS, and those are doorings.
See
‘s comment about this here, or below.“Before we get into looking at this graph, I want to point out something. I set the Y intercept at zero, even though I only have data points down to around 4 homicides per 100,000. It wastes white space on the graph, but showing the origin prevents you from being misled about the data. That trick — moving the origin around to confuse the reader about the data — is tremendously popular with modern media outlets, particularly cable news. Fox News gets rightfully trashed for this routinely, but CNN and other sources do it too. Watch closely for it, and when you see it, be especially critical of any commentary that accompanies the doctored graph. People who doctor graphs are usually trying to sell you something.”