Is Burning Man a Hotbed for Suicide? Not Really.

I don’t go to burns, but my brother and sister in law are pretty heavily baked into the local southeastern burner culture. And since burners talk about burns about as much as Crossfit people talk about Crossfit, I get to hear a lot of chatter about the subject. So when this article by hit the social media scene about seven Burning Man employees committing suicide, it caused quite a stir. Since then, the story has been picked up by liberal rag and by conservative rag Daily Mail, both of which make the case that the Burning Man work environment is innately evil, because that’s how Big Media rolls. Find the freakouery and feed it.

So I did some math, because that’s basically my hobby, and I think it’s largely just a bunch of handwaving freakoutery. In short, it seems to me that Burning Man employees aren’t killing themselves at a rate any higher than the rest of the country is. Follow along, and I’ll try to explain why.

Explaining Burns to Muggles

Burns are neat. I like the idea. Functionally speaking, a “burn” is when a bunch of like-minded hippies (or similar) go camping and share art, which might include sculptures, music, or performance. It goes for several days, then they burn a big effigy. Pretty straight forward. They have special social rules that burners are required to follow, which generally fall under the umbrella of collectivism, kindness, and a strict ban on commerce. There are “regional burns” across the world, which vary in size. My brother and sister in law have worked as camp organizers for a local one. Most states have these, some sanctioned, some not. But the Mecca of burns, which is going on right now in Nevada, is simply called “Burning Man,” and boy is it big.

Burning Man got started in the mid-1980s when some San Francisco folks started burning effigies on the beach, and then merged with a similar event in the early nineties in Nevada, which was a kind of a Dadaist performance art gig in Black Rock Lake, a pretty inhospitable place north of Reno Nevada. It started to grow through the 90s, eventually reaching the point where it was getting dangerous without some kind of voluntary government infrastructure. They now have streets for emergency vehicular access, safety standards, speed limits, and a variety of safety related bans on dangerous activity. As if a bunch of anarchists cooked up their own government. Intriguing stuff. It also leads to interesting permit disputes with the actual government, such as Choco-Taco-Gate, a grand bru-ha-ha between Burning Man and the US Bureau of Land Management over the necessary amenities for government employees working the festival. Choco-Taco-Gate was eventually resolved at a political level, rumored to be fueled by the economic impact of losing the nation’s largest commerce free zone. Irony thy bite be palpable.

The site is known as Black Rock City, and for one week out of the year it houses 70,000 very like-minded, very free people.

And to make it work, it takes staff. There’s an LLC, a board of directors, a host of contract employees, and around 2,000 volunteers, all necessary to pull the thing off. But they’re not just punching clock, they’re ideologically aligned with the thing, so when stories start getting spread about suicides among the staff, the burner culture at large gets very anxious.

Examining the Article (and Salon, and Daily Mail)

The article in question isn’t particularly deep, but it touches a deep nerve in the burner community. It opens like this:

Nevada’s premier desert gathering is starting to take some heat after multiple reports of suicide are slowly beginning to surface. Between the years of 2009 and 2015, seven Burning Man employees have died by suicide.

When I read that, my data alarm bells started ringing. We’ve examined this trick before, when Vox or Fox does it. Why cut the sample period off at 2015? It’s currently 2018. Liberal outlets do this with their gun articles, and conservative outlets do it with their global warming articles — arbitrarily establishing start and end dates to skew the data.

They move on with their thesis:

When the reports first surfaced, we all were devastated and demanded answers. It’s been a pretty slow process but more information surrounding the deaths of these current or former employees are now being released. The deaths are directly being attributed to unfair pay, mistreatment, stressful and extremely dangerous work conditions.
Caleb Schaber, a former photojournalist in Iraq and Afghanistan is known for his organized protest in San Francisco outside of the company’s headquarters. Caleb was a full-time employee who claims the company flat out didn’t take care of their employees and wanted to fight for worker’s right.
Here’s what he had to say:
“They don’t help out the workers that are injured, quite often, and they just try to get them to work for the most by giving them the least and then discard them. They seem to feel that it’s OK to exploit workers like they’re some kind of resource that’s just there to take and not help out. They’re a multi-million dollar corporation that has franchises, and they’re not taking care of their workers.”
Two years after the protest, Schaber was found dead by suicide. According to friends and former co-workers, they truly believe the way he was treated after the protest contributed to his death.
Here’s what they had to say:
“He just wanted to be paid a fair day’s wage, and he wanted the crews that he worked with to be paid the same,’ said a staffer who declined to give their name. ‘He wanted it to resemble a community and a job at the same time.”

Suicide is a big problem across the board in the USA. Any suicide is bad. Seven of them are seven times worse than one of them. But seven of them in eight years is less than one a year, whereas the American Foundation for Suicide Prevention clocks the nation in at 123 suicides per day. This brings up the question of rates. If Burning Man were simply a representative sample of the US population, how many suicides would we expect them to have over an eight-year period?

This is a complicated mathematical question, which requires some assumptions, but I took a stab at it and the results of my analysis surprised me. Skip ahead to my analysis if you like, but first let’s grab a tidbit from the article, from which the Daily Mail article is basically derivative.

Salon found that in the seven years between 2009 and 2015, there were seven DPW worker suicides in the department.
That number is statistically significant enough to be alarming, according to Dr. Sally Spencer-Thomas, a psychologist and the lead of the Workplace Task Force for the National Action Alliance for Suicide Prevention. “To give you a benchmark, in a community of 1,000 people we would expect one suicide death in one decade,” she explained. Spencer-Thomas noted that the construction industry in the U.S. does have an elevated suicide rate.

This is abject bullshit, and anybody with a calculator can check it. The crude suicide rate in the United States is about 22 per 100,000, which works out to a 0.00022 chance of any given person committing suicide per year. In a community of 1,000, that’s an expected value of 0.22 suicides per year, and an expected value of 2.2 suicides per decade, not 1. Her statement is closer when looking at age-adjusted rates, which are 13.26 per 100,000, but it’s not appropriate to use age adjusted rates to compare different standard populations. To do it properly, you have to look at the population breakdown in the sample, which we do in this article below.

The Salon article goes on, and tries to go into more depth, but it fails at fact checking its numbers. Here, it quotes a document by Ricard Romero, a guy who got fired by Burning Man staff, took them to court, etc. Here’s the quote from Salon:

There is a great deal of concern about the high frequency of depression and suicide among Black Rock City LLC (BRC) workers. While several factors contribute to depression and suicide, and correlation is not causation, the fact remains that 3 suicides (in a year) is an astonishingly high rate for virtually any population so small, and more so because, while these deaths are mourned, they are not entirely unexpected.
To put this in perspective, the US Army in 2011 reported a peak of 22.9 suicides per 100,000 soldiers, which was the highest rate seen in a decade. Per 100,000 appears to be a standard metric for this sort of thing. Assuming the combined numbers of Gate, DPW and Rangers to be approximately 1,000 strong, that would mean a suicide rate of 300 per 100,000. Statistically speaking, Black Rock City’s staff are 13 times more likely to kill themselves in the off-season than veterans returning from active combat duty. Even in a “slow year”, where only one BRC worker commits suicide, that is still 4 times the Army’s highest recorded suicide rate.

His numbers about the Army are bullshit too. Suicide among veterans and active service military are 28.7 per 100,000 among women, and 32.1 among men, averaged over the entire decade that included 2011, so he’s goosing the numbers to make his case. Next, he’s limiting his sample size to one year with three suicides, not eight years with seven suicides. That’s a data cherry picking problem tantamount to calculating the annual mass shooter deaths for the USA extrapolating from the week the Vegas shooting happened. And his assumption about the total number of staff appears on face to be wrong as well, by at least a factor of two, if not four or more. More on that as we unpack the thing. Read on.

The Data

I don’t have access to Burning Man’s employee records, but I do have access to quite a bit of data on Black Rock City, because they do an annual census that is fantastic. They have age breakdowns, income breakdowns, gender, race/ethnicity, and all other sorts of data. We’ll open with our first big presumption, that the demographic breakdown of Burning Man attendees is similar to the breakdown of Burning Man employees and volunteers. Some snapshots which are important to our analysis, and are also excellent examples of clear, beautiful, meaningful data visualization:

We can see from the visualizations, that the racial demographics of Burning Man are 10% “whiter” than the general population, and whites have around four times the suicide rate of every other ethnicity other than the native American population. Burning Man attendees are 58% male, and males have a suicide rate generally three to four times that of females. And the age demographics of burning man lie predominantly within a band of 25 to 54, which is where suicide peaks across both genders. (suicide among men specifically peaks in the elderly)

The next bit of data we need, is suicide breakdowns by age bracket, which we can get from the National Institute of Mental Health, here.

The last bit of data we need, is the number of Burning Man employees. Their website indicates they usually use about 2000 volunteers, and have around 100 year round staff. But they also employ a lot of subcontractors and part time staff, which are very hard to estimate. Around 425 people list Burning Man on their LinkedIn profiles, and I figure approximately five times that many are probably associated with the program without being on LinkedIn. That’s a number out of a hat, that I’m not pleased with, but it’s the best I can do with what’s available. (if you know a better number, reply in the comments section and I can adjust the math) Add those up, and we estimate 4650 total staff.

The Analysis

In a perfect world, I would have been able to find racial breakdowns of each age band in Black Rock City, and similarly fine-grained data for suicides, but I wasn’t, so I had to throw the impact of BRC’s marginal shift towards whiteness out. This will make my estimates for this analysis artificially low, but probably not unreasonably so. Another thing which definitely deserves mention, is rates of suicide among transgender individuals are alarmingly high. Unfortunately, I don’t have age-based breakdowns of this for suicide, so I didn’t include it either. I don’t anticipate that will affect the overall numbers much, because attendees identifying as ‘fluid’ and such only make up 1.6% of attendance.

To determine how many suicides we should “expect” within the Burning Man staff, presuming the staff exhibits rates in line with national averages, we estimate the total number of staff by each gender and age band, multiply by the “rate per 100,000,” and divide by 100,000 to get the units right. This tells us the expected value of annual suicides for each pot of employees. Add them up, and we get an expected value of total staff suicides. It looks like this:

If Burning Man employees follow the same demographic profile as attendees, follow the same suicide profiles as the rest of the country, and if our estimate of total staff is correct, we would expect to see 0.865 suicides among the staff per year. This isn’t necessarily due to anything Burning Man is doing. The reasons for suicide are long and complicated, and vary from person to person, and I’m sure some people commit suicide due to work related stress. But to make a claim that Burning Man is driving staff to commit suicide at a rate higher than the national average, someone would have to show that Burning Man’s actual suicide rate among staff was higher than the national average.

What is Burning Man’s actual suicide rate? Seven suicides in eight years is 0.875 suicides per year. Almost exactly the expected value above.

Are seven suicides worse than six? Yes. Are six worse than one? Yes. Is zero better than any? Maybe, but maybe not. That’s actually a very complicated ethical question, into which I will not dive at all. Were the working conditions at Burning Man responsible for some of the seven suicides mentioned in the article? It’s certainly possible, but the biggest problem with studying suicides, is you really have no idea why they did it unless they left you a note. Could Burning Man do more for their employees? Perhaps, but you’d have to ask them.

But in my mind, there’s no clear mathematical case to be made that Burning Man staff have a suicide problem any more than the rest of the USA has a suicide problem.