Introduction to the 2013 Census Blog Post Series

Volunteering for the random sample team can be a great playa experience. You see the city from a different angle, go beyond the gates with your team, and you are the first person that incoming citizens meet when they enter. This year, the Black Rock City Census team had a great time on the playa! We continued our operations from the Census Lab, initiating some new projects and engaging dozens of fantastic new volunteers.

The 2013 Black Rock City Census consisted of two main components:

  1. We randomly sampled burners at the gate before, during, and after the event. Randomly sampled burners completed a short, 8-question version of our census survey.  In total, we sampled 1,562 individuals in 674 vehicles, which was a great improvement over the 1,050 individuals in 470 vehicles randomly sampled in 2012. As a trial, we also conducted an exodus random sample for the first time in census history!
  2. This year, we did not distribute an on-playa census survey. Instead, burners who participated completed the long version of our census survey online after the event. At first, we were worried that burners might forget to fill out the census after the burn. However, more than 11,919 Burning Man participants completed our census online this year (1 in 5.5 citizens)!

We used the data collected during the random sample to statistically weight the data collected from the voluntarily online (convenience) sample. The implementation of this online census saved months of data entry and yielded a similar amount of reliable data. These methods, together, ensured that our findings would avoid potential biases, accurately representing the Burning Man population.

In the next few weeks, we’ll publish a series of blog posts presenting the most important results of the 2013 census. In each post, a member of the Census Lab will present and comment on the results from one section of the survey, so stay tuned for results! Feel free to contact us at if you have any questions, comments, or if you are interested in being involved with our census team.

One of our teams of random samplers

Team Photo

Written by:

Autumn Albers and Steven Crane aka Indiana (in collaboration with Dominic Beaulieu-Prévost aka Hunter)

Edited by:

Tabitha Palmer aka Tabicat


7 responses

  1. Nice work! I do worry that your random sampling of vehicles may be significantly off though. In 2013 you saw 1,562 people in 555 vehicles, for an average of 2.81 people per vehicle. According to statistics collected by Traffic Works LLC (a consulting firm paid by Burning Man to study participant traffic in 2013) and the Nevada Department of Transportation, the typical vehicle coming into the event carries 1.9 people. The fact that these numbers are so off worries me that there may be significant bias in your sample. How was the sampling randomized?

    • Hi Colin!

      Definitely something to follow-up on. We randomize the sampling times based on historic vehicle entry data. We also randomize the collection points within lanes. We have a subsequent post on passengers and arrival data and it clearly indicates that around 10% had 1 person in their vehicle and 42% had 2 passengers. The implication is that 48% had 3 or more. What I suspect might be happening is someone projected a midpoint value and used the imprecise word “typical” which could easily be mistaken for “average”. Given that data, it would be pretty unlikely that the average would be less than 2. Note, the 95% confidence interval is +/-1%. Look for the upcoming post on arrivals data to understand more.

      Thanks for the post!

      – Scribble

      • I come across as kind of a dick here. But I like what you are doing. I just am pedantic about numbers…

        I am pretty sure your confidence interval math is wrong – the sample size is 555 for calculating people per vehicle, not 1,562. Also, the underlying sample isn’t Gaussian, etc.

        But the bigger problem is that you are letting the math mislead you. Would you bet $50 that the actual average number of people per vehicle in 2013 was between (let’s be generous) 2.7 and 2.9? Even when there is another study that shows 1.9? What about when Blue Cross (previous Burning Man Traffic Manager) says the number is close to 2.3 and:

        “As far as my direct-read numbers tell, the density of people/vehicle has remained within 0.1 people/vehicle for the past many years. So, on my analysis, we aren’t seeing less people per vehicle. Now, that is not to say that we are not seeing a higher percentage of large-capacity vehicles with 2 people in them. That wouldn’t skew the numbers, but does change (at least) the impact (environmental anyway, maybe not traffic…that is more a function of the raw number of cars).” –

        What about when your own 2012 numbers (reported above) show 2.23? Did things really change that much between 2012 and 2013? I worked gate both years, and I didn’t see a huge increase in passengers per vehicle.

        Bottom line – I think your random sample is wrong (at least for people per car).

      • Hi Colin,

        Your questions are valid ones so it is totally appropriate to discuss. We at the Census Lab are pretty focused on getting the data right so that the derived insights are actionable with positive effect.

        Dominic’s comments are correct in that I was citing individual responses. However, when I take the data for each bucket we have with those reporting arriving in a vehicle by themselves with a vehicle-count of “1” and those arriving in a vehicle with 2 passengers as one-half and those arriving in a vehicle with 3 passengers as 1/3rd, et cetera, across all our responses, adjusted with the random sampler weights, we get an average of 2.3 passengers per vehicle. I got roughly the same answer from the 2012 data as well.

        I am also following up with the Traffic Works people to understand their data better.

        Keep reading and posting!

        – Scribble

      • One thing that might be different – the Traffic Works study is based on road traffic, so would include everyone who enters through point one (deliveries, law enforcement, porto service, etc.). Almost all of those vehicles are single occupant.

    • Hi Collin!
      I double checked the data after reading your comment and here’s more info. First, thanks for pointing out the issue. There is indeed an error in the post (which should be corrected asap). We sampled 674 vehicles, not 555. According to our estimates, the nb of people per vehicle was between 2.2 and 2.4. The difference with the Traffic Works estimate is thus not as big as you estimated, but our estimate is still bigger than the one provided by Traffic Works. It is hard to know which of the two estimates is the more accurate since we have no info about the method used by Traffic Works or their margin of error.
      On our side, we sampled vehicles from Friday (preevent) to wednesday in seven 3h-shifts (happening at various hours of the day and night) and adjusted the estimated proportions according to the total nb of entries per day. Also, the number of people per vehicle was relatively stable between shifts and never below 2, ranging from 2.1 to 2.6 for individual shifts. I would thus be tempted to trust our estimates.

      One last thing to mention, the proportions reported by Scribble are accurate but cannot be directly compared to the mean number of persons per vehicle because they are based on people counts and not vehicle counts. Ex: in a population of two vehicles, one with one person and the other with two persons, the nb of persons per vehicle is 1.5. However, one person will say that s/he was alone while two will say that they were two. Thus, half of the vehicles were carpooling, while 2/3 of the persons were carpooling. Scribble results are about peoples, while the nb persons per vehicle is about vehicles.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s