This information isn’t being sold, correct?

Report this comment

Report this comment

“Last year 42% of the people who participated in the Black Rock City Census considered themselves artists, and an additional 35% considered themselves artists sometimes (read prior stats from the Census). Do these numbers seem high to you? Low? We don’t know if the people who voluntarily filled out the Census were truly representative of the citizens of BRC.”

It reminds me of two stories. One is of a woman who told her daughter that she was taking classes to learn how to draw. Amazed (and a little shocked) the daughter asked, “When did you forget?”

The second concerns a man who went into a kindergarten classroom, walked up to the board and, taking a piece of chalk, put a dot in the middle of the board. Standing back, he asked, “What is it?” He got about a hundred answers. “It’s a cloud.” “It’s a bird.” “It’s a snowflake.” An airplane, a star, a speck of dust… the list went on an on.

The same wan went into a 2nd yr university physics class, went up to the board and put a dot in the middle of the board. He stood back and asked, “What is it?” There was complete silence in the room for several minutes. Finally one of the students raised a hand, and when asked, volunteered “It’s equidistant from the four corners.” Everyone else in the class was immediately relieved that the answer had been found.

EVERYONE is an artist.

Hugs,
Mike.

P.S. The original authors of those two stories are unknown, but much appreciated.

Report this comment

Sounds like a sound study plan within the limitations that exist. I do think sampling on Friday would be important to capture the “weekend warrior” crowd who only come for the burns. I’m on another work crew so I am not available to volunteer that day, but I do think sampling that day is important. Thanks for the complete response. You’ve made happy my geeky heart very happy. I’m currious if you have baseline data on previous years demographics, including experienced burner versus newbie ratios. It would be interesting to see what the effect of ticket shortages are on those numbers.

@ Sticky

A truly creative person can spell a word three or four diferent ways…but I like your implication…

Report this comment

Report this comment

Report this comment

Report this comment

For purposes of discussion, let’s just call the traditional “census” the “long form.” Several of the concerns voiced here (e.g., “I never go to Center Camp”) are the exact motivation for the sampling, i.e., let’s get less biased tabulations from tabulations of variables from the long form. We can’t get a random sample for the entire long form because it takes a while to fill out, so we looked for other methods. This is what we came up with:

The sampling described in this post will be multistage (some number of shifts within each day, 1 respondent from each sampled vehicle). We’re working now on how we sample vehicles, but it will probably be some variation of systematic random sampling at entry during each sampled shift. The variance projections are coming along, and they’ll go on the blog when we finalize them.

Data will be captured by handheld devices (Android phones), synced at the end of each shift, and new stratum distributions estimated. We don’t know what the best stratification will be yet, so we’re capturing about 14 mostly demographic variables and we’ll test it as data come in.

The point of the sampling will be to develop stratum weights to adjust tabulations done from variables in the long forms; we’ll do the adjustment via raking, matching the proportions estimated for each stratum by the sample. A stratum is a combination of variables we’ll capture, e.g., 20-24yo F from the non-California Western US in their 1 or 2 year attending who got tickets in the STEP program (age x sex x home region x playa age x ticket origin is one possible combination).

We’ll estimate the population proportion for each stratum, then use that to re-weight the comparable stratum in the long form responses. So if we estimate from the sample that 2% of the city is in the stratum above, but 4% of the long forms are filled out by people in that stratum, long form responses in this stratum will have a weight of 0.5.

The estimates we get this way reduce bias that is caused by stratum imbalances in the long form’s convenience sample, which is where we think most biases come from. But, as one AAPOR commenter wrote recently, “raking is not magic!” This method won’t get us to results as reliable as either a true census (which would be intrusive and burdensome) or as a random sample of the entire long form (which would be cumbersome and very slow to administer). So we think this is a good balance, and we’re eager to find out how it goes in practice.

We’ll get some direct estimates from the random sampling, but our real focus is on adjusting the “census” long form.

The universe to which we are generalizing will be Burners who enter during the days we sample. We could do another shift or two on Friday, though it’s not my impression that many new people turn up on Friday. If you’re worried we’ll miss people, volunteer for a few shifts sampling, and we’ll add Friday to the schedule!

Report this comment

Report this comment

Report this comment