"There are lies, damned lies, and statistics."
-- Mark Twain, misquoting Benjamin Disraeli.
I get messages like this occasionally:
It is interesting to read your current tally of attendance at the world's planetariums. It would be interesting to compare this number to previous year's tallies. Seeing numbers from previous years, especially from a source using the same method, would enable trends to be observed. Such trends would be as important as the number for this year. Could you post a comparison?
I could, but I would not want to claim trends could be observed, because of the methodology (or lack thereof) involved.
When I post a new attendance tally here on the Web site, it overwrites the old. I manually captured the comparison above from previous postings.
In my database, each dome theater record has one data field for "Annual Attendance". When a figure is provided to me, I enter it — and it wipes out previous data (if any) that was there. When I generate the attendance projection, it is merely a snapshot of the state of the database at that time. If the figure for 1995 is the last one I received, that's what gets stored for the rest of eternity. It was valid once, anyway :-). I do not wipe out the data each year and start over with a clean slate. My projection is based on accumulated, not yearly data.
Moreover, I do NOT keep track of the year for which the attendance figure is given. However, in 2005 I did start keeping track of the year the attendance report was received. I set a look-back filter to calculate the projection using only reports received over the last 5 years. This helps keep the estimate relatively current.
Also, keep in mind I'm using volunteered data (from you kind volunteers out there). An accumulated reportage from less than 10% of theaters isn't exactly comprehensive, though pollsters frequently project from far smaller samples. On the other hand, to my knowledge there are no other such tallies available; I daresay this guess is as good as you're likely to find. It wouldn't hurt to have more data points, of course!
Additionally, my counting program has evolved a little over the years, and some anomalies have caused the totals (and planetarium count) to bounce around some. For example, some time ago, I had entered an international planetarium's attendance figure of 50,000 — but the planetarium itself had an unknown dome size in my database. Based on that one data point, the program estimated that every "unknown" dome size had an average attendance of 50,000, which was clearly not the case — but it inflated the projection by more than 10 million people! Woops. In trying to correct it, I rewrote the program to leave out facilities for which the dome size was unknown. Woops again; those facilities then got dropped from the planetarium count altogether; it was just the attendance figure that should have been omitted. Eventually I got it straightened out.
In September 2002, one sole attendance report had a significant impact. A large U. S. facility reported an annual attendance of 1.2 million people. This skewed the average for the size category from 220,000 up to more than 320,000; and thus the projection for that category by nearly 1 million people. In 2015, they reported the same attendance numbers. Hayden continues to be a data point way off the bell curve, affecting the projections significantly. Would that every 21-meter dome have such great attendance! But none come close.
In 2005, I restructured my database, and separated out institutions from the domes they operate. Before, if an institution was operating multiple dome theaters (portables, for example), their total attendance had been counted for their large, fixed theater dome size. So, I re-defined the term "planetarium" to mean the dome(s) instead of the singular institution. Then I discovered that I'd been using a fixed-dome theater's attendance report count for any portables they operated, too; in other words, the number of responses for the latter category had been getting over-counted. Now the figure for "Responses" in those small size categories (portables are all in the 5-7 meter diameter range) is more correct. But, the average (total visitors divided by responses) became a much larger figure, and when multiplied by the large number of domes of that size, the projected attendance for that category tripled. This significantly boosted the bottom line attendance projection by some 7 million. One cannot conclude that worldwide attendance increased so much from the previous report; the estimating method became more refined.
That increase was offset in 2013, when I happened to notice another programming gotcha — the "look-back" filter I'd programmed in 2010 to calculate the projection using only reports received over the previous 5 years. Before, I had hard-wired the look-back period by simply typing the 4-digit target year; I came up with new coding to use the system date and calculate back 5 years automatically. That part was working fine -- but I'd forgotten to remove the hard-wired date, which was overriding the calculated date. Woops. So all the projection averages from 2010 to 2013 were based on accumulated data from 2004, instead of the expected 5 years. Once the offending code was repaired, the projected total dropped by 22 million people. You win some, you lose some.
In summary, then, I wouldn't feel comfortable saying there's a "trend" in attendance one way or the other. The most I can report is, some are up, some are down, most are within the error bars for maintaining. I cannot say with certainty (like Ann Taylor used to on NPR): "Declines led advances by a margin of 3 to 2".
On the other hand, while some older facilities are closing down, others are renovating and new theaters are being built. Overall, as the number of domes continues to increase inexorably, so does attendance. I think we can safely say with certainty the planetarium field is not moribund.