So the 2020 Australian Go Census has come to an end. The anonymised data can be found here.
Honestly, I’m not thrilled with this dataset. I was trying to streamline the process, but looking at it I think I lost more than I gained. The Limesurvey system is fine, but the way it handles data exporting and managing is awkward at best. Maybe I’ve been spoiled by Qualtrics, I don’t know.
But more precisely what bothers me about this dataset comes down to the size. In usual, when you run a survey like this, you get a number of completed responses, and maybe double that number in partial responses – people do the first section, then drop off due to time, interest, whatever. Which is fine – that’s why you put the most important stuff up first. But with this, despite running it for two months, I got about 85 completed responses – already less than I’d like, I generally consider less than 100 completed responses to be more or less a waste of time because you’ve almost certainly missed some kind of important group – but only maybe a dozen or so partial responses, for maybe a hundred responses in total, for all of Australia (and actually less, once you remove the respondents from Switzerland and Japan. Bless their hearts, but the census was very precisely aimed at people currently residing in Australia). Looking at the gender and state distributions, they’re roughly what you’d expect to see, so this might not have happened?
I have some thoughts as to some of the reasons why the response rate for this was so low, and I am trying to work out ways to make next year’s more useful, but suggestions are always welcome.
But more to the data we do have. Eyeballing it, it looks broadly right to me – about 10% female-identified, QLD has a few more responses than I’d expect but seems broadly in keeping with past censuses so at least it’s consistent.
More interestingly, respondents seemed to be quite formally educated. Over 40% claimed a Bachelor’s degree, and over 12% claimed a PhD. For reference, roughly 25% of people have a Bachelor’s in Australia, and about 1 to 2% have a PhD. So this is clearly out of whack. I haven’t compared the other educational outcomes, but it wouldn’t surprise me in the slightest if it was similarly disproportionate.
OGS was the most popular server identified by far. This does not necessarily mean much – the survey was linked to on the OGS forums, and thanks to the help of OGS moderators, was also a popup ad for a time on OGS. So it’s plausible that it was just more visible on OGS. The link was the Message of the Day for several weeks in the Australia Room on KGS, but I don’t know how much traffic that sees. In addition, OGS was the default server for the Australian Digital Tournament, which was running concurrently to this.
Lastly, the most important question: what do you prefer to call go/baduk/weiqi? Unsurprisingly, the most common answer by a mile was “go”, but I was not expecting baduk and weiqi to be about the same, with baduk edging out weiqi slightly. I don’t know, maybe it’s just the cultural oddities of my club, but I see way more Chinese-culture players than Korean-culture players, so I kind of assumed that weiqi would be more popular. Guess that’s a good wake-up call for me to remember to be careful about generalising from my experience.
Anyway, that’s my brief musings on the data and survey. I encourage you to get the data above and examine it for yourself! There was obviously a lot more questions asked, and there’s a lot more interesting analysis that can be done – looking at potential relationships between variables etc. But I wouldn’t want to take all the fun away from you!