A few days ago, I made the conjecture that the age distribution of twitter would be a bell curve. Indeed, I actually went so far to stake $5 on it. Fortunately, I didn’t say who I had to pay $5 to because it looks like I might be wrong. Or maybe not.
Over the course of a couple of days, I conducted an informal survey consisting of begging twitterers to send me their ages. I promised to keep direct message responses anonymous.
Eventually, I got bored with the whole thing and decided to wrap it up. Below, I present an analysis of the data. There’s a very strict caveat here: I know nothing about statistics. So please don’t try to draw any actual conclusions from this.
27 people responded to my survey. I follow 159 people and have 201 follow me. So my sample represents 17% and 13% of the populations I’m interested in. It’s estimated that twitter will have 1000000 users by the end of the month so my sample is .003% of that. I haven’t bothered computing confidence intervals for all of this (because, frankly, I don’t remember how), but I imagine that “statistically insignificant” would probably be a phrase that would come up. Since I’m not trying to call a major national election, I’ll reiterate that you shouldn’t trust these numbers.
It’s actually even worse than that. My sample was self-selected as the “type of people that will respond to a random guy on the Internet asking their age” group. I don’t know how this affects the results, but I bet it does: in my experience people in a certain age range tend to not broadcast their ages to the world. So ultimately, my data represents People on twitter who James finds entertaining and who also find James entertaining enough to follow, and they’re willing to share their age with James (someone who they have likely never met); and also speak English because James only follows people who mostly speak in English
I’m sure you’ll agree that it’s something of a narrow demographic. Still, I’m armed with a spreadsheet so I will press on.
The minimum age in my sample was 21 and the maximum was 24. The median age was 20 and the mean was 30.3. The standard deviation was 6.89. At 25, I am .769 standard deviations from the mean.
Looking at a graph of the distribution, I do not feel confident saying that the data represents a bell curve; but it might if I had a bigger sample size. Obviously, it’s hard to say. To me, the most interesting thing is that it looks like the age distribution would eventually demonstrate a long tail with a bunch of people clustered at the younger end of the graph (which is to be expected on the Internet, I think) and then a lot of older people who don’t share ages in common. Or, maybe that’s the sort of thing any sociologist would expect…but it surprised me.
Another thing that surprised me is the large age range of the people I’m friends with on twitter. I’m fairly young and still not used to being an adult: that I can have things to talk about with a 21 year old and a 46 year old at the same time is amazing. That the 46 year old isn’t already annoyed with me is nothing short of miraculous.
I was also a little disappointed at the lack of response. A quick “@willia4 I’m 25″ doesn’t exactly take long. I guess not everyone finds collecting data as much fun as I do. Ah well. ;)
