average or just a median
April 27th, 2007I watched the much anticipated Human Footprint on Channel 4 last night. Although watching TV might not sound like much of a sacrifice, I did give up the Catherine Tate Show and Roman’s Empire (and even Graham Norton!) so there was a nontrivial amount of interest there.
In summary, it was the average number of things we (and I get the impression that this meant Britons, since some of the statistics explicitly made mention of the demographic) do in our lifetimes. The average number of times people have sex (4000 odd, which seemed low. Even more so considering how marriage rates are plummetting). The average volume of tears shed in our lifetime (120 liters? Really?). The average number of words spoken in our lifetime (123 million? Hmm.) and so on.
All accompanied by some wonderful imagery – the number of apples we eat in our lifetime was illustrated by a gigantic apple made up of … apples. In a slightly more squeamish twist (not if you chuckle at toilet humour), the loo roll use and crap we emit illustration was as graphic. The number of times people have sex was illustrated by condoms hanging from a tree and so on. It was faux art, I suppose – but my inner Philistine can’t tell the difference anyway so I found it all rather cool.
One thought did occur to me though – and it’s a thought that I’ve been pondering (in a different context) for a couple of days. Everything was expressed as averages – presumably an estimated arithmetic mean. As interesting, although perhaps significantly more difficult to gather, I would have liked to see the median in there too.
There is a simple reason for this desire. It is in fact explained in the median link above – but I recently saw a more entertaining explanation which I intend to reproduce via paraphrase (can’t remember where I saw it though).
Imagine that Bill Gates walked into a bar full of penniless, jobless, hobos. Suddenly the average (arithmetic mean) earnings of every person in the bar jumps into the giddy millions – but this does not represent “a real number”. The median calculates the probability distribution instead – so the influx of one high number does not skew all the remaining low numbers in the distribution.
The arithmetic mean would have one believe that the earnings potential of everyone in the bar jumps into the millions when Bill Gates (an outlying anomaly) walks into bar. The median provides a more realistic picture. Knowing as much as I do about statistics, I would guesstimate (ha!) that calculating the median based on a small sample size is much harder though.
And in a brief and unrelated footnote, Twitter has an awesome easy-to-use API (I wrote a PHP messaging API for Ach in literally half an hour) – but no means of keeping contacts private. Jaiku has a horrible API in comparison, but they do offer privacy for contacts. In other words, neither of the two services provides me with exactly what I need for my next step in wurld dominayshun. Curses! The Loverlord is not happy.