Math report

A number of people replied to some of the things I said about Graeme Smith and Jacques Kallis with the mantra “Statistics do not lie.” Oh yeah?

Until the end of 2007, James Anderson was a second-choice bowler for England, only getting in the side when someone was injured. During this period he averaged 39.21 with the ball. This year, he was picked ahead of both Matthew Hoggard and Steve Harmison for the second Test against New Zealand, thus graduating to first-choice bowler. Since then, he has taken his wickets at just under 27 apiece [at the time of writing]. He now has a career average of 34.

Mark Butcher had two stints in the England side. In the first, lasting 27 matches, he scored 1253 runs with two centuries at an average of 25.06. In the second, lasting 44 matches, he scored 3035 runs with six centuries at an average of 41.01. Overall, he has a career average of 34.58.

What truths do these averages of 34 tell about Anderson and Butcher? In my view, none. On the contrary, in fact: what they tell is lies. They allege that Anderson and Butcher are or were mediocre players, when the truth is that they have had periods of being consistently awful and periods of being consistently quite good without ever really being mediocre.

England’s collapsible top order have been the subject of considerable disquiet in recent months, but the Team England camp keep intoning that they all have career averages in the 40s (apart from Pietersen at 50+), thus asserting their right to keep their places. Yet Andrew Strauss, Alastair Cook and Paul Collingwood are all averaging under 40 over the last 12 months (and Michael Vaughan’s 12-month average was under 30 before he fell on his sword). Those healthy career averages are mostly telling lies about how good these players are *today*. And while Bell’s average over the last year is a reasonably impressive 47, when has he scored runs against a good attack on a vaguely helpful pitch? Unanalysed averages in his case probably serve to hide his being a bully on flat wickets or when faced with popgun attacks but pretty much useless when the chips are down.

As should be clear by now, I do not think that statistics are out-and-out liars. What they do is answer the precise question you have asked, but that is not always the question you were trying to ask, and that makes them awkward and untrustworthy unless you pin them to the floor and beat the truth out of them.

And the most untrustworthy is the career average, which means different things for different players. Some players are only picked at their peak and perform well for the four years they are in the side. Others, usually coming from weaker countries, have eight-year careers but get picked two years before they are ready and hang around for two years after they have stopped being good enough, simply because there is no-one else. And their career averages are correspondingly worse even though they are intrinsically just as good as the players who could only get in a side when they were actually good enough.

That is why I am always deeply suspicious of contributions to cricket debates which effectively say “X’s career average was 35 and Y’s was 40 and that proves it” (whatever “it” might be). As has already been observed in various posts to Different Strokes, comparing them across eras is fraught with difficulty, and the point I’m making here is that it’s not always a straightforward matter to do so even between contemporaries.