When I have important tasks to get on with, I instead make graphs. In reference to Sunday’s post, here is an illustration. It assumes you accept as a given that Conservative-inclined seats have higher percentage turnout. It’s trivially obviously true so I can’t be bothered proving it, but if anyone wishes to dispute that point in the logic chain, I will get my calculator out. In what follows I have left out the most extreme seats on each chart – Staffordshire South, Isle of Wight, Highlands and Islands, etc, so as to make the overall point more clearly.
The bloggers in question were complaining about the upward slope on a graph in which the x-axis is, by proxy, “turnout”. Their maths was, however, apparently based on a situation where the y-axis is “votes cast”. It is fairly clear that even if constituencies are scrupulously fair, this graph will have a rising trendline – the higher the percentage of people eligible to vote who do so, the more people in total will in fact vote. And this is indeed what we see.
The graph about which they can in my view more legitimately complain, is one where percentage turnout is once again on the x-axis, but the y-axis is ‘constituency size’.
This does have an upward sloping trendline too (not terribly obvious at first glance perhaps, but there is one) although of a much smaller size – the crunch question is whether this upward slope is resolved by the current boundary changes on which the next election will be fought, or not. I intend to have a go at working that out after the election – someone poke me if I haven’t done it by, say, August.