Turned out nice again

This weekend the Electoral Reform Society added to the worries around the Police and Crime Commissioner election when they published an estimated turnout of 18.5% for November's election. While Policing Minister Nick Herbert had famously spent an earlier part of the week on the Today programme refusing to give Evan Davis a figure for what would constitute a good turnout, it is unlikely that the Society's prediction will have brought him much cheer.

However, it may be worth looking a little deeper, at the Society's own document on how they arrived at their figures.

Firstly, they started with a baseline of 'c. 34%' because 'Recent Local Election Turnouts are in this region'. I don't disagree with that, or with the decision to assume a turnout baseline similar to a local election rather than a national one, but being used to reading opinion polls and research reports, the description of the baseline was lacking a certain precision that I have come to expect.

They deduct 6% because the election is in November rather than May, based on data from Rallings and Thrasher on variable turnout at local by-elections – seems fairly sensible. Rallings and Thrasher are acknowledged good-eggs/experts in this sort of thing.

Then they deduct 5.5% to reflect the fact that turnout is lower when there is no free mailshot, and boosted when there is one.

Then they deduct 4% because there aren't going to be any Party Political Broadcasts (technically Party Election Broadcasts, but I'll not fuss too much about that).

But, WAIT A COTTON-PICKIN MINUTE, did they really just deduct 5.5% for there being no free mailshot from a local election baseline? Local Elections don't have free mailshots, so surely the lack of one is factored in already in the 34% they started with? Taking it off again is a form of double-counting.

Now, there are other factors they mention but couldn't measure, and others they perhaps didn't mention, but I noticed that while my references to their 18.5% turnout were retweeted, my query as to the double-counting did not fare so well. Sometimes we are not truly happy unless we are being grumpy.

Update: – the Electoral Reform Society have responded to this article in the comments section of an article on their website. They explain the reasons behind 34% (which is fine), and say the 5.5% deduction is about measuring campaign intensity rather than the presence of a mailshot per se. Thanks to them for the response – have a look everyone and see what you think.


This entry was posted in Media coverage, Perspectives, Resources. Bookmark the permalink.

2 Responses to Turned out nice again

  1. Chris W Drew says:

    The wider – and almost pleasing – message here is that, as I so many areas of forecasting (political, economic, euro, weather, share prices, the gee gees), the “EXPERTS” are flying by the seat of their pants just as much as the rest of us.

  2. steveb1960 says:

    So we end up with a 24% turnout on their estimates rather than 18.5. A slightly better estimate but still on the low side?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s