Tuesday, 13 September 2016

The electoral register is the most accurate source for adult population data

Right now the electoral registers are probably the most accurate estimate of the UK's adult population. There are two reasons for this - one a credit to the Coalition government and the other to the insistence on data sharing by local authority chief executives. What disappoints me is that opposition parties at Westminster continue to present the idea that there is somehow a better estimation of adult population available for the drawing of constituency boundaries.

Let's look firstly at the the new electoral registers based on a process of individual registration.This new system was introduced for various reasons including the persistence of register stuffing (that is registering people to an address who aren't living there or in some cases don't even exist), inaccuracies such as dead and gone way folk (nixies as we call then in direct marketing) staying on the register, and a whole load of people who for various reasons weren't registered at all.

Had we just stuck with introducing a system of individual registration then it is beyond doubt that the numbers registered would have fallen and that fall would have disproportionately been among the young and the poor. But the new system didn't take that approach but instead allowed existing verified name and address data within government systems to be shared for the purpose of registration. As a result somewhere between 65% and 80% of registrations were concluded using existing information and this information, given much of it was for benefits recipients, disproportionately focused on the poor. Anyone receiving a benefit is automatically registered - that's every mum, all the registered unemployed, those receiving in-work benefits, retired people and those receiving sickness or ill health support.

We can add to this the use of council tax records, the DVLA database, MoD staff records and some higher education records. All-in-all a pretty comprehensive effort to ensure that most of the registration process didn't require expensive and time-consuming paperwork. To fill the gap, local authorities undertook a variety of different approaches including traditional mailshots, door drops, advertising campaigns and canvassing. The result is a system of registration that is over 90% accurate even before the addition of a further 2 million electors - an extra 4% on the total - during the first half of 2016 (controversially excluded from the drawing of new boundaries).

On the basis of this successful (but still not perfect) system it seems sensible to use the electoral registers as the basis for drawing up boundaries for parliamentary constituencies. Some seem to differ, arguing that we should use the "whole adult population" as the measure. The basis for drawing boundaries using a 'whole population' measure has to be the 2011 census and this is already five years out-of-date. We should also note that the census is not itself a precise identification of every person (although it strives to do this) and the data is adjusted to account for under-reporting in some areas and this data was challenged by some local authorities. Again, the census is better than 95% accurate which is good enough for most purposes but in this it isn't much better - at the point of collection - than electoral registration. Its advantage isn't a better identification of individuals at addresses but rather the richer data associated with the individuals actually identified.

And this is all out-of-date so we would need to make some adjustments to the numbers so as to be accurate. At the national level this is pretty straightforward - add and subtract births and deaths then adjust for net migration to get a pretty accurate estimate. For local authorities the first part (births and deaths) is easy but migration isn't as we've no requirement - such as is the case in Spain - for registering residency. We'd have to use proxy measures such as the council tax base, the Post Office change of address file and (irony) the electoral register. And this would still only get us a measure at the local authority level rather than the ward level data needed to draw up boundaries. None of this makes for a more accurate estimation of ward-level population than the electoral register.

There's a valid criticism arguing that those 2 million 2016 registrations should be included in the review but the effect of this would be pretty marginal unless the distribution of that 2 million is very skewed in terms of geography. It's effect would be to shift the quota range from 71-78,500 to 73.6-81,300 a result likely to further disadvantage smaller inner city seats - there's no evidence that excluding these voters would have anything other than a very marginal impact on the new seat distribution.

As a last thought, we can note that the last full review of English local council ward boundaries (2003) was on the basis of a future population projection rather than the census or electoral register. It is clear that, for many places, this was a pretty unsatisfactory measure - you only need to look at the population change in Leeds Council's Headingley ward to recognise that current population is a better basis for boundaries. And right now the most accurate source of current adult population numbers at ward level is the electoral register.


1 comment:

Dr Evil said...

I don't agree. I think the population as a whole needs to be deduced from the sale of food and also the amount of sewage that is being processed. Subtract from that the likely number of children under 18 years old (ONS) and that should give the approximate number of adults in the UK. That will include all the illegal ones of course who, I hope, are not on the electoral role.