27
Feb
2017
2

Everything seems simple until you think about it

One of the joys of working for OS is that you get asked to give authoritative answers to all sorts of geographic questions. ‘Classic’ questions such as how long is the coastline of Great Britain? often crop up. Which, if you read our recent blog on which English county has the longest coastline, you’ll know isn’t as easy to answer as you might think. Often, seemingly simple questions have no definitive solution. For me that doesn’t matter. The joy comes from thinking through the problem to come up with the best answer possible.

The coastline question reminded me of a problem I tried to tackle myself last year. Listening to a news story on the radio, it described “27 million households across the country”. Over the next 48 hours the story was repeated across a broad selection of media outlets and every time the same statistic came up. After mulling on this for a while I decided I didn’t like this number. It’s too imprecise. So, I decided to delve a little further and try and work out a more accurate number for myself.

What is a household?

The first question is what is a household? I went online to look for definitions. I found many of the online definitions problematic:

  • “a house and its occupants regarded as a unit” – what about flats?
  • “those who dwell under the same roof and compose a family” – what about single people? Aren’t they a household?
  • “A domestic unit consisting of the members of a family who live together along with nonrelatives such as servants” – which century is this written for?

I decided I liked this one the best:

  • “All persons living under one roof or occupying a separate housing unit, having either direct access to the outside (or to a public area) or a separate cooking facility.”

Under one roof and occupying a separate unit is the key criteria I decided to focus on.

Finding the data

AddressBase Plus

Next, I needed to find a list of households to count. The obvious answer was to use an address register, of which there are two main ones in Great Britain: Royal Mail’s Postal Address File (PAF) and Ordnance Survey’s AddressBase family of products. I decided to start with AddressBase Plus as this is built using all of the addresses from PAF, with additional addresses sourced from Local Authorities and OS. All of the addresses in AddressBase Plus are allocated one of 563 different classification codes meaning it should be possible to judge whether an address is a household.

Working with the classifications

The simplest query to run is to count all the addresses in AddressBase Plus. This gives us: 35,111,589. The next step is to exclude any categories that clearly can’t be considered households. For this, I decided to use the Primary Classification Codes (of which there are nine) and exclude addresses that are:

  • Commercial – 2,816,508
  • Land – 919,243
  • Military – 4,579
  • Other (features surveyed by OS such as ponds) – 366,187
  • Parent Shells (the shell of a building that is subdivided into smaller units) – 1,891,196
  • Unclassified – 211,206
  • Objects of interest (includes visitor attractions and places of worship) – 65,389

AddressBase Plus: property classifications

This leaves us with everything with a Primary Classification of ‘R’ which is Residential: 28,780,574. OK, I was getting closer to the 27 million figure.

The next step was to start examining the secondary and tertiary classifications (sub categories in plain speak) within the overall ‘Residential’ category. I started by identifying the easy ones that I’m sure everyone would agree are households. This list includes:

  • Residential, Dwelling, Detached, Semi-Detached, Terraced, Self-Contained Flat (Includes Maisonette / Apartment), Sheltered Accommodation: 28,081,423

Remembering the definition above about separate units under a single roof, I decided to include addresses where it was reasonable to expect the residents led individual lives:

  • Houses of Multiple Occupation, Residential Education, Care /Nursing Home:, 308,268

I also decided it would be reasonable to include more communal establishments:

  • Residential Institution, Religious Community, Communal Residence: 18,703

Then I had to consider the remainder: I decided not to include ‘Privately Owned Holiday Caravans / Chalets’ as these are unlikely to be a primary residence, but I did decide to include the other categories that could be:

  • Caravan, House Boat: 121,367

Lastly, I have added in Dual Use: 56,707. These are premises that have a mixed work/live classification (think flats above pubs) and therefore should be included.

Adding all of that together, the grand total is 28,586,468, higher than the published figure of 27 million.

In conclusion

Arriving at a precise total like this was satisfying, so I shared it with my colleagues and sat back to bask in the glory. Within ten minutes I had this reply from a colleague:

“Sorry, but I think you have fallen into a semantic quagmire. The Government’s definition of a household is premised (a pun!) on demographic behaviours not spatial entities. It is a demographic definition set by ONS and deployed in the 2011 Census definition. That Census definition underpins subsequent ONS estimates of household numbers as 26.7 million households in the UK in 2014. I think a more pertinent question might be ‘Where in our 28,586,468 addresses are the 26.7 million households to be found?’…and I do not believe that question can be answered.”  

Somewhat deflated, I sat back and remembered the words of author Audrey Niffenegger: “Everything seems simple until you think about it”.

You may also like

Going beyond Great Britain with AddressBase Islands…
Why you need an Address Master Data Management strategy
Announcing the end of an addressing era
Get more addresses with AddressBase

2 Responses

  1. Adrian

    On top of the response you got, there’s also the issue of local authorities completing the secondary and tertiary classifications codes. i’ve found that varies wildly!

Leave a Reply

Your email address will not be published. Required fields are marked *

Name* :

Email* :

Website: