Understanding UK address referencing

Data from many sources, including administrative records and surveys, are increasingly being recorded in association with full address information. In previous decades, many such datasets contained only a postcode or placename. Address information is of great value for geographical referencing but the potentially complex structure of addresses needs to be understood in order to maximise potential benefits for mapping or data linkage.

Addresses in practice

It is important to recognise that the locational information recorded in most postal addresses does not equate in any direct way to the geographical references attached to most social science datasets. This can be illustrated by way of some examples.

Address example 1:

Address	Address Elements
39 Acacia Avenue	Property number and street address
Silhurst	Post town
SH15 6BP	Unit postcode

In this very simple example, all the required elements are present to make a valid postal address. These include a property identifier (in this case a house number, although in other circumstances it may be a house name), thoroughfare (street) name, post town and postcode. These elements are entirely sufficient for everyday purposes such as delivering mail or visiting the address. However, few of these elements are likely to relate unambiguously to the geographical units and codes used for most social science data. House number ‘39’ may equate exactly to one property which is a unit of study. However, if the researcher is really interested in households, individuals or businesses, it is important to recognise that number 39 may not be sufficient to unambiguously identify the units of interest. ‘Acacia Avenue’ will not equate directly to any statistical unit – indeed, a long street will often cut through many different geographical areas with different codes and available statistical data. ‘Silhurst’, a post town, is a geographical entity defined for mail delivery purposes and is most unlikely to correspond exactly with any census ward, super output area, local authority district or other standard geographical area. The unit postcode ‘SH15 6BP’ does not in its own right comprise a statistical area, but can be associated with most official geographical units using standard directories. It is probable that address, in its entirety, could be successfully matched to one or more entries in the major national address lists. It is equally possible that it may fail to match uniquely for various reasons – for example if there is more than one subdivision of the property at 39 Acacia Avenue, or if there is more than one Acacia Avenue in Silhurst. The additional certainty provided by the unit postcode, which typically relates to around 15 addresses explains the utility of the postcode as a locational reference.

Address example 2:

Address	Address Elements
Geo-Refer Enterprises Inc.	Business name
Flat B	Property subdivision
39 Acacia Avenue	Property number and street address
North End	Locality address
Silhurst	Post town
Loamshire	County
SH15 6BP	Unit postcode

In this example, four additional elements have been added to address example 1, which demonstrate additional aspects of address structure. Geo-Refer Enterprises Inc. appears to be a business name. This is likely to be important to some researchers, but it is not formally part of the address. This may be the registered address of many different businesses and business names are often subject to frequent change while the property itself remains the same. 'Flat B' suggests that the property at '39 Acacia Avenue' is subdivided. Some residential addresses are also used as business addresses and it can be very difficult to separate out these different types of address for analytical purposes. Similarly, it is difficult to confirm whether a record relating to 'Flat B, 39 Acacia Avenue' actually corresponds to the same address as a record for '39 Acacia Avenue'. In this example, a locality name 'North End' has been included. The locality is a necessary part of the Royal Mail postal address if there is more than one 'Acacia Avenue' in the 'Silhurst' post town. The locality presents no stronger association with administrative or statistical geographies than the post town. The County name is frequently included in postal addresses, but is not a requirement and it cannot be guaranteed that the county will correspond with the administrative county by the same name.

Address example 3:

Address	Address Elements
501a Halfway Street	Property number and street address
Sidcup	Post town
Kent	County
DA15 7XZ	Unit postcode

Although this exact address is fictitious, its major elements are genuine and it serves to demonstrate the mismatch between postal and administrative geographies. The post town of Sidcup does not fall within the administrative county of Kent but within the London Borough of Bexley. The DA15 postal district relates to Dartford, which is actually in Kent. Halfway Street falls mainly within the London Borough of Bexley but extends a small distance into the London Borough of Greenwich. It should thus be clear that none of the postal address elements should be used directly as a basis for geographical referencing. Either the full, structured address or the unit postcode should be matched against a trusted directory in order to provide corresponding geographical codes at an appropriate level of resolution.

Wherever possible, researchers should seek to record, store and manipulate address information in a structured format, even if a fully standard-compliant address cannot be produced. Splitting addresses into separate fields and requiring each of the major elements to be clearly demarcated will always increase the utility of the address information and aid matching. Text string matching of addresses is complex and there are many causes of potential confusion including variant spellings, multiple languages and variant abbreviations. These are much reduced when the information is clearly structured. Researchers needing to undertake large-scale address list matching should consider using commercial address matching software or services.

Details of postal address structure can be found at the the websites of Royal Mail [http://www.royalmail.com/] and Post Office [http://www.postoffice.co.uk/].