Frequently Asked Questions
Answers to the questions posed about GeoConvert most often
- Who can use GeoConvert?
- What does GeoConvert contain?
- How has GeoConvert changed?
- Is it possible to misuse GeoConvert?
- Why doesn't GeoConvert match all of my postcodes?
- Why do sums of populations of postcodes within output areas not agree with the published output area populations?
- What level of accuracy/confidence can I expect from GeoConvert data conversions?
- What input and output areas will give me the best results?
- Is my data held within the GeoConvert application?
- Why don't the columns of data that I supplied in addition to zone codes appear in the outputs file?
- I only have part of a postcode - what is this called?
- I only have part of the postcode, can I still get IMD scores (or other metadata)?
Who can use GeoConvert?
Anyone can use GeoConvert. Registration/login is not required
What does GeoConvert contain?
GeoConvert uses geographies used in the recent 1991, 2001 and 2011 Censuses (which can be used for matching and converting):
- 2011 Postcodes, Postcode Sectors, Postcode Districts, Postcode Areas
- 1991, 2001, 2011 Counties (English only)
- 1991, 2001, 2011 Districts (Local Authority, Unitary Authority, Council Areas in Scotland, Local Government Districts in Northern Ireland etc)
- 2001 and 2011 Lower Super Output Areas and Data Zones
- 2001 and 2011 Middle Super Output Areas and Intermediate Zones
- 2001 and 2011 Census Output Areas and Small Areas
- 1991 Enumeration Districts
- English regions
By uploading postcodes, people can find out data related to them such as:
- English Index of Multiple Deprivation rank and score (IMD 2010 and IMD 2015)
- Welsh Index of Multiple Deprivation rank (WIMD 2014)
- Scottish Index of Multiple Deprivation rank and score (SIMD 2012)
- Northern Ireland Multiple Deprivation Measure rank and score (NIMDM 2010)
- Output Area Classification Group, Subgroup, and Supergroup
- Urban/Rural classifications
- Postcode populations
- Postcode easting and northings
How has GeoConvert changed?
During August 2015, we changed GeoConvert. Changes included:
- Open to all - no login required
- Inclusion of 2011 Census geographies (as well as 2001 and 1991)
- Deprivation data from other parts of the UK
- Some other geographies that we previously provided access to could not be included (for licensing reasons)
The old version of GeoConvert used postcode delivery points as a population proxy, whereas the new version uses estimates of the number of usual residents per postcode from the 2011 Census. The change in the underlying data means that the new version has a more accurate estimate of the population, and also allowed us to make GeoConvert open to everyone. There is a disadvantage, as some of the geographies we previously provided has to be removed due to the different methodology and also for licensing reasons. As the new GeoConvert uses census postcode estimates as the underlying data, obviously any postcode that was introduced after the 2011 Census will not appear in GeoConvert.
Is it possible to misuse GeoConvert?
GeoConvert is based on the principle of proportional assignment of total usually resident populations between areas. All conversions of data are performed on this basis. The conversions will be most appropriate/accurate where the variable being converted is related to, and evenly distributed across the population (eg. sex (males/females). It will be less appropriate for variables that are unrelated to, or highly spatially-skewed within populations (e.g. average monthly rainfall, number of dairy cattle, ethnicity, income, etc.)
Why doesn't GeoConvert match all of my postcodes?
The current version of GeoConvert is based upon the sets of enumeration postcodes provided by the three UK census agencies. Enumeration postcodes are a subset of the complete set of live postcodes at the time of the 2011 Census time (mostly May 2011), and are aggregated to create census output areas, which are themselves aggregated to create most other census geographies. Only postcodes with at least one resident person are included. Many postcodes, such as those assigned to businesses, don't have any resident populations, and so won't appear in GeoConvert. Postcodes are quite volatile. New postcodes are created and old ones are terminated regularly. Existing/live postcodes can also change through the addition or removal of delivery points. The ONSPD records all live and terminated postcodes. Each postcode has a date of introduction and, if relevant, a date of termination. Things are further complicated because postcodes can be re-used, so a postcode can be terminated, and then reappear with a new date of introduction, replacing/removing the record for the previous instance of the postcode. Postcodes which weren't current at the time of the census also won't appear in GeoConvert. Missing postcodes won't affect the quality of matching and conversion operations between geographies.
Why do sums of populations of postcodes within output areas not agree with the published output area populations?
In GeoConvert, postcodes are assigned to other geographies on the basis of their centroids. Centroids are geographic, snapped to the nearest address point (is this true across UK?). It's possible that a postcode's centroid can be outside the actual area of the postcode. In these cases, the postcode centroid can fall outside the OA to which the agency assigned the postcode, creating an anomaly (postcode population sums for OAs don't agree with published figures for the OAs themselves).
- ONS information on output area derivation
- NISRA policy for 2011 geographies
- NRS Census 2011 geographies guide
What level of accuracy/confidence can I expect from GeoConvert data conversions?
Postcodes and their populations are allocated all-or-nothing to areas on the basis of their centroids. This can sometimes give rather arbitrary/uncertain allocations where postcodes straddle two different areas, with populations in both, and the centroid falls close to a boundary between two areas, or in situations where the centroid might even be outside the postcode area. These problems occur most where the postcode is close to a boundary between areas. As a general rule, these 'edge effects' are more pronounced for smaller areas (higher edge to volume ratio). The uncertainties are doubled in conversions (double uncertainty for both source and target allocations), and most pronounced where both source and target geographies are small. The best conversions are between pairs of geographies that both have areas that are large relative to the size of postcode areas.
What input and output areas will give me the best results?
To get a good result when converting from smaller to larger areas it is best to have a complete set of the smaller nested geography zone. For example trying to aggregate two output areas to a whole country will give a bad result. But if you had a complete set or almost complete set of output areas that make up the country then you would get a good result.
The best data conversions are from 2011 geographies to other geographies. Matches and conversions are based upon analysis of the population recorded in the 2011 Census. Conversions of data contemporary to 2011 will, therefore, be more reliable than, for instance, conversions of 1991 variables (eg from the 1991 Census) as there may have been considerable change in the distribution of population in the intervening time.
Is my data held within the GeoConvert application?
Yes, user-supplied input files are held within an upload directory to make it available to the GeoConvert application for processing. No-one outside the GeoConvert team has access to the upload directory, and files are automatically deleted at regular intervals. However, if you have any concerns about the confidentiality of your data, it's best to anonymise it by separating sensitive components (eg names and addresses) and non-sensitive components (eg numbers), linking related items in the two sets using unique identifiers, and only submitting the non-sensitive components for processing by GeoConvert
Why don't the columns of data that I supplied in addition to zone codes appear in the outputs file?
GeoConvert does not 'pass through' any additional data from your input files other than the original source code (i.e. output area, ward code, etc.), except in the 'Convert Data from One Geography to Another' function, where data you input is converted to reflect the new population base of the output geography.
Postcodes do vary in length, and the position of the space is important. However for GeoConvert you can enter the codes with or without the space, but you have to ensure you select the correct input geography (e.g. postcode sector). Here are some examples:
For the postcode M1 3GF
M is the Postcode Area
M1 is the Postcode District
M1 3 is the Postcode Sector
For the postcode M16 0AA
M is the Postcode Area
M16 is the Postcode District
M16 0 is the Postcode Sector
For the postcode: SK1 4AG
SK is the Postcode Area
SK1 is the Postcode District
SK1 4 is the Postcode Sector
For the postcode SW1A 2AA
SW is the Postcode Area
SW1A is the Postcode District
SW1A 2 is the Postcode Sector
I only have part of the postcode, can I still get IMD scores (or other metadata)?
Levels of socio-economic deprivation can be very localised, and vary enormously over relatively small distances. Deprivation scores, such as the Indices of Multiple Deprivation - are developed to be available for small areas (in the case of the IMD, down to LSOA) in order to try to identify local variation and heterogeneity.
Postcode sectors for example can have very large geographical extents which may overlap several of the deprivation score geographical units with widely differing deprivation score values. If deprivation scores are calculated for larger areas in order to overcome this problem, then their ability to identify local heterogeneity is lost.
However, if you only have part of a postcode (e.g. postcode sectors), then you can use GeoConvert to generate a list of postcodes that belong to the postcode sector (along with the IMD score for the postcodes LSOA).
To do this, use the "Match one geography to another" function
For the "Source geography" select "Postcode Sector"
For the "Target geography" select "Postcode"
Click "Display Metadata", tick the relevant boxes (e.g. IMD scores/ranks)
and then upload you file of postcode sectors
This will give you a list of all postcodes for the sector along with the metadata you selected.