Taking leave of the census

Last month, CACI announced that it was moving away from using data from the UK census in the development of its geodemographic segmentation tool Acorn, which categorises the UK’s population into 62 different consumer types. The new approach “recognises that census data is no longer the primary and most up-to-date source of information about neighbourhoods,” says CACI. Instead, the company now draws on a range of data sources, including commercial and public sector open data and administrative data.

The change comes amid an ongoing review of the census, with the UK government having tasked the Office for National Statistics (ONS) with considering a range of alternatives that would be more cost effective than a full population count every 10 years.

Many of the options on the table envisage a greater use of these same public and administrative data sources that Acorn is now using.

So Research asked CACI partner John Rae about whether possible changes to the census influenced the revamped Acorn approach, and to share his thoughts on the ONS’s Beyond 2011 consultation.

Is the change to CACI methodology a result of proposed changes to the census?
John Rae: No, not really. The basis of our approach has been developed, and in part applied, to Acorn over the past decade. We were already intending to change our approach prior to feedback from the ONS Beyond 2011 census initiative.

Specifically, feedback to the MRS Census and Geodemographics Group led us to believe that for economic reasons there is a very high likelihood that future census statistics will not include small area data useable for geodemographics. This supported the decision to implement our new approach to data. It seems extremely likely that all geodemographic segmentations will have no alternative but to take a similar decision, either now or before publication of the next census.

Is CACI essentially road-testing for the census?
JR: The broad concept of Beyond 2011 – that one can amalgamate sources of administrative data as a surrogate for the census – lies behind our new approach to geodemographics. Building a geodemographic in this way, however, is by no means the same as creating a full census from administrative sources. CACI is not road testing for the census, either formally or informally.

We have offered to supply data to the Beyond 2011 project and if, following their analysis, the team feel there is merit in using it we will be happy to provide this data.

What format do you see the census in the future?
JR: Small-area census statistics have been very useful to many businesses and CACI would certainly prefer to see this data source continue in the future. However we realise that in the current economic climate the cost requires justification.

As more information becomes ‘open data’, particularly for small areas, there is a growing set of alternatives to some of the information provided by the census. While a census or similar surrogate dataset is traditionally a key dataset, we can imagine a time when, with sufficient expansion, open data might form a viable alternative for our purposes.

Our new approach to geodemographics does not require there to be a census as such. For CACI, and the many businesses we support, the requirement is for good sources of small-area data. We have used a mixture of open data, other administrative data, and private sector datasets.

Similarly, we would find it useful to have access to the raw data used in Beyond 2011. Some form of access to these datasets, with suitable privacy controls, perhaps in a secure environment, might well allow interesting and beneficial forms of geodemographic analysis.

Although it clearly might not meet the needs of every census user we welcome the growth of open data, and would welcome ONS in a role protecting access to many raw administrative datasets. We feel such a role would strongly support the provision of more regular census-like statistics for larger geographic areas.

How do you think the shake-up of the census will pan out? 
JR: We are greatly encouraged by the ideas behind the ONS Beyond 2011 project. The approaches seem to offer many opportunities for the analysis of small areas. Against this there may be certain uses of census data that cannot be supported by the proposals, specifically detailed evaluation of small sub-segments of the population. Such trade-offs seem inevitable.

• A technical document outlining the changes to Acorn’s methodology is available here.