OPINION17 March 2011

The 2011 census – have we learned the right lessons?


Last year’s US census set a great example of how research can be used to make sure hard-to-reach groups are properly counted. But has the Office for National Statistics missed opportunities to achieve the same in the UK?

The next census for England and Wales is taking place in ten days, on 27 March. In providing population data at national, regional and local levels as well as geographic demographics and shifts over time, census data is at the heart of most market research work.

The US Census took place a year earlier, on 1 April 2010, offering a timely learning opportunity for the UK’s Office for National Statistics.

We live in a world where individuals willingly share ever more personal information via social media and shop loyalty cards, but are increasingly wary of sharing this same data with official organisations. This is an issue that the ONS must overcome if the census is to provide the accurate population statistics that users require.

One way to counter this lack of trust is for official organisations to be as open as possible about the way they operate and their motivations. In this, the US Census Bureau set a positive example, commissioning several pieces of high-quality research to improve the return rate.

“The open attitude that the US Census Bureau demonstrated from the outset is something that the ONS, which faces many of the same challenges and controversies, can still learn from”

Lessons from abroad
The US Census Bureau’s work focused on hard-to-count populations, and started by determining who they were and why they were less likely to complete census returns. The Bureau commissioned three waves of research to identify the weaknesses in its communication and enumeration strategies, using data from annual ‘mini-censuses’. It identified correlations between demographic data and non-return of census forms, created social models of hard-to-count groups and quantified them as a proportion of the total population.

The Bureau built on its research with a survey on the factors that motivate certain groups of people to take part or not take part. Instead of relying on crude identifiers such as ethnicity, they were able to pinpoint small sectors of society that had distinct reasons for not filling out the census. The Bureau shares its research with hundreds of government, non-profit, corporate and community organisations, showing its commitment to transparency. This enables local government and smaller-scale local groups to engage with and support the census process.

The open attitude that the US Census Bureau demonstrated from the outset is something that the ONS, which faces many of the same challenges and controversies, can still learn from. But such transparency has at times been lacking in the ONS’s dealings with local authorities and community organisations keen to support a successful census.

Census population statistics have a direct link to the amount of government funding local authorities receive, so it is in their interest to encourage the highest possible and most accurate response rates. Yet the ONS has proved reluctant to release research data that would enable local authorities to better tailor census messages to their local populations.

As a starting point we know that the massive undercount of Westminster and Manchester during the 2001 Census was due in part to small pockets of high non-response rates. Notwithstanding some methodological issues with the way the census was run in these areas, identifying these pockets and tailoring communication and enumeration strategies to them, as was done in the US, could have improved the accuracy of the final count and reduced the need for controversial methods of filling in missing data.

The ONS has not yet published any research into people’s propensity to take part in this year’s census, and despite claims that it has undertaken extensive research on hard-to-reach groups, very little data has been shared on that either. It has focused on groups such as the elderly and ethnic minorities, but has not said how it identified these groups as hard to reach in the first place.

Freedom of Information requests have been submitted to try to get hold of this research, but the ONS has said it is unable to provide any more data. It said that it had mainly used information from previous censuses and other surveys, as part of “a very wide range of information sources and experiences”. The majority of documents it has cited are several years old and focus on issues that arose in the 2001 census – but the attitudes of 2001 are vastly different from those of 2011, on such salient issues as the behaviours of internet users or concerns over privacy and data security.

Understanding the population
As media interest in the census grows, comment has focused on the cost of the exercise and the award of the contract to US defence contractor Lockheed Martin. People have even questioned the need for a census at all. It is worrying that from the research available in the public realm there is no evidence the ONS has considered the effect of such issues on national response rates or tested communication strategies to offset any negative impact.

The US Census Bureau recognised the need for original research and tackled such issues head on, helping it to deliver the census on time and under budget. In contrast, the lack of work in the UK to gauge propensity to complete the census among the population as a whole could turn out to be significant. The England and Wales census could be in a precarious position if even those groups who could traditionally be relied upon to fill in their forms without question were not to comply.

The importance of the transparency and community engagement demonstrated by the Census Bureau is not, however, limited to logistical efficiency and maximising response rates. The UK’s census is now less than two weeks away, so the bulk of the exercise is a fait accompli, but the ONS could stand to gain from being more transparent in the publication of census data and any subsequent challenges to it. If the users of census data are to have confidence in the results when they begin to be published next year, they need to be able question anomalies or unexpected figures and the ONS needs to make sure a fair resolution process is available.

Thus far it has resisted calls from local authorities whose populations were undercounted in the last census to create a formal resolution process for disputed statistics. In contrast, the US Census Bureau has been open in its approach to post-publication of results, with Bureau director Robert Groves responding in his blog to inaccurate press stories on census statistics and misinterpretation of data.

With the reductions in the funding available for local government in the coming years, authorities are likely to examine local census population statistics more keenly than ever before, and be more vocal when they find anomalies.

The ONS is clearly confident that its approach will deliver accurate, reliable census results that represent value for money. Local authorities will only share that confidence if it agrees to an open, transparent process through which to settle disputed results.

Rachel Antony-Roberts is research and customer insight manager and assistant census liaison manager in Westminster Council’s communications and strategy unit


9 years ago

Good article that. I know that the for the last census young males in particualr seemed to be 'missing' in high numbers. A new question regarding overnight stay has been introduced to counter this but i'm not sure it will help I wrote a blog last week on how the census highlights similar issues regarding self completion surveys that may be of interest to you http://murrayconsultancy.blogspot.com/2011/03/census-highlights-problems-with-postal.html

Like Report

9 years ago

Why and How is a date such as March 27th chosen? Why not the first of a month? Is there some historiucal reason? Dennis Barton

Like Report