Frequently asked questions

About the first release of 2011 Census data

1.1 Are the unrounded estimates published in September 2012 a revision to those published in July 2012?

No. The unrounded estimates in the September release were a planned update to those published 16 July 2012. The unrounded estimates were made available once the final few stages of data processing were complete. The estimates released in July 2012 were published to provide users with access to the first release of census data and also allowed for them to be used in the preparation of the mid-year population estimates. At the time, ONS explained the need to round these estimates, and began plans to release the unrounded estimates in time for the publication of the mid-2011 population estimates. This release was announced in the ONS release calendar and on the UKSA publication hub in late July.

The differences between the estimates published 16 July and those published 24 September are minimal. The differences for every age/sex estimate at local authority level are always smaller than 100. In fact only three per cent of these estimates are different by more than 50.

1.2 Why were the unrounded estimates published two months after the rounded figures?

At the time the rounded estimates were published on 16 July 2012, a few of the final stages of data processing were yet to be completed, including statistical disclosure control which ensures that no personal information is disclosed and that no individual can be identified in census results.

1.3 Why were the estimates that were published in July 2012 rounded to 100?

The estimates that were published on 16 July 2012 were rounded to the nearest 100 because at that time a few of the later stages of processing had not yet been completed. Rounding protected against any changes that might arise from later stages of data processing, such as statistical disclosure control.

1.4 If I apply rounding to the unrounded estimates published in September 2012, will I notice any differences in the figures compared to those published in July 2012?

The differences between the estimates published 16 July and those published 24 September are minimal. The differences for every age/sex estimate at local authority level are always smaller than 100. In fact only three per cent of these estimates are different by more than 50. As the estimates in the July release were independently rounded, no local authority total is different by more than 100.

1.5 What data has ONS used for comparisons with previous censuses in the first release statistical bulletins?

Comparison with 2001 and 1991 is based on the mid-year population estimates for those years, comparison with 1981 and earlier is based on census results.

For households, comparisons have been made with household estimates from Department of Communities and Local Government (DCLG) for England and the Welsh Government (for Wales). This data is consistent with the 1991 and 2001 mid-year population estimates. For 1981 and earlier, data on households from the relevant census has been used

1.6 Why has ONS not compared the 1991 and 2001 Censuses in the first release statistical bulletins?

The published mid-year estimates are our best estimates of the population at national/local level within the census years for 1991 and 2001 as they contain adjustments made following the 1991 and 2001 Censuses and therefore provide a better indication of population growth.

For households, for 1991 and 2001, the 2011 Census estimates are compared with the household estimates for 1991 and 2001 produced by Department of Communities and Local Government (DCLG) for England and the Welsh Government (for Wales) as these correspond to relevant mid-year population estimates.

For 1981 and earlier, the difference between the census and the mid-year estimates is much smaller than for 1991 and 2001 and so census results have been used for comparison. The 1981 and earlier census are 'population present rather than usual residents'.

About the 2011 Census

2.1 Who was responsible for the 2011 Census?

The census in England and Wales was run by the Office for National Statistics (ONS). Simultaneous, yet separate, censuses took place in Scotland and Northern Ireland. All three censuses took place on 27 March 2011.

More information about the 2011 Census in England and Wales can be found on the ONS website.

2.2 Who did the 2011 Census include?

The 2011 Census set out to include everyone. The main statistical outputs are in respect of usual residents in England and Wales on census day, 27 March 2011. ONS worked with organisations and groups to support key population groups, such as the homeless, in completing their census questionnaires. Visitors and short-term residents (people here for at least three months but less than a year) were also counted

2.3 Why are census statistics called estimates this time?

As with any census there will always be a small percentage of people who will have been missed and an even smaller proportion who will have been counted twice (perhaps because they live in more than one home, for example, students). This is also true of censuses internationally.

The Office for National Statistics (ONS) has used statistical procedures to deal with this, including information from the Census Coverage Survey and other quality assurance measures. Where people have been missed, these measures enable us to estimate their numbers and characteristics and produce estimates of the whole population

2.4 How successful was the 2011 Census?

The 2011 Census in England and Wales was the biggest statistical operation ever undertaken in the UK. It was a great success, delivered on time, on budget and to a high quality.

We made particular efforts to make this the most inclusive census to ensure that we understand more about the characteristics of all sections of the community.

Many new features of the census were designed to improve the quality of the data, and as a result response rates were higher than in 2001 in those areas that are traditionally difficult to enumerate. Statistics from the 2011 Census are available on the ONS website.

2.5 What difference will this census make?

Census statistics provide valuable information for public and private organisations to plan services nationally and in the community over the next 10 years, for example:

• An accurate population count helps the Government to calculate the grants it allocates to each local authority and health authority.

• Data collected and analysed about the age, social and economic make up of the population, and on general health and long-term illness, enables the Government and local authorities to plan and fund health and social services.

• Information about housing and its occupants indicates where accommodation is inadequate and helps in planning new housing.

• Knowing how many people work in different occupations helps government, local authorities and businesses to plan jobs and training policies.

• Information about travel to and from work and car ownership highlights the pressures on transport systems and how road and public transport could be improved to meet local needs.

• Information about ethnic groups helps central and local government to plan and fund programmes to meet the needs of these minority groups.

• Population statistics enable licensed census distributors to create business planning software products.

• In Wales, information on Welsh language speakers helps government and many other organisations, to take account of Welsh language in planning appropriate services for local areas.

• Census statistics helps research organisations to decide how, when and where to capture representative samples.

• Population statistics help businesses to decide where to locate or expand their premises to reflect local demand and the available workforce.

More statistics will become available throughout 2013. Information about the statistics which are being released can be found in the 2011 Census Prospectus.

2.6 How much did the 2011 Census cost?

The 2011 Census cost approximately £480 million. This breaks down to less than £1 per person per year over the 12-year planning and operational cycle of the census. This is very good value and compares favourably with other censuses around the world.

The cost of the census has to be set against its value in helping central and local government to allocate annually many billions of pounds of funding to communities. Without census data, the ability to make informed decisions would be severely hampered.

2.7 How many people completed online?

Some 16.5 per cent of census questionnaires were completed online. For more information see "Providing the online census" report and "Internet take-up rates" on the How did we do in 2011 evaluation page.

About the release of 2011 Census data

3.1 Why does it take more than a year to produce the results?

We have processed more data than any previous census. We have to be sure that the data are of the highest possible quality so we scrutinise them carefully before publication and correct any inconsistencies we find (for example, if the data show that someone is three years old and is a brain surgeon). We also adjust for people who didn’t complete a form. This quality assurance work takes time.

3.2 Why are more detailed results not available sooner?

Although the processing of the data that was collected is now complete, and the estimates of the population have been fully quality assured, there is still work left to do to produce detailed results.

The current phase of statistics covers the full range of topics from the census - providing the most detail possible about each individual topic covered - for a range of areas down to the most local level, the output area.

A number of checks are also needed to further assure the quality of the final detailed results that will be published, and to ensure that the protection of personal information is guaranteed when detailed information is prepared for small local areas. Work is well underway to complete these final steps and begin to publish detailed results in phases from November 2012.

3.3 When will more results from the 2011 Census be published?

ONS will be producing further, more detailed information from the 2011 Census in phases. The phase ends in February 2013. This phase includes the first statistics for local areas. Further phases will include the release of more detailed results later in 2013. Further information about the phasing of releases is available in the 2011 Census prospectus.

About where to find 2011 Census data

4.1 How do I get statistics about my local area?

2011 Census statistics are published on the ONS website.

www.ons.gov.uk/census

The statistics currently available include:

• usually resident census population estimates for England and Wales at regional and local authority level by single year of age and sex

• estimates of occupied households at country, regional and local authority level

• estimates of short-term residents by sex and local authority, and

• estimates of the number of people with second addresses in local authorities in England and Wales. by broad age groups, sex and the reason for the second address reason (working, holiday or other reason)

More statistics will become available throughout 2013. Information about the statistics which are being released can be found in the 2011 Census Prospectus.

About 2011 Census data for the UK

5.1 When will the unrounded population figures for the United Kingdom be released?

ONS is responsible for delivering population estimates for England and Wales, and for collating a UK population estimate.

Population figures for Northern Ireland are produced by Northern Ireland Statistics and Research Agency (NISRA) and for Scotland by National Records of Scotland (NRS).

More statistics will become available later in 2013. Information about the statistics which are being released can be found in the 2011 Census Prospectus.

About differences between the 2011 and 2001 censuses

6.1 How does the 2011 Census compare with the 2001 Census?

The 2011 Census was different to the 2001 Census in several ways. Questionnaires were, in the main, posted to addresses, and people could complete their answers online. There were some new questions, and lots more facilities to help people complete their questionnaire (such as online help, support materials in many languages and a census telephone helpline). All of these helped make the 2011 Census as inclusive as it could be.

More detailed information about the changes is available in the User Guide Comparability Report.

About the 2011 Census in Wales

7.1 Was anything different about the 2011 Census in Wales?

People in Wales were able to choose to complete their questionnaire (on paper or online) in Welsh, if they wished. Enquiry lines, advice and online guidance were also available in Welsh. An extra question was asked of people living in Wales, about the Welsh language.

ONS worked with many organisations in Wales during the development and operation of the 2011 Census, such as the Welsh Language Board (now the Welsh Language Commissioner), local authorities, businesses and voluntary organisations and established a Census Advisory Group for Wales.

More information about the 2011 Census in Wales is available on the ONS website.

About the confidentiality of 2011 Census data

8.1 How does the census ensure confidentiality?

Personal census data are kept confidential at all times and are not shared with other organisations. The data are even exempt from the Freedom of Information Act under section 40 of the Statistics and Registration Service Act 2007 (SRSA).

Information security is fundamental to the 2011 Census. All personal census data are owned by ONS, remains in the UK, and is processed only by ONS or UK/EU companies. Personal census data are protected by law and are not shared with any other government departments, local councils or marketing agencies.

To ensure the confidentiality of personal information, all 2011 Census systems, processes, staff and contractors are bound by the Data Protection Act, the 1920 Census Act and the SRSA.

Some specialist researchers may be allowed limited, supervised access to some records from which all personal identifiers has been removed.

After scanning, all the paper questionnaires were shredded and recycled. Census records are digitally archived and stored in a secure location for 100 years.

About the 2011 Census topics

9.1 About second addresses

9.1.1 Why was information on second addresses collected?

Second address information provides more detail about how we live in the early 21st century. An increasing number of people in the UK have more than one residence in the UK, for example, children of separated or divorced parents; people with a second address for work; and people with holiday homes. This situation led to the need for a new question to collect information on second addresses. Potential users of the data include central government departments, local authorities, academics and businesses. For example, having information on the numbers of people who live in one place but have a second address elsewhere can help local authorities to plan local services.

9.1.2 Why is the information on second addresses being released now?

In addition to the information already published on usual residents and short-term residents, this provides further information about the number of people that may require services in a local authority. Further information will also be published on workday populations, the number of people that travel to work in particular areas, next year.

9.1.3 What information is covered by the statistical release on second addresses?

The 2011 Census captured information on people who stayed at another address for more than 30 days in a year. This includes students living away from home, children who spend time with separated parents, members of the armed forces on military bases who were counted as usually resident at their family home and people who have holiday or second homes.

The statistics cover those who have a second address in a different local authority area to the one in which they usually live. They are grouped under three headings; ‘Working’ (an armed forces base address or another address when working away from home), ‘Holiday’ (a holiday home) or ‘Other’ (a student’s home address, another parent or guardian’s address or other).

As this is the first time that this information has been collected there is no comparative data from earlier censuses.

9.1.4 How do these figures relate to the statistics you published in July and how do they affect the population estimates?

The population statistics published in July 2012 relate to the number of ‘usual residents’ (those who have a main home) in each local authority. Second addresses are addresses that they stay at for more than 30 days a year, that is not their place of usual residence.

People with a second address in a local authority are additional to the usually resident population. For example, The City of London contains the highest proportion of people who say that their main home is elsewhere in England and Wales and that their address within the city is a second address. If all these were to stay there at the same time as all the usual residents in the area then the population of the City of London would increase by 18.52%.

9.1.5 Will ONS publish information about the number and type of second homes?

The 2011 Census asked about people rather than homes. The data do not contain estimates of the number of second homes because more than one person can record the same second address. For example, these addresses could include wholly-owned second homes but could also be the usual address of a relative that someone regularly stays with.

9.1.5 Didn’t ONS publish numbers of holiday homes in 2001?

The 2001 Census did not ask questions about holiday homes. However, the statistics that were published following the census included estimates of the number of holiday homes based on assessments made by census enumerators as to whether an empty property might be a holiday home.

9.1.6 What about people who have a second address within the same local authority area as the one in which they live?

These second address figures do not include information on people who have a second address within the same local authority area. This is being considered for publication at a later date.

9.1.7 Could the statistics tell me where, for example, someone living in Wiltshire had a second address?

No. But flow data, showing where people in a particular local authority had a second address, is being considered for a future release.

9.1.8 What about people with more than one second address?

The census questionnaire only had space for one additional address per person, so information on people with more than one is not available.

9.1.9 Does the "more than 30 days a year" have to be consecutive days?

No, the 30 days can be any time in the year prior to census day.

9.1.10 Can you tell from the 2011 Census how many days people stay at their second address?

No. Although a question to collect the length of time spent at the second address was considered when the questionnaire for the 2011 Census was being developed, it was not recommended for inclusion because extensive testing found that the data collected would not be of sufficient accuracy or consistency.

9.1.11 Can you tell from the 2011 Census how far away people travel to their second address?

This information is not currently available, but is being considered for a later release.

9.1.12 How does this information on second addresses compare with other sources of data on second addresses?

There are no comparable data regarding people with second addresses - these questions were included in the 2011 Census to help meet new user requirements for such information.

9.1.13 How accurate is the data on second addresses?

We are confident that these data are accurate. Census response rates were high, with around 19 out of every 20 people in England and Wales completing their questionnaire, and extensive testing in the years before the census demonstrated that these questions, and all others on the questionnaire, were well understood and answered by respondents.

9.1.14 When will UK figures about second addresses be available?

UK statistics about people with second addresses will not be available because this information was not collected in the censuses in Scotland and Northern Ireland.

The censuses in Northern Ireland and Scotland are the responsibility of Northern Ireland Statistics and Research Agency (NISRA) and National Records of Scotland (NRS) respectively, and their user consultation prior to the 2011 Census identified a greater requirement for information on other topics.

9.1.15 Will information about second address abroad be available?

This information may be published at a later date.

9.1.16 Why has ONS not published the number of people with second addresses for more age groups?

The second address data were divided into broad ages to distinguish between children, people of working age and people more likely to be retired. This information will meet key user requirements, but a more detailed breakdown is being considered for publication at a later date.

9.1.17 Why has ONS not published a more detailed breakdown of those with "other" second addresses?

The second address data were divided into broad categories to distinguish between people using an address for work, for holidays or for another purpose. This information will meet key user requirements, but a more detailed breakdown is being considered for publication at a later date.

9.1.18 Why is information on those not usually resident in England and Wales (including those from Scotland or Northern Ireland) who have a second address in England and Wales not included?

This information is not collected.

About 2011 Census small area statistics

10.1 What is the geographical hierarchy of small areas?

Output Areas (OAs) are the lowest level of geography at which census statistics are produced. They are composed of aggregations of unit-level postcodes. Aggregations of approximately five OAs are then used to produce a Lower Super Output Area (LSOA) layer. Like OAs, they have minimum and target populations (and households) and aggregating OAs ensures they are consistent.

Aggregations of approximately five LSOAs are then used to produce a Middle Super Output Area (MSOA) layer. Like OAs and LSOAs they have minimum and target populations (and households) and aggregating LSOAs ensures they are consistent.

Together OAs, LSOAs and MSOAs form the geographical hierarchy of small areas. It is also known as the statistical building block hierarchy. They are designed so that statistics can be published at the smallest geographical level possible (OA, LSOA, MSOA) without the risk of identifying a person or a household.

The entire small-area hierarchy nests within the Local Authority District (LAD) boundaries to allow exact-fit estimates to be produced at the LAD level.

10.2 What is an Output area (OA)?

Output areas (OAs) are the lowest level of geography at which census estimates are produced. OAs contain approximately equal numbers of population/households, and are intended as geographies that allow reporting of statistics across time on a consistent geographical base.

Using 2001 Census data, 175,434 OAs were created for England and Wales with an average of 297 persons and 123 households in each OA. To prevent statistical disclosure when releasing estimates at OA level, each OA was designed to have a minimum population (100 persons) and number of households (40 households) within it. OAs were intended to be a stable geography and did not change until the 2011 Census when around 2.6% of the 2001 OAs were changed to reflect the change in populations since the 2001 Census. There are now 181,408 OAs in England and Wales. More information is available in A Beginner's Guide to UK Geography.

10.3 What is a Lower Layer Super Output Area (LSOA)?

Lower layer super output areas (LSOAs) are groups of around five outputs areas (OAs)

10.4 What is a Middle Super Output Area (LSOA)?

Middle layer super output areas (MSOAs) are groups of around five Lower Layer Super Output Area (LSOAs).

10.5 What is a ward?

Wards are electoral areas represented by one or more local government councillors. In Wales, the Isle of Wight and several of the new unitary authorities created as part of the LGR in 2009, the equivalent areas are legally termed 'electoral divisions', although they are frequently also referred to as wards. Wards change so that they contain roughly equitable numbers of electors in each. This can make time series comparisons difficult.

10.6 Why are output areas (OAs) needed?

In the UK, there is a high level of change to the administrative geography. For example wards and parliamentary constituencies change on a rolling programme so that they contain roughly equal numbers of electors. These changes make it difficult for users to compare data for small areas over time. OAs were designed in such a way to provide users with a stable geography base that would allow reporting of statistics across time on a consistent geographical base.

OAs have been designed around target populations (and households), and have a minimum size (100 persons, 40 households) to prevent disclosure of confidential information in estimates released for them. They are also designed to try to group together households with similar characteristics (using census variables “tenure” and “accommodation type”).

As they are a small, non-disclosive, socially homogeneous geography, purpose built for the publication of statistics, they are consistent, comparable and stable to an extent not provided by any other level of UK geography.

All national statistics for any geography should be best-fit from OAs. Therefore OAs are the core statistical geography of national statistics, and underpin the Geography Policy for National Statistics.

10.7 What do electoral wards/divisions have to do with OAs?

When output areas (OAs) were created in 2003, there was still a requirement to analyse some statistics at the electoral ward/division level. Ward boundaries were therefore used as a constraint in the OA design to allow exact-fit estimates to be produced by aggregating OA level estimates to the ward that they fitted within.

This was known to be a short-term solution, as the high level of change at the electoral ward/division level means that it would not be possible to keep OAs and wards aligned, and maintain the core value of stability for time series. For 2011 Census therefore, OAs will not be aligned with electoral ward/division boundaries and there will be no relationship between the two geographies other than a residual one from 2003.

10.8 What was the average size of an OA/LSOA/MSOA in 2011?

A 2011 OA contains an average 309 persons (297 in 2001), a 2011 LSOA had 1614 persons (1514 in 2001), and a 2011 MSOA had 7,787 persons (7235 in 2001).

10.9 What was the average household count of an OA/LSOA/MSOA in 2011?

A 2011 OA contains an average of 129 households (123 in 2001), a 2011 LSOA contains an average of 672 households (630 in 2011), and a 2011 MSOA contains an average of 3,245 households (3011 in 2001).

10.10 Why are OAs more stable than other geographies?

Unlike other geographies for which statistics are produced, OAs, LSOAs and MSOAs were purpose built by ONS for statistics. So while other geographies change (for example wards and parliamentary constituencies change on a rolling programme so that they contain roughly equal numbers of electors) OAs and LSOAs are kept stable so that statistics released for them can be better compared over time.

10.11 If OAs are supposed to be stable, why have they changed?

The vast majority (97.4%) of 2011 output areas (OAs) remain the same as they did in 2001, ensuring that their geography continued to be stable. Those OAs that did change have populations that are broadly in line with the rest of the OAs. OAs were designed to be a ‘stable’ geography for the publication and analysis of time-series statistics when they were created in 2003. This was part of a broader vision to produce a geography that is also consistent across England and Wales and comparable. OAs are still considered to be stable, but changes in the population and geography of England and Wales during the period between censuses means that they are no longer as consistent and comparable as when they were created in 2003. A decision was therefore taken to change only those where the population had changed significantly so that they could continue to be consistent and comparable as well as stable.

10.12 How have the small area geographies changed since 2001?

Around 2.6 per cent of 2001 output areas have changed in the 2011 Census OAs, along with 2.5 per cent of LSOAs, and 2.1 per cent of MSOAs. The average number of persons in an OA has increased from 297 in 2001 to 309 in 2011, and the average number of households in an OA from 123 in 2001 to 129 in 2011.

Of 348 local authorities, 22 had changes to 5 per cent or more of their OAs, 54 had changes to 5 per cent or more of their LSOAs, and 61 had changes to 5 per cent or more of their MSOAs. Six local authorities had no changes to any of their OAs, LSOAs, or MSOAs.

A total of 161 OAs and SOAs were changed because they were considered unsuitable for statistical outputs (see question “What makes a geography unsuitable for statistical outputs?” for more information). To see the link between 2001 and 2011 small-area geographies, and the reason for their change, select the appropriate output-area lookup file available online from ONS Geography.

10.13 How has the confidentiality of data at small areas been protected?

When publishing statistics for small areas, low counts may be present in the data which could risk an individual being identified. To avoid this, statistical disclosure control (SDC) has been applied to protect against disclosing information about any individual. SDC uses a method of record swapping. Every household has a probability of being selected for record swapping and some records are created by imputation by the statistical methodology used to create census estimates. The method has been designed to ensure there is sufficient doubt as to whether a value of one in a census table is a true value or one that has been created by imputation or swapping persons in or out of that cell.

The use of the best-fit method also contributes to the confidentiality of data by ensuring that all statistical outputs use aggregations of whole OAs as the basis for their estimates. This means that slivers created by overlapping geographies do not reveal confidential data. More information is published in An overview of best fitting.

10.14 Can you give advice on time-series comparison of small areas between 2001 and 2011?

Changes to the 2001 OAs were kept to a minimum with only 2.6% of OAs being changed using the 2011 Census data. Of the 175,434 OAs created in 2001:

• Unchanged: A total of 170,865 (97.4%) of the 2001 OAs remain unchanged. This means that direct comparisons can be made between these 2001 and 2011 OAs.

• Split: 3,239 (1.8%) 2001 OAs have been split into two or more 2011 OAs. This means direct comparisons can be made between estimates for the single 2001 OA and the estimates of the two or more 2011 OAs, aggregated together.

• Merged: 1,115 (0.6%) 2001 OAs have been merged with one or more other 2001 OAs. This means direct comparisons can be made between the estimates from the multiple 2001 OAs, aggregated together, and the single 2011 OAs estimates. Each original 2001 OA that was merged to other OAs into a new 2011 OA was counted towards the merge statistics (e.g. three 2001 OAs merged into one 2011 OA counted as three affected OAs).

• Complex correction: 215 (0.1%) of 2001 OAs have been redesigned because of local authority boundary changes, and/or to improve their social homogeneity. These can’t be easily mapped to equivalent 2011 OAs, as they may also have been subsequently changed because of population changes, or may have been split and merged at different processing stages. Therefore like-for-like comparisons of these 2001 and 2011 OAs and SOAs are not possible.

To see the link between 2001 and 2011 small-area geographies, and the reason for their change, select the appropriate output-area lookup file available online from ONS Geography.

10.15 Where do I find the details of the areas that have changed?

To see the link between 2001 and 2011 small-area geographies, and the reason for their change, select the appropriate output-area lookup file available online from ONS Geography.

10.16 What makes a geography unsuitable for statistical outputs?

If there's a demand for it, statistics can be produced for any geography, but there are a number of things to consider:

• If a geography is too small, e.g. postcode, estimates released for it may be disclosive.

• If it changes a lot, e.g. wards, it may be difficult to compare change over time.

• If a geography doesn't have a defined set of boundaries, or an agency responsible for them to define their currency, e.g. the 2012 set, 2011 set, then statistics released on them may be different as there is no official definition of the geography.

• If the geography is not supported by ONS, then referencing to an unsupported geography may be done inconsistently between statistics producers and therefore the statistics may not be comparable and consistent.

10.17 How is average household size calculated?

Average household size is the total number of household residents divided by the total number of households occupied by at least one usual resident.

10.18 Output areas (OAs) should contain at least 100 people to protect against disclosure. Are all the OAs above this threshold? If not, what are you doing to protect against disclosure in these areas?

There is a very small number of OAs that have fewer than 100 usual residents. All of these are above the threshold of 40 households. All census data have previously been subject to record swapping, targeted to records more likely to be prone to disclosure. Where OAs have small population sizes, the percentage of records swapped is likely to be larger in order to offer greater protection to the expected higher percentage of records likely to be risky.