Official Statistics in Sociology

Last Updated on October 10, 2026 by Karl Thompson

Official statistics are numerical data collected and published by the government and its agencies, such as the Office for National Statistics (ONS), the Home Office and the Department for Education. Sociologists use them as a source of secondary quantitative data.

This post covers:

1. What official statistics are, and where they come from
2. The difference between “hard” and “soft” statistics
3. How positivists, interpretivists, Marxists and feminists see them
4. The theoretical, practical and ethical strengths and limitations of official statistics
5. Worked examples of how valid different official statistics are, with a ranking exercise
6. Whether official statistics can be seen as “social facts”
7. How official statistics come up in the exam

Official statistics are one of the main methods you need to know for the research methods part of A-level sociology. They can come up in the theory and methods section of Paper 3 and Paper 1, and as the method in a methods-in-context question on Paper 1.

Where official statistics come from

Official statistics are collected in two main ways.

Registration. The law requires some events to be registered: births, deaths and marriages, for example. Because registration is compulsory, these records cover almost the whole population.

Official surveys. The government also runs surveys. The biggest is the Census, which is sent to every household in the country every ten years. The first official census in England and Wales was in 1801, and the 1841 Census is usually seen as the first “modern” one, because it was the first time each household filled in its own form. Other official surveys are carried out on samples, such as the Crime Survey for England and Wales (CSEW), which asks people about crimes they have experienced.

A third source is administrative data: information the state collects anyway as it runs public services. Police recorded crime, exam results and the school census all come from records kept by police forces, exam boards and schools.

Not all statistics are official. Charities, pressure groups, think tanks and businesses also produce statistics. These are non-official statistics, and many of the same strengths and limitations apply to them, with the added question of what the organisation wants the numbers to show.

Accredited official statistics

UK official statistics carry a label. Accredited official statistics have been independently reviewed by the Office for Statistics Regulation (OSR) and found to meet its standards of trustworthiness, quality and value. Before September 2023 these were called “National Statistics”. Official statistics without the accreditation have not passed that review, and official statistics in development are new or changing statistics that are still being worked on.

The label is a useful point to make in an answer, because the state itself is admitting that some of its statistics are more trustworthy than others. Two examples:

• In January 2014 the UK Statistics Authority removed the National Statistics status from police recorded crime, saying there was accumulating evidence that the underlying data may not be reliable. Police recorded crime is still published only as official statistics, not accredited ones.
• In November 2023 the OSR removed accreditation from estimates based on the Labour Force Survey, the main source of unemployment and employment figures, because falling response rates had damaged their quality.

Hard and soft statistics

Sociologists often distinguish between “hard” and “soft” official statistics.

Hard statistics measure things that are relatively easy to define and count, where there is little room for anyone to interpret what counts. Births, deaths, marriages and the prison population are good examples: a person is either in custody on the day of the count or they are not.

Soft statistics measure things that are harder to define, or where the figure depends on decisions made by people along the way. Crime, unemployment, poverty and special educational needs are examples. A crime only becomes a statistic if a victim notices it, reports it, and the police decide to record it. Each step is a decision someone makes, and each one can go differently.

The distinction is not absolute. Even the “hardest” statistics involve some interpretation (what counts as a “household” in the Census, for example), but it helps explain why some official statistics are much more trusted than others.

Sociological perspectives on official statistics

The positivist view

Positivists favour official statistics. They see them as objective social facts: features of society that exist independently of individuals and can be measured, much as a scientist measures the natural world.

Official statistics give positivists what they want from research: large, representative data sets that are collected in a standardised way, which makes them reliable and allows patterns, trends and correlations to be identified. Positivists then use these patterns to test hypotheses and look for cause-and-effect relationships.

The classic example is Durkheim’s study of suicide (1897). Durkheim used official suicide statistics from several European countries to show that suicide rates varied consistently between social groups, for example between Protestants and Catholics, and argued that these differences were caused by social factors such as levels of social integration and regulation. For Durkheim, the suicide rate was a social fact, to be explained by other social facts. See Positivism and Interpretivism in Sociology and What are Social Facts? for more.

The interpretivist view

Interpretivists argue that official statistics are not objective facts but social constructions. They are the end product of a long chain of decisions and interpretations made by the people who produce them.

Atkinson (1978) studied how coroners decide whether a death was a suicide. He found that coroners used their own common-sense ideas about what a “typical” suicide looks like, such as whether there was a note, the way the person died, where it happened and their recent life history. A death that fitted the coroner’s picture was more likely to be recorded as suicide. On this view, suicide statistics tell us how coroners classify deaths, not what the “real” suicide rate is.

Douglas (1967) made a related point: families may try to prevent a death being recorded as suicide, and some social groups may be better at this than others. If more integrated communities are better at hiding suicides, then Durkheim’s finding that integration reduces suicide could be produced by the statistics themselves.

The same argument applies to crime. Police recorded crime depends on what the public reports and what officers choose to record. According to the CSEW, only an estimated 36% of crimes it measures (excluding fraud and computer misuse) were reported to the police in the year ending March 2026. Interpretivists would add that statistics can tell us how many, but not why. They say nothing about the meanings behind the numbers.

The Marxist view

Marxists argue that official statistics are not neutral. They are collected by the state, and Marxists see the state as serving the interests of the ruling class. That shapes the statistics in two ways.

First, the state decides what to measure and how to define it. Irvine, Miles and Evans (1979), in Demystifying Social Statistics, argued that statistics should be treated as social products, shaped by the purposes of those who produce them. What is not measured cannot easily become a political issue.

Second, statistics can serve an ideological function. Marxists point out that corporate and white-collar crime is largely absent from the crime statistics, which helps reinforce the idea that crime is mainly a working-class problem. See Official Statistics on Ethnicity and Crime for how this argument applies to the criminal justice system.

Hindess (1973) offers a more moderate position. He argued that statistics are produced through “conceptual and technical instruments” (the definitions and procedures used to collect them), and that sociologists can use them as long as they understand and allow for how those instruments shaped the figures.

The feminist view

Feminists argue that official statistics have often ignored or under-counted women’s experiences. Domestic abuse is the clearest example. Much of it never reaches the police, so police figures understate it, and feminists have campaigned for better ways of measuring it, such as the CSEW’s self-completion questions on domestic abuse. Feminists also point out that unpaid work in the home, most of which is still done by women, is not counted in standard economic statistics such as GDP.

Theoretical factors

Theoretical advantages

They give an overview of society. Official statistics make it easy to get a picture of social life in Britain: how many people there are, how they live, what they earn, how well they do at school. The ONS website is the obvious starting point.

They allow comparisons between social groups and regions. The Census is a good example. Because it covers almost every household (the return rate from occupied households in England and Wales was over 97% in 2021), it allows comparisons between social groups and local areas that no sample survey could match.

They allow historical comparisons. Many official statistics go back a long way. The Census has been carried out every ten years since 1801 (except 1941), the British Crime Survey (now the CSEW) was first carried out in 1982, and school performance tables have been published since 1992.

Some large data sets would not exist without the state. Individual researchers and universities do not have the money to run something on the scale of the Census, and private companies would only collect data that is profitable for them.

They are representative and reliable. Large samples, or the whole population, make official statistics highly representative. They are collected in a standardised way by trained staff following set procedures, which makes them reliable, and other researchers can check the same figures.

Positivists favour them. For the reasons above, official statistics allow researchers to spot trends, find correlations and make generalisations, while the researcher stays detached from the data. That means less room for the researcher’s own bias to affect the results.

Theoretical disadvantages

Some official statistics lack validity. Crime statistics are the obvious example. Many crimes are never reported to the police, and some crimes, such as rape and domestic abuse, are notoriously under-reported. The gap between what happens and what is recorded is known as the dark figure of crime.

The way things are measured changes over time, which can make historical comparisons difficult. Police recorded crime is a good example. When the National Crime Recording Standard was introduced in April 2002, the number of crimes recorded rose, so the ONS mainly reports police trends only from the year ending March 2003 onwards. Recording has changed again since, for example in how fraud is counted.

The state may present statistics in the most favourable light. Because governments are judged on what the statistics show (crime, unemployment, waiting lists, exam results), there is pressure on the people producing them. The 2014 decision to remove National Statistics status from police recorded crime followed concerns that recording standards had been slipping.

Marxist and feminist sociologists argue that official statistics reflect the interests of powerful groups. Data tends to be collected on the things the state is interested in. Marxists argue that corporate and financial crimes of elites get little attention, while feminists argue that domestic abuse was not taken seriously by the state for a long time.

Official statistics may reflect the biases of the agencies that produce them. Black people are over-represented at most stages of the criminal justice system, which could be read as showing higher levels of offending. However, labelling theorists and critical criminologists argue that over-representation is at least partly produced by the way the system works. In the year ending March 2023 there were 24.5 stop and searches for every 1,000 Black people in England and Wales, compared with 5.9 for every 1,000 White people. If the police focus more stop and search on some communities, more offences will be found there, and the arrest and prison statistics will reflect where the police looked as much as who offended. See labelling theory for more on this argument.

Statistics tell us “what” but not “why”. Even valid statistics only describe patterns. They cannot tell us the meanings behind them, which is the interpretivist objection.

Practical factors

Practical advantages

Many official statistics are freely available to researchers and the public. This is a clear advantage over privately collected data. Facebook and Amazon hold huge amounts of data on individuals, but they are not going to share it for free! Official statistics are more likely to be shared because they are paid for by taxes.

They are quick and easy to access. Most are published online, through the Office for National Statistics for example, and you can do comparative research from home without needing any “people skills”.

They allow “before and after” studies. Because many statistics are collected regularly, researchers can compare figures before and after a change in policy to see whether it had an effect.

Practical disadvantages

They are free to use but very expensive to collect. The Census and the large official surveys cost the state a great deal of money and staff time.

They are collected for the state’s purposes, not the sociologist’s. The categories and definitions used may not match what a researcher wants to study. Social class is a good example. The Department for Education does not record pupils’ social class at all. It uses eligibility for free school meals instead, which is a measure of poverty, not of class in the sociological sense.

Definitions differ, and change. Different agencies may define the same thing differently, and definitions change over time. That makes comparisons harder, both over time and between countries. See Cross National Comparisons.

Survey response rates are falling. Official surveys depend on people agreeing to take part. The falling response to the Labour Force Survey, which led to its accreditation being removed in November 2023, shows that even large government surveys can become unrepresentative.

Ethical factors

Ethical advantages

There are few ethical problems in using them. The data has already been collected and published, and it is anonymised, so the researcher does not need to gain consent or deal with participants directly.

They are collected in the “national interest” rather than for one researcher’s interests, or for profit, as with privately collected data.

They make public bodies accountable. Official statistics let us check how public services such as the police and schools are performing, and whether taxpayers’ money is being spent well.

Ethical disadvantages

Collecting some statistics can have harmful effects. The publication of school performance tables has been criticised for encouraging teaching to the test, narrowing the curriculum and making school more stressful for both pupils and teachers.

Collecting statistics can be a form of surveillance and control. Data on school performance allows the state to control teachers, while data on individual pupils allows “problem pupils” to be identified and managed from a young age.

Anonymity is not always guaranteed. When statistics are broken down into very small groups or local areas, it can become possible to identify individuals.

Summary table: strengths and limitations of official statistics

Factor	Strengths	Limitations
Theoretical	Large, representative data sets; standardised, so reliable; allow comparisons between groups, areas and over time; favoured by positivists	Many are socially constructed (interpretivists); soft statistics lack validity, e.g. the dark figure of crime; definitions change over time; may reflect the interests of the powerful (Marxists, feminists); show “what” but not “why”
Practical	Free and quick to access; already collected; cover the whole population; allow before-and-after studies	Collected for the state’s purposes, not the sociologist’s; categories may not fit the research; expensive for the state to produce; falling survey response rates
Ethical	Anonymised, so no consent issues; collected in the national interest; make public bodies accountable	Collecting them can harm those measured (e.g. league tables); can be a tool of surveillance; small breakdowns can identify individuals

How valid are different official statistics? Worked examples

The validity of official statistics varies enormously between different types of statistic. Some are hard statistics, with little disagreement over how to measure them. Others are soft statistics, where there is much more disagreement over the definitions of what is being measured. Below I look at some of the main official statistics you are likely to use, and how far you can trust them.

Ethnic group in the 2021 Census

In the 2021 Census, 81.7% of usual residents in England and Wales identified as White (74.4% as White British), and 9.3% identified as Asian, Asian British or Asian Welsh.

How valid is this? To an extent, ethnicity is a fairly objective matter. However, the Census only records which box people ticked. It tells us nothing about what that identity means to them. For some people their ethnic identity is central to how they see themselves, and for others it means very little. The categories themselves are also decisions made by the ONS, and they have changed from one census to the next.

Religion in the 2021 Census

In the 2021 Census, 46.2% of people in England and Wales described themselves as Christian, down from 59.3% in 2011. This was the first census in which fewer than half did so. 37.2% said they had no religion, up from 25.2% in 2011.

How valid is this? Statistics on religious identity may lack validity as a measure of religious belief. What does ticking “Christian” actually mean? Church attendance is much lower than 46%, so the “Christian” box covers everything from devout believers to people who tick it out of habit or cultural identity. The religion question is also voluntary. 94% of people answered it in 2021, so around 6% of the population is not counted at all. See the Beliefs in Society module for how this affects the secularisation debate.

Employment and unemployment

Employment and unemployment figures come mainly from the Labour Force Survey, a large household survey.

How valid are these? Measuring whether someone is in work sounds straightforward, but these figures depend on people taking part in the survey and on how “employed” and “unemployed” are defined. In 2023 response rates fell so low that the ONS suspended publication for a time, and the OSR removed the survey’s accredited status. Employment figures will also miss people working cash-in-hand in the informal economy.

Crime: the Crime Survey for England and Wales

The CSEW is a victim survey, carried out by interview with a large sample of households. It estimated 9.6 million incidents of crime in the year ending March 2026, including fraud and computer misuse. Police recorded crime, by comparison, was around 5.2 million offences, excluding fraud and computer misuse.

How valid is this? The CSEW is almost certainly more valid than police recorded crime, because it includes crimes never reported to the police. It is an accredited official statistic, while police recorded crime is not. HOWEVER, the CSEW still misses a great deal. It does not cover homicide, crimes against businesses, “victimless” crimes such as drug possession, or people living in institutions such as care homes. Domestic crimes may also be under-reported because the offender may be in the house during the interview. For much more on crime statistics, see the Crime and Deviance module.

The prison population

The prison population in England and Wales was 87,342 at the end of March 2026.

How valid is this? It is hard to argue with the validity of prison population figures. Someone is either held in custody or they are not, so this is a good example of a hard statistic. That does assume open and due legal process. In some countries, prison statistics are far less trustworthy.

Net migration

The ONS estimated net migration to the UK (immigration minus emigration) at 171,000 in the year ending December 2025, down from 331,000 the year before.

How valid is this? Migration figures are among the softest official statistics. These estimates are based largely on Home Office visa and travel records and other administrative data. They are labelled “official statistics in development”, they stay provisional for a year, and earlier estimates are regularly revised, sometimes by large amounts. The ONS gives a range of 145,000 to 197,000 for this one. People who enter or stay without the authorities’ knowledge are, by definition, not counted.

Ranking exercise

Rank the following official statistics in terms of validity. Which is closest to reality, and which is furthest from it? Give a reason for each.

• The prison population in England and Wales was 87,342 at the end of March 2026
• Net migration to the UK was 171,000 in the year ending December 2025
• The CSEW estimated 9.6 million incidents of crime in the year ending March 2026
• The police recorded around 5.2 million crimes in the year ending March 2026
• 46.2% of people in England and Wales described themselves as Christian in 2021
• 81.7% of people in England and Wales identified as White in 2021

You might also like this official statistics starter activity, which tests how quick and easy official statistics really are to use.

Are official statistics social facts?

This is one of the most important debates about official statistics, and it has come up as an exam question (see below).

Positivists say yes. Official statistics, especially hard statistics, measure real features of society that exist independently of individuals. They are collected in standardised ways and can be checked and repeated, so they can be treated as social facts and used to find causes, as Durkheim did with suicide.

Interpretivists say no, for two main reasons.

First, official statistics are socially constructed. They are the product of decisions made by the people who collect them: coroners deciding whether a death was suicide (Atkinson), victims deciding whether to report a crime, and police officers deciding whether to record it. The statistics measure those decisions as much as the behaviour itself.

Second, they are shaped by the purposes of the state. The state decides what to measure, how to define it and when to change the definition. Marxists would add that what the state chooses to count, and what it ignores, reflects the interests of the powerful.

A middle position. Many sociologists today take a position closer to Hindess. Some official statistics are much more valid than others, so the question is not “facts or constructions?” but “which statistics, measured how?”. Researchers can use them carefully, alongside other methods, as long as they understand how they were produced. The accredited official statistics label is itself an admission by the state that not all of its statistics are equally trustworthy.

Evaluating official statistics: a checklist for any example

In the exam you may be asked about official statistics on a topic this post doesn’t cover. These questions work for any example:

• Is it a hard or a soft statistic? How much room is there for interpretation in deciding what counts?
• How was it collected? By registration, by survey, or from administrative records? Is it the whole population or a sample?
• Who decided what counts? Victims, police officers, coroners, teachers, survey respondents? What could lead them to decide differently?
• What does it leave out? Is there a “dark figure”? Who is not covered?
• Has the definition or method changed over time? Can you really compare this year with ten years ago?
• Whose interests might it serve? Would anyone benefit from the figure being higher or lower?
• Is it accredited? Has the Office for Statistics Regulation reviewed it?
• What can’t it tell you? What would you need qualitative research to find out?

Official statistics in the exam

Official statistics can come up in several places on the AQA A-level papers:

• The theory and methods 10-mark question on Paper 3 or Paper 1 (“Outline and explain two…”). There are no marks for evaluation on these questions, so develop two points fully rather than listing several.
• The methods-in-context question on Paper 1, where you apply official statistics to a particular issue in education. See Official Statistics on Educational Achievement.
• Essay questions on the relationship between positivism, interpretivism and choice of method.

Two recent AQA 10-mark questions on official statistics:

Outline and explain two reasons why some sociologists choose to use official statistics in their research. (10 marks) (Paper 3, June 2024). Good reasons to develop are the large, representative data sets that allow comparisons between groups and over time, and the practical advantage that the data is free and already collected. Explain why each would appeal to a sociologist, and link it to positivism. For a full worked answer, see this 10-mark question post.

Outline and explain two reasons why official statistics should not be seen as social facts. (10 marks) (Paper 1, June 2022). Use the two interpretivist reasons in the section above: statistics are socially constructed through the decisions of the people who produce them, and they are shaped by the purposes and definitions of the state. Develop each with an example, such as Atkinson’s coroners or the police decision to record a crime.

For more on how to answer these questions, see my exam advice page.

Theory and Methods A Level Sociology Revision Bundle

If you like this sort of thing, then you might like my Theory and Methods Revision Bundle – specifically designed to get students through the theory and methods sections of A-level sociology Papers 3 and 1.

[IMAGE: existing bundle cover, linked to https://revisesociology.sellfy.store/p/2qyB/. Alt text: “Theory and Methods A-level sociology revision bundle cover”]

1. 74 pages of revision notes
2. 15 mind maps on various topics within theory and methods
3. Five theory and methods essays
4. “How to write methods in context essays”

Signposting and Related Posts

This post has been written primarily for students studying research methods as part of the AQA’s A-level sociology course.

Official statistics are a type of secondary quantitative data. You might also like to read about secondary qualitative data, which is usually taught alongside them.

For how official statistics are used to research education, see Official Statistics on Educational Achievement. For crime, see Official Statistics on Ethnicity and Crime.

Cross National Comparisons looks at how official statistics are used to compare countries, starting with Durkheim’s study of suicide.

For the wider debate about methods, see Positivism and Interpretivism in Sociology and Types of Research Methods in Sociology.

Return to the Research Methods hub.

Please click here to return to the homepage – ReviseSociology.com

---

Sources

• ONS (2026) Crime in England and Wales: year ending March 2026
• ONS (2026) User guide to crime statistics for England and Wales
• ONS (2022) Ethnic group, England and Wales: Census 2021
• ONS (2022) Religion, England and Wales: Census 2021
• ONS (2026) Long-term international migration, provisional: year ending December 2025
• Ministry of Justice (2026) Offender management statistics quarterly: January to March 2026
• Ethnicity facts and figures (gov.uk) Stop and search
• Office for Statistics Regulation Accredited official statistics
• Office for Statistics Regulation Statement on Labour Force Survey and Annual Population Survey derived estimates
• ONS Story of the Census