correlations - ReviseSociology

Nine criticisms of the theory that school exclusions are to blame for knife crime

Last week, senior police chiefs wrote to Theresa May arguing that there was a link between the increase in the number of formal school exclusions and ‘off-rolling’ (where heads informally get parents to withdraw their children, without them being formally recorded as ‘permanently excluded’) and an increase in knife crime.

The theory is that those excluded or off-rolled are more likely to ‘drift’ because they are less effectively cared for and monitored in alternative provision institutions. The problem is believed to be especially bad for those who are off-rolled. When a pupil is off-rolled, the parents are responsible for finding alternative provision, and it is their kids who are much more likely to end up out of education altogether.

If we look at the stats, there does seem to be a correlation between the increase in school exclusions and the increase in knife crime:

School exclusions have been increasing since 2013

Knife crime has been increasing since 2015

And regionally:

HOWEVER, it is a well-known mantra in sociology that correlation doesn’t mean causation, and there is very good reason to think that this is the case here.

Numerous commentators (see below for links) have criticised the police for suggesting there is a causal link between the increase in exclusions and the increase in knife crime, and here’s a summary of why we should be critical…

Nine criticisms of the ‘school exclusions cause knife crime’ theory

For starters, even with the above crude statistics there isn’t a perfect correlation – it’s true that London has a higher exclusion rate and knife crime rate than any other city, but then the West Midlands has a higher knife crime rate than Yorkshire and Humber, but a lower exclusion rate.
The above data only includes formal exclusions, not off-rolling, so we don’t get a full picture (there are validity issues) – true, it might be more likely that someone who is off-rolled turns to knife crime compared to someone who is formally excluded, but I these figures don’t show us the off-rolling.
This government report from June 2018 which examined the relationship between educational background and knife crime found that ‘knife possession rarely followed exclusion’.

There may be another cause behind both ‘being excluded from school’ and ‘being convicted of knife crime’ – possibly rooted further in the past of these individuals, such as their having come from a troubled family and/ or having experienced neglect or abuse during their childhood.
It is unfair to blame schools for excluding children in greater numbers as they have been hit by 10 years of Tory funding cuts – schools actively educate about not getting involved in knife crime, but have become less effective at dealing with ‘troubled kids’ because they now have fewer resources to help them do so.
The fact that someone has previously been excluded from school may make it more likely that they are going to get a knife-crime conviction – being excluded from school puts you on the police radar and doesn’t sit well with judges and juries. It could be that there are proportionally just as many people who have not been excluded from school who commit knife crime, but they just don’t appear in the official statistics because they are less likely to get caught and convicted.

Back to underlying causes, it’s possible that a ‘deeper’ reason lying behind why people who are excluded from school are also more likely to appear in the knife crime conviction figures is because they are victims of discrimination by the system – males, the poor, and African Caribbean children are more likely to appear in both the exclusion figures and the knife rime conviction figures – it could be that both are caused by a sense of injustice at being excluded in the first place.
The stats available to us tell us nothing about the life-histories, or the journies people take from being excluded (or not) to knife-crime. This could be a more complex few years than we imagine, and these possibly diverse journies are simply not going to be unveiled by crude statistical analysis. The data simply isn’t there!
Finally, there are number of other variables that cause knife crime to increase – the changing nature of drug-dealing (county lines), and cuts to police funding come to mind as being two of the most obvious. These would somehow need to be factored in to any ‘causal’ equation.

In conclusion it’s a well-known mantra in sociology that correlation does not mean causation, and this particular topic is a great one to use to illustrate this.

To my mind there are so many problems with maintaining the causality argument here that the only possible reason anyone would try to make it in the first place is to distract attention away from all the other social problems that correlate with the increase in knife crime – the kind of problems government policies either exacerbate or can do little to combat.

Relevance to A-level Sociology: this is a great topic that bridges education, methods and crime!

Sources not already linked above

Huffington Post – Don’t blame school exclusions for rise in knife crime

The Guardian – Knife crime and exclusions are a symptom of wider malaise.

Evaluating the Usefulness of Official Statistics

Official Statistics are numerical data collected by governments and their agencies. This post examines a ranges of official statistics collected by the United Kingdom government and evaluates their usefulness.

Click the image to search 13, 848 official statistics produced by the U.K. government

The aim of this post is to demonstrate one of the main strengths of official statistics – they give us a ‘snap shot’ of life in the U.K. and they enable us to easily identify trends over time.

Of course the validity and thus the usefulness of official statistics data varies enormously between different types of official statistic, and this post also looks at the relative strengths and limitations of these different types of official statistic: some of these statistics are ‘hard statistics’, they are objective, and there is little disagreement over how to measure what is being measured (the number of schools in the U.K. for example), whereas others are ‘softer statistics’ because there is more disagreement over the definitions of the concepts which are being measured (the number of pupils with Special Educational Needs, for example).

If you’re a student working through this, there are two aims accompanied with this post:

Before reading the material below, play this ‘U.K. official statistics matching game’, you can also do it afterwards to check yer knowledge.
After you’ve read through this material, do the ‘U.K. official statistics validity ranking exercise’.

Please click on the images below to explore the data further using the relevant ONS data sets and analysis pages.

Ethnic Identity in the United Kingdom According the U.K. 2011 Census

U.K. Census 2011 data showed us that 86% of people in the United Kingdom identified themselves as ‘white’ in 2011.

How valid are these statistics?

To an extent, ethnic identity is an objective matter – for example, I was kind of ‘born white’ in that both my parents are/ were white, all of my grandparents were white, and all of my great-grandparents were white, so I can’t really claim I belong to any other ethnic group. However, although I ticked ‘white’ box when I did the U.K. Census, this personally means very little to me, whereas to others (probably the kind of people I wouldn’t get along with very well) their ‘whiteness’ is a very important part of their identity, so there’s a whole range of different subjective meanings that go along with whatever ethnic identity box people ticked. Census data tells us nothing about this.

Religion according to the U.K. 2011 Census

In the 2011 Census, 59% of people identified as ‘Christian’ in 2011, the second largest ‘religious group’ was ‘no religion’, which 25% of the U.K. population identified with.

Statistics on religious affiliation may also lack validity – are 59% of people really Christian? And if they really are, then what does this actually mean? Church attendance is significantly lower than 59% of the population, so the ‘Christian’ box covers everything from devout fundamentalists to people that are just covering their bases (‘I’d better tick yes, just in case there is a God, or gods?’)

The British Humanist Society present a nice summary of why statistics on religious belief may lack validity…basically based on the ‘harder’ statistics such as church attendance which show a much lower rate of committed religious practice.

The United Kingdom Employment Rate

The employment rate is the proportion of people aged from 16 to 64 in work.

The lowest employment rate for people was 65.6% in 1983, during the economic downturn of the early 1980s. The employment rates for people, men and women have been generally increasing since early 2012.As of December 2016, the employment rate for all people was 74.6%, the highest since records began in 1971

Critics of the above data point to the existence of an informal or shadow economy in the United Kingdom which is worth an estimated £150 billion a year – people who are working and earning an income, but not declaring it. In reality, the actual paid-employment rate is higher.

Household Income Distribution in the United Kingdom

Household income statistics are broken down into the following three broad categories:

original income is income before government intervention (benefits)
gross income is income after benefits but before tax

disposable income is income after benefits and tax (income tax, National Insurance and council tax).

In the year ending 2016, after cash benefits were taken into account, the richest fifth had an average income that was roughly 6 times the poorest fifth (gross incomes of £87,600 per year compared with £14,800, respectively)

Reasons why household income data may lack validity

While measuring income does appear to be purely objective (you just add and minus the pounds), the income data above may lack validity because some people might not declare some of the income they are earning. Cash in hand work, for example, would not be included in the above statistics, and some money earned via the ‘gig economy’ might not be declared either – how many people actually pay tax on their YouTube revenue for example, or from the goods they sell on Ebay?

The United Kingdom Crime Rate

Below I discuss data from the Crime Survey of England and Wales (CSEW), which is a victim-survey conducted by structured interview with 35 000 households. It seems pointless discussing the crime rate according to police recorded crime because it’s such an obviously invalid measurement of crime (and the police know it), simply because so many crimes go unreported and hence unrecorded by the police.

Latest figures from the Crime Survey for England and Wales (CSEW) show there were an estimated 6.1 million incidents of crime experienced by adults aged 16 and over based on interviews in the survey year ending December 2016.

The green dot shows the figure if we include computer based crimes and online fraud, a new type of crime only recently introduced to the survey (so it wouldn’t be fair to make comparisons over time!) – if we include these the number of incidents of crime experienced jumps up to 11.5 million.

Reasons why even the CSEW might lack validity

Even though its almost certainly more valid than police recorded crime – there are still reasons why the CSEW may not report all crimes – domestic crimes may go under-reported because the perpetrator might be in close proximity to the victim during the survey (it’s a household survey), or people might mis-remember crimes, and there are certain crimes that the CSEW does not ask about – such as whether you’ve been a victim of Corporate Crime.

The U.K. Prison Population

The average prison population has increased from just over 17,400 in 1900 to just over 85,300 in 2016 (a five-fold increase). Since 2010, the average prison population has again remained relatively stable.

Prison Population Statistics – Probably have Good Validity?

I’ve included this as it’s hard to argue with the validity of prison population stats. Someone is either held in custody or they or not at the time of the population survey (which are done weekly!) – A good example of a truly ‘hard’ statistic! This does of course assume we have open and due process where the law and courts are concerned.

Of course you could argue for the sake of it that they lack validity – what about hidden prisoners, or people under false imprisonment? I’m sure in other countries (North Korea?) – their prison stats are totally invalid, if they keep any!

United Kingdom Population and Migration Data

Net migration to the U.K. stood at 248 000 in 2016, lower than the previous year, but still historically high compared to the 1980s-1990s.

There are a number of reasons why UK immigration statistics may lack validity

According to this migration statistics methodology document only about 1/30 people are screened (asked detailed questions about whether they are long term migrants or not), on entering the United Kingdom, and only a very small sample of people (around 4000) are subjected to the more detailed International Passenger Survey.

Then of course there is the issue of people who enter Britain legally but lie about their intentions to remain permanently, as well as people who are smuggled in. In short the above statistics are just based on the people the authorities know about, so while I’m one to go all ‘moral panic’ on the issue of immigration, there is sufficient reason to be sceptical about the validity of the official figures!

Ranking Exercise:

You might like to rank the following ‘official statistics’ in terms of validity – which of these statistics is closest to actual reality?

Immigration statistics – Net migration in 2016 was 248 000

Prison statistics – There are just over 85 000 people in prison
Crime statistics – There were around 6 million incidents of crime in 2016
The richest 20% of households had an average income of around £85 000 in 2016

The U.K. employment rate is 75% in 2016.
59% of the population were Christina in 2011
86% of the population was white in 2011

Related Posts

Official Statistics in Sociology

Education Statistics – 12 things Department for Education data tell us about the state of education in England and Wales today (forthcoming)

Family and Household Statistics – seven interesting statistics about family life in the U.K.

Sources

Please click the pictures above to follow links to sources…

The United Kingdom Census is a survey of every person in the United Kingdom, carried out every 10 years, the last one being in March 2011. It asks a series of ‘basic’ questions about sex, ethnicity, religion and occupation. It is the only survey which is based on a ‘total sample’ of all U.K. households. You might also like this summary – What is a Census?

U.K. Prison Population Statistics – House of Commons Research Briefing

Scientific Quantitative Methodology in Sociology

Positivists prefer to the limit themselves the study of objective ‘social facts’ and use statistical data and the comparative method to find correlations, and multivariate analysis to uncover statistically significant ‘causal’ relationships between variables and thus derive the laws of human behaviour.

This post explores the Positivist approach to social research, defining and explaining all of the above key terms and using some examples from sociology to illustrate them.

Social Facts

The first rule of Positivist methodology is to consider social facts as things which means that the belief systems and customs of the social world should be considered as things in the same way as the objects and events of the natural world.

According to Durkheim, some of the key features of social facts are:

they exist over and above individual consciousness
they are not chosen by individuals and cannot be changed by will

each person is limited (constrained) by social facts

According to Durkheim what effects do social facts make people act in certain ways, in the same way as door limits the means whereby you can enter a room or gravity limits how far you can jump.

Positivists believed that we should only study what can be observed and measured(objective facts), not subjective thoughts and feelings. The role of human consciousness is irrelevant to explaining human behaviour according to Positivists because humans have little or no choice over how they behave.

For a more in-depth account of social facts, have a look at this blog post: What are Social Facts?

Statistical data, Correlation, and Causation

Positivists believed it was possible to classify the social world in an objective way. Using these classifications it was then possible to count sets of observable facts and so produce statistics.

The point of identifying social facts was to look for correlations – a correlation is a tendency for two or more things to be found together, and it may refer to the strength of the relationship between them.

If there is a strong correlation between two ore more types of social phenomena then a positivist sociologist might suspect that one of these phenomena is causing the other to take place. However, this is not necessarily the case and it is important to analyse the data before any conclusion is reach.

Spurious Correlations

Spurious correlations pose a problem for Positivist research. A spurious correlation is when two or more phenomena are found together but have no direct connection to each other: one does not therefor cause the other. For example although more working class people commit crime, this may be because more men are found in the working classes – so the significant relationship might be between gender and crime, not between class and crime.

Multivariate Analysis

Positivists engage in multivariate analysis to overcome the problem of spurious correlations.

Multivariate Analysis involves isolating the effect of a particular independent variable upon a particular dependent variable. This can be done by holding one independent variable constant and changing the other. In the example above this might mean comparing the crime rates of men and women in the working class.

Positivists believe multivariate analysis can establish causal connections between two or more variables and once analysis is checked establish the laws of human behaviour.

Positivism – Establishing the Laws of Human Behaviour

A scientific law is a statement about the relationship between two or more phenomena which is true in all circumstances.

According to Positivists, the laws of human behaviour can be discovered by the collection of objective facts about the world in statistical form and uncovering correlations between them, checked for their significance by multivariate analysis.

Positivism and The Comparative Method

The comparative method involves the use of comparisons between different societies, or different points in time

The purpose of using the comparative method is to establish correlations, and ultimately causal connections, seek laws and test hypotheses.

The comparative method overcomes the following disadvantages of experiments:

Moral problems are not as acute

The research is less likely to affect the behaviour or those being studied because we are looking at natural settings
The comparative method is superior to the experimental method because allows the sociologist to explore large scale social changes and changes over time

However, a fundamental problem with the comparative method is that the data you want may not be available, and you are limited to that data which already exists or which can be collected on a large scale via social surveys.

Positivism and Interpretivism in social research

Social Action Theory – criticises the positivist approach to social research, arguing that human consciousness is too complex to reduce to numbers.