Chasing Up Another Fake Statistic About Exclusions

The Guardian frequently prints false information about suspensions and exclusions from school. I’ve just found an example from earlier this month. In this article they claimed the following:

74% of girls in youth custody have previously been permanently excluded from school compared with 63% of boys.

This struck me as unlikely, as there has been very recent research, from the Department for Education and the Ministry of Justice, on the link between exclusions and offending . That showed that for roughly 5000 young people who received a custodial sentence before the age of 18, the majority had never been permanently excluded, although most had been suspended at some point.

This is broadly consistent with other sources. For instance, FFT Education Datalab, found fewer than 20% of 11000 young people who had been in custody between 16 and 18 had been permanently excluded. I’m not aware of any credible source that gives a higher figure, although because suspensions were for many years called “Fixed Term Exclusions” there are sources (eg. this) that give a figure for “exclusions” that almost certainly includes suspensions and is, therefore, much higher.

So where did the Guardian’s figure come from? As with so much bad information in the media on this issue, it comes from a report by campaigning organisations. This one was from a report by Agenda and Alliance for Youth Justice. (The aptly named, charity Agenda previously featured in another blog post I wrote about false claims about exclusions.) In their report they claim:

74% of girls in youth custody have previously been permanently excluded from school compared to 63% of boys.

The source they give for this is a report on Education in Youth Custody from 2016, published by The Parliamentary Office of Science and Technology (POST). This report states that:

Around 90% of children in custody had been excluded from school at any one time before entering custody and 63% of boys had been permanently excluded (74% for girls).

This gives two sources, which is a bit of a warning if one is using statistics to make comparisons. The first is an article from British Journal of Community Justice from 2015. Researchers distributed questionnaires to just 85 young people in YOIs and got 47 responses (not all complete) and found that:

Of 45 respondents, nine out of 10 (89%, 40) had been excluded from education, 63% of which were permanent exclusions.

The POST research appears to have considered a survey of 45 to be a reliable data point. Also, the writers of that research appear to have missed that it is 63% of 89% (about 56%) not 63% who have been permanently excluded. The second source is this 2012 report from HM Inspectorate of prisons. This included a survey of around a thousand inmates in YOIs. This included the following.

…Seventy-four per cent (n=19) of young women reported having been excluded from school at some point…

That “n=19” refers to the number of responses from young women to the question about exclusions. Only 25 young women were surveyed, so presumably 6 did not answer this question, and 5 said they hadn’t been excluded. The use of the phrase “at some point” indicates this refers to suspensions (then called “Fixed Term Exclusions”), as does the fact that elsewhere it states that:

The majority (88%) of young men said that they had been excluded from school at some point…

So, it would appear that POST’s statistics include multiple errors; ludicrously tiny sample sizes; surveys not completed by all recipients, and it ignored a data point that clearly contradicted one of the claims being made. POST’s parliamentary website states:

The core objective of POST is to supply trusted and impartial analysis to the UK Parliament.

To do this, our advisers are in constant contact with experts from academia, industry, government, the third sector and beyond. Experts help us scan the horizon, identify literature, contextualise research evidence, and peer review our work.

Not very impressive. Also, it’s not impressive that their obviously false statistics were repeated by two further sources. All I can say is: Never believe any claim about exclusions that appears in the media or comes from a charity or campaign group. They are frequently incorrect.

Advertisement

A Report on Education and Offending

Early this month, it was pointed out to me that there was a planned release from the Department for Education that would be “Examining the education and children’s social care background of children who have been cautioned or sentenced for an offence”. I feared the worst, because almost everything written on children’s offending seems to find reason to blame schools. Usually, it’s claimed that schools are at fault for expelling or suspending pupils, and not acting as amateur prisons keeping the most dangerous young criminals in classrooms where they can’t scare the public and can only harm teachers and children.

I was wrong. What has actually been published is not the usual attack on schools. “Education, Children’s Social Care and Offending Descriptive Statistics” is a rather thorough analysis which is careful not to promote the usual narrative.

I have been writing about behaviour and exclusions for some time now, and there is a lot in the report that is relevant to those discussions. I think this report will be informing a number of future blogposts as well as this one. I will be looking at a number of points that I think will be of particular interest to teachers. Please be aware though, that I am selecting a few points of interest, not attempting to summarise the report.

The research is produced jointly by the Department for Education and the Ministry of Justice. It looks at the data for 3 categories of children selected from the 3 cohorts who were in year 11 in the years 2012-2015. It explains:

….three offending groups are identified in this publication: approximately 77,300 children who had been cautioned or sentenced for an offence, which is equivalent to 5% of the total pupil cohort; approximately 18,000 children who had been cautioned or sentenced for a serious violence offence (equivalent to 1.1% of the total pupil cohort), and approximately 12,300 children whose offending had been prolific (equivalent to 0.8% of the total pupil cohort).

This is not a sample, but as far as possible, an attempt to collect all the data from the pupils in those cohorts who fit the relevant criteria.

The first thing to notice is that, when you consider the size of an average state secondary school is about 1000, these pupils are not uncommon. The data on offending covers the children when they are in the 10-17 age group, but most first offences are within the age group 14-16. Ignoring changes in crime rates since then, do we secondary teachers think of our pupils as a population where 1 in 20 will have been cautioned or sentenced before the age of 18? Do we realise there is likely to be 1 violent young criminal in every 90 children? We might teach that many children before lunch some days. Are we aware that we can expect there to be one prolific young criminal in every 125 pupils? That’s likely to give you an average higher than 1 in each year group. If you account for the fact that these pupils will not be evenly distributed and you can expect to have a lot more of them in a boys school or a school in a disadvantaged area, its actually quite frightening. It’s also worth considering the fact that these are only the young criminals who were caught and sentenced or cautioned. Secondary school populations can include a lot of young people who have committed, or will soon commit, criminal offences, including both violent offenders and habitual criminals. This should be a fact teachers should be reminded of in every CPD session on behaviour and I intend to blog further thoughts about this at some point in the future.

The second thing I noticed about the report was how thoroughly it made an effort to discourage readers from making assumptions about causation.

Care should be taken when interpreting this analysis as the findings do not imply a causal link between the educational or children’s social care characteristics and being cautioned or sentenced for an offence. Future work using these data will aim to build upon this analysis to better understand the relationships between the outcomes and characteristics in this publication.

The education variables included in this paper have generally been analysed independently of each other. It is important to note that there may be links between these key variables which have not been factored into the analysis, and other factors which could not be taken into account.

Children who had been cautioned or sentenced for a serious violence offence and children whose offending had been prolific represent a small, atypical group of young people; their results should not be assumed to be representative of all children who have been cautioned or sentenced for an offence or young people more generally.

It is rare for any writing in education to be this clear in stating that a risk factor does not indicate causation. It also makes it clear that just because the risk of an outcome is increased for one group of pupils it doesn’t mean that the outcome is very likely. For instance, it points out that while 76% of those cautioned or sentenced for a serious violent offence have been eligible for FSM (Free School Meals), only 2% of those who have been eligible for FSM have been cautioned or sentenced for a serious violent offence. Not only is FSM not demonstrably a “cause” of violence, it does not allow us to predict who will offend in that way. Unless people understand that distinction, they will be prone to thinking they have found the cause, or causes, of criminality.

Another way in which this report is clearer than most is the way it spells out the difference between suspensions and permanent exclusions.

A suspension is where a pupil has been temporarily removed from the school, whilst a permanent exclusion is when a pupil is no longer allowed to attend a school.

Perhaps it is just that a low bar has been set previously, but it is a relief to see anyone making a clear distinction between the two and analysing them separately rather than talking about “exclusions” and leaving the reader to look for contextual clues as to which is being referred to. This enables the report to reach clear conclusions about the correlations between offending and suspensions and permanent exclusions.

There is a clear connection, but there is no inevitability about it. There is no school to prison pipeline here. The suspended may be more likely to be cautioned or sentenced, but they are not likely to be cautioned or sentenced. The permanently excluded are likely to be cautioned or suspended (unsurprising when you consider that only the most extreme 1 in a 1000 pupils are permanently excluded each year), but the vast majority will not be cautioned or sentenced for a serious violent offence or for a prolific number of offences.

Another point that perhaps should inform discussion is that the offending figures above for permanently excluded pupils are less than for pupils in Alternative Provision.

Looking only at the pupil cohort which had ever been registered at a state or non-state funded AP setting, 41% had ever been cautioned or sentenced for an offence. (This rises to 45% for those that were registered at state funded AP). The rates for the other offending groups are much lower: 14% of those at any AP setting had ever been cautioned or sentenced for a serious violence offence, and 15% of those whose offending had been prolific.

In some of the narratives about a “school to prison pipeline” AP is one of the villains, with claims that attending AP is a mechanism by which permanently excluded children are turned into criminals. These figures (almost a third lower than among permanently excluded pupils) should be unsurprising because not everyone in AP is there for behaviour, and where they are, AP is meant to provide specialist help suited to those pupils. However, it seems important to note that pupils in AP are not at greater risk of offending than permanently excluded pupils.

Another issue raised by the report is Special Educational Needs.

80% of those who had been cautioned or sentenced for an offence, and 87% of those cautioned or sentenced for a serious violence offence, had been recorded as ever having SEN. 95% of those whose offending had been prolific had been recorded as ever having SEN. 45% of the all-pupil population had been recorded as ever having SEN at some point up to the end of KS4.

As mentioned in a previous blogpost, FFT datalab had looked at the scale of SEN labelling in a blogpost entitled More pupils have special educational needs than you might think, Looking at the cohort of students who were in year 11 in 2016/17 they found that “44% of the cohort had ever been classified as having SEN by the time they reached the end of Year 11”. Here we have a similar figure for the combined cohorts for three earlier years. This is a system where it is very easy to be labelled SEN, but also one where the risk of being labelled seems connected with the risk of offending. Again, this is something I will hope to look at in a future blogpost.

An unsurprising part of the data is the effect of gender on offending.

Male pupils were over-represented amongst children who had been cautioned or sentenced for an offence, with children whose offending had been prolific containing the highest proportion, at 84%. This is marginally higher than children who have been cautioned or sentenced for a serious violence offence, which is also 84% to the nearest whole number. In comparison, 76% of all children who had been cautioned or sentenced for an offence and 51% of the pupil cohort was male

I think this raises two issues. Firstly, why does this risk factor for criminal behaviour, often get ignored, while others (FSM, SEND, exclusions) are used not only as explanations for offending, but reasons not to hold offenders responsible? Secondly, there is increasing pressure to use the word “gender” to mean gender identity not biological sex. This data predates the push to use the word “gender” this way, but it seems important to make sure that in the future we know precisely which one we are referring to.

I’m not going to go into the key findings on ethnicity for two reasons. Firstly, there seems to be a lack of useful detail in the report. There may be some in the accompanying data, and I may look at that at a later date. Secondly, school demographics have changed a lot since these cohorts, so I’m not sure how useful the analysis would be to schools at the moment.

Finally, I should point out there is a lot more interesting data about the connection between offending and exclusions and suspensions. I think this will have to be revisited in a future blogpost.

Update 3/4/2022: It’s been pointed out to me that in the paragraph about how many young offenders there are, I mentioned the size of an average secondary school, but then talked about the frequency of offending using the whole cohort figures which would have included those in AP of one sort or another. I don’t think it substantially changes my argument, and it is impossible to modify the frequencies accurately. However, we can use the figures on page 26 showing how many of those cautioned have ever attended AP to get an upper bound for how much I could be out.

I asked:

…do we secondary teachers think of our pupils as a population where 1 in 20 will have been cautioned or sentenced before the age of 18?

This was based on 5% of the pupil cohort having been cautioned or sentenced. It remains correct if “population” refers to the cohort rather than the population of an average secondary. But we can consider how it would affect the figure if we only consider those not in some form of AP. According to the report, 26% of those who have been sentenced or cautioned attended AP at some point. If every one of those 26% had been in AP for their entire time at secondary school that would reduce the figure from 1 in 20 to 1 in 29. According to page 41) at least 54% of those pupils who attended AP did so after their first offence, so it seems likely the 1 in 20 figure is a lot more more accurate than 1 in 29, but the true figure is presumably somewhere in between.

I also asked:

Do we realise there is likely to be 1 violent young criminal in every 90 children? We might teach that many children before lunch some days.

Again, the first sentence remains true when considering the whole cohort. However, if we assume we are talking about pupils not in AP, then (again according to page 26) up to 37% of those who are suspended or cautioned for serious violence have attended AP at some point. Again, it’s unlikely that will be anywhere near 37% at one time, but if it was that would change the figure from 1 in 90 to 1 in 143. Although this is now likely to be a drastic under-estimate (two fifths of those who are cautioned or sentenced for serious violence and attend AP, only attended AP after their first offence) we can observe that 143 is not that many pupils, and many teachers regularly teach 143 children in a day.

Finally I said:

Are we aware that we can expect there to be one prolific young criminal in every 125 pupils? That’s likely to give you an average higher than 1 in each year group.

This is the figure that probably holds up least well. Prolific criminals are the group most likely to have been attended AP. Page 27 says that 57% of them have attended AP. Again, it is unlikely that 57% would all be in AP at one time, but if that were the case that would change the figure from 1 in 125 (actually that was with a bit too much rounding, it should have been 1 in 133) to 1 in 309. That would bring us nearer to 1 per key stage in an average secondary rather than 1 per year.

As I said, I don’t think this affects any of my points in this, or my later post, but I thought I’d better acknowledge it. Thanks to Kat Stern for pointing it out.

Did Glasgow cut youth crime by ending suspensions and expulsions? Part 3

It has recently been claimed that Glasgow has, by following a policy of reducing suspensions (or possibly expulsions) from schools, cut youth crime dramatically. I am exploring this in a series of 3 posts of which this is the last. In my first post, I put the changes in the wider context, observing that, compared with Glasgow, youth crime fell by proportionately more in Scotland as a whole, and that similar changes appear to have happened in England and Wales. In my second post, I looked at how the change in the number of suspensions seemed too small to account for the change in the number of offences. Here I will consider a couple of other points.

1. Whether changes in expulsions might explain the fall in crime.
2. Whether there’s a reason these stories focus on Glasgow, not the whole of Scotland.

To begin with, an extract from my previous post to remind you about the stories I’m addressing, and a note about my use of terminology.

---

A BBC report last October gave credence to the idea that it is possible to prevent crime by reducing exclusions. The following was claimed about Glasgow:

Ms McKenna’s [Glasgow’s Director of Education] approach fitted with the work and ethos of Scotland’s Violence Reduction Unit (VRU), which was started just before she took up her post. It aims to treat crime as a public health issue and look at root causes of the problem.

There has been an 88% reduction in school exclusions in the past 10 years, at the same time there has been a 50% reduction in youth crime.

Several English councils and Sadiq Khan, the Mayor of London, have sent representatives to learn from Ms McKenna how the approach to exclusions could be having a positive impact on reducing violence.

In a recent video made by The Economist, even more definite claims were made about Glasgow.

One city in Britain has done something about reducing exclusions. This is Saint Roch’s school in Glasgow. 15 years ago, it excluded around 300 pupils a year….

But then a new policy was implemented with the agreement of the headmaster. And in recent years only a handful of students have been excluded…

“Exclusions for us are very  much an absolute last resort. And the reason for that is because  we don’t think that exclusions solve any problems.”

The school has invested in training  its teachers to better understand and support students  displaying behavioural issues.

“We have a real need to find out what is lying behind poor behaviour. And if we can understand what is  leading to that, then we can tackle the root cause rather than the  symptom of the bad behaviour in school. You can kind of see some of the  kids who come in the morning from challenging  backgrounds or kids who have low confidence and low self-esteem  may struggle early in the morning”

This approach was implemented systematically across Glasgow’s schools in 2010. The results for children of all  ethnic backgrounds have been positive. Between 2010 and 2015, youth crime in the city fell by 50%.

“I don’t have any evidence to say,  well, because you’ve reduced exclusions, that’s what’s  happened with youth crime. But they must be linked, if we have  more young people getting more qualifications, able to make more  positive choices in their lives able to be the next generation  of positive families then everybody wins, don’t they?”

If you look at the whole of these stories you can see that they include one of the most common faults in stories about exclusions: a failure to distinguish between expelling a pupil from school permanently (“permanent exclusion” in England; “removal from roll” in Scotland), and removing a pupil from school for a short period of time (now “suspension” in England, previously “fixed term exclusions”; “temporary exclusion” in Scotland). This is an ongoing problem with reporting and campaigning on exclusions: with claims about the likely harms from exclusion consistently based on the serious nature of expulsions, but claims about the frequency of exclusions, and the trend in exclusions over time, often being based on suspensions. In the above stories, we see figures for exclusions that appear to be about suspensions, yet If you look at the full stories there are a number of points about expulsions. I will be very careful to look at both and use the words “expulsion” and “suspension” for clarity.

---

As stated above, there is often confusion between suspensions and expulsions, because the word “exclusion” has been used to refer to both. This has often allowed campaigners against expulsions in England to use claims that are actually about suspensions to refer to both. As I noted before, the two media reports mentioning Glasgow I have been looking at, also mention expulsions, even though the substantial claims appear to be about suspensions. As it is unclear how a change in suspensions could have resulted in the fall in crime, it’s probably worth checking that the fall in expulsions could not be the cause.

When we try to explain changes in youth crime in Scotland by looking at expulsions, we have all the same problems we have with suspensions. The numbers involved are too small, and the changes in youth crime seem to have happened in England and Wales too. But we also have the problem that expulsions were never as common in Scotland as they were in England and they were extremely low by the period mentioned in the stories.

The change in expulsions in Scotland has been even more extreme than that for suspensions, but the numbers are far smaller to begin with.

Did Glasgow cut youth crime by ending suspensions and expulsions? Part 2

It has recently been claimed that Glasgow, by following a policy of reducing suspensions, or possibly expulsions, from schools, has cut youth crime dramatically. I intend to explore this over 3 posts. In my last post, I put the changes in the wider context, observing that youth crime fell by proportionately more in Scotland as a whole than it did in Glasgow, and that similar changes appear to have happened in England and Wales. Here I will look at how the numbers involved in changes in school suspensions and exclusions don’t seem to fit the numbers involved in changes in crime data.

To begin with, an extract from my previous post to remind you about the stories I’m addressing, and a note about my use of terminology.

---

A BBC report last October gave credence to the idea that it is possible to prevent crime by reducing exclusions. The following was claimed about Glasgow:

Ms McKenna’s [Glasgow’s Director of Education] approach fitted with the work and ethos of Scotland’s Violence Reduction Unit (VRU), which was started just before she took up her post. It aims to treat crime as a public health issue and look at root causes of the problem.

There has been an 88% reduction in school exclusions in the past 10 years, at the same time there has been a 50% reduction in youth crime.

Several English councils and Sadiq Khan, the Mayor of London, have sent representatives to learn from Ms McKenna how the approach to exclusions could be having a positive impact on reducing violence.

In a recent video made by The Economist, even more definite claims were made about Glasgow.

One city in Britain has done something about reducing exclusions. This is Saint Roch’s school in Glasgow. 15 years ago, it excluded around 300 pupils a year….

But then a new policy was implemented with the agreement of the headmaster. And in recent years only a handful of students have been excluded…

“Exclusions for us are very  much an absolute last resort. And the reason for that is because  we don’t think that exclusions solve any problems.”

The school has invested in training  its teachers to better understand and support students  displaying behavioural issues.

“We have a real need to find out what is lying behind poor behaviour. And if we can understand what is  leading to that, then we can tackle the root cause rather than the  symptom of the bad behaviour in school. You can kind of see some of the  kids who come in the morning from challenging  backgrounds or kids who have low confidence and low self-esteem may struggle early in the morning”

This approach was implemented systematically across Glasgow’s schools in 2010. The results for children of all  ethnic backgrounds have been positive. Between 2010 and 2015, youth crime in the city fell by 50%.

“I don’t have any evidence to say,  well, because you’ve reduced exclusions, that’s what’s  happened with youth crime. But they must be linked, if we have  more young people getting more qualifications, able to make more  positive choices in their lives able to be the next generation  of positive families then everybody wins, don’t they?”

If you look at the whole of these stories you’d see they include one of the most common faults in stories about exclusions: a failure to distinguish between expelling a pupil from school permanently (“permanent exclusion” in England, “removal from roll” in Scotland), and removing a pupil from school for a short period of time (now “suspension” in England, previously “fixed term exclusions”; “temporary exclusion” in Scotland). This is an ongoing problem with reporting and campaigning on exclusions, with claims about the likely harms from exclusion consistently based on the serious nature of expulsions, but claims about the frequency of exclusions, and the trend in exclusions over time, often being based on suspensions. In the above stories, we see figures for exclusions that appear to be about suspensions, yet If you look at the full stories there are a number of points about expulsions. I will be very careful to look at both and use the words “expulsion” and “suspension” for clarity.

---

One of the main problems with confusing expulsion and suspension in one’s analysis of exclusions is that while expulsion might have a dramatic effect on a pupil’s life, suspensions are not obviously that big a deal. This is critical here. Suspensions in Scotland usually last for a day or two, and rarely more than a week.

As in similar debates in England, it is unclear how these relatively small amounts of time out of school will give a great opportunity for criminality. In order to justify the claim that a day or two out of school due to a suspension can have a drastic effect on criminality, anti-exclusions campaigners often seem to assume that a suspended pupils can be expected to be wandering the streets getting into trouble that they wouldn’t have otherwise become involved in. This is an assumption that is not usually accompanied by any attempt to justify it. A small study in the 90s (too small and old to be generalised, but enough to give us reason to question assumptions) found the following about a sample of 56 children with considerable experience of suspension or permanent exclusion:

Self-report interview questions reveal that whilst forty of the young people had offended, 90% (36) reported that the onset of their offending commenced prior to their first exclusion. Moreover, fifty (89.2% of the total number of young people in the sample), stated that they were no more likely to offend subsequent to being excluded and thirty one (55.4%) stated that they were less likely to offend during their exclusion period.

It is often the case that correlation does not equal causation; it is entirely possible for both suspensions and youth crime referrals to fall at the same time without one causing the other. When we consider how little time out of school is a result of suspensions (compared with, say, school holidays and absences) there is every reason to doubt that suspensions are the main driver of youth crime.

I would even speculate that in order to maintain the plausibility of the idea that cutting suspensions cut youth crime, the extent to which youth crime has apparently reduced in Scotland has been underplayed by these stories about Glasgow. As noted last time, in Glasgow, the proportionate decrease in offence referrals was not as great as for Scotland as a whole. Looking at Glasgow underplays the extent of the change in youth crime in Scotland, and that’s not the only way it is being underplayed. Both versions of the Glasgow story describe youth crime halving in Glasgow between 2010 and 2015. Actually, the number of offense referrals in Glasgow went down from 3760 in 2009/10 to 1423 in 2014/15. I think the recording period starts on April 1st so perhaps the relevant change is from 2556 in 2010/11 to 994 in 2015/16. Either way, the reduction is more dramatic than merely halving offences. The appeal of a simple story about a city reducing “exclusions” and achieving a reduction in youth crime of 50% might be that it just seems more believable than what the data actually shows: an entire nation reducing suspensions and achieving a reduction in offence referrals of over 80%. The apparent change in Scotland’s youth crime is so large that the claim that it might have been caused by reducing suspensions would be hard to believe even if we didn’t know similar changes in youth crime happened in England and Wales without the same policy on suspensions.

It becomes even harder to believe if we directly compare how many fewer suspensions and how many fewer offence referrals there were. If we measure from the peak of suspensions in Scotland (2006/7) to the last pre-Covid figure 2018/19, the number of suspensions has gone from 44536 to 14987, a fall of 29549. Over the same period, offence referrals fell from 34385 to 7763, a fall of 26622. For Glasgow, the fall in suspensions in that period is 6590 and the fall in offence referrals was 4185. If the reduction in suspensions had caused the reduction in referrals in Glasgow, it would be the case that for every 5 suspensions that were stopped, 3 fewer offences would be referred. And if the same logic was applied to the whole of Scotland, for every 10 suspensions that were stopped, 9 fewer offences would be referred. When you consider that offence referrals are the cases where perpetrators have been identified and reported to the courts, this is likely to mean the actual number of offences being prevented is much greater than that. Does anyone believe that suspended pupils are such prolific (or incompetent) criminals that these figures could make sense? That for every 10 suspensions, Scottish pupils are not just committing 9 crimes, but being caught too? The problem here is not that reduced suspensions in Scotland haven’t coincided with a fall in offence referrals, but that the scale of the fall is way beyond what even the most ardent opponent of suspensions can reasonably claim could be caused by the fall in suspensions.

It would be convenient if I could tell you what did cause the change in offence referrals, but I’m not aware of any simple answer, particularly knowing that youth crime fell dramatically in England and Wales too. However, any credible explanation of the change in referrals will need to involve more than just the issue of suspensions. Once we have noted that there must be reasons for the change in offense referrals, other than suspensions, then there is no reason to assume that reducing suspensions had any particular effect on offence referrals. Reducing suspensions might be the policy that’s in fashion, but it simply doesn’t explain what happened in Scotland.

Did Glasgow cut youth crime by ending suspensions and expulsions? Part 1

It has recently been claimed that Glasgow has, by following a policy of reducing suspensions, or possibly expulsions from schools, cut youth crime dramatically. I intend to explore this over 3 posts. Here I will look at how the claims ignore the wider context.

A BBC report last October gave credence to the idea that it is possible to prevent crime by reducing exclusions. The following was claimed about Glasgow:

Ms McKenna’s [Glasgow’s Director of Education] approach fitted with the work and ethos of Scotland’s Violence Reduction Unit (VRU), which was started just before she took up her post. It aims to treat crime as a public health issue and look at root causes of the problem.

There has been an 88% reduction in school exclusions in the past 10 years, at the same time there has been a 50% reduction in youth crime.

Several English councils and Sadiq Khan, the Mayor of London, have sent representatives to learn from Ms McKenna how the approach to exclusions could be having a positive impact on reducing violence.

In a recent video made by The Economist, even more definite claims were made about Glasgow.

One city in Britain has done something about reducing exclusions. This is Saint Roch’s school in Glasgow. 15 years ago, it excluded around 300 pupils a year….

But then a new policy was implemented with the agreement of the headmaster. And in recent years only a handful of students have been excluded…

“Exclusions for us are very  much an absolute last resort. And the reason for that is because  we don’t think that exclusions solve any problems.”

The school has invested in training  its teachers to better understand and support students  displaying behavioural issues.

“We have a real need to find out what is lying behind poor behaviour. And if we can understand what is  leading to that, then we can tackle the root cause rather than the  symptom of the bad behaviour in school. You can kind of see some of the  kids who come in the morning from challenging  backgrounds or kids who have low confidence and low self-esteem  may struggle early in the morning”

This approach was implemented systematically across Glasgow’s schools in 2010. The results for children of all  ethnic backgrounds have been positive. Between 2010 and 2015, youth crime in the city fell by 50%.

“I don’t have any evidence to say,  well, because you’ve reduced exclusions, that’s what’s  happened with youth crime. But they must be linked, if we have  more young people getting more qualifications, able to make more  positive choices in their lives able to be the next generation  of positive families then everybody wins, don’t they?”

If you look at the whole of these stories you’d see they include one of the most common faults in stories about exclusions: a failure to distinguish between expelling a pupil from school permanently (“permanent exclusion” in England, “removal from roll” in Scotland), and removing a pupil from school for a short period of time (now “suspension” in England, previously “fixed term exclusions”; “temporary exclusion” in Scotland). This is an ongoing problem with reporting and campaigning on exclusions, with claims about the likely harms from exclusion consistently based on the serious nature of expulsions, but claims about the frequency of exclusions, and the trend in exclusions over time, often being based on suspensions. In the above stories, we see figures for exclusions that appear to be about suspensions, yet If you look at the full stories there are a number of points about expulsions. I will be very careful to look at both and use the words “expulsion” and “suspension” for clarity.

When I looked into the data about Glasgow, I was able to confirm that there has been a dramatic reduction in suspensions and a dramatic reduction in offence referrals to the Scottish Children’s Reporter Administration, the only measure of youth crime that is easy to find. The claim that youth crime halved turns out to be an apparent understatement about Glasgow. I will look at possible reasons for underplaying the fall in my next blogpost. Here I will argue that, in the focus on Glasgow, there is a complete lack of context. Despite The Economist’s claim that “one city in Britain has acted”, Glasgow is in line with the rest of Scotland with its policy on expulsions and suspensions and is not an outlier for having had a fall in youth crime. The policy of reducing suspensions and expulsions is Scotland-wide. The fall in youth crime is, if anything, even more dramatic for Scotland as a whole than for Glasgow.

For some reason in recent years, expulsion and suspension figures in Scotland appear to be published only every other year, but assuming schools and LAs don’t change their practices in the off years, the policy of reducing suspensions is exceptionally clear. The bar chart below shows the number of suspensions per thousand of the school population.

(Figures for Scotland taken from the first spreadsheet here and for Glasgow I have calculated them from same spreadsheet and also the school census here). I think Glasgow’s figures may include expulsions as well as suspensions, but these should have little effect on the data. As can be seen, suspensions were massively reduced in Scotland, but in Glasgow suspensions fell particularly sharply.

There was a reduction in youth crime at the same time. In Scotland, crimes committed by young people are referred to the SCRA (Scotland’s Children’s Reporter Administration) and Glasgow and Scotland can be found here (although you will have to click to get numbers per 1000) and are displayed like this.

As can be seen from the first bar chart, Scotland as a whole, and Glasgow in particular, reduced suspensions dramatically. As we can see from the second bar chart, they also had dramatically reduced offence referrals at the same time. However, we can also see that while Glasgow’s suspensions moved from much higher than average to below average for Scotland, its rate of youth crime referrals, despite being much reduced, not only remains much higher than Scotland as a whole, but showed no signs of becoming less of an outlier. If we look at Glasgow’s figures relative to the whole of Scotland, suspensions and offence referrals are telling a different story (I won’t put them on the same axes, as we don’t have suspension figures for every year).

If anything, Glasgow’s referral rate has grown relative to Scotland’s as a whole. In fact, by Scottish standards, Glasgow seems to be a poor choice for an exemplar of reducing youth crime. Instead of making the claim that youth crime in Glasgow halved (according to the BBC this was in the last ten years; according to the Economist it was in the years 2010 to 2015) one could instead observe that (over either period), offence referrals in Glasgow went from about one and a half times the national average, to twice the national average. This is not evidence that lowering suspensions doesn’t reduce crime, but it does highlight that if it does, Scotland as a whole would be a better example of this effect than Glasgow alone.

If we consider Scotland as a whole, we can acknowledge that crime fell after (or at the same time as) suspensions fell. It is the post hoc ergo propter hoc fallacy to believe that because one thing happened after another, then the first even caused the second. Two trends can coincide by chance. Nobody should assume without a good causal explanation that one thing caused another, and even then it should be noted that the role of the other factors, or chance, might have played in this data. In my next blog post I will look at the problem with causal explanations for the link between the changes in suspensions and offence referrals. Here though, I will point out the biggest single reason for thinking that Scotland’s (or Glasgow’s) fall in youth crime is not a result of its policy on youth crime: it happened in England and Wales too.

The following graphs are from England and Wale’s Youth Justice Statistics 2019/20 (the following year’s data have been published last month but were not available at time of writing). The offence referral system is not directly comparable to the measures of crimes used in England and Wales, so I’m just going to give you the first 4 graphs that seemed relevant.

As these are different types of data I won’t attempt to comment on whether this is objectively a greater fall in youth crime than in Scotland, I will simply claim that it is equally dramatic. Looking at older data indicates that youth crime in England and Wales peaked around the same time as in Scotland. It is actually remarkable how similar the trends in youth crime in Scotland are in England and Wales, when you consider the completely different criminal justice system.

If this was a blog abut crime rather than education, I might look into whether this change was seen in other countries, or try to work out the causes. However, as this is a blog about education, I will simply note that the policy regarding suspensions in Scotland cannot be given credit for a change that appears to have happened in England and Wales too. If the fall in youth crime in Scotland (or Glasgow) was a result of Scotland (or Glasgow’s) policy of reducing suspensions, why did it happen in England and Wales too?

Part 2 is now available here.