Why Do Some Ethnic Groups Have Higher Exclusion Rates? Part 2

August 21, 2022

I have been looking at the the differences in the exclusion rates for different ethnic groups. In Part 1, I observed that ethnic minority pupils generally don’t have a higher exclusion rate than white British pupils, and that those ethnic groups that consistently have higher exclusion rates tend to make up only a small proportion (11%) of ethnic minority pupils. The way we divide people into ethnic groups seems to create more groups with disparities than would be expected. In that previous post, I identified 5 ethnic groups that consistently have a higher exclusion rate than white British pupils.

Ethnic Group Number of Pupils % of pupils Number of exclusions (2020/21)
Ethnicity Minor Any other black background 64,064 0.77% 43
Ethnicity Minor Black Caribbean 82,726 1.00% 70
Ethnicity Minor Gypsy Roma 25,836 0.31% 46
Ethnicity Minor Traveller of Irish heritage 6,140 0.07% 6
Ethnicity Minor White and Black Caribbean 129,193 1.56% 153

These groups are small, and unrepresentative of ethnic minority pupils as a whole. One way to picture this is to imagine teaching a class of 27 pupils who are representative of the entire state school pupil population. In that class you would have 18 white British pupils and 9 ethnic minority pupils. Of those 9 ethnic minority pupils, 7 would be consistently less likely to be excluded than the white British pupils. 1 would be more likely to be excluded than the white British pupils in some years, but less likely in other years. And 1 would be consistently more likely to be excluded than the white British pupils. The debate about exclusions and ethnicity seems to be entirely focused on that 1 pupil.

Because the exclusion rate is very low (particularly in 2020/21), and these groups of pupils are very small, the higher rate of exclusions for these groups actually represents very small numbers, as can be seen from the table above. Given that there are well over  twenty thousand schools in England, it should be noted that relatively few headteachers have permanently excluded pupils from any of these groups. Even if some of those headteachers have made the wrong decision in some of these cases, it would be odd to extrapolate from them to the whole education system. Yet a lot of the discussion assumes these figures represent a systemic problem with exclusions, or with prejudice in exclusions. If you are convinced these exclusions must include some that are unfair, it would make more sense to ask the DfE to scrutinise the individual exclusions than to change policy regarding exclusions. There is certainly no reason to think most exclusions are unfair because of these 5 groups.

Opponents of exclusions use the exclusion figures for those 5 groups to generalise about all ethnic minorities and about the entire education system. There is also a tendency to generalise about the experiences of pupils in these 5 groups. Perhaps the most important thing to remember is how little exclusion rates actually tell us about an ethnic group. The conventional way to consider the effects of ethnicity on exclusions is to look at the exclusion rates of ethnic groups with high rates of exclusion and compare them to white British pupils (or all white pupils or all pupils). So, for instance, this chart emphasises the disparities.

However, most pupils in every ethnic group are unlikely to be excluded. The every day reality of school for most pupils in every ethnic group is perhaps better given by a graph showing the proportion of pupils not excluded.

The differences in exclusion rates are barely visible. All pupils in all ethnic groups are extremely unlikely to be permanently excluded. The disparities in exclusions do not tell us about entire ethnic groups. The exclusion rate for an ethnic group tells us about a fraction of 1% of pupils in that ethnic group. A difference in inclusion rates is not a judgement on an entire ethnic group. Trying to explain disparities in exclusion rates, using our knowledge of the entire group, is likely to be a waste of time. The tiny scale of the issue may be why so much of the debate consists of people misrepresenting the extent to which exclusions happen, and exaggerating the likely consequences of being excluded.

Having said all this, it might still be possible to say more about the causes of disparities in exclusion rates between different ethnic groups if we are willing to look at all the data, rather than just those groups that have a higher exclusion rate than white British pupils. It has never seemed plausible to me that ethnic groups who differ in location, social class, family structure, level of parental education, peer groups and so many other things, should be expected to all have identical levels of exclusion. It would be interesting to see if we can find any drivers for exclusion that explain those differences.

There are no shortage of campaigning groups out there who claim, often on very limited evidence, to know what causes exclusions. When it comes to race, explanations usually involve the claim that ethnic minorities are suffering some form of injustice that results in a higher rate of exclusions. As already discussed, in reality, we have very limited data and most disparities between ethnic groups are in the opposite direction. However, there is one factor that does seem to drive differences in exclusions between ethnic groups. If I combine the exclusion data from here and the Free School Meal data from here I get the following.

There is a strong positive correlation (r=0.76) between the exclusion rate for an ethnic group and the percentage of pupils having Free School Meals. The one point on that scatter graph that is an obvious outlier is travellers of Irish descent, the smallest ethnic group (6012 pupils) we have figures for. The small numbers involved (6 pupils excluded) make it unsurprising that the rate doesn’t fit the pattern.

Now, it is already known that Free School Meals pupils have a much higher exclusion rate (0.12 to 2 decimal places) than those who don’t (0.03), but the numbers here are greater than can be explained by the proportion of FSM pupils in each group. However, the FSM percentage is usually considered a good proxy for disadvantage. The simplest explanation here is that the exclusion rates for ethnic groups depend to a large extent on how disadvantaged that ethnic group is. Disadvantage that is concentrated in an ethnic group has more of an effect on exclusions than disadvantage that is spread throughout the population.

We now have two reasons for why some ethnic groups have a higher exclusion rate than white British pupils. In this post, we have observed that differing levels of disadvantage seem to have a big influence. From the previous post, the way the data has been subdivided into small groups, seemed significant. I suspect the two are connected, with disadvantage being a major reason why a small ethnic group would be considered in isolation in the statistics.

Returning to the five ethnic groups with highest exclusion rates, they were also the groups with highest FSM rates. (For comparison, 20% of White British pupils have Free School Meals.)

Ethnicity FSM% 
Any other Black background 32.8
Black Caribbean 37.6
Mixed – White and Black Carribean 38.8
Gypsy/Roma 47.2
Traveller of Irish heritage 63.5

We might need additional factors to explain why some groups have a lower exclusion rate than their FSM numbers would suggest. We might find out more about the causes of exclusions if we could explain why Bangladeshi pupils, black African pupils or mixed race pupils with white and black African parents are less likely to be excluded than white British pupils despite having proportionately more FSM pupils (although still below the 5 groups mentioned above). There may be factors around the history, geography or sociology of these groups that explain why disadvantage seems to affect some ethnic groups less than others. However, I’m not sure we actually have data that would enable us to do that*. I suspect our efforts might even drift into relying on stereotypes (“oh, this ethnic group is a bit more studious and compliant than this ethnic group”).

Overall though, when it comes to explaining why some ethnic groups have an exclusion rate much higher than white British pupils, there seems little need to look for any explanation beyond economic disadvantage and the habit of government statisticians to subdivide disadvantaged groups more than other groups. This should not be the end of the story. This does not answer the question of why entrenched disadvantage in a small ethnic group will result in more exclusions. More should be done to address disadvantage. However, it seems unlikely that schools can do this alone, and there is no obvious case that it is schools that are to blame for the pattern of exclusions.

*More data relevant to these questions may be available in the future when more information from the 2021 Census is published. There is also a discussion to be had about what data schools collect. One reason why we end up discussing ethnicity and disadvantage so much in education is because every effort is made to collect and process detailed information on ethnicity and indicators of disadvantage. One wonders whether we might have completely different perspectives and priorities if, say, information on family structure and level of parental education was the priority.



  1. […] Why Do Some Ethnic Groups Have Higher Exclusion Rates? Part 2 […]

  2. […] Why Do Some Ethnic Groups Have Higher Exclusion Rates? Part 2 […]

  3. […] likely to be factors that correlate with disadvantage. Some ethnic groups are more disadvantaged, and some of the most disadvantaged ethnic groups have the highest exclusion rates. Academic achievement might seem like a risk factor and evidence of vulnerability in its own right, […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: