h1

How misleading was today’s Guardian article on exclusions?

March 25, 2021

As I said in this morning’s post, I spent a lot of time on Twitter from November to February interrogating the claims, particularly the statistical ones, of the conspiracy theorists alleging that the result of the 2020 November US presidential election was won through fraud. When it came to looking at the numbers, the same tactics were used again and again.

Ignore the big picture.

The 2020 election was won by the favourite. The only surprise in the result was that he didn’t win by quite as much as the polls suggested and lost some states he might have been expected to win. Yet conspiracy theorists would ignore this and talk as if Joe Biden’s victory was inexplicable.

Scour the data for anomalies in the details.

An anomaly is something improbable in the data. All elections have anomalies and there are always unprecedented features for any election result (as XKCD show here). Because of the scale of US elections then just at random there will some unlikely events. If a million things have happened then you would expect to find even a one in a million fluke occurring somewhere. If you look at the smallest possible subdivisions of the electorate, counties and precincts rather than whole states, then you get a lot of opportunities to find unlikely events. Conspiracy theorist statistics showing fraud were often based around results at county level (and occasionally precincts). It would actually be more suspicious if there weren’t any unusual results and, of course, they did not consider anomalies that worked in the favour of the losing candidate to be evidence of fraud by his side.

Assume all anomalies are explained by fraud

Once an alleged anomaly is found it was assumed to be evidence of fraud. Other explanations, like changing demographics or the effects of high turn out were ignored. Even results that were largely the same as 2016 were treated as suspicious this time. Errors made by election officials were given as evidence of fraud. Even errors made by journalists reporting on the election were seen as evidence of fraud.

It’s not that an anomaly couldn’t be evidence of fraud, but even the most serious anomalies are more like a smoke alarm going off than seeing your house burn down. It is necessary to look at all explanations. It is also necessary to ask “even if we don’t have an explanation for this anomaly, does fraud actually explain it?” Many anomalies were simply irrelevant to the result of the election, but were seen as evidence that there was fraud everywhere.

Uncritically repeat false or misleading claims.

Finally, and this is probably the most frequently used tactics, false claims (like Pennsylvania counting more mail in ballots than they sent out) or misleading information (like exit polls results in an election where a huge proportion of voters voted by post) would be repeated without any effort being made to see if they were false or misleading. Even mathematical errors, like an incorrectly calculated percentage, would be repeated without checking.

What does this have to do with education?

I was reminded of all this when I saw today’s Guardian story about exclusions being racist. I should point out it is about Fixed Term Exclusions i.e. pupils be sent home for a short time, often just a day.  I am neither bothered about FTEs as an issue (ultimately schools have alternatives if there is political pressure to reduce FTEs) or convinced there is no evidence of racism in the pattern of FTEs. However, given that the headline does not mention that the article is about FTEs, I do fear that this may impact the debate on permanent exclusions which I do care about, so this is worth commenting on to see if the article is accurate. And what I see is all the same tricks that drove me to distraction on MAGA twitter.

Ignore the big picture.

The big picture on FTEs and race is that in the most recent data the rate of fixed period exclusions (number of FTEs as a percentage of the headcount of pupils) is 3.91 for minority ethnic pupils and 5.80 for white pupils. In itself, that would be grounds for doubting racism, but it’s fair enough to look more closely. The single minority ethnic statistic hides that there’s a big difference between Asian and Black pupils, with FTE rates of 2.03 and 5.54 respectively. And if we break down data further we will find significant discrepancies in FTEs rates between ethnic groups, although we would still find most ethnic groups have a lower FTE rate than white pupils and particularly white British pupils. Nevertheless, some groups do have a much higher rate. The big picture here is that there is no clear pattern of racism, but there is inequality that cannot be simply explained by any one cause I’m aware of.

Scour the data for anomalies in the details.

If you have to describe the method used by the Guardian article it is this:

  1. Find the ethnic groups with the highest FTE rates.
  2. Find the LAs with the highest FTE rates for those ethnic groups (ignoring those with less than 100 pupils from those ethnic groups).
  3. Report the above information as a comparison with white pupils where possible.

This is a deliberate strategy of looking for the most anomalous results (favouring the racism hypothesis) and reporting only those. There is no way this method would not find dramatic discrepancies apparently favouring white pupils, even if the number of FTEs was distributed completely by chance. It is cherry picking, with even the choice of white pupils (FTE rate 5.80) over white British pupils (FTE rate 6.01) being made to increase the discrepancies found. I should point out that the data was released on February 24th, but it has apparently taken a month to subdivide it sufficiently to get the story the Guardian was looking for.

Perhaps some will say this is a fair method of finding evidence of racism, so I’m going to give them a challenge. Use the exact same method but replace “white” with “Asian” or “black”. Do you get equally convincing discrepancies showing FTEs favour Asian or black pupils? Unless the decision to ignore LAs where the relevant ethnic minority has a headcount less than 100 scuppers it, this would almost certainly find even greater discrepancies than carrying out this procedure for white pupils.

Assume all anomalies are explained by racism.

It is entirely possible that some of the discrepancies could be explained by racism. There could be some LAs where exclusion policies operate in racist ways. However, the article’s only concession to the fact that other explanations are available is an admission that it’s most extreme anomaly involving Black Caribbean pupils (the figure for Cambridgeshire) might be partially explained by the “relatively small number of Caribbean students”. No attempt is made to work out if any discrepancy is explained by social class, even though London is known for having a lot of wealth alongside its disadvantage. No attempt is made to look at the reasons for exclusions, despite these also varying quite noticeably between ethnic groups. I have no idea why the largest category for exclusions for black and Asian pupils is physical assault against another pupil while it is only the fourth highest category of exclusions for white pupils, but it seems like it might be relevant to the analysis and certainly likely to be relevant to policy suggestions to address discrepancies. But assault is not mentioned, and yet “uniform policies” are mentioned as a reason for exclusion.

Uncritically repeat false or misleading claims.

This to me is the worst part of the article. For all the cherry picking, the groups with the highest FTE rates might be discriminated against. But, when you then use this as evidence for discrimination against groups that don’t have high FTE rates, you are at best mistaken and at worst lying. The article quotes people getting this wrong without correcting it.

We are told of:

an “incredible injustice” for schoolchildren from minority ethnic backgrounds

a “PRU [pupil referral unit] to prison” pipeline for working-class black children.

BAME … children being disproportionately excluded

an “incredible injustice” for schoolchildren from an ethnic minority background [again]

Yet as pointed out above, none of the categories I’ve emphasised here are at greater risk of FTEs than white British pupils. The writer of the article knows this as these claims are all attributed to others (with the repeated claim being attributed to both “experts” and a politician) rather than claimed by the Guardian. Yet they are reported uncritically despite being at best misleading and, at worst, dishonest.

There is no excuse for reporting that consists of cherry picking statistics to fit a hypothesis that contradicts the big picture, presenting only one explanation and then repeating objectively false claims uncritically to ensure people are misled. If this Guardian article alone convinced you that exclusions are racist, then you should be equally convinced that Democrats stole the US election. If you aren’t, you should ask yourself why, before our schools end up looking like the riot on Capitol Hill.

All statistics cited by me in this blogpost come from here.

4 comments

  1. If they cause six times as much trouble in classrooms, this can only be a good thing.


    • That’s not what exclusion rates show. They are the number of exclusions divided by the headcount of the ethnic group. So a small ethnic group in an LA could have a really high exclusion rate and minimal impact on classrooms. This is why subdividing into LAs is such a bad idea for analysing the stats, even with the decision to ignore headcounts under 100.


  2. It’s all a little too much “Whose line is it anyway?” . where everything’s made up and the points don’t matter.


    • By that I meant that the chosen narrative (of the Guardian in this case) is all that matters.



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 )

Google photo

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

Twitter picture

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

Facebook photo

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

Connecting to %s

<span>%d</span> bloggers like this: