Antisemitism and the misuse of statistics

Last week saw the publication of a very strange report entitled ‘How serious is the threat of antisemitism in Britain today?’ and published on the website of the controversial Jewish Voice for Labour group. It was written by Alan Maddison, an ‘independent strategist’ and associate member of the group with a history of previous attempts to discredit claims of antisemitism. Maddison begins by noting some recent media headlines about antisemitism, and then proceeds to set out the following position:

A review of the latest evidence, presented here, suggests that these headlines are unnecessarily alarmist. The pro rata risks for assaults are lower for Jewish people than for those from other races or religions. The increase in antisemitic hate crimes reported to the police is around half that reported for other victim groups. Finally, a Jewish person is 50 times more likely to be the victim of a general assault than one motivated by antisemitism

Having set out this position, Maddison promises to ‘review the latest evidence behind these more reassuring statements, and place antisemitic hate crime in a broader perspective’. Unfortunately, the way in which he does so discredits the entire enterprise, and calls into question Jewish Voice for Labour’s motivations in publishing his report.

I’ll go through Maddison’s arguments one by one.

Rates of antisemitic hate crime

Maddison’s first argument is an attempted refutation of the Community Security Trust’s claims of a 30% rise in antisemitic incidents. He makes this argument by citing evidence appearing to show that there has only been a rise of 15% in antisemitic crime. Even if we accept Maddison’s figures, how can a 15% rise in antisemitic crime be reassuring to British Jews?

Perhaps realising that he won’t get far with this approach, Maddison then presents a chart showing the number of hate crimes suffered per 1000 members of each victim population. This appears to show that Jewish people are the group that is at the second highest risk of hate crime (the highest risk being for members of groups targeted because of their race). However, Maddison focuses on the rates of hate crime involving violence. Judging by his chart, the rate of violent hate crimes against Jews appears to be lower than the rate of violent hate crimes against LGBT people and people targeted for their race – but higher than the rate of violent hate crimes against the disabled and about the same as the rate of violent hate crimes against followers of other religions. It is unclear how this is supposed to be reassuring for Jews (or anybody else).

Maddison then makes one of the strangest statistical arguments I have ever seen, observing that ‘hate crimes involving violence… against Jewish people… only represented 10% of antisemitic hate crimes… compared [with] an average of 33% for all other victim groups’. Although this claim appears mathematically plausible from his chart (one cannot be quite sure because the actual numbers are not given), it only reflects the fact that rates of non-violent hate crime against Jews were so much higher than rates of non-violent hate crime against all other groups except for those targeted for their race (at least going on what seem to be Maddison’s numbers). For example, the rate of violent hate crimes against disabled people is (to judge by Maddison’s chart) somewhat lower than the frequency of violent hate crimes against Jews, but because the total rate of hate crimes against disabled people is (again judging by Maddison’s chart) much lower than the total rate of hate crimes against Jews, the percentage of hate crimes that are violent comes out higher in relation to disabled people than in relation to Jews.

Think about that for a moment. If we take it seriously, what the 10% versus 33% comparison actually means is that while Jewish people are at least as likely to be the victims of violent hate crime as members of all other victim groups in Maddison’s chart except those targeted for their race, they are – on top of that – far more likely to be the victim of non-violent hate crime than members of all other victim groups except those targeted for their race.

Is that really reassuring? Trying to reassure Jews that ‘only’ 10% of antisemitic hate crimes are violent would be like trying to ‘reassure’ people in malaria-risk areas that ‘only’ one in every three to six hundred malaria victims actually dies of malaria. The ‘one in every three to six hundred’ statistic merely reflects the fact that, in addition to killing vast numbers of people every year, malaria also causes non-fatal suffering to still vaster numbers. That is the opposite of reassuring. And so it is for the rates of antisemitic hate crime, because nonviolent hate crime is in addition to violent hate crime, and does not somehow dilute it or make it more bearable.

Having made this baffling argument, Maddison returns to the idea that there is a victim group whose members are more at risk of hate crime than Jews, i.e. people targeted for their race. But why does he do so? If he thinks Jews should find it ‘reassuring’ that hate crime against them is less common than hate crime against people targeted for their race, then would he also expect disabled people to find it ‘reassuring’ that hate crime against them is (on his own figures) less common than hate crime against Jews?

One would hope not. Following that line of reasoning to its logical conclusion would lead us to a really dark place.

’50 times more likely’

Maddison’s next assertion is that ‘[a] Jewish person is 50 times more likely to suffer an assault unrelated to antisemitism [than an antisemitic assault]’. This claim is completely illegitimate because he has no data at all on how likely a Jewish person is to suffer such an assault.

Indeed, such data do not exist. Maddison’s ’50 times’ figure was arrived at by comparing a British Jew’s risk of suffering an antisemitic assault with a member of the general British population’s risk of suffering an assault of any kind. That is comparing apples with oranges.

Unfavourable opinions

Maddison’s fifth point is that ‘Unfavourable opinions about British Jews are much lower than for people from other races or religions’, and for this he cites Pew Centre figures showing the frequency with which survey respondents self-reported having an ‘unfavourable view’ of various ethnic and religious groups. This research found 7%, or what Maddison calls ‘a small minority’, of people in the UK holding unfavourable views of Jews. Again this is apparently supposed to be reassuring. But figures such as these are likely to understate prejudice against Jewish people. Statistician and antisemitism researcher Daniel Staetsky writes as follows:

Attitudes towards ethnic and religious groups are a sensitive topic in contemporary Britain and elsewhere in the West. As a general rule, negativity towards certain groups is not a sentiment that is easily admitted to and/or readily expressed. There is often a degree of apprehension about holding and vocalising indiscriminate negativity towards whole groups defined by religion, ethnicity or lifestyle. Within the context of this survey, that means that the respondents may have been somewhat cautious about revealing the true nature of their feelings toward certain groups, and may have given responses that were socially acceptable instead, i.e. responses that were unlikely to result in them being negatively judged. In survey science jargon the outcome of such under-reporting is called social desirability bias.

For the reasons that Staetsky explains in the above quote, a number of surveys (including the one that Staetsky led) have additionally or instead given respondents a series of anti-Jewish and pro-Jewish statements and asked them whether or not they agreed. Such surveys have found considerably higher levels of agreement with anti-Jewish statements than Maddison’s quoted figure of 7% might lead one to expect. For example, 30% of respondents to Staetsky’s survey of British adults, carried out by Ipsos MORI in 2017 for the Institute for Jewish Policy Research and commissioned by the Community Security Trust, agreed with at least one of six anti-Jewish statements and/or reported having a negative opinion of Jews, while a YouGov survey commissioned by the Campaign Against Antisemitism the same year found 36% of British adults agreeing with at least one of seven anti-Jewish statements. Yet Maddison writes as if this body of research did not exist – despite the fact that it is more recent than the Pew survey that he cites, and that his article promises to ‘review… the latest evidence’.


Considering all these points together, it is hard to avoid the conclusion that evidence has been deliberately downplayed (the 15% increase, which Maddison appears to consider reassuring because it is less than a 30% increase), cherry-picked (the 7% claim, which ignored more recent studies), distorted (the bizarre 10% versus 33% comparison), and completely made up (the ’50 times’ figure in relation to non-existent data on non-antisemitic violent crimes against Jews). But perhaps most worrying is the research question that Maddison sets out to answer in the sub-heading of his post: ‘Are British Jews the target for more hate crimes than other victim groups?’

Strangely, Maddison’s figures do appear to suggest that British Jews are more at risk of hate crime than several other victim groups – but I am not aware of any organisation that has claimed that British Jews are the target for more hate crimes than other victim groups, either in the specific or in general. Speaking for myself, I would make no such claim and I would attach no significance to it if it were made by somebody else. Such a position certainly has not been taken by either of the two organisations tagged in Maddison’s post, i.e. the Campaign Against Antisemitism and the Community Security Trust. It is an invented position set up only to be knocked down by the eccentrically interpreted ‘evidence’ that Maddison assembles. And it has no relevance for how we as a society should deal with hate crime, antisemitic or otherwise. In posing and then answering a question of this nature, Maddison is not only committing the ‘straw man’ fallacy but competing – or encouraging others to compete – in what is colloquially known as the ‘Oppression Olympics’.

What does it matter whether British Jews are more or less likely to be the target of hate crimes than any other specific victim group? It does not matter at all. What matters is that Jews are the target of hate crimes and other antisemitic incidents, that the frequency of such crimes and incidents is increasing, and that something has to be done.

2 thoughts on “Antisemitism and the misuse of statistics”

  1. What numbers are in the denominators of Maddison’s ‘pro rata risks’ chart? ‘Race’ should be the total BAMI population; ‘LGBT’ the total LGBT population. But it is wrong to use the total Jewish population in the denominator for antisemitism. Unlike the BAMI population, most Jews are not identifiable as such! The correct number to use is those who are visibly Jewish (including for example regular walking to synagogue on Shabbat) or who choose to manifest their Judaism in other ways, for example by speaking out at anti-Israel meetings! (a point which David Collier makes). Why does Maddison even go down this ‘pro rata’ road? It can only be for what Dr Daniel Allington calls the ‘Oppression Olympics’. A pointless exercise. (In fact he uses a figure of 4 million for the ‘religious’ population – where does this come from?).

    Now look at his table ‘pro rata risks’ (NB the title is wrong, column 2 shows only violent incidents), Maddison concludes that the chance of assault for a resident of England and Wales (he says ‘Britain’ but that’s wrong) is ‘50 times greater than the risk for a British Jew being the victim of an antisemitic assault.’ And ‘the Jewish community is exposed to only one fifth of the pro rata risk for hate crime assaults compared with victims of racial hate crimes (0.41 versus 1.9).’ This is a false comparison. It’s ‘apples and pears’. Not only are Jews vulnerable to the assaults in the first row of the table as well as those in the fourth row, you cannot compare for example a robbery with violence in the street to an antisemitic assault. They are simply different offences. Being Jewish carries an extra risk – not a lower risk.

    And of course there is masses of evidence of antisemitism which Maddison ignores in his rush to assume the problem away. Take the experience of Jewish students at UK universities for example. The JPR’s 2011 National Jewish Student Survey found that 42% of Jewish students had witnessed antisemitism or been a victim of it. A shockingly high figure!

    Jonathan Hoffman

    1. Excellent points! It looks as if the only guiding principle in the preparation of Maddison’s JVL report was the imperative to come up with the lowest numbers possible.

Comments are closed.