There was an apparent rise of 36 per cent in antisemitic incidents recorded in 2016 — a total of 1,309 such incidents. It was, stated a headline in this newspaper, “the worst year on record.” “It seems,” maintained the JC, “that Jew hate is simply becoming more prevalent and more open.”

Maybe. But not so fast. As the CST acknowledges, antisemitic incident data should be read carefully. Like all forms of hate crime data, they rely on people coming forward to report incidents when they occur. So before jumping to conclusions about trends, one critical question must be asked: has anything happened around how incidents are reported or recorded that might account for any change observed?

In the case of antisemitic incident data, the answer is yes. Here’s why.

When an incident occurs, victims have a choice. First, they have to decide whether to report it or not. In most cases, they do not — according to a 2012 EU study conducted by JPR, about seven in 10 incidents go unreported. Thus, as is the case with every annual figure, the 1,309 figure is an undercount, considerably lower than the real count. Yet reporting rates fluctuate, and the evidence suggests that active encouragement through advertising and awareness-raising yields positive results.

Second, having chosen to report, victims then have to decide to whom? The most obvious options are the police and/or the CST. Until quite recently, these two channels had no established system of sharing information, so CST only recorded those counts reported directly to them from victims or witnesses. However, to its credit, CST has signed several information sharing agreements with the police since 2011: initially, with individual regional forces, and then, in 2015, nationwide. So, now, as well as recording incidents reported directly to them, CST also records incidents reported to the police. This is a huge breakthrough in terms of accurate recording, but from a statistical perspective, it constitutes a break from the past. Figures previously based on information coming through one channel are now based on two. Straightforward comparisons of recent counts with those from previous years, particularly pre-2011, become impossible.

However, even after accounting for this factor, 2016 still appears to have been a notably bad year. Not the worst on record, but certainly one of them. Yet there is another reason to question the “worst year ever” hypothesis. It is very difficult to ascertain empirically whether reporting rates have improved over time, and if so, by how much.

The CST, again to its credit, has gone to great lengths to encourage and facilitate incident-reporting in the community, with the support of government funding. In 2016, the police promoted reporting especially strongly, particularly in the aftermath of the referendum, and concluded that part of the increase in incidents recorded can be accounted for by a significant increase in the number of incidents reported.

Given these interfering factors around the annual figures, is it possible to make any kind of assessment of the trend over time? To understand a trend, one needs a consistent data source. Arguably, the best one that exists is incidents reported directly to CST. Focusing exclusively on these, we see that they fluctuate year on year, but the trend since 2000 shows a slight increase of about 15 incidents per annum. Bearing in mind all that CST has done over many years to encourage and facilitate reporting, it is distinctly possible that at least part of that rise can be accounted for by this factor.

CST incident data comprise one of the most important sources of information about antisemitism in the UK, and people who discount them often have another agenda. Yet, without very careful assessments and adjustments, the raw data should not be used to track trends. There are simply too many interfering factors at play. For similar reasons, in 2014, the UK Statistics Authority removed “National Statistics” designation from police figures based on recorded crime data because, while acknowledging their value, they concluded that they are “not suitable for long-term trend analysis”.

What can be said, with a high degree of confidence, is that the counts CST reports nowadays are much closer to the actual number of incidents that occur in any given year than they used to be. But whether they demonstrate that 2016 was the worst year since records began is an entirely different matter, and demands much closer investigation.

Jonathan Boyd is Executive Director of the Institute for Jewish Policy Research