Last month four teenagers were murdered in London. The fact their deaths occurred in the space of ten days only highlighted the tragedy.
In recent years teen murders have been seen as a barometer of ‘Broken Britain’. An increase indicates a breakdown in society and its values, a decrease… well, let’s gloss over the decrease.
The public and media uproar peaked in June 2008 with the murder of 16 year-old Ben Kinsella, the brother of Eastenders actress (and now anti-knife crime campaigner) Brooke Kinsella.
One of the measures introduced by the Labour Government was an increase in the minimum term for murders carried out using a knife brought to the scene, from 15 to 25 years.
But by the time this came into effect in March 2010 (following a review, then the official announcement in November 2009) the number of teen murders had decreased. In fact it more than halved, from 29 in 2008 to 13 in 2009.
In 2010 it rose to 19, although the total number of murders continued to fall.
The table below can be illustrated by any number of colourful graphs and charts but the reality is that crime appears to come randomly in waves. Trends can only be seen over longer periods.
Four teen murders in a month is not that unusual – it last happened in April 2010 – and there were five in both June 2007 and May 2008.
But imagine that you only gathered statistics between October and December 2008, the peak year for teenage murders. There was only one.
At the risk of stating the obvious, one month does not make a trend.
A similar point was made following reports of four homicides in London in a single day, July 10, 2008. Statistical analysis revealed that this was actually a predictable event, rather than an alarming development.
According to a study by David Spiegelhalter
We can’t predict individual murders, but their pattern is highly predictable. This should mean we can be ready for events that appear to be good (a long gap between murders) or bad (3 or more murders on the same day) – both events are to be expected by chance alone. But by knowing what pattern to be expect, then we should also be able to spot when something really unusual is happening.
He also makes the point that “there is no evidence for homicide rates to depend on the month, but there is a significant ‘Saturday effect’ of around 60% increase in homicide rate compared to all other days of the week combined.”
For those interested in colour charts, here are two based on the above table.