Monthly Archives: December 2008

Whatever happened to bird flu?

Now we’re in full flu and norovirus season, I wondered what had happened to bird flu. In 2006 and 2007 it seemed to be the biggest threat to the world, never mind the UK.

Checking the World Health Organisation figures, I realised how few people had been killed by the ‘deadly’ H5N1 virus. 32 in 2004, 43 in 2005, 79 in 2006, 59 in 2007, 30 in 2008. This is globally. You are about a thousand times more likely to be struck by lightning (24,000 deaths a year).

I know that there was and still is a theoretical risk that it could be much worse. However, I’m interested in the fact that now there’s a recession on, bird flu is no longer newsworthy. People are still dying, and birds with H5N1 are still turning up in various places around the world. Germany had a case in October, and the H7N2 bird flu was in the UK in June, but no-one’s really interested anymore. Strange.

1 Comment

Filed under News, Politicians, Statistics and simplicity

Knife crime

I woke up today to hear a very hyped story about murders involving knifes. The Conservatives, needing to make Labour look soft on law and order, told us that ‘Knife killings have soared to a record high in England and Wales… citing police figures obtained under freedom of information laws.’ (Reuters)

As with all reporting of government statistics, there is some confusion as to what they mean. Reuters reported that there were 277 killings between April 2007 and March 2008, whereas the Times said this figure was ‘so far this year’. Reuters also got it right when it reminded us that the figures include all ‘sharp instrument deaths’ (including bottles and glasses, spears(?) and so on, whereas the papers assumed they were all stabbings with a knife.

There is no argument that this number is a record high (since records began in 1977). However, for this to represent evidence for a breakdown in society (‘underlying causes of crime such as drugs, family breakdown and gang culture’ Conservative home affairs spokesman James Brokenshire) it’s important to make sure that the change is significant. After all, the Times also reported that:

In spite of the increases, as a proportion of all homicides, deaths caused by a knife or other sharp instrument have remained broadly stable for the past 30 years.

Does this mean that knife murders and all murders have rocketed?

Not really.

Chart of homicides in England and Wales, 1964-2008

Homicides have been increasing steadily, but have been fewer in the past few years. It looks like they’ve trebled since the 60s, but once population increases are stripped out they’ve only doubled. However, there’s still an increase over all.

Although we can’t see the full figures yet, my bet is that the homicide rate has fallen, but knife killings have increased. Because all these movements are quite small (e.g. 277 this year versus 259 last year, I think), the proportion is ‘broadly stable’. Indeed, because killings are such a rarity in the UK, a small change  of 18 more is a big percentage increase, but easily within an expected range.

Furthermore, the changes in methods may indicate better policing (and so could be a good news story). If ‘sharp instrument’ killings are going up as killings go down, does this mean that shootings have fallen? Is the increase in knife crime due to guns being taken off the streets?

Finally, it’s worth pointing out that the England and Wales homicide rate is really very low at 1.37 murders per 100,000 population. This could be compared to:

Scotland 2.56
Europe 5.4
USA 5.7

As always, it needs to be said that any killing is a tragedy for those involved. But there’s no ‘soaring’, no problem growing dangerously out of control. The story of knife crime is hyped by the two main political parties (to appear tough) and by the media (to beat the politicians).

Leave a comment

Filed under News, Politicians, Statistics and simplicity

British National Party membership: the null hypothesis

A few weeks ago the British National Party membership was leaked online. Although this was a breach of individuals’ privacy, the media seized upon the list and provided a great deal of analysis, to satisfy the curiosity of those who want to know ‘who are the BNP?’ (see the Telegraph,  the Guardian, the Times, and many more). We’ll leave aside the question as to whether the media and public would have been so curious if it was a mainstream party, and focus on the assertions that the BNP are either ‘ordinary’ or ‘different’.

What interested me, from a ‘bad social science’ perspective was how much of the analysis was missing a check as to whether the data was actually significant, i.e. whether this analysis showed the members to be any different to the larger population or from political activists. An uncontroversial example would be the finding that BNP members were mainly men, when this would also be the case in other political parties too. Indeed, less than a fifth of our MPs are women.

Similarly, the Telegraph article notes that ‘at least 30 people named on the list have criminal convictions’. However, at least a third of all men in the UK have a criminal conviction by the time they are 30, and in poorer areas this is likely to be even higher. So the members might be less criminal than the public at large. For comparison, we’d also need to know how many members of other political parties have criminal convictions.

More amusingly, one Guardian journalist wrote:

And there are, of course, the serious headbangers: BNP leader Nick Griffin may boast that this list proves his members are not “skinhead oiks”, but there are still martial arts fanatics, people suspended for “inappropriate tattoos” and at least seven email addresses incorporating the number 88, which is neo-Nazi code for HH, or Heil Hitler. Other email addresses are lordhawhaw, saxondelight, darkenedangel and napalmdeath. Someone gives his pastime as “World War II reenactment”.

Now these people might well be ‘serious headbangers’, but this evidence doesn’t tell us anything. The tattoos could be on someone’s face and, like many employers, the BNP feel it doesn’t fit with the image they wish to portray. But many people, some ‘headbangers’, some not, have tattoos like this. The ‘napalmdeath’ email address might have been chosen as a tribute to Napalm Death, a band whose anarcho-punk origins, cover of the Dead Kennedys’ Fuck Off Nazi Punks, and contributions to anti-racism activism, make them an unlikely ally of the BNP. Perhaps the member with this address doesn’t know this history, or has switched sides (it does happen). And to note that seven email addresses contain ’88’ requires comparison to other numbers too. It could be the year of someone’s birth or connected to another ‘memorable name’ etc.  If a similar number of email addresses included ’87’ or ’89’  I’d be wondering whether the ’88’s were just coincidental. It would be also useful to look at the context… there’s a world of difference between AryanNation88 and dave.bloggs88 or dave.bloggs120388.

What the analysis should do is test the theories against a null hypothesis: is this sample of people, email addresses and so on, different to the wider population, or could those characteristics arise from any sample of UK citizens.

Leave a comment

Filed under bad social science, media, Politicians