Facts are stubborn, but statistics are more pliable*

Eight months ago, Oxfordshire County Council (OCC) needed to find a saving of £600,000 from its annual budget. They weighed up the pros and cons and after a brief consultation with themselves over tea and a chocolate biccy, they decided to ignore the easy option of sacking a few very expensive senior staff or axing one of the health and safety battalions (dustbin inspection division) and, instead, followed the example of Swindon Borough Council to withdraw the funding for the County’s speed cameras.

Within 3 months, after pressure from what the Oxford Mail called (in the usual understated language of a parochial newspaper) a ‘storm’ of protest from road safety groups (a.k.a. the police) and members of the public (Private Eye’s Sid and Doris Bonkers), OCC announced that they planned to reinstate the cameras.

You’ll note that this volte-face was in response to ‘road safety’ campaigners. There were no statistics to justify the decision and nor could there have been after a statistically insignificant three months. Today the cameras have gone back on and lo – we have some statistics that magically support the decision that was taken 5 months ago. According to the BBC:

Data released by Thames Valley Police revealed in the six months after they were switched off there were 70 slight injuries, 13 serious injuries and no fatalities from a total of 62 accidents at fixed camera sites.

In the same period the year before [August 2009 to January 2010] there were 55 slight injuries, 13 serious injuries and no fatalities from a total of 60 accidents.

Away from camera sites there were 867 collisions with 982 slight injuries, 179 serious injuries and 18 deaths during the period the cameras were turned off.

In the same period the year before there were 885 collisions with 999 slight injuries, 160 serious injuries and 12 deaths.

There are difficulties with these reports – this bit is necessarily dull, I’m afraid. I’m going to be bored typing it so please try and stick around to keep me company.

The first thing you will notice is that the data comes from Thames Valley Police, the very party that was lobbying for, and benefits from, the reinstatement of the cameras.

Secondly, there is no causation data – these accidents could have occurred for a number of reasons including the treacherous December weather that occurred during 2010: they do not necessarily relate to speed. The data are for a six-month period, again statistically insignificant, but not unreasonable given that it is directly compared with the same period for the previous year. However, neither data set can be particularly valid as an indicator for the effectiveness of speed cameras.

Another concern is the method by which these statistics are gathered. The cameras will photograph a speeding vehicle but they do not, as far as I’m aware, have any clever circuitry that detects and records an accident to conveniently produce an accident/injury database. Immediately, this leaves the record open to a degree of human error/interpretation and the possibility of reporting bias: whether or not an accident and/or an injury gets reported will, to a large extent, depend on whether the police are present at the time, the degree of damage relative to the value of the vehicles involved, whether the vehicles involved are legal and roadworthy and the litigiousness of the individuals involved.

And if anybody knows, it would be interesting to hear how pedestrian casualties are reported – presumably, to make road accident data meaningful, they must be included if a vehicle is involved but how are those people who trip over kerbs reported? If they happen to be close to a camera site, are they included in the road accident statistics?

And lastly on the methodology of data-gathering, who determines the limits of the ‘camera site’? Is it within 50 metres of a camera? 100? 200? Is it always the same or is this, too, liable to reporting bias?

Ignoring the slight injuries for the moment, there were 62 accidents at camera sites in the six months after the cameras had been switched off. This compares to 60 accidents while the cameras were working (we are not told but have to assume that this figures was also for camera sites), an increase of 2 (3.3%). We have no data about traffic growth over the year and this may well have increased by a similar percentage. As we’ve already determined, the causes of the accidents are not given – so we do not know if weather played a part and nor do we know if confusion had any effect. In any event, the increase is marginal, at best.

Serious injuries at 13 and fatalities at 0 remained unchanged after the cameras went off.

The increase of 15 slight injuries for the period, an increase of almost 27.3% over the previous year’s figure of 55, seems suspiciously high for such a minimal increase in accident rates. It’s possible that more cars with full occupancies were involved in 2010 compared to 2009 or perhaps there was a bus accident? As discussed above, if the statistics include pedestrian casualties, there could have been a significant number of slips and falls due to the severe winter that we experienced.

When we turn to the figures away from camera sites, there is a subtle shift in reporting terms. We’re no longer talking about accidents – we are talking about collisions. As collisions are a subset of total accidents – and as they imply more destructive forces – my suspicious mind would immediately question the general seriousness of the accidents at camera sites. How many of them involve a collision? Are we deliberately recording accidents in a different way at camera sites in order to make them appear more dangerous than they are? Or perhaps the total accident figures away from camera sites highlight an uncomfortable truth about the comparative safety of modern vehicles that the vested interests do not want us to know?

Whatever the truth, the number of collisions in areas not controlled by cameras is down by 18 (2%) which might suggest a number of things, including the fact that cameras are not necessarily a determining factor in accident reduction. If they were, I suggest that Oxfordshire’s cameras are misplaced as all the fatal accidents are obviously occurring at non-camera locations.

Overall then, the data provided are effectively meaningless but have been presented in a fashion that superficially appears to support the case for switching the cameras back on. What the data don’t tell us – and what the anti-car lobby do not want you to ask – is whether £600,000 of your money is a reasonable price to pay to prevent the equivalent of 4 accidents per year, none of which result in serious injury and none of which are necessarily the result of speed – indeed, there is no information here to confirm that a single one of these accidents at the camera sites is related to the cameras’ stated purpose.

Which brings me to a conundrum: if, as the anti-speed lobby says, speed cameras are an effective means of reducing speed, there can be no accidents due to speed and, therefore, quoting accident statistics is effectively meaningless. The corollary of this is that if we are meant to believe the accident statistics are significant, then speed cameras clearly do not do the job they are meant to do.

It is very hard to avoid the critics’ view that speed cameras, generally, are no more than revenue-raisers. Their existence clearly depends on some extremely pliable statistics.

PS As an interesting coda, Swindon’s much longer experience seems to support the view that speed cameras have no positive effect in accident reduction.

*Mark Twain

This entry was posted in Adventures in Time Travel, Big Brother, Health & Safety, Politics, Roads, Uncategorized and tagged , , , , , . Bookmark the permalink.

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