Blue Monday cycling blues (2)

Since I published my comments this morning on motor insurer LV=’s survey into cycling-related accidents in Britain, I’ve been contacted by a senior company press officer. Emma Holyer has provided a little more information than is contained in the press release and Independent news report, but I remain totally unconvinced by the arguments presented.

The survey was carried out by market researchers YouGov, and Holyer appears to be arguing that this in itself lends credibility to the exercise. It doesn’t. A company such as LV= is unlikely to have the resources and expertise required to carry out a large-scale statistical survey, and it is standard practice to contract the work out to professional pollsters.

In what could be construed as an attempt to sow divisions within the cycling community, Holyer says that LV= has been working with Cycling Plus magazine, and adds it is “unfortunate” that CTC chose to publish its reaction to the story when Police-reported injuries and collision figures are not directly comparable. LV=’s figures relate, she says, to collisions amongst all road users – pedestrians, other cyclists and motor cyclists – as well as motorists. The argument is that the numbers cannot be compared as the majority of collisions go unreported.

As for the statistical survey itself, Holyer outlines how the reported figures were calculated…

Among all road users (of which there are 33,809,493, according to the Office of National Statistics), 15% (5,071,423) have at some point been involved in an incident with a cyclist. Of the 608,570 accidents in 2008, some 264,570 occurred in the first half of the year, and 344,000 in the second. It is this that gives rise to the 29% increase quoted in my first comment.

The total sample size in the LV=/YouGov survey was 2,193 adults, and the exercise was carried out online during the period 18–20 November 2008. Holyer says that the figures “have been weighted and are representative of all GB adults”.

CTC is correct in stating that the 29% figure can be accounted for as a seasonal variation. Being a former research scientist, I know a thing or two about statistics, and LV= is clearly misusing them in this case. It would take much to convince me otherwise, and this would require a thorough audit of both the entire dataset and study methodology. In statistics it is often said that the questions asked are more important than the results obtained. YouGov knows what it’s doing when it comes to market research, but the specific questions in this case will have been framed by LV=.

One must of course be careful when comparing and contrasting accident figures of different types, but LV=’s claim for a “surge” in cycling-related accidents is way off the mark. The data simply do not show this. As for the reference to the majority of cyclists being unfamiliar with the Highway Code, I have no doubt that this document goes unread by many. Just as it is among motorists.

I suspect that many minor accidents involving motor vehicles go unreported owing to insurance concerns, and when quoting data in support of arguments for legislative change, lobbyists have a responsibility to cite only those that are both reliable and meaningful. It may make the PR message more difficult to convey, but relevant caveats should also be included.

LV=’s arguments have been torn to shreds not only by cycling lobbyists, but also Independent readers who commented online following the Press Association piece. This misuse of statistics and poor presentation is bad PR for LV=, and it is no wonder that people are thinking the worst of the company for it.