On Friday, February 16, 2018 at 9:33:11 AM UTC-8, sms wrote:
On 2/16/2018 9:03 AM, jbeattie wrote:
On Friday, February 16, 2018 at 7:49:50 AM UTC-8, Joerg wrote:
On 2018-02-15 12:16, sms wrote:
On 2/15/2018 9:35 AM, Joerg wrote:

snip
... Of course
we will soon see claims that this is all self-interest by Trek, whose
sole aim is to sell more lights.


Everybody should know that such articles aren't very suited to foster
sales of their own products but lights in general.

True. But it's a convenient excuse to dismiss the findings of such
article. Even though Trek is not a major supplier of bicycle lights, you
can already predict the narrative that will spew forth. You saw it
already with the Odense study.


Yep, afterwards the usual denialist stuff roll in.

Any study showing that a little mag-dyno blinky decreased daytime solo accidents by 27% is suspicious on its face. If that doesn't raise an eye-brow, you don't have eye-brows.

I think, that as a lawyer, you likely understand that the fact that the
DRL is powered by a magnetic dynamo is irrelevant in terms of its
effectiveness. You probably also understand, unlike some others, that
correlation and causation are not the same thing.

O.K., let's put it this way -- a very low-powered flasher located at hub height.

I have been saying correlation is not causation all along -- and thus my comments about the Odense study. They assume that the data establishes causation -- except when the data doesn't pass the smell test. Here is the explanation given in the report for the reduction in solo accidents:

"The self reporting of accidents is on the other hand somewhat problematic. Prior to the study, it was expected that the bicycle running lights would reduce the occurrence of multiparty accidents involving cyclists. The initial results suggest that this is a very likely outcome, as the accident rate is 45% lower for the treatment group than for the control group, when all reported accidents are taken into account, and 61% lower when only accidents with personal injury is taken into account. The bicycle running lights were, however, not expected to affect the occurrence of solo accidents, but the initial results show; having made sure by closer examination of the accident descriptions that the accidents in question are in fact solo-accidents, that the accident rates for solo accidents are 24% (all accidents) and 27% (person injury accidents) lower for the treatment group than for the control group; the effects close to being significant.

It is likely that this apparent effect on solo accidents of the bicycle running lights actually reflect a systematic under-reporting of accidents in the treatment group due to an inherent bias in favour of the bicycle running lights amongst the members of the treatment group. During the project, additional questionnaires were carried out in order to evaluate the design and functionality of the bicycle running lights. From the data gathered here, it is evident that the members of the treatment group were very fond of the running light as they found the bicycle running lights very convenient, e.g. they did not have to buy batteries any more, they did not have to fear being stopped by the police for having forgotten their bicycle lights, they felt very safe with the bicycle running lights etc. As a consequence it is likely that the treatment group has been somewhat strategic in their reporting of accidents by omitting some of the minor bicycle accidents; as reflected by the apparent under reporting of solo accidents in the treatment group. The apparent effect for solo accidents is almost the same for relevant subgroups of solo accidents, see Table 9, which suggests that the underreporting is general and not associated with certain solo accident types.

O.K., so could the same light-loving study group be under-reporting multi-rider accidents?

61% reduction in PI accidents by using a weak hub-height blinky? Hmmmm. I ride and drive around bikes all the time during the day, and I've never seen a bike because it had a little Knog light. Not during the day.

Also, the Reelights are practically useless off-angle (in candela):

Vertical angle 0° Horizontal angle
−80° −20° 0° 20° 80°
Front light outer diode 0.02 0.22 4.43 3.74 0.05
Front light inner diode 0.05 0.59 5.50 2.15 0.05

The table doesn't even address vertical angle. I guess the cars are really low to the ground in Denmark.

-- Jay Beattie.