Terrible "Bikes are to Blame" _SF_Chronicle_ Article

=v= Today's _San_Francisco_Chronicle_ has a front page story
about "bicyclists to blame," which is a shockingly insensitive
response to the recent tragedy where an entirely-to-blame
motorist killed two bicyclists.

Bicyclists blamed twice as often as drivers
http://www.sfgate.com/cgi-bin/articl.../MNU3VOB22.DTL

=v= I am at least grateful that the headline used "blamed"
rather than "at-fault," though the latter was used throughout
the article. Basically they regurgitated California Highway
Patrol statistics, which are based on police reports.

=v= The article purports to be a "data analysis" but utterly
fails to analyze any data.

=v= The data itself is classic GIGO, anyway. "At fault" in
a police report is the opinion of the officer on the scene,
subject to his or her biases. Many of us who've interacted
with police around the Bay Area have seen this personally.
In addition to anti-bike bias, there's also an anti-work
bias: no real investigation, no forensics, lost evidence,
less paperwork. Of course an investigation and/or hearing
can and does result in different "at fault" determinations
than the police report, but that's not in this data.

=v= (Also not in the data are incidents where the police
discouraged or outright refused to do a report.)

=v= In the case of fatalities where no independent witnesses
were present, absurd narratives from the killing motorists
about cyclists coming out of nowhere, swerving against the
laws of physics, etc., are taken at face value. Cyclist is
"at fault," no charges are filed. The group Right of Way
did a thorough analysis of car/bicycle collision police reports
in the New York City area and found that the lack of witnesses
had a devastating effect on the "at fault" statistics.

=v= Statistical crosschecks can find trends across population
samples, and Right of Way couldn't find anything correlating
with the absence or presence of witnesses. Taking out the
uncorroborrated, unreliable data and controlling for the other
variables, they concluded the exact opposite: the motorists
are usually at fault. The *type* of motorist usually at fault
had demographic indicators.

=v= Then they did the same thing for car/pedestrian collisions.
And got the same results.

=v= THAT'S data analysis. The _Chron_ ran some demographics
(sex and age) to make a totally unenlightening pie chart and
bar graph, then quoted a bunch of people's opinions.
_Jym_