ICBC Fudged their math!!!!!!!!!!!!!

[rockcrete]

This is an email I received from an as yet un-named source.

In the August 2007 article : "The Safety of Right-hand-drive Vehicles in
British Columbia", by : "Peter Cooper, Wayne Meckle, Glenyth Nasvadi, and
Sandi Wiggins", it was stated that the "Odds Ratio (for Table 2) is
EQUIVALENT to the Relative Risk = 1.272 / 0.886" (Page 9). From my
recollection of elementary statistics, Relative Risk for Table 2, should be
calculated as p1 / p2, where p1=201/(201+158)=0.5599, and
p2=187/(187+211)=0.4698, giving rise to p1/p2 = 0.5599/0.4698 = 1.19. This is
obviously NOT equivalent to 1.44 as quoted. The only situation where O.R. is
close to R.R. is when both p1 and p2 are small ..... definitely not for
Tables 1 and 2.

I am utterly surprised that for someone familiar with SPSS, SAS, survival
analysis and cox regressions, and working for a crown corporation, would made
such a blatant mistake in the calculation of R.R, and use it to justify
conclusions such as "The estimated relative risk of 1.46 (in Table 7) for RHD
drivers was very similar to the value estimated in the relative crash
culpability analysis reported earlier." (Page 14, Table 7), and "The results
of the all three analyses in terms of relative risk for crash involvment
associated with RHD vehicles were very similar." (Page 14, Discussions and
Conclusions)

I would doubt the validity and reliability of the entire analyses, as results
can be twisted to match each other.

[Shadao]

wow more fuel for our fire....

i knew just by looking at the nubmers in there report that they were wrong, i just lack the skills to analize them... thanks and good work for whom ever did this

[dieselDaryl]

I've been in contact with some statisticians/risk analysts at UBC and SFU. ICBC can try and hide their numbers behind big words and fancy model names but I dont think it will take much for a University professor to blow holes in the report..

Daryl

[aerocabin]

You should ask your source to challenge the authors of the report directly:

peter.cooper@icbc.com
wayne.meckle@icbc.com
glenyth.nasvadi@icbc.com
sandi.wiggins@icbc.com

And then post the results back here

[KAMworks]

You should ask your source to challenge the authors of the report directly:

It a good idea, but given some of the heat generated, I'd be surprised if there is a reply. Also think it should be a BC resident doing this.

[Shadao]

it shouldnt matter who does it as the out come will affect all canadians alike

[crushers]

this is very true, if ICBC is succesful then all private and government insurance companies will look to this result as a "factual" template to follow...
<shiver> as if ICBC hasn't loaded the figures to back a preconcived result...