andrel wrote:
> Let's keep this in mind and see what happens if we *force* a linear
> scale. E.g. assume that there are only a finite number of cars and that
> the government has decided that only the best drivers get a licence.

Alright, this will be interesting.

> First we have to create a committee that can judge drivers capacity.

Okay

> For this a natural first group is that group that knows about cars e.g.
> because they own a couple and repair them themselves. This is based on
> the logical assumption that people who know how cars work also know how
> to drive them well. Presumably they will come up with a test like how
> fast drivers can negotiate an obstacle course without damage, to test
> the ability of the drivers. That would indeed create an objective linear
> scale.

Only, that's not a logical assumption. The people I know who own a
couple cars and repair them drive like maniacs and admit it.

> So we have as our main ingredients a complex multidimensional concept, a
> need to make it objective and one dimensional, and a group of
> knowledgeable men and what comes out is a very reasonable measure that
> somehow and unplanned is not going to be gender-insensitive. And as long
> as new members of this committee are recruited from the 'best' drivers,
> it will stay that way.

There is the undefined jump from 'statistically safe driving habits'
that, if everyone followed, would reduce total accidents to the 'locally
safe driving habits' that may cause more accidents, on the whole, but
are safer in certain circumstances where other drivers are making random
decisions. 'Best' drivers, those who can put a car through it's paces on
a closed course, are not by default the 'best' driver on a crowded
street. Besides, these already is a committee that decides best driving
practices and tests people on it. Around my part's the the Department of
Motor Vehicles, and a few other federal agencies.

How does your test by committee become gender biased? Simply because
there are more male race-car drivers, or what ever other pool you pick
the judges from? I do not see the connection there.

> This is of course an imaginary scenario and the fact that so many women
> are going to fail the test is a dead give away that something is wrong.

The women fail the test because you imply a gender bias is somehow
unplanned, but still present in the test. Why would a government
committee, that picks the best drivers to devise the test, settle for a
unplanned for bias? And what is the bias, anyways?

Now, if you said the government picked a group consisting of civil
engineers, a few material science folks, some chauvinistic gear heads
who race in demolition derbies on the weekend, and a mathematician, then
I would believe the proposal you set out.

> You might even argue that no government is going to do something so
> simplistic for such a complex problem. On the other hand many people
> apparently fail to see the fact that driving ability is a
> multidimensional problem to begin with. If you don't, it may seem
> logical that it is a fair and adequate test. There is an even better
> argument that this kind of fallacy is common: it works this way in science

Some governments already do have a single dimension score for driving;
around here you have to pass a test with a certain score. Too low of a
score, and no license.

Post a reply to this message