More California

In article .com,
Bret wrote:

On Aug 22, 9:27 am, wrote:

Berkeley was really a hotbed of
electronics and biotech development. The numbers of startups there
were probably the highest in the nation and maybe the world. Now there
are essentially no startups there, all of the electronics firms and
all of the biotech firms are gone.

What did Berkeley actually do? Are you sure this wasn't a result of
market pressures? It must be expensive to set up shop in Berkeley. I
know someone who thinks their favorite BBQ restaurant in Longmont went
out of business because of a plot to Boulderize Longmont.

One company I worked for was
threatened closure for having "dangerous chemicals on site". What were
these "dangerous chemicals"? Two unopened cans of motor oil used in
the machine shop to lubricate the lath and mill bearings.

This isn't a commie plot, it's life in a litigious society and
someone's just covering their ass. When I worked at IBM in the late
70's, 3-in-1 oil was supposed to be handled with goggles and gloves.
When the department moved to a new building, it took a ride in the
"chemical safety truck".

Alternately, this could have been a bit of overstating the actual event (I know
that'd be hard to believe). The county (and sometimes the fire department) come
around and do inspections to see how people are handling various materials, like
oils. They recognize that oil is (1) flammable and (2) possibly damaging to the
ground water and/or the soil if it's spilled (both reasons for oil qualifying as a
"dangerous chemical"). So they prefer that it's in a fire cabinet or on containment
platforms (plastic tray that will catch oil spills). It's highly unlikely that those
people would threaten anyone with being "threatened closure" on a first visit.
Perhaps after a few repeated infractions they might but it's still unlikely. At least
that's been my experience in all the machine shops I've worked at (in three different
counties here in the Bay Area) and that of friends in other shops. The inspectors are
very flexible.

--
tanx,
Howard

Never take a tenant with a monkey.

remove YOUR SHOES to reply, ok?

SLAVE of THE STATE wrote:
"If you torture the data long enough, it will confess to anything." --
BF, 1759

BF must have been aboard the starship Enterprise at some stage.

On Aug 23, 8:33 pm, Dan Connelly
wrote:
SLAVE of THE STATE wrote:

On Aug 23, 2:10 pm, Dan Connelly
wrote:

But the problem is there is no information for how long those born in 2007 are expected to live.

That isn't a problem unless (possibly) one is an actuary. But if
voluntarily paid actuaries aren't doing it, then you can make a
reasonable assumption it is problem not worth solving.

Why would you even care about such a thing?

Because I care how long I am expected to live, not previous generations.

If you really want a 100% accurate answer to that
question, I know a non-statistical method, but I don't
think you're going to like it.

Ben
Humming the M*A*S*H theme

read the herein materail carefully, Joseph. These are Californians
talkin' not Norwegans

On Aug 23, 2:27 pm, I wrote:
On Aug 23, 2:10 pm, Dan Connelly wrote:

From:http://en.wikipedia.org/wiki/Life_expectancy

Oh dear. I'll have to edit that when I get a chance.

Last night I asked more than 1% of the world's PhD demographers to
rate the section of the Wikipedia entry that Dan cited for content and
accuracy. The average rating was a just a bit higher than 6 out of 10.

wrote:

Humming the M*A*S*H theme


My threshold for pain is even lower, so this method doesn't work for me.

Dan Connelly wrote:
My threshold for pain is even lower, so this method doesn't work for me.

Its a problem if testing your lactate threshold exceeds your
pain threshold.

On Aug 23, 8:33 pm, Dan Connelly
wrote:
SLAVE of THE STATE wrote:

On Aug 23, 2:10 pm, Dan Connelly
wrote:

But the problem is there is no information for how long those born in 2007 are expected to live.

That isn't a problem unless (possibly) one is an actuary. But if
voluntarily paid actuaries aren't doing it, then you can make a
reasonable assumption it is problem not worth solving.

Why would you even care about such a thing?

Because I care how long I am expected to live, not previous generations.

There might be other things you could look at to give just as good an
idea, and maybe better, since sweeping data can as much obscure (for
the individual) as reveal.

I think I'm ready to start signing Ol' Man River.

Ah gits weary,
An' sick o' tryin',
Ah'm tired o' livin',
And skeered o' dyin',
But Ol' Man River,
He jes' keeps rollin' along!

wrote:

It's not quite the average age of everyone who died in the last year
because populations can grow and (sometimes) shrink. The period life
expectancy is related to the weighted average age of death last year
(or any other period of time you're interested in).

Consider:

p(t, t0) dt = number of people born between time t0 and t0 + dt who are alive at time t

then I can evaluate:

rd(t, t0) = -del p(t, t0) / del t (the present death rate of people born at time t0)

(del: partial derivative)

then the probability that someone born at time t0 will die between t and t + dt is:

f(t, t0) dt = [rd(t, t0) / p(t, t0)] dt

If I assume this rate is time-independent, I assume it represents a universal death rate for age (t - t0). Then

f'(age) = f(t, t - age)

Then I can calculate a life expectancy:

y = integral(0 to infinity) age * f'(age) d age

Which obviously differs from the life expectancy of someone born at any moment of time, including today.

Dan

On Aug 24, 10:49 am, Dan Connelly
wrote:
If I assume this rate is time-independent,

y = integral(0 to infinity) age * f'(age) d age

Which obviously differs from the life expectancy of someone born at any moment of
time, including today.

You assumed this rate is time-independent.

But your math is, of course, correct. We distinguish between a cohort
life expectancy and a period life expectancy and the context is
generally so clear that we don't specify it. When we write "the life
expectancy at birth in 2000" and "life expectancy at birth in 2001"
we're talking about period rates. When we write "the life expectancy
at birth for the cohort of 1857" we're talking about the cohort rate.

It's sort of like the convention for saying "life expectancy" instead
of "life expectancy at birth." Notice in your equation "integral(0 to
infinity)". If you took integral(X to infinity) that would be
expectation of life at age X, conditional on having survived to age X.
That's why I referred to life expectancy as a conditional expected
value.

Your partial derivative is the hazard rate. Your p(t,t0)dt is the
survivorship function. Life expectancy calculations are the same as
MTBF calculations. Your issue is congruent with what are called
censored observations.

I think we went over this a couple of years ago.