Not a lightbulb joke
I was trying to explain the concept of a Poisson distribution
to Kristin, and of course one of the examples that immediately
jumped to mind was light bulbs.
And in fact if you Google appropriately, you will find that
light bulb failure rates and Poisson distributions do seem to
crop up together.
But do light bulbs really fail using a (reasonable approximation to a)
Poisson distribution?
Further googling could find no obvious studies or references
to studies that attempted to determine this rather critical
piece of information (because, after all,
Statistics teachers don't really want curry eternal damnation by lying all
the time, do they?)
I would tend to think that entropy, and possibly other physical
processes, would prevent light bulbs from failing in a tight approximation
to a Poisson distribution after not-all-that long a time (and I wouldn't
expect anything like a PD after years of continuous usage, but what
do I know)
You'd think that someone, 100 years ago, would have set up 100,000
light bulbs in some sub-basement of a General Electric building
and kept them on for 50 years. Or maybe they did, and some of the
bulbs are still playing Energizer Bunny, and they're still waiting
to publish the paper.
to Kristin, and of course one of the examples that immediately
jumped to mind was light bulbs.
And in fact if you Google appropriately, you will find that
light bulb failure rates and Poisson distributions do seem to
crop up together.
But do light bulbs really fail using a (reasonable approximation to a)
Poisson distribution?
Further googling could find no obvious studies or references
to studies that attempted to determine this rather critical
piece of information (because, after all,
Statistics teachers don't really want curry eternal damnation by lying all
the time, do they?)
I would tend to think that entropy, and possibly other physical
processes, would prevent light bulbs from failing in a tight approximation
to a Poisson distribution after not-all-that long a time (and I wouldn't
expect anything like a PD after years of continuous usage, but what
do I know)
You'd think that someone, 100 years ago, would have set up 100,000
light bulbs in some sub-basement of a General Electric building
and kept them on for 50 years. Or maybe they did, and some of the
bulbs are still playing Energizer Bunny, and they're still waiting
to publish the paper.