RFI
[Top] [All Lists]

Re: [RFI] CFL bulbs - NO WAY

To: rfi@contesting.com
Subject: Re: [RFI] CFL bulbs - NO WAY
From: W0UN -- John Brosnahan <shr@swtexas.net>
Date: Sat, 20 Oct 2007 11:01:16 -0500
List-post: <mailto:rfi@contesting.com>
At 11:46 20-10-07, Jim P wrote:
>"But wait a minute.  Some studies point out the reduced
>   activity of birds and bees within a 100 mile radius of
>   some of these sites."
>
>Need a *cite on just one of those studies, please.
>
>Jim P // WB5WPA //

Jim,

This is precisely the correct response -- show me the DATA!

(And in many cases of data analysis -- BPL comes to mine --
show me the MONEY trail, in order to understand how the data
may have been manipulated.)

Rigorous science is very difficult to do, no matter what the field.

I have had a few dozen CFLs and I am sure that NONE of them
have lasted even half of the claimed lifetime -- 5 to 7 years.
Can I make the statement that all CFLs burn out much sooner
than their advertised lifetime?  No!  Because my sample may
not be representative -- I may have gotten a bad batch, or maybe
I bought them too early on the learning curve, or maybe my
installation is suspect (ceiling fan mounted ones for instance).

Or can I take ALL of the stories about their lifetime on this
reflector as indicative?  No!  Because maybe the only ones
to bother with posting here are the very people who have had early
failures.  The satisfied customers may not reside on an RFI
reflector or at least not bother to post.

Lifetime analysis issue difficulties apply equally to RFI analysis
generation from the bulbs.  Just because some people are satisfied
with their RFI performance does not mean that everyone will be
satisfied.  It is just not a well controlled experiment to take a few
personal stories and then extrapolate them into a generalization.

What one CAN do is to read the comments, reviews, etc and
get some general idea of what to look out for.  For instance,
there may be an increase in RFI during the early stages of
bulb failure.  This could be a good alert that something is wrong.
And it might warn of what appears to be a real over-heating
issue during the final phase of failure.  I am certainly interested
in this for safety reasons, both because of fire and because
of the release of the unknown chemicals that stink up the house
from this type of failure.

Designing rigorous science on REAL WORLD CFLs is difficult in
many of the same ways that designing an experiment to accurately
characterize antenna performance, or for many other issues such as
receiver performance or global warming (there I go again) or sunspot
predictions or even the life expectancy of birds and bees n any given
area.

A graphic example of the issues of unbiased data taking is made
obvious by an experiment that the National Science Foundation
once did (no, I don't have the reference, just the word of my partner
who was a program director at NSF at the time -- for all I know it
was never published for fear of embarrassment of the participants).

This experiment was to look at the incidence of cancer surrounding
nuclear power plants.  And as many suspected (read -- BIAS) there
WAS an increase of cancer surrounding nuclear power plants.  But
someone had the bright idea that what they should do is perform the
same experiment in the areas surrounding PROPOSED nuclear power
plants.  And ZOWIE -- there was an increase in cancer surrounding
plants that had not even been built yet!  So, apparently, just the mere
THOUGHT of installing a nuclear power plant was enough to have a
negative impact on the local cancer rates.

Well, obviously (I sure hope it is obvious) the methodology for taking
the data for the experiment was suspect.  These sorts of problems
PLAGUE complex science like global warming, but they also are
factors in ANY sort of analysis of any data set from CFLs, to RFI
measurements, to antenna gains.

John  W0UN


_______________________________________________
RFI mailing list
RFI@contesting.com
http://lists.contesting.com/mailman/listinfo/rfi

<Prev in Thread] Current Thread [Next in Thread>