Re: Request for feedback on statistical method

From: Shepherd Moon (shepherdmoon_at_yahoo.com)
Date: 11/04/04 

Date: 4 Nov 2004 12:02:18 -0800

clemenr@wmin.ac.uk (Ross Clement) wrote in message news:<e5b5d377.0411030237.2f6ac453@posting.google.com>...
> shepherdmoon@yahoo.com (Shepherd Moon) wrote in message news:<cf4333c3.0411011525.3b032134@posting.google.com>...
[...]
>
> Given that a claim that the speed of light is not constant must be
> fairly contentious, I'm sure that there have been publications where
> people have argued against this hypothesis. What do those papers say?
>
> Cheers,
>
> Ross-c

First, thank you for your reply. And yes, you are correct that this
claim is quite contentious. That may be why only a small handful of
statisticians have commented on it, or even know about it. But it
seems to be an instructive statistics exercise of how data can be
manipulated.

In terms of papers arguing against the hypothesis, I can't give a
comprehensive list, but here are a few papers/sites I have consulted:

Dealing with Creationism in Astronomy
by W.T. Bridgman
http://homepage.mac.com/cygnusx1

Young Earth Argument: Is The Speed Of Light Decreasing ?
by Don Lindsay
http://www.don-lindsay-archive.org/creation/speed_of_light.html
http://tinyurl.com/6nga3

Is the Speed of Light Slowing Down?
by Frank Steiger
http://www.fsteiger.com/light.html

The Decay of c-decay
by Robert P. J. Day
http://www.talkorigins.org/faqs/c-decay.html

For the most part, these sources deal with the physics part of the
claims involved (which I will take to sci.physics as has been
recommended), but here's a passage on the statistics from the Day
article: 

---
The final blow to Setterfield's credibility is his statistical
analysis of the results, given in Appendix 3, in which he discards 3
of the 41 data points shown in an earlier table, and claims a
coefficient of determination r^2 of "1 to nine significant figures,
indicating a near perfect fit to the data" (emphasis added). As anyone
with even the most basic knowledge of analysis will know (and as
Setterfield will later learn the hard way), a coefficient of
determination of 1 can only be realized if the data points lie
precisely on the curve in question, yet Setterfield shows a pathetic
ignorance of this fact by following the above claim with, "All told,
17 values were above the curve and 21 below, the r^2 value indicating
a perfectly balanced distribution of the cluster of points as well as
close proximity to the curve."
---
http://www.talkorigins.org/faqs/c-decay.html
I'll report back with anything I find out in the sci.physics post that
may bear on the statistics.
Regards,
Shepherdmoon

Relevant Pages

  • Re: 10 Dice Experiment is PERFECTLY NORMAL!
    ... >curve with the correct theoretical std. ... are those of the Statistics Department or of Purdue University. ... Herman Rubin, Department of Statistics, Purdue University ...
    (sci.math)
  • Pls advise: "best fit" statistics?
    ... I looked at the price of homes sold in the county during ... bell-shaped curve with a very long right tail. ... graph a normal curve that fits those statistics, ... Which set of statistics best represents the data: ...
    (sci.stat.edu)
  • Re: Null test, Wiless work
    ... > Here's an attempt at getting some interest by suggesting a null test ... That is more related to statistics, ... Mathematical proofs are set up in a way that they give the desired ... showed that in an important subset of cases such a curve could indeed ...
    (sci.math)
  • Re: prior distributions of estimated parameters
    ... I also have an "experimental" curve ... > one can only assume some "error bars" for each point. ... > which assumes that the distributions ... I am not expert on statistics, so that I am not even ...
    (sci.stat.math)