18
$\begingroup$

I will need to present a project to non-statisticians in an upcoming conference and I would like to include a quote that I have read at one point that makes the comparison between statistics and a machine which processes ore. The reason is that I want to emphasize a certain connection between the capabilities of statistics and the goals of the audience.

I cannot find the quote, and was hoping someone would be able to point me to a reference for it. It is something along the following lines (but not exactly this):

Consider the case of extracting gold from ore. The expectation of a good gold extraction machine is that it cleanly separates the gold from the ore with no waste. We would not criticize this machine for failing to extract gold if none was originally present in the ore, nor would we judge it too harshly if it failed to extract gold from ore with only minute quantities. Similarly, statistics is a machine which extracts information from the data. Statistics cannot create information, the data must contain that.

I believe the quote is due to Fisher, and that he goes on to note the resemblance between himself and a machine operator. I think the point he makes is that he doesn't need to be particularly brilliant to get good results as long as his statistical machinery is smart.

Any help would be appreciated. Please provide the full quote if possible.

DETAILS ON LOCATION OF QUOTE (added after answer was found):

After following the lead of the selected answer, I found the paper with the quote. It is:

  • Fisher, R. A. (1947). Development of the Theory of Experimental Design. Proc. of the Int. Statist. Conf., Washington, 3, 434-439.

It can be found for free here, in the digital archives of Fisher's collected works. It appears to be a conference speech he made in 1947.

The actual quote in this 1947 source is:

It was rigorously demonstrated that the amount of information extracted in the process of estimation could never exceed the quantity supplied by the data. This finding, combined with the practical fact that directly available processes of computation would extract almost always a very large fraction of the total available, shifted the moral balance. The statistician was no longer responsible for the accuracy or precision of the results of his labors. His business became much less like that of a conjurer who is expected to work wonders, and more like that of a chemist who undertakes to assay the proportion of gold in a sample of ore. He need not be ashamed if the assay is low, or elated with pride if it is high. Consequently, provided he used available methods of proved competence, and did his arithmetic right, the statistician no longer had any responsibility for the value, or for the worthlessness, of his estimates. The weight of his responsibility was thrown back on to the process by which the data had come into existence.

[italics in original]

Improve this question
$\endgroup$
6
  • 3
    $\begingroup$ There is a Fisher quote relating to extracting gold (starting something vaguely along the lines of 'A statistician is not an alchemist ...' and ending something like '...his job is only to produce what it contains'), but I can't locate it on the internet. $\endgroup$
    – Glen_b
    Commented May 14, 2014 at 11:06
  • $\begingroup$ This is quite close to what I had in mind. Perhaps someone knows what @Glen_b is referring to more precisely? $\endgroup$
    – Deathkill14
    Commented May 14, 2014 at 11:10
  • 5
    $\begingroup$ @Glen_b: The quote you're referring to is the following: "The statistician is no longer an alchemist expected to produce gold from any worthless material offered him. He is more like a chemist capable of assaying exactly how much of value it contains, and capable also of extracting this amount, and no more". I can't find the original source though, only a quotation in J. N. Spuhler's review of "Advanced statistical methods in biometric research" (Radhakrishna Rao, John Wiley & Sons, Inc., New York, N. P. 1952.) $\endgroup$
    – EagleV_Attnam
    Commented May 14, 2014 at 13:27
  • 1
    $\begingroup$ @EagleV_Attnam thanks, that certainly sounds like it's the one I've seen. $\endgroup$
    – Glen_b
    Commented May 14, 2014 at 13:39
  • $\begingroup$ Yours is a beautiful analogy that I would not mind quoting for some of my talks. I find it more effective to non-statisticians than Fisher's or Rao's versions. $\endgroup$
    – Thomas Speidel
    Commented May 14, 2014 at 14:21

1 Answer 1

Reset to default
7
$\begingroup$ 
Title   Statistical Design
Springer Texts in Statistics
Author  George Casella
Edition illustrated
Publisher   Springer, 2008
ISBN    0387759654, 9780387759654

... makes reference to this simile in the first page of the Preface. He cites Fisher, 1947. I only have google books access, and the reference section is excluded from the Casella reference above. So, I was not able to track down which Fisher, 1947 but one imagines it is in The design of experiments.

Improve this answer
$\endgroup$
7
  • $\begingroup$ Does it seem like the simile mentioned in the answer below by @fgnu, or is it more in accordance with the description I provided above? $\endgroup$
    – Deathkill14
    Commented May 14, 2014 at 10:54
  • $\begingroup$ In accordance with the description above. Ore. Gold. $\endgroup$
    – russellpierce
    Commented May 14, 2014 at 11:26
  • $\begingroup$ Yes! This appears to be a good lead. The paper given here, which is an introduction to a longer speech (it seems) is what I was looking for. It also has flavors of @glen_b's recollection of alchemy. $\endgroup$
    – Deathkill14
    Commented May 14, 2014 at 11:46
  • $\begingroup$ And I confirm after getting Castella's book that this paper is the source he cited. $\endgroup$
    – Deathkill14
    Commented May 14, 2014 at 12:04
  • 2
    $\begingroup$ For reference would you please provide the full reference as provided by Castella? It is an interesting quote. $\endgroup$
    – russellpierce
    Commented May 14, 2014 at 12:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.