Thanks for several thoughtful contributions. There does not seem to be an explicit condemnation of *'s - perhaps statisticians are too polite to colleagues who borrow our tools. Several sources - Stern et al (see Ruth Butler below), ICH "Statistical Principles in Clinical Trials" (thanks S Senn) make comments such as "precise p-values should be reported rather than making exclusive reference to critical values."
The * notation might be defended by those who have a definite experimental plan and pre-set the alpha value, though in practice this will be by default (alpha=.05) not by thought. And do they really accept on that basis that 1 in 20 of their reported results should be wrong?
The discussion of hypothesis testing and significant in Stern et al seems to me particularly clear and well written.
Allan
-----
Extract from original posting
On 7/5/08 13:39, "Allan Reese (Cefas)" <[log in to unmask]> wrote:
> Dear colleagues
> I recently commented to a journal editor that the * notation was regarded as
> outmoded and widely deplored, and he responded that he'd not seen any
> condemnation in the places he read. I'm sure he is right, and the same
> probably goes for most other editors. In the allstat archive (20 July 2000),
> there is a summary of statisticians' comments on the reporting of p values.
>
> QUESTION: can anyone recommend a cogent and authoritative reference for
> editors that will persuade them that current practices on the reporting of
> statistical results can and should be improved?
-----
> From: kornbrot [[log in to unmask]]
I think Lee Wilkinsonıs report to Ame Psyc Assoc, APA is really excellent
BUT APA journal editors do not necessarily follow the guidelines:
Wilkinson, L. (1999). Statistical methods in psychology journals -
Guidelines and explanations. American Psychologist, 54(8), 594-604.
Http;//dx.doi.org/000081919300009
There are also BMA type recommendations for RCTs, etc. that include
statistics reporting
The most update source for editorial guidelines etc comes form the Equator
Network
http://www.equator-network.org/index.aspx?o=1037
To the extent that articles recommend exact p-values rather than significance level, they are condemning ****
But you are right. This needs to be made more explicit
In some of my Mss, I give exact p-values, but make Significant figures
bold for 0.01 and/or italic for 0.05
It helps reader pick put salient effects without theoretical commitment to
significance levels, on my own spreadsheets I use garish colours
Professor Diana Kornbrot
School of Psychology
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
-----
From: Nick Cox [[log in to unmask]]
I agree with you but this is all whistling into the wind. My experience
elsewhere is that when I raise issues like this, there are no
counter-arguments, unless it is something like "But this is what
journals I want to publish in insist upon".
One of the most amusing variants of this disease is that people ask for
P-values after correlations because they supposedly quantify strength of
relationship -- which naturally is precisely what correlations are
intended to do. Indeed sometimes a correlation can have some merit as a
interpretable numerical measure even when the assumptions on which a
P-value would be calculated (independence, for one) demonstrably do not
apply.
-----
From: Silva P.d.N. [[log in to unmask]]
I suggest consulting the chapter 1 of Hsu, J. C. (1996). Multiple comparisons: theory and methods, from Chapman & Hall.
It is not specific about the asterisk notation, but it addresses the same issue and perhaps might assist with your discussion. See especially the example in page 4.
Pedro Luis do Nascimento Silva
Southampton Statistical Sciences Research Institute
University of Southampton
Highfield
Southampton, SO17 1BJ, UK
Tel: 44-23-80597169
-----
From: John Bell [[log in to unmask]]
Guidelines for reporting statistical results .
The Journal of Prosthetic Dentistry , Volume 85 , Issue 1 , Pages 5 - 6
http://ajpgi.physiology.org/cgi/content/full/287/2/G307 (opens a link to article on my PC but might depend on connection)
http://web.psych.washington.edu/writingcenter/writingguides/pdf/stats.pdf
http://aje.oxfordjournals.org/cgi/reprint/141/10/896.pdf
John F Bell
Head of Statistics
Research Division
Cambridge Assessment
1 Regent Street, Cambridge CB1 2EU
Telephone: +44 (0) 1223 553849
www.cambridgeassessment.org.uk
-----
From: Ruth Butler [[log in to unmask]]
I'm not sure whether this is sufficiently authoritative, but there is a
small paragraph relevant to this in
R.D. Stern, R. Coe, E.F. Allan, I.C. Dale (Editors) Good Statistical
Practice for Natural Resources Research (CABI publishing)
chap p237
Ruth Butler
Biometrician
Crop & Food Research
Private Bag 4704 email: [log in to unmask]
Christchurch 8140 phone: +64 3 325 6400 (extn 3501) direct phone:
+64 3 325 9501
NEW ZEALAND fax: +64 3 325 2074
***********************************************************************************
This email and any attachments are intended for the named recipient only. Its unauthorised use, distribution, disclosure, storage or copying is not permitted. If you have received it in error, please destroy all copies and notify the sender. In messages of a non-business nature, the views and opinions expressed are the author's own and do not necessarily reflect those of the organisation from which it is sent. All emails may be subject to monitoring.
***********************************************************************************
|