Re: Re: st: replicating 2 X 2 data from a paper

[Nick Cox <njcoxstata@gmail.com>]

I'd like to pick up on this point. As a (relatively) statistical
reviewer, I've often experienced my advice being ignored by journals
within my field (loosely, various Earth sciences and/or geography) --
even when I've been commissioned by editors who know me personally as
more statistical than most in the territory.

Protocols and procedures differ, but cogent statistical criticism may
be ignored if

1. The critical reviewer is in a minority of one and other reviewers
are very positive. (That does not mean that they understand the
statistical issues; it's more common that they ignore them.)

2. The Editor can't distinguish what is damning from what is merely a
preference for a different analysis.

3. The authors set up a smokescreen, e.g. accepting that the advice is
good but arguing that what they are doing is easier for people to
understand, the standard method, or on the advice of some unnamed
person they trust.

4. There is a counter-argument. For example, statistical people hold
very different views on the merits of working on logarithmic scales,
non-parametric tests, etc., etc.

Of course, this is just my version, and details of my experiences
should remain obscure.

Nick
njcoxstata@gmail.com


On 4 April 2013 01:30, David Hoaglin <dchoaglin@gmail.com> wrote:

> As a statistician, I think you expect too much of the "statistical
> reviewers" or of peer review more generally.  Many papers that I read
> contain rough edges and errors that, to my amazement, the authors and
> the reviewers somehow failed to notice.
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/