The English language dominates modern scientific communication, as a lingua
franca - a doubly ironic term - but our world is a multilingual one, in which
English is for many a second language, with great effort expended to reach
anything like the degree of fluency that Ray and I would take for granted. In
response to Ray's article, I pointed out that one language with which I'm
familiar, Welsh, is fortunate to have two words for 'statistics', viz. ystadegau
(a collection of figures - like the English word 'data', rather ambiguous
whether plural or collective, as Honey's usage clearly shows) and ystadegaeth
(singular, the academic discipline of quantitative inference). I have always
been at pains to ensure that when my department's title gets translated, it is
the latter that is used. Having said that, there is little recognition that two
distinct words exist - the two words are so obviously closely related, both when
written and pronounced, and many fluent first language Welsh speakers didn't
have the benefit of the level of education in the language that would make them
aware of the distinction.

There is a very important point behind all this, of course. Those with a
mathematical training are familiar with concepts of 'one-to-one correspondence'
and 'many-to-one correspondence'.  There isn't a one-to-one correspondence
between words or phrases and concepts, nor between words or phrases in different
languages. There is a saying in English 'It's not what you know but who you
know' - which no doubt originated in the era in which those with hereditary
privileges predominated society. This would be untranslatable into any of
French, German or Welsh (and no doubt many other languages also) in which the
verb 'know' would be translated differently in the two senses. Indeed, even in
older registers/dialects of English, a distinction was made between to 'wit'
(savoir/wissen) and to 'ken' (connaitre/kennen). 'Lost in translation' trips off
the tongue easily, but it's a very real issue. I have seen a statistical paper
translated, in which what was obviously 'median' in the original was translated
'mean'. Without wanting to disparage professional translators, whose job is a
difficult one because the subject matter they're presented with could be
anything under the sun - nevertheless a subject specialist who knows both
languages should check, and back-translation is also a useful safeguard. 

I'm sure that Honey and other allstat members could easily comment on the usage
in their own first languages, though I suspect that I'd get some flak if people
from even a tiny proportion of the world's thousands of languages (v. complex to
enumerate/taxonomise) all responded directly to allstat. I suggest that if
anyone would like to respond, please reply to me, then I'll briefly summarise to
Allstat in a couple of weeks.

To get back to some of the points Honey raised, the singular term 'statistic'
is of course correctly used in several senses, a hypothesis testing statistic, a
sufficient statistic etc. The popular/media usage 'He became a statistic' as a
euphemism for accidental traumatic death is of course something we deplore.

For a 'qualitative' variable (which is better called 'unordered categorical'),
the mode is an obvious summary. The word 'qualitative' is best reserved for a
quite different type of research to the 'quantitative' discipline which is the
concern of Allstat. In the qualitative paradigm, maximal diversity / saturation
is generally more important than strictly verifiable representativity. The data
are generally verbal in nature, and are analysed in terms of themes and meaning,
not generally by quantitative methods. Qualitative research concerns what can
happen, quantitative concerns what generally happens. There is increasing
recognition that the two are complementary. Over the past couple of years
several Royal Statistical Society meetings have related to the interface between
the two disciplines. Also note the recently announced Association for Survey
Computing one-day conference 'Words instead of Numbers. The status of software
in the Qualitative Research World' on Friday 29th September 2006 at Imperial
College, London. http://www.asc.org.uk 

Robert G. Newcombe
Professor of Medical Statistics / Athro Ystadegaeth Feddygol
Cardiff University
[log in to unmask] 


>>> ray thomas <[log in to unmask]> 14/06/06 09:04 >>>

There was an excellent leading article on precisely this topic entitled
'Pstatistics and Mstatistics' in RSS News in May 2003.

I know because I wrote the article.  If anyone has difficulty in getting
hold of a copy email me and I'll dig out a copy from my hard disk.

Ray Thomas  ([log in to unmask])


-----Original Message-----
From: A UK-based worldwide e-mail broadcast system mailing list
[mailto:[log in to unmask]] On Behalf Of honey vincent valle
Sent: 14 June 2006 08:17
To: [log in to unmask] 
Subject: statistic vs statistics, singular and plural senses

Hi List members,

I'd like to get your opinion regarding this "debate" that's going on among
some members of our department, including me. This is about the meaning of
statistics in the plural and singular senses. I know that statistics
(singular) is a science that deals with the collection, organization,
etc.... of data. Statistics (plural) are the data themselves and/or
computations derived from that set of data and also sample characteristics,
like the sample mean, median, etc. In short, statistics (plural) refer to
numerical/quantitative data and numerical derivations from these data. The
bone of the contention I think lies in the data itself. What if the data is
qualitative (not quantitative)? Are these considered statistics? The only
statistic I can get from these is the total number of elements in the data
set. 

Pls. email me for your replies, comments, reactions, corrections, etc. Thank
you very much.

H. V. V. Valle
Philippines