At 10:27 04/11/05 +0000, Ray Thomas wrote:
>I am sceptical of the idea that forms of presentation should be designed for
>groups identified on the basis of knowledge of statistical method, or
>frequency of use of statistics. Understanding of statistical materials
>depends primarily on familiarity with the subject matter of the statistics,
>not upon knowledge of statistics.
>I suspect that the bookies at my local dog track are statistically
>uneducated. But I imagine that the level of performance of these bookies,
>however measured, is well above that achievable by a statistician who has
>just strayed into this area.
I think this is essentially semantic. Those bookies (and plenty of other
groups one could cite) have 'knowledge of' and 'frequently use' the aspects
of statistics or probability theory which concern them, even if not more
generally, and formally, 'educated' in statistics.
I would re-word the statement you are challenging, by suggesting that a
presentation should always be designed with regard to (amongst other
things) the level of knowledge/familiarity the intended audience was
expected to have OF THE SUBJECT MATTER BEING PRESENTED. You go on to say:
>Again I would argue that the best forms of presentation should depend upon
>the nature of the subject matter not upon the type of audience.
As above, I believe it should be BOTH. It should depend of the level of
knowledge/understanding/familiarity the particular audience is expected to
have of the subject matter in question. A presentation entirely
appropriate for an audience of bookies would quite probably be totally
inappropriate for an audience of farmers.
>Charts are great for very simple points, for example. But tables can
>convey far more information and statistical detail.
I have to say that I'm a great believer in 'a picture being worth a
thousand words (or numbers)', no matter what the nature of the audience -
and I'm therefore a great believer in graphical presentations, wherever
possible. True, one can put much more 'statistical detail' in a tabulation
of numbers, but (particularly for non-Statistician audiences) this is
rarely what is required of a presentation; numerical detail should
generally go into reports or 'handouts', not a presentation. Many of the
worst presentations I've seen have been those which have attempted to
communicate a lot of numerical detail.
Merely determining which of a pile of numbers is the largest is much
quicker and easier with a graphical presentation. Seeing how distributions
or CIs relate to one another is so much easier graphically. The tabular
equivalents of scatter plots are almost impossible to interpret by
Statisticians, let alone others. Illustrating correlations with graphics
is very hard.... etc. etc.
>One commandment should certainly be to avoid statistical jargon.
I think we can agree with that, if we're talking about an audience of
non-Statisticians. However, it's merely a specific case of the general
fact that one SHOULD always tailor one's presentation (in terms of
vocabulary as well as other things) to the audience in question. To
deliberately avoid statistical 'jargon' when presenting to Statisticians
would be silly and counter-productive, just as avoiding medical 'jargon'
would be silly when presenting to doctors. Indeed, one should even be
prepared to learn and ADOPT the 'jargon' of the intended audience (even if
it is normally unfamiliar to oneself), or at least adopt alternatives more
like to be familiar to them, if that helps communication - e.g. those
bookies would be much more familiar with odds than probabilities.
Just a few thoughts ....!!
Kind Regards,
John
----------------------------------------------------------------
Dr John Whittington, Voice: +44 (0) 1296 730225
Mediscience Services Fax: +44 (0) 1296 738893
Twyford Manor, Twyford, E-mail: [log in to unmask]
Buckingham MK18 4EL, UK [log in to unmask]
----------------------------------------------------------------
******************************************************
Please note that if you press the 'Reply' button your
message will go only to the sender of this message.
If you want to reply to the whole list, use your mailer's
'Reply-to-All' button to send your message automatically
to [log in to unmask]
*******************************************************
|