John/ colleagues everywhere

Very interesting to me for the last 30+ years!  Quite a lot of material and
the problem is what to choose.  Thanks for your own contribution.

I would commend "Plain Figures" (2nd ed. published 1996 by The Stationery
Office ISBN 0-11-702039-7).  It includes references to quite a lot of the
main research work as well as other guidance.  Much of the guidance in my
view works for screens as well as paper (but see below).  I organised a
discussion meeting at the Royal Statistical Society back in the 1990s to
share it, so have a bias.

The big gap to my mind is the research work that is needed for presentation
on a screen and using the features of the web.  I see lots of fads but I
don't see a picture emerging of the evidence for particular modes.

For official statistics web sites, you might be interested by my colleague
Ed Swires-Hennessy's commentaries and principles at
http://www.surfingwithed.org.uk/

There is a UK National Statistics protocol on Data Presentation etc which
covers some of the principles as you have.  But it's oriented to official
statistics again and doesn't go into more detail, which is left to more
detailed guidance.  Protocol is at
http://www.statistics.gov.uk/about_ns/cop/downloads/datapresentation.pdf

Interested to see suggestions and comments from others.  I am posting this
on Allstat also to see if there's interest there - but that doesn't have
email discussion like Radstats.

best wishes
Greg

Greg Phillpotts
Regional and Local statistics
Office for National Statistics
1 Drummond Gate, LONDON SW1V 2QQ

tel 020 7533 5480




Date:    Tue, 1 Nov 2005 14:17:55 +0000
From:    John Maslen <[log in to unmask]>
Subject: How should we present statistics?

In line with a forthcoming seminar run by the Association of Regional
Observatories entitled "Developing best practice in presenting statistics"
(programme soon to be published at
www.regionalobservatories.org.uk/aro_events.html) I wondered whether the
group is aware of any research and key publications in this area?
Obviously
it is a high profile subject that hits the headlines in relation to public
health, school league tables etc etc.

My relatively limited trawl has not picked up much.  I am one of the many
fans of Edward Tufte's publications and there's some relevant stuff on his
discussion board such as:
http://www.edwardtufte.com/bboard/q-and-a-fetch-msg?msg_id=00009X&topic_id=1&topic=Ask+E%2eT%2e


I may be trying to simplify things too much but how about coming up with a
set of 10(+) Commandments For Presenting Statistics - these could be split
into what I like to think are 3 semi-distinct user groups:-

(1) Expert Professionals (Researchers, senior analysts, statisticians)
(2) Regular Users (Policy advisors, some managers, technicians)
(3) Informal Users (citizens, members of community groups/voluntary
organisations etc)

This relates to other published papers by this group which introduced
'barefoot statisticians' and 'parastatisticians' and the need to educate
citizens and members of organisations so they can make informed decisions
about the statistics they are presented with.

To get the ball rolling here's a few I made-up earlier (which are biased to
web-based publishing I'm afraid):-

(1) Understanding your audience, what they want to know, and what decisions
they are likely to want to take is fundamental.  It is not always easy!
(2) Publishing on the Internet is a major challenge given the uncontrolled
nature of the potential audience.  In some cases, for example a more
generic
report for Internet publication, some descriptive text on who the intended
audience is likely to be can help, especially if this can be picked up in
search results.  There is not always scope for tailoring outputs to all the
different audiences.  This limitation should be recognised and, where
possible, some measures adopted to provide different types of user with
different 'channels' to information.
(3) Use presentation techniques for your statistics that are appropriate to
your data (obvious but needs to be stated) - visual techniques can be
excellent although novice users may still prefer to read a text-based
synopsis.
(4) Use presentational techniques that are appropriate to the medium which
you are using - for example don't automatically replicate a hardcopy output
for web distribution if you can bring new insights and understanding
through
adoption of web browser-based technologies (using user-click interactivity
for example).
(5) If publishing on the Internet, especially when targeting more novice
users, it can be more effective to present statistics in a controlled
manner
that includes some form of neutral interpretation of specific 'data views'.
(6) If publishing on the Internet be aware of usability and accessibility
issues and ensure you provide at least one 'accessible' channel to your
data.
(7) Always use metadata to qualify your statistics and the way you present
them.  This should be at multiple levels from the dataset as a whole down
to
 individual records if necessary.  It should also include metadata on how
you present them if appropriate.
(8) Metadata should not be 'hidden' from the user as a secondary priority -
data users need to be encouraged as much as possible to understand the
strength and weaknesses of the data.  As such it should be relatively
explicit.
(9) Metadata needs to be appropriate to the type of user - too often it is
riddled with jargon that is only understandable to a data expert.  Where
possible adopt a recognised metadata standard as a basis for managing
metadata behind-the-scenes and tailor this information to the audience.
(10) Try to adopt recognised metadata standards like Dublin Core, DDE and
SDMX - be aware that standards do overlap/compete and generally evolve.

Please throw any of these out and/or add your own!  Perhaps you have some
examples you could share that you regard as good practice?  Are the
mainstream press consistently poor in making these judgements?  Should we
ever be even attempting to publish complex raw statistics to a largely
unknown, relatively novice audience - do they just need interpreted
reports?

If you're interested in the ARO seminar you can register now at
www.aro-stats-seminar.org.uk.

Thanks a lot all.  It's my first post so I hope this is a topic that is of
interest to the group.
John
GeoWise







For the latest data on the economy and society consult National Statistics at http://www.statistics.gov.uk

*********************************************************************************


Please Note:  Incoming and outgoing email messages are routinely monitored for compliance with our policy on the use of electronic communications
*********************************************************************************


Legal Disclaimer  :  Any views expressed by the sender of this message are not necessarily those of the Office for National Statistics
*********************************************************************************


The original of this email was scanned for viruses by the Government Secure Intranet (GSi) virus scanning service supplied exclusively by Energis in partnership with MessageLabs.

On leaving the GSi this email was certified virus-free