Bill Venables
What do users really want in a statistical computing environment?
*****************************************************************
The talk will review from a fairly personal perspective the conflicting
needs users have for a statistical computing environment and the varying
levels of engagement users have with the software systems. The issues to be
reviewed include the conflicts between the needs for highly interactive
users and those more comfortable with a more remote programming style, as
dictated by their application. We speculate on the direction that
statistical practice will evolve in the near future, in particular the
balance between classical and modern techniques, the rise and rise of random
effect models and the differing, and converging, needs of frequentist and
Bayesian users. Finally we consider the question of to what extent should a
statistical support system such as R for data analysis and graphics attempt
to be educative: should we give the users what they ask for without question
or should we arrange our priorities so that some of the more egregious
excesses in bad statistical practice are at least naturally discouraged?