Keynote Lectures |
| |
James Steiger
 |
|
Things We Could Have Known: Some Thoughts on Seeing the Future and Avoiding Regret in Data Analysis and Model Selection |
|
Modern graduate statistical education in the social sciences often
reflects an implicit view that statistics is a set of fixed tools to be
assimilated quickly, so as not to interfere with a student's rapid progress
toward generating "substantive research." The notions that statistical
procedures themselves have properties that vary with experimental conditions,
and that these properties might be explored as part of the research, are lost
on the majority of students, their advisors, and, indeed, authors of their
textbooks. With our current computing power, this need not be so. Should we
care? I'll try to convince you that we should by exploring a variety of data
analytic disasters, wrong conclusions, and oversights that could have been
predicted, and avoided, if researchers had asked a few simple questions, and
used modern computer power to answer them.
|
|
| |
| David Rindskopf |
|
Some Neglected Relationships Between Cognitive Psychology and Statistics:
How People Try to Think Using Advanced Statistical Methods Without Realizing It
|
|
|
Teachers of statistics often try to increase students' motivation
by making statistics relevant to everyday life. But unless a students'
everyday life consists of running research studies, most such attempts fail.
In this talk I propose that many simple everyday decisions (subconsciously)
involve attempts by people to use statistical methods that are much more
complicated than is generally realized. I show how these methods relate to
other methods (also seldom considered in cognitive psychology) that provide
a basis for much of everyday human decision-making. |
|
|
Questions?
