|
An examination
of the statistical issues involved in ensuring that an
experiment yields relevant information. Topics include randomized block,
factorial, fractional factorial, nested, Latin square and related designs.
Further topics as time permits. The emphasis is on applications.
Good experimental design is the cornerstone for the generation of
good data. It can be viewed as selecting the best feasible experiment
to achieve some specific objective. This includes the choice of
treatments,
and the random allocation of experimental units to them. Method of
estimation is an important component of the determination of a
design. One
must consider how the data will be analysed after the experiment is
carried out and data are collected. With the analytical procedure in
mind, a proper choice of experiment is then determined to achieve that goal.
Various designs will be discussed in this course through
definition of objectives, analytical procedures, and feasibility of
experimental constraints.
Statistical programs for sample size and power analysis
will also be introduced.
For the text and references, please see the web site: http://www.math.yorku.ca/Who/Faculty/Ng/menu.html
The final grade may be based on assignments (15%), a project (10%),
a midterm test (30%), and a final examination (45%).
Prerequisite:For 1999/2000: A second 6 credits in statistics,
including either AS/SC/MATH 3033 3.0, or both
AS/SC/AK/MATH 3230 3.0 and AS/SC/AK/MATH 3330 3.0;
or permission of the course coordinator.
For 2000/01: AS/SC/AK/MATH 3034 3.0
or permission of the course coordinator.
Coordinator: P. Ng
|
