Required for next week: January 30, 1996
- ``Ecological regression in voting rights cases'', S.P. Klein and D.A. Freedman,
Chance 6(3), 38--47 (1993).
Come prepared to discuss this article and ask questions about the
parts you didn't understand.
- The final part of short project 4, your essay proposal, is due.
Sample proposals are on my home page (and were handed out in class).
Further reading on athletic performance
- Wainer, H. (1993). How much faster can humans run than swim?
Chance 6(1), 17--21. World record times in swimming and
track are analysed to try to estimate the relative
efficiency of running and swimming. The short answer is that
humans can run about three to four times as efficiently as they
can swim.
- Wainer, H. and DeVeaux, R.D. (1994). Resizing triathlons
for fairness. Chance 7(1), 20--25. The authors argue
that the swimming portion of triathlons is too short, and the cycling
portion too long, relative to the best performances available in
swimming, running and cycling, and as a result swimmers are at a
disadvantage, and probably tend not to compete in todays triathlons.
For example, the Olympic triathlon has a 1.5 km swim, a 10 km run, and
a 40 km cycle. The suggested distances are a 2.7 km swim, 10 km run, and
a 22.4 km cycle.
Least squares regression
The relationship between two variables in a set of data is often described by
plotting one variable against the other: salary against years of experience,
for example. It is not uncommon to summarize the relationship by
a line or curve (just as we sometimes summarize data on a single
variable by the mean or median). Very often, this line or curve is computed using
an algorithm called least squares; the idea is that the resulting line
or curve provides the `best possible' fit to the data. The phrases
``least squares line'' and ``regression line'' are
often used interchangeably to describe the summary line (or curve),
and ``least squares'' and ``regression'' are also used interchangeably
to describe the method for computing the summary line.