Corrections
and clarifications
(There
is
a separate section below for corrections to
solutions to the problems)
(The
notation is: line n means n lines from top,
including equations; line -n means n lines
from the bottom, excluding footnotes. )
p182,
Example 9.3: The 95% confidence
intervals for the normal approximation has
correctly used a 2.5% quantile, but the
intervals for the exact t-distribution
have been calculated using a 5% quantile and
are incorrect. Using the correct value of
2.00, instead of 1.67, gives an
interval [4.36, 5.64] for a sample size of 60,
i.e. 1.28, just slightly larger than the
normal interval. The relevant table is C5 and
not C4 as stated. For the second case, using a
sample size of 8, the new intervals are [3.30,
6.70], with a width of 3.40 (not 4.40 as
printed), for the normal approximation and
[3.00, 7.00], with a width of 4.00, for the t-distribution.
So the difference in the interval widths
decreases as n increases.
Corrections
to
problems and their solutions
(For
completeness I have listed even trivial
typographical errors. The notation is: line n
means n lines from top, including equations;
line -n means n lines from the bottom, excluding
footnotes. Comments are in [ ] brackets.)
There
are
no reported corrections at present.
Minor
typographical
errors
(For
completeness
I have listed even trivial typographical errors.
The notation is: line n means n lines from top,
including equations; line -n means n lines from
the bottom, excluding footnotes. Comments are in
[ ] brackets.)
p12,
second equation: The terms summed should be
k^2 s^2 and not n^2 s^2.
Comments
There are no
comments at present.
Last
updated February 2012