- Page ix: "Part ??" should be "Part I", and "Part I" should be "Part II".
- Page x, line 12b: "development" should be "develop"
- Page 9: In the equation at the bottom of the page, the square root symbol is missing over the "2\pi".
- Page 11, header: "1.2" should be "1.3" (this is a LaTeX problem, but there is a fix for it).
- Page 11, line 11b: Insert at beginning of line, "if the y_i are independent realizations of random variables Y_i with CDF P(\cdot),".
- Page 14, line 27:
The range on nu should be [0,1].
(
*Thanks to Li Li!*) - Page 17, line 1:
The inequality for p should be greater than or equal, rather than
strictly greater than. The value p=1 yields
a very important Lp estimator, namely the
least absolute values estimator.
(
*Thanks to Herbert Lee!*) - Page 17, equation (1.14):
should have a minus sign on the right-hand side.
(
*Thanks to Li Li!*) - Page 19, equations (1.16) and (1.17) and the unumbered equation following:
all should have a factor of 2 on the right-hand side.
(
*Thanks to Li Li!*) - Page 20, line 9b: Add: "This estimator is not, strictly speaking, a least squares estimator of \sigma^2. It is based on least squares estimators of the other parameters." Delete the last sentence in the last complete paragraph.
- Page 21, line 12:
"H_r" should be "H_s". (
*Thanks to Xun Wang!*) - Page 25, paragraph containing equation (1.32): The expected values should be taken for all quantities in the equations and the first part of the first sentence should read, "If theta is a scalar, the expected value of the square of the first derivative is the negative of the expected value of the second derivative,..."
- Page 27, line 13b:
"left censored" should be "right censored".
(
*Thanks to Antonia Turkman!*) - Page 28, line 3b: minus sign missing in exponential (both in numerator and in denominator).
- Page 28, line 1b: in the equation, perentheses should enclose everything following the fraction 1 over n+m.
- Page 30, line 21: "properties estimators" should be "regularity properties".
- Page 36, Exercise 1.7(e): Should say "...and computes the MLE of..."
- Page 37, Exercise 1.9(b): For clarity, should say "...(as in equation (1.13)..."
- Page 46, line 15, 17, 18:
The range on y should be e^{-x^2} (line 15); the conditional distribution of
Y is U(0,e^{-X^2}) (line 17); and the range of the conditional distribution
of X is (-sqrt{-log Y}, sqrt{log Y}).
(
*Thanks to Li Li!*) - Page 55, equation (2.9):
The right-hand-side should be divided by m; that is, the denominator
should be m(m-1). (
*Thanks to Petri Koistinen!*) - Page 57: the web site mentioned in the fourth paragraph is no longer maintained.
- Page 74, equation (3.2):
"S_1" should be "S_2". (
*Thanks to Cliff Sutton!*) - Page 77, equation (3.8) and previous equation:
"n" should be "r". (
*Thanks to Cliff Sutton!*) - Page 77, line 2b:
"V_J" should be "hat{V(T)_J}" as in equation (3.9).
(
*Thanks to Cliff Sutton!*) - Page 79, equations (3.11) and (3.12): "\bar{T}^*" (which is not defined) should be "\bar{T}_{(cdot)}" (from equation (3.5)).
- Page 89, line 23: "T*-t" should be "T*-T".
- Page 97, Exercise 4.5: the notation is confusing. When expected values are taken, we clearly mean that the y's are to be considered random variables, rather than realizations of random variables. Also, in the solution given on page 380, "x" is used in place of "y".
- Page 98, Exercise 4.8:
delete (or work again -- this is the same as Exercise 4.4).
(
*Thanks to Li Li!*) - Page 100, line 10: period should be deleted.
- Page 100, line 20: period should be a colon.
- Page 101, line 7b: opening perenthesis under the second square root sign should be omitted.
- Page 101, line 4b:
in the second expression, "(xQ)^T" should be "(Qx)^T".
(
*Thanks to Weijie Cai!*) - Page 107:
The matrix displayed near the middle of the page is the transpose of T.
(
*Thanks to Li Li!*) - Page 117, line 4b:
should be "...elements are the reciprocals of the square roots ..."

and in equation (5.20) the argument to "diag" should be reciprocated. (*Thanks to Fred Webber!*) - Page 121, line 14:
"15" should be much larger, say, "100".
(
*Thanks to Herbert Lee!*) - Page 135, equation (6.11):
omit "=0".
(
*Thanks to Li Li!*) - Page 137, equation (6.16):
The multiplier of the (k-2)th ploynomial should be (k-1), rather than
k.
(
*Thanks to Gerrit Stemerdink!*) - Page 173, line 9:
"x_1" should be "the first observation".
(
*Thanks to Herbert Lee!*) - Page 203, Exercise 8.5:
all exponents of e should be preceded by minus signs.
(
*Thanks to Antonia Turkman!*) - Page 210, line 3:
should be "...expectation of the histogram estimator...".
(
*Thanks to Roger Shores!*) - Page 219, equation (9.32) and second line following the equation:
"p(y-Vt)" should be "p(y-Vu)".
(
*Thanks to Weijie Cai!*) - Page 220, line 4: the integration is just over the reals, that is, d=1.
- Page 221, line 2:
the equation should be "k_{rs} = r/(2B(1/r,s+1))".
(
*Thanks to Kyle Caudle!*) - Page 228, Exercise 9.5:
the inequality is going the wrong way (should be "0<=y").
(
*Thanks to Kyle Caudle!*) - Page 259, line 3:
the website should be http://www.qhull.org/ .
(
*Thanks to Mark Lukens!*) - Page 265, line 17: in the equation the argument to "diag" should be reciprocated.
- Page 273, line 7: the left-hand side should just be X-\bar{X}
- Page 313, line 12:
"alway" should be "always".
(
*Thanks to Herbert Lee!*) - Page 316, line 16b:
The equation should be "rho(t) = |t|" (instead of "t^2").
(
*Thanks to Fred Webber and to Aki Vehtari!*) - Page 316, line 6b, and page 398:
The reference to Hawkins and Olive (1999) is missing. It is

Hawkins, Douglas M., and David Olive (1999), Applications and algorithms for least trimmed sum of absolute deviations regression,*Computational Statistics & Data Analysis*32, 119--134. (*Thanks to Fred Webber and to Aki Vehtari!*) - Page 322, line 7b, and page 398:
The reference to Hawkins and Kass (1982) is missing. It is

Hawkins, Douglas M., and Gordon V. Kass (1982), Automatic interaction detection,*Topics in Applied Multivariate Analysis*(edited by D. M. Hawkins), Cambridge University Press, Cambridge, United Kingdom, 267--302. (*Thanks to Cliff Sutton and to Aki Vehtari!*) - Page 334, Exercise 11.3:
The subscript i is missing on x_1 and x_{q+1}.
(
*Thanks to Li Li!*) - Page 334, Exercise 11.4:
"xtmp" in the code should be "xdat".
(
*Thanks to Li Li!*) - Page 344, line 26:
should be "as |\beta_1| increases".
(
*Thanks to Johan Bjursell!*) - Page 359, line 2 in Figure B.3:
instead of "rnunif", should be "runif".
(
*Thanks to Petri Koistinen!*) - Page 369, line 19 (definition of little o sub p):
"Pr(|X(n)-f(n)|>e)" should be "Pr(|X(n)/f(n)|>e)".
(
*Thanks to Petri Koistinen!*) - Page 370, lines 1 and 2: delete "the transpose of". (The symbol represents just the Jacobian itself.)
- Page 378, line 17 in equation for expectation of x_{12}:
delete "x_1" in denominator.
(
*Thanks to Allen Harvey and Victoria Plamadeala!*) - Page 379, line 14:
A good estimate of \theta is \sqrt{m_2/2}, instead of m_2/(2(n-1)).
(
*Thanks to Johan Bjursell and Michael Manti!*) - Page 380, solution to 4.5d:
in second line the terms in the sum should be divided by n, rather than
multiplied by n.
(
*Thanks to Romain Parsad!*) - Page 382, in solution to Exercise 9.16b, line 15:
The integrand should be y instead of y^2.
(
*Thanks to Li Li!*) - Page 408, line 6: page numbers should be 281--295.
(
*Thanks to Cliff Sutton!*)