Fox 2006: Applied Regression Analysis and Related Methods: Errata

From MathWiki

Back to SCS:_Seminars_2006-07

Use this page to report errors. Use Comments for comments and discussion.
Table of contents

1 Parts IV and V of ms., chapters 14-22

September 22, 2006 version of ms, pp. 1-370

page 18

Second paragraph under 2.1 Preliminaries: "utopian" not "upotian"

[Fixed, with thanks. (I'm not sure how my spell-check missed this one.) John]

page 19

Third line, "x=4", should be "X=x4", fourth line "x=5" should be "X=x5".

[Fixed: thank you. John]

page 44

Second paragraph under 3.1.4 Boxplots: "Figure 3.12):" not "Figure 3.12:"

[Fixed, with thanks. John]

page 45

Figure 3.11: "Sierra Leone" not "Sierra.Leone"

[I'm not sure what to do about this one: The observation labels in the data file have periods between words, though that wouldn't have to be the case. This applies to some other observation labels as well, including another one in this graph. Thanks for pointing this out. John]

page 55

In the first paragraph on the page, there are two subject-verb number disagreements: "If the conditioning variable are discrete..." should be either "If the conditioning variable is discrete..." or "If the conditioning variables are discrete..." and "If a conditioning variables is continuous..." should be "If a conditioning variable is continuous..."

[Fixed, with thanks! John]

page 66

Last sentence of the first paragraph is missing a comma. Should read: "This is, in my experience, an effective device..."

[Fixed, with thanks! John]

page 80

Equation 4.3 omits the Jacobian factor (in contrast with the formula on p. 321, exercise 12.9)

[I almost missed this one. I believe that the following corrects Equation 4.3:

p(\mathbf{x|}\boldsymbol{\mu}\mathbf{,\Sigma,}\boldsymbol{\lambda})=\frac {1}{(2\pi)^{p/2}\sqrt{\det\mathbf{\Sigma}}}\exp\left[  -\frac{1} {2}(\mathbf{x}^{(\boldsymbol{\lambda})}-\boldsymbol{\mu})^{\prime }\mathbf{\Sigma}^{-1}(\mathbf{x}^{(\boldsymbol{\lambda})}-\boldsymbol{\mu })\right] {\prod\limits_{j=1}^{p}} X_{j}^{\lambda_{j}-1}

Then the log-likelihood is

\log_{e}L(\boldsymbol{\lambda}\mathbf{,}\boldsymbol{\mu}\mathbf{,\Sigma|X)}  =-\frac{np}{2}\log_{e}(2\pi)-\frac{n}{2}\log_{e}\det\mathbf{\Sigma}-\frac {1}{2}\sum_{i=1}^{n}\left[  (\mathbf{x}_{i}^{(\boldsymbol{\lambda} )}-\boldsymbol{\mu})^{\prime}\mathbf{\Sigma}^{-1}(\mathbf{x}_{i} ^{(\boldsymbol{\lambda})}-\boldsymbol{\mu})\right]   +\sum_{j=1}^{p}(\lambda_{j}-1)\sum_{i=1}^{n}\log_{e}X_{ij}

Does that appear correct? Thank you: John.]

page 89

First paragraph: "\widehat{Y}_i = A + BX_i" not "\widehat{Y} = A + BX_i"

[Fixed -- thank you. John]

page 91

End of first paragraph: "\mbox{fitted values }\widehat{Y}" not "\mbox{fitted values }\widehat{Y}'"

[Fixed: thank you. John]

page 201

Perhaps I'm confused, but for the overparameterized model, isn't the model matrix of rank m+1, such that the number of columns is one *more* than the number of groups in the ANOVA? One column for the intercept, and then m columns for the levels of X? [This model matrix has m + 1 columns, but it is of rank m, because any one of the columns can be expressed as a linear combination of the others, while the remaining m columns are linearly independent. John]

page 289

The fourth line: likelihood for the model

"L(\beta, \sigma_\epsilon^2)=\frac{1}{(2\pi)^{n/2}\left |\Sigma \right | ^{1/2}}exp \left({-\frac{1}{2}(Y-X\beta)^T\Sigma^{-1}(Y-X\beta)}\right )\!"

not "L(\beta, \sigma_\epsilon^2)=\frac{1}{(2\pi)^{n/2}\left |\Sigma \right | ^{1/2}}exp \left({-\frac{1}{2}(Y-X\beta)^T\Sigma(Y-X\beta)}\right )\!"

[Fixed -- thanks: John. This typo was also in the first edition!]

October 6, 2006 version of ms, pp. 355--697

page 673

Exercise 22.1 (b): Just in case some literal-minded students miss the point, it might be safer to say 'largest absolute t-values' instead of 'largest t-values'

[Done. Thank you. John]

Parts IV and V of ms., chapters 14-22

page 407

Footnote 4 ends with a comma instead of a period. [Fixed. Thank you, John.]

3rd sentence in footnote 4: "identity" not "identify" [Fixed. Thank you, John.]

page 621

1st line of the Summary: "Missing data are" not "Missing ata are" [Fixed. Thanks, John]

page 628

middle of page: "\ n^n = 256" not "\ n^2 = 256" [Fixed]

page 657

2nd line: The last \quad \tilde{\boldsymbol{\theta}}_j needs to be in bold. [Fixed]

Equation 22.7: The last \quad \tilde{\boldsymbol{\theta}}_j needs to be in bold. [Fixed]

Middle of page: "integral in Equation 22.7" not "integral in in Equation 22.7" [Fixed. Thank you, John]

page 631

At the end of footnote 9, the square brackets don't seem right: "...is an imperfect estimate of [the true standard deviation of the sample mean, SD(Y-bar)], however..." [I've removed the square brackets -- I agree that the passage reads better without them. Thanks, John.]

page 575

Middle of page, under second expression for LAV criterion: "The first component includes" not "The first components includes" [Fixed: thanks, John.]

page 507

Under the polynomial regression equation, "and so on." not "and so." [Fixed. thanks, John.]