faraway

Errata for Linear Models with R, 2nd Edition

Errata are corrected as found and incorporated into newly printed copies of the book. This means that not all the errata listed below exist in your copy of the book. If you see any errors not listed below, please let me know.

• throughout: A change to the ggplot2 package means that xint and yint are no longer accepted abbreviations. These are now replaced with xintercept and yintercept.

• p6. The response for the boxplots is diabetes function not diastolic as stated in the figure caption.
• p21. l4 The Z should be an X
• p21. (first displayed equation) There is a missing beta near the end. It should read: \(RSS = \|Q^T y - Q^T X \beta \|^2 = \left\| \left( \begin{array}{c} f \\ r \end{array} \right) - \left( \begin{array}{c} R \\ \mathbf{0} \end{array} \right) \beta \right\|^2\)
• p24. Figure 2.2 caption: The solid arrowed line is unknown x variation. The dotted line is the known x variation.
• p43. Displayed equation. Subscript i missing from second beta.
• p46. line 4 “univariate tests would fail to reject”
• p54. R code for plot of airline data should have ylab=”Passengers”. It is not logged.
• p76. R code for nonlinear residual plot example is better with cos(2*x*pi). Also sd=1 is the default so not needed.
• p93. Figure 6.12. Annotation of contours on the Cook plot has been truncated - should be 0.5, not zero
• p103. Simex method uses SD not variance. Should read simex(lmod,"speed",sqrt(0.5), B=1000) which gives an estimate of 3.99
• p105. Change to ggplot2 means that p+coord_flip()+xlab("Predictor")+geom_vline(xintercept=0,col=gray(0.75)) is now needed.
• p116. Weighted Least Squares, third point: weights should be inversely proportional to the variance of the response, not the SD.
• p136. Sentence should read: The Box–Cox method is not the only way of transforming the response.
• p150. Line 1, structured is misspelled.
• p154. Last line minimum is misspelled
• p152. The first model output on the page is the same as the one on the previous page and does not correspond to the model with Area removed.
• p160. Q5. stack.loss is the response
• p161. List items 1-2 should read var(Xu₁) and var(Xu₂).
• p162. z_i = Xu_i
• p168. In the caption of the figure, the second PC is the dashed line and the third PC is the dotted line.
• p176. Figure 11.10 - hat on the beta of the y-axis label has been cut off.
• p177. LASSO is an acronym. It stands for “Least Angle Shrinkage and Selection Operator”.
• p192. Figure 12.7. Annotation of contours on the Cook plot has been truncated - should be 0.5, not zero
• p204. The galamiss dataset is missing from the package. Here is the rda data file which you can load into R using load("galamiss.rda")
• p210. The intercepts for the lines do not correspond to the model output above. Should be:

  > abline(3.696, 0.764)
  > abline(3.696+6.861, 0.764-0.314,lty=2)
  

• p229. Displayed equation should have a square root on (1/J_i + 1/J_j)
• p252. $\beta_j$ is the blocking effect (line below displayed equation)

This site is open source. Improve this page.