This is the site for the third edition of the text and is no longer maintained.

Click this line to go to the site for the 4th edition
.

E•r•R•a•t•A

NOTE: The page numbers of the hard copy and the electronic copy differ a little, so consider the page numbers below as approximate.

Chapter 2

• Footnote 1, p52: Lk is the max-ed likelihood, not the logged value... but you knew that already.
• Example 2.2, p56: The regression shown on the top of page 56 is for trend = 1:length(cmort) and not the value specified on line 9 of the code below the equation.
• Problem 2.10, p81: The model in the problem statement is supposed to be a restatement of the model in (2.46); that is, the sum should be to n/2 instead of n.

Chapter 3

• p149 - end of first paragraph: "... MA(2) model in (3.152)." should be "... MA(2) model in (3.151)."
• p152: The figure on this page seems to have another ghost figure on top of it (at least in the hardcopies I've seen), but it's supposed to look like this: page 152.

Chapter 4

• Equation (4.16), p184: The equation is missing an eye. It should read ... (e−2 π i ω + e2 π i ω) ...
• Problems: The Section 4.6 problems (4.17 and 4.18) on page 260 should be in Section 4.8. In addition, Problems 4.25 - 4.27 in Section 4.8 problems on page 262 should be in Section 4.6. (This is the case for the 2nd edition, too).

Chapter 5

• Section 5.8: Γ is k × r in both equations (5.86) and (5.98), not p × r.

Chapter 6

• p354: equation (6.103): εt+1 should be εt   and   equation (6.104): At+1 should be At
• p337, last sentence: ... optim call in line 18 ... instead of line 19
• p344, line 6: ... Deistler, 1988 ... instead of ... Diestler, 1988 ...
• p369: the line below equation (6.145) should have the sum over j, not i: Σ m j=1 πij = 1
• p383, just below (6.183): ... normal density N(., σj2) ... should be ... normal density N(., Σtj )...
• Example 6.19, pp385+: The words say ... residuals from an MA(2) fit ... but the code on line 4 says fit = arima(gnpgr, order=c(1,0,0)), so I'm guessing the code is correct and the analysis is on the AR(1) residuals. There shouldn't be much difference either way.

Elsewhere

• Example A.2, p509: In the last line of the inequality below (A.11), change γx(0) to [ γx(0) + μx2 ]