Cardinal RN, Aitken MRF (2006). ANOVA for the behavioural sciences researcher. Lawrence Erlbaum Associates,
New Jersey, USA (ISBNs 0805855858, hardback; 0805855866, paperback). [Web
links for this book]|
Errata for Cardinal & Aitken (2006)
|
Back to statistics
|
Cardinal RN, Aitken MRF (2006). ANOVA for the behavioural sciences researcher. Lawrence Erlbaum Associates,
New Jersey, USA (ISBNs 0805855858, hardback; 0805855866, paperback). [Web
links for this book]
Print run numbers are on the bottom of page iv, below the library cataloguing data. The last number printed is the print run. For example, if the book shows "10 9 8 7 6 5 4 3 2 1", the print run is number 1.
| Page | Erratum | Status |
| Front matter; Contents | ||
| xiii | Subsections of section 9.8 should be numbered (9.8.1, 9.8.2, ...). | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| Chapter 1: Quick summary | ||
| 1 | Post hoc tests are covered in chapter 3, not chapter 4. | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| 3 | Typo: the list numbered 1-9 has "1" instead of "6". | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. Bug in Word; requires manual fixing after printing every time (text touchup of PDF). |
| Chapter 2: Understanding the basics | ||
| 8 | The table of fictitious results has incorrect group variances (calculated erroneously by missing out the first datum in each set). They read 13.53, 13.69, 17.78, 12.75, and 8.00, but should read 12.04, 12.49, 15.82, 12.68, and 8.71. This error propagates onto page 11 (see below). | 1: Content.Discovered in print run 1. Fixed in master 20Aug2006. |
| 11 | The equation for the pooled estimate is written as (13.53 + 13.69 + 17.78 + 12.75 + 8.00) / 5 = 13.15, but since the group variances were wrong (see erratum for page 8), this should read (12.04 + 12.49 + 15.82 + 12.68 + 8.71) / 5 = 12.35. | 1: Content. Discovered in print run 1. Fixed in master 20Aug2006. |
| 18 | The ANOVA table presents correct numbers (including F = 5.93), but then the F being discussed suddenly changes to 9.09 and the second table (which should be identical to the first apart from the addition of the column labelled p) is full of different and wrong numbers. The p value is correct for the original F4,45 of 5.93. | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
|
21 22 24 |
It might be slightly more consistent to use N rather than n for the discussion of linear regression (but this is not a major point). | 2: Clarity. Noted in print run 1. Clarified on p. 21 in master 21Dec2005. |
| 22 | "The error in this prediction will [be] related..." | 3: Cosmetic. NOT YET FIXED. |
| 23 | The numbers used in the pictorial representation of regression ANOVA bear no relationship to the cholesterol/dietary fat intake example being discussed, which isn't as clear as it should have been. | 2: Clarity. Noted in print run 1. Clarified in master 21Dec2005. |
| 34 | The picture has incorrect means for the B subgroups (B1 and B2), which have been transposed in places, so there are errors in the third ("predicted by B") and fourth ("predicted by A and B together...") panels. The quoted dferror should be 20, not 22. | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. [Replacement page PDF] |
| 41 | The tables should each use a consistent number of decimal places. | 3: Cosmetic. Noted in print run 1. Fixed in master 21Dec2005. |
| Chapter 3: Practical analysis | ||
| 69 | Not an error, but just to be clear: if Y = aXb, then log(Y) = log(a) + b log(X), which is why a logarithmic transformation yields a linear relationship. | 2: Clarity. Noted in print run 1. Clarified in master 21Dec2005. |
| 91 | There's an inadvertent space before the "17" of footnote 17. | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| 92 | "[this will always be the same as the preceding t test, with F1,k = (tk)2; see p. 103]" - the reference to p. 103 (section 3.10.2) should be to p. 105 (section 3.10.4). | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| 93 | "In our example, the standard error in [the] estimated marginal means table..." | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| 93 | "The standard error of the difference (SED)... (p. 103)" - the reference to p. 103 (section 3.10.2) should be to p. 98 (section 3.8.3) or to p. 105 (section 3.10.4). | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| 98 | "A full explanation is given on p. 103" - the reference to p. 103 (section 3.10.2) should be to p. 105 (section 3.10.4). | 2: Clarity.Discovered in print run 1. Fixed in master 21Dec2005. |
| 101 | "... as described on p. 77 [onwards]" | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| Chapter 4: Pitfalls and common issues | ||
| 113 | "This is a common statistical fallacy; we discussed it on p. 79." This reference (to section 3.7.3, p. 79) would be clearer to section 3.7.4, on p. 80. | 2: Clarity.Discovered in print run 1. Fixed in master 21Dec2005. |
| Chapter 5: Using SPSS for ANOVA | ||
| 125 | "The data is [are] entered..." | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| 138 | "... the sizes of the groups being compared (pp. 91, 103)" - the reference to p. 103 (section 3.10.2) should be to p. 98 (section 3.8.3) and p. 105 (section 3.10.4). | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| 144 | "As for any t test, F1,k = tk2 ... (see p. 103)" - the reference to p. 103 (section 3.10.2) should be to p. 105 (section 3.10.4). | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| Chapter 6: Contrasts and trends | ||
| 154 | In the equation for "treatment effect" contrasts, there is a misprint: "TM = µCON,S ..." should read "TM = µCON,M ..." | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| 155 | The coefficients for the quadratic trend ("Are people happier in the middle of the week?") should be +2, –1, –2, –1, +2 (not +2, –1, –2, +1, +2). | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| 158 | "using the syntax outlined above (p. 150)" - the reference to p. 150 (section 6.1.3) should be to p. 151 (section 6.1.4). | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| Chapter 7: Advanced topics | ||
| 165 | "... when we discussed it earlier on pp. 43 and 161" - the reference to p. 161 (section 7.1) should be to p. 164. | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| 179 | "... when calculated in the usual (Type III) way, SSA + SSB + SSAB = t + x + z does not equal SStotal, which is t + u + v + w + x + y + z" should read "... when calculated in the usual (Type III) way, SSA + SSB + SSAB + SSerror = t + x + z + SSerror does not equal SStotal, which is t + u + v + w + x + y + z + SSerror". | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| 196 | It's not clear why footnotes 41 and 42 are identical and immediately next to each other! :-) Both footnotes entered the draft simultaneously in the version saved at 18:23 on 21 June 2005. | 3: Cosmetic. Discovered in print run 1. Not fixed, as unimportant, and to fix it would alter the numbering of subsequent footnotes. |
| Chapter 8: Specific designs | ||
| 218 | SSfactor is used instead of SSA at one point. | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| 225 | "where a, b[,] and c" | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| 233 | "... a further digit span test when they're on dry land and sober [should read drunk] (U1 V2)..." | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| 234 | The definition of αβij is missing: αβij is the contribution of the interaction of treatment Ui with treatment Vj: αβij = µUiVj – (µ + αi + βj). | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. This alters pp. 234-237 inclusive. |
| 234 | The error is referred to as εijkl, which should read εijk. | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| 250 | "or alternatively" should read "An alternative and better syntax, making the nesting explicit (which is vital if your subjects do not all have identifying codes in your data file that are unique across all A groups), is this:" (as on pages 254 and 259). | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| 259 | "... unique across all AB groups..." should read "... unique across all A groups..." | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| 262 | In the AxBxCx(UxVxS) design, the C, AxC and BxC terms are missing from the between-subjects list, the UxC, UxAxC, UxBxC, UxAxBxC, VxC, VxAxC, VxBxC, VxAxBxC, UxVxC, UxVxAxC, UxVxBxC, and UxVxAxBxC terms are missing from the within-subjects list, and all within-subjects error terms should end "... S/ABC", not "... S/AB". | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| 263 | In the AxBxCx(UxVxWxS) design, the BxC term is missing from the between-subjects list. | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| 295 | Ideally, "–log10[H+]" shouldn't be split across a line. | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| 296 | SScovariate is used in the sources of variance; SSregression and MSregression are used later on p. 296 and p. 297; they refer to the same thing. | 3: Cosmetic. Noted in print run 1. Fixed in master 21Dec2005. |
|
317 318 |
In the df partitioning on p. 317 and in the ANOVA table on p. 318, the total df are given as "abcn – 1", which should read "anuv – 1". | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| 318 | BDMP8V should read BMDP8V. | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| 321 | "αβij, αγik, βγjk and is the contribution of the AiBj, AiCk, and BjCk interactions, respectively" should read "αβij, αγik, βγjk, and αβγijk are the contributions of the AiBj, AiCk, BjCk, and AiBjCk interactions, respectively". | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| 340 | Missing line break before "Calculate the corresponding MS by hand." | 3: Cosmetic. Discovered in print run 1. Not an error in the original; problem in PDF conversion process. Not fixed. Affects pp. 340-341 inclusive, depending on where the software puts the page break. |
| 347 | In the model description, a "+" is missing before εijkmp, and the description of k is missing. | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| Chapter 9: Mathematics | ||
| 367 | "...the sample mean and SD are good estimator[s]..." | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
|
398 399 400 |
Subsections of section 9.8 should be numbered (9.8.1, 9.8.2, ...). | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| Chapter 10: Statistical tables | ||
| Glossary; Further reading; References; Index | ||
| 410 | Definitions should perhaps be given for n (the number of observations, usually in a single treatment condition) and N (the number of observations, usually the number of observations in total). | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
| 411 | BDMP8V should read BMDP8V. | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| 415 | "... σx is the variance [should read standard deviation] of x, and σy is the variance [should read standard deviation] of y..." | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| 424 | "... each subject is only tested at one level of the between-subjects factor(s)." Either "the" or "(s)" should be removed. | 3: Cosmetic. Discovered in print run 1. Fixed in master 21Dec2005. |
| 430 | A one-sample t test is equivalent to testing the intercept term of an ANOVA only when that one-sample t test is used to compare the mean to zero. | 1: Content. Discovered in print run 1. Fixed in master 21Dec2005. |
| 436 | Goldstein (1995): Web link updated; see links page. | 2: Clarity. Discovered in print run 1. Fixed in master 21Dec2005. |
Private area for publisher's use.
