Discovering Statistics Using SPSS: Errata

What do you get if you let an idiot write a statistics textbook? Mistakes, and this page is dedicated to them.

Errors in the first print run:

If your copy was parchased before December 2005 it is likely to contain these errors:

• Files on the CD-ROM:
  • Some syntax files are missing, these can be downloaded from the companion website: http://www.sagepub.co.uk/field/field.htm
• Page 19, Figure 1.10:
  • Two of the confidence intervals don't include the mean when they should do (samples 22 and 25). It should look like this.
• Section 3.4:
  • The data file SPSSExam.sav on the CD-ROM is wrong. The correct file is available from the companion website (see above) or by clicking here.
  • The confusion between an old and new version of the data file also created some mistakes in section 3.4. To download a PDF of this section with the corrections made click here. Please note the page numbers do not correspond to the final typeset version of the book. The differences to the printed version are highlighted in yellow (anything not highlighted is the same as the printed version).
  • The bars on one of histograms on page 92 are also missing (see link to PDF file above).
• Page 275:
  • The two figures on this page are the wrong way around.
• Section 8.5:
  • The equation for omega-squared and resulting calculations are incorrect. To download a PDF of this section with the corrections made click here. Please note the page numbers do not correspond to the final typeset version of the book. The differences to the printed version are highlighted in yellow (anything not highlighted is the same as the printed version). [In case your interested, the original equation I used was for the Random effects version of omega-squared; but for most people using the fixed effects version will be more appropriate, so the correction involved replacing the random model equation with the equation for the fixed model ... I bet you're glad you asked!].
• Page 539:
  • In SPSS Output 13.5, Rank table, 3rd line says 'Tiles', should be 'Ties'.

These errors were corrected in the Reprint (December 2005). If your book says 'Reprinted 2005' on the page before the contents then the errors above do not apply.

Errors surviving past the first print run:

Hooray, some new errata have been spotted!

• Page 15:
  • Explanation of the standard error. The sentence 'Therefore, the standard error could be calculated by taking the difference between each sample mean and the overall mean, squaring these differences, adding them up, and then dividing by the number of samples', strictly speaking described the variance of sample means, not the standard deviation of sample means (the standard error). It should read 'Therefore, the standard error could be calculated by taking the difference between each sample mean and the overall mean, squaring these differences, adding them up, and then dividing by the number of samples. Finally, the square root of this value would need to be taken to get the standard deviation of sample means, the standard error.'. (Mike Kenfield)
• Box 1.2 Page 28.
  • The sentence before #3, should read (correction in bold) "So, significance testing can never tell us that the null hypothesis is true...". (Brittany Cornell)
• Box 3.3, page 85
  • Last sentence 'ends will a full stop' should read '...end with a full stop.' (Dr. Mizanur Rashid Shuvra)
• Table 5.1, p. 167:
• value should be -1.90, not -1.09 (Jane Webster)
• Page 197:
• The reference to Chapter 5 should be to Chapter 6. (Jane Webster)
• Table 6.2, Page 246
  • '... 5% should lie between +-1.96 and 1% between +-2.58.' should read '... 5% lie outside +-1.96 and 1% outside +-2.58
    
• Page 365:
  • The highlighted rows for the mean and SD of the placebo group are a line too high. So, the means for participants and partners libido's respectively are 3.22 and 4.11, and their associated SDs are the values in brackets (1.79 and 1.76). (Suh-Ing (Amy) Hsieh).
• Page 376:
  • 34.57 (value for SSM) should be 34.75 (Suh-Ing (Amy) Hsieh).
• Page 382:
  • The text states that the ANCOVA assumption of homogeneity of regression slopes is tested by making an interaction between the covariate and the dependent variable. This should say independent variable. The example itself is correct. (Joop Hox and Ruben van Genderen).
• Page 384:
  • The degrees of freedom for the contrasts should be 26, not 28 as stated (see page 372). (Laura Hutchins-Korte)
• Page 605
  • SPSS Output 14.7: the two rows in the table both say "Level 1 vs Level 3". The data for the second row is for "Level 2 vs Level 3" and should say so (Anthony Fee).
• Page 646
  • Reference to section 15.2.4. should be to section 15.2.3.
• Page 655:
  • The communalities are shown in SPSS Output 15.5 and none exceed .7"; should say "The communalities are shown in SPSS Output 15.5 and only one exceeds .7" (Jane Webster)
• Page.705, Section 16.6
  • "...to delete the animal variable, for this to be valid, the animal x training x dance should be non-significant, and either the animal x training or the animal x dance should also be non-significant." (Jim Parkinson)
• CD-ROM
  • Chapter 8 Answers: the calculations of omega are incorrect because I used the random-effects equation rather than the fixed-effects equation from Chapter 8 (the equation was originally incorrect in Chapter 8, see above, but this got corrected but unfortunately the answers on the CD-Romn did not. A corrected pdf file of answers to chapter 8 can be downloaded here. (Ned Palmer)
  • Chapter 13 answers: the effect size calculation in Task 2 is wrong because I used N = 40 instead of N = 64 (the folley of the cut and paste function ...). To downloaded a corrected PDF of the chapter 13 answers click here. (Paul Rogers)

Thanks to the various people who have spotted these and alerted me to them.