Chapter Image

Errata for the 7th edition


This is an ongoing listing of errors in the 7th edition. I give credit below to those who point out errors. When I don't know the name of the person who pointed out the error, I am still grateful even if I can't give credit.

Because of the way I expect to distribute this list, I cannot be sure that subscripts and Greek symbols will come across. Subscripts are usually shown in parentheses [as in s(1)], superscripts as s^2, and Greek letters are spelled out.

The Text

Page Line Correction
27 Fig 2.9 The leaf for males/stem = 6 should be 6, not .6. Thanks to Marie Thomas at Cal St U San Marcos
30 11 lines from bottom I intended (sigma X)^2 not sigma X^2. Of all that places to make that error, this was the dumbest. Thanks to Robert Fierro at Pitzer College.
42 End 3rd paragraph Jeanette Hunter, at the University of Canberra, pointed out that the limits should be $35,000 - $43,000, not $25,000 - $43,000. Just to make me feel worse, about 4 lines above there is a decimal point when there should be a comma.
78 Figure 3.10 Andrew Waters pointed out that both of the Q-Q plots have reversed the labels on the X and Y axis. The label "expected quantiles" should be on the X axis and vice versa.
90 middle Mark Yates, at the Univ. of South Alabama, pointed out that I wrote "p < .5" when I should have written "p < .05."
96 End paragraph 2 Jennifer Pollock at Texas Tech pointed out that the paragraph should end with "different means" rather than with "same mean." That was a dumb error.
102 4th line from bottom Replace the Greek letters eta and nu with "h" and "n". Thanks to Marie Thomas at Cal St U San Marcos
104 First full para--4th line from bottom The Greek sigma should be a Greek alpha. Thanks to Marie Thomas at Cal St U San Marcos
124 Last equation p(P|NU) = .10, not .15, so the result would be .513 instead of .413. Thanks to Michael Warren at Claremont Graduate University.
125 First two paragraphs
First paragraph of Second Example
Change .413 to .513.

Delete the second sentence from the end in that paragraph. Next, change p(g|B) to p(b|G).
214 Near middle Mark Yates made the following observation. "I noticed that where you list "Min(n1-1,n2-1) <= df'" the upper bound of df' <= n1+n2-2 is missing." He is correct and that should be added.
Ex 7.49 The reference should be back to Exerice 7.27, not Exercise 7.25.
227 Figure 8.1 The arrow labeled "critical value" should point to the vertical line between the light and dark shaded areas. Chris Green at York pointed this out to me in an earlier edition, but I failed to make the change as I should.
??? Figure 9.1 Justin Fuller at Ohio University offered some interesting suggestions to explain the positive relationship between infant mortality and the number of physicians. It is very possible that this is a reporting problem--the more physicians, the better the reporting. It is also possible that physicians increase the rate of live births that then die soon after birth and are counted in the mortality rate.
263 near bottom Mark Yates, at the University of South Alabama, pointed out that the percent reduction in error is equal to r2, not to 1-r2. If, for example, the variance of Y was 75, and the residual variance after we predict Y from X is 40, then we have reduced the error by 75 - 40 = 35, which is a reduction of (75 - 40)/75 = 47% = r2.
268 Footnote The formula is incorrect. Delete the "1 +" under the radical. Sometimes I even find my own errors.
348 Calculation of RMSSE Karl Wuensch pointed out that the sum of the squared deviations is 35.152, which produces a RMSSE of 0.953. Rafael Klorman at Rochester pointed out the same error.
365 Middle of page "FW is in the general vicinity of c*alpha'," not c*alpha.
372 Middle of page The value of phi is -1.25, not -2.5. Thanks to Amanda le at Harvard.
388 Calculation of d2 The label is wrong. It should read "M-S, M-M, S-S versus S-M, Mc-M". Thanks to Amanda le at Harvard.
396 Line 9 FEW should be FWE.
398 Table 12.8 Someone, I forgot who, noted that Dunnett's test, which compares each mean with a control mean, is almost by definition an a priori test. Good point!! But it is traditionally listed as a post hoc test.
428 Line 7 I wrote that the "Total" in the SPSS summary table is Σ X2/N. That is silly. Delete the N. I seem to have been thinking about the correction factor, which is (Σ X)2/N Thanks to Nathan Smith at McGill.
436 SPSS code In that code the number "5" appears everywhere that "=" should have been. I think that is related to different fonts that the printer used, but I don't know why there should be such a problem. Thanks to Karl Wuensch for pointing this out. Karl also questioned the syntax for SAS and I agree with him. The code should say
Proc GLM data = GenderTherapist;
    Class Gender Therapist;
    Model dv = Gender Therapist(Gender);
    Random  Therapist(Gender)/test ;
    Test H = Gender E = Therapist(Gender);
Notice that the random statement has been changed in two ways.
436 Table 13.9 Joshua Wiley at UCLA pointed out that the total SS should be 2667.79 and not 550.775. That's what happens when you try modifying old tables and aren't careful enough.
533 First par. under Sample Sizes The expression that I give for the expected value of R for random data is actually the expected value of R2. The expected value for R is approximately the square root of that.
537 Figure 15.4 There is supposed to be a box around that diagram. 100% of the variance is assumed to be contained within that box. David Scott and the University of Chester pointed this out. Chris Martin at William and Mary pointed out that even with the box, the brackets are very unclear. Rather than telling you how to redraw the brackets, simply take the description from the text. But even there we have a problem because the proportion of residual variation should be 1.000 - .886 = .114, not .059.
545 Last equation Sara Saeed at the Univ. of Ottowa pointed out that I had divided by the MSresidual(reduced) when I meant to divide by the MSresidual(full) = 680.403. The result should be 0.456.
565 Third line Chris Martin at William and Mary pointed out that the reference is to Figure 15.10, not to 15.8.
592 First line of text. Chris also pointed out that the reference to Exhibit 16.2 should be to Table 16.3.

Answers Exercise 7.25 This is all screwed up. The correct answer is t = 3.186, p = .002
Answers Exercise 7.27 This is all screwed up too. The correct answer is t = 5.00
Answers Exercise 7.49 This is all screwed up because earlier ones were. The correct answer is d = 0.47, using the pretest standard deviation. Katherine Long, at Fordham University, pointed me to the last three errors. No wonder she couldn't get the answer that I gave.
Answers Exercise 15.31 I gave the wrong value for e.488. It is 1.649 and the relationship is positive, as it should be. Mark Darby at UNC Greensborough pointed this out long ago but I failed to correct it at the time.


Data Sets

There is a self-extracting archive of all of the data found in Simply click on this link and tell the resulting dialog box where you want to store the download--the desktop is fine. After the file is downloaded, click on the downloaded file and indicate where you want the data files to be stored. Once the data are extracted, you can safely delete the archive file itself.


Last revised: 3/31/2009