Please email me if you notice any errors.
Mistakes
- Page 129, answer to second question. Replace the second (incorrect) and third (irrelevant) sentence with, "It depends on the scientific context."
- Page 148, end of first paragraph. The phrase "statistically significant" should be "not statistically significant".
- Page 227, first sentence of the third paragraph. Change from "Here is a conclusion that is easy to remember." to "Here is a rule-of-thumb that can be used when the two sample sizes are equal or nearly equal."
- Page 227. second sentence of fourth paragraph. Change "do not overlap, the P" to "do not overlap, and the sample sizes are nearly equal, the P".
- Page 229, legend of table 30.3. Replace entire legends with: "Rules of thumb for interpreting error bars that overlap or do not overlap. These rules only apply when comparing two means with an unpaired t test, and when the two sample sizes are equal or nearly equal."
Minor corrections
- Page 104, line 3. The word "value" is extraneous. It should read, "The SEM computed from one sample is your best estimate..."
- Page 107, legend to Figure 14.1. The SEM and CI error bars don't actually portray variability -- they show how precisely you have determined the population means.
-
Page 132, Figure 17.2. The 95% CI should be centered on the Observed Mean and the label n=130 shouldn't be split on two lines. Correct figure:
- Page 154. Replace the paragraph after the two bullet points with: "Each hypothesis is flipped to form a one-sided null hypothesis (that the mean ratio is less than 0.80, and that the mean ratio is greater than 1.25). If both one-tail P values are smaller than a preset threshold, you reject both null hypotheses, and so conclude that the mean ratio must be between 0.80 and 1.25. Thus the results demonstrate equivalence."
- Page 161, first heading. Change "not needed" to "not essential".
- Page 161, sixth paragraph. Change "No corrections were needed," to "The reader does not need to try to informally correct for multiple comparisons, "
- Page 165, second to bottom paragraph. Change "for 90%" to "for at least 90%", and "10% to be" to "no more than 10% to be".
- Page 229, caption to Figure 30.5. Change "less than 0.005" to "much less than 0.05." With large samples, in fact the P value must be less than than 0.005, as it is here. But a more general statement is to say that non-overlapping 95% confidence intervals mean the P value is much less than 0.05.
- Page R-2. Missing reference (cited on p 174): Berry, D. A. (2007). The difficult and ubiquitous problems of multiplicities.Pharmaceutical Statistics , 6, 155-160.
- Page R-8. Missing reference (cited on page 353): Wilcox, R. (2001) Fundamentals of Modern Statistical Methods: Substantially Improving Power and Accuracy. New York: Springer. ISBN=0387951571
Trivial
- Page 112, line 3. Change "fraction" to "proportion".
- Page 112, middle of page, first sentence under "how many P values were calculated?". The sentence should end with a period, not a question mark.
- Page 114, line 2. Change "1/10" degree Fahrenheit to "1/2".
- Page 153, middle of the 5th paragraph. “Figure 21.1” should be 21.2, and “Figure 21.2” should be 21.3.
- Page 154, second paragraph. “Juggling two null hypothesis …” should be “Juggling two null hypotheses”.
- Page 162 seems to start midsentence as if something is missing. Nothing is missing, but a sentence is hidden between Table 22.1 and 22.2.
- Page 162, Table 22.2, middle of the last row. Insert “is” between “hypothesis” and “really”.
- Page 162, second to bottom line. "Table 22.3" should be "Table 22.2".
- Page 166, middle of the page: Change "2/200" to “2/100”.
- Page 316, second line. Remove the word "equals".
- Page 358, bottom. The first paragraph under "Bayes revisited" is incomplete. Change "discussed in Chapter" to "discussed in Chapter 18."
- Page 384, second line. "Weight" should be "width".