Below you will find help with selected exercises from the book.

11-5,17
11-10, 12
11-16, 4
11-19, 5

11-5, 17. According to two reports in the New England Journal of Medicine, oil from fish can prevent heart disease.

(b) causal factors in populations. Watch out for a phrase like "heart disease," which is stated in the singular. In such instances a singular noun indicates a group. (See Chapter 2 on grouping ambiguity.)

11-10, 12. "A new study shows that the incidence of cancer tumors in rats exposed to high doses of X-rays dropped dramatically when the food intake of the rats was cut by more than half. Dr. Ludwik Gross . . . noted that this study is the first to demonstrate that radiation-induced tumors can be prevented by restricting diet.

"The experimenters exposed a strain of laboratory rats to a dose of X-rays that produced tumors in 100 percent of the rats allowed to eat their fill—about five or six pellets of rat food a day.

"When the same dose of X-rays was given to rats limited to two pellets of food a day, only nine of 29 females and one of 15 males developed tumors, the researchers reported. . . ." (Paul Raeburn, Sacramento Bee)

The book's answers to most of the questions here are self-explanatory. But a few are worth commenting on. First, although the answer to b. is "rats," do not use the subjects of such studies as a way of dismissing their results. Rats resemble human beings closely in their development of cancer and responses to treatments. All things being equal, we have good reason to expect a result about rats to apply to humans.

c. Controlled cause-to-effect experiment. This one is easy as soon as you notice that a study was conducted experimentally. Any experiment must be cause-to-effect.

e. As above, the desired effect is not having a tumor. This effect exists in none of the control group. While we don't know the size of that group, we do know it has a 0 percent frequency of effect. In the experimental group, 10 rats developed tumors, so 34 did not. Dividing that number by 44, we get about .77, or 77 percent.

f. Assume that the control group is roughly the size of the experimental group, that is, near 50. Then d must exceed 19 to be significant. Since d = 77, we have no problem with statistical significance.

11-16, 4. The reason Collins was ill is that she ate and drank more than she should have.

Explanation. Context matters. The example can be an argument if the speaker is trying to convince you that Collins was ill, or if the speaker wants to convince you not to eat and drink too much ("Look at what happened to Collins"). But if you know that Collins was ill and you want to know why – which is the natural way to read this example – it is an explanation.

11-19, 5.What is the frequency of the effect in the experimental and control groups, including size?

First, make sure you've identified the effect. Since we can assume that the point of a vaccine is to prevent infection, rather than cause it, we call the effect "remaining uninfected." One hundred percent of the experimental group enjoyed this effect. As for the control group, infection in 39 of the 446 children means that 407 remained uninfected; divide this number by 446 to get about .91, which means 91 percent. Then d, incidentally, is 9. We round the experimental group size up to 500 (for simplicity's sake) and learn from Table 11-1 that this d exceeds the figure of 6 demanded by the group size. So the result is statistically significant.