{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

ch.2 - _IV'I".IIII Homework jfhz fiolul’tm’t...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Background image of page 2
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: __IV'I".IIII-__. Homework jfhz fiolul’tm’t; ANSWERS TO REVIEW EXERCISES PART I. DESIGN OF EXPERIMENTS Chapter 2. Observational Studies False. You need to know how much people fly by commuter airlines vs. scheduled carriers. Then you can compare rates. Per aircraft mile flown, the scheduled cairiers are safer titan the commuter airlines. Either way, air travel does seem quite satire-compared for instance to travel by car. (:1) False. There were a lot more Bcrcttas on the street. You need to look at rates: 300147.598 is about6 per l000, and 134/ I 8.938 is about 7 per 1000; if anything. thieves prefer Corvettes—a much snappicr car. (b) False. Looking at rates adjusts for the difference in production figures. 'Ihat’s why statisticians use rates and percents. . No. In the Salk trial, the parents who consented were on the whole better off than the parents who did not consent. and their children were more at risk to begin with (p. 4). (a) They were controlling for age and sex as possible confounders: this is discussed on p. 13, with respect to a specilic disease—lung cancer. (b) This is the wrong conclusion to draw. Bit-smokers are a self-selected group, and many people give up smoking because they are sick. So recent cit-smokers include a lot of sick people. (Other epidemiological data suggest that if you quit smoking. you will live longer.) No. The data from the dauble-blind study are more reliable, and suggest that the results from the single-blind were biased. Subjects who did not improve during the first part of the trial probably concluded that they were on the placebo (whether they were or they weren't) and would be switched to the “real“ medication during the second part of the trial. This expectation made them improve—the placebo effect. (a) This is an observational study, so confounding may be a problem. (b) Rates of cervical cancer go up with age; women of different marital status have different patterns of sexual activity, and are therefore exposed to different kinds of risk; similarly for education. In other words. these are potential confounders. (c) Pill users are more active sexually than non-users. and have more partners. That seems to be what makes the rate of cervical cancer higher among pill users. (This is like example 2 on p. 16 or exercise 1 1 0n p.23.) (d) No; see (c). 9, (21) False. (b) 'l‘rue. (c) False—that is the whole point of experiments. Discussion. People who eat lots of fruits and vegetables are different from the rest of us in many other ways. Some other aspect of diet or life style may be protective. Of course. the observational studies might be right; something in the fruits and vegetables other than the vitamins might be the protective factor. 10. (a) Observational study. (b) Yes. (c) Yes. (d) No—the gene would also have to be associated with controlling behavior by the mother (p. 20). (e) Mothers who see their children over—eating might react in a way that psychol- ogists would interpret as “controlling" behavior—Johnny, stop eating! (f) No. The Chronicle seems to have over-reacted. 11. (a) The treatment group consists of those who finished boot camp. The control group consists of other prisoners—including those who do not volunteer, or those who volunteer but do not complete the program. (b) This is observational. The prisoners decide whether to volunteer for boot camp and whether to stay in the program or drop out. That is the problem—those who volunteer and stay the course might be quite different from the rest. (c) False. Comment. An experiment could be done either like the polio trial (p. lff) or the HIP trial (exercise 9 on pp.22—23): Like the polio trial. Ask for volunteers. Randomize some of the volunteers to treatment (assignment to boot camp) and some to control. Compare the recidivism rate for the two groups—but include the dropouts in the treatment group. (Otherwise, you still have the problem of self-selection.) 1 1 Like the HIP trial. Take a group of prisoners. Randomize some to treatment l (invitation to participate in boot camp) and some to control. Compare the recidivism rate for the two groups—but include in the treatment group those who decline to participate and those who drop out. (Again, this is to guard against the problem of self-selection.) 12. False; the conclusion does not follow. This is just like the admissions study (pp. 17ft). The Democrats may be concentrated in wards with low turnouts. Here is an example. with only two wards (and sec exercise 13 on p.24). DEMOCRATS REPUBLICANS Total number Number voting Total number Number voting Ward/X 1000 100 100 5 WardB 100 60 [000 500 I“ A} percent. l’ Jemocm l7 :- J‘s" :402 > Permul’ Pele/bl [00¢ ‘ - - i: : {Z 45 g' ' - W“ ._ a 0 ,. r ' _ u 1'“ 8/ M7 5'60! m) _ §°°:fo£ 6‘ i ' i “in: t 300+ o _ - a" f ' Jun. A! {A MIL—alt P¢((_¢ul’ Jflmctfdtb : ‘ “l“- 'H-\ /.'1 i'<: 24F J loci-f .v 5'05"-4r : --"""- -‘i' ‘ i100 e ...
View Full Document

{[ snackBarMessage ]}