Unit 2A

7. The population is all students who will vote in the election (students who do not vote do not affect the outcome of the election). The sample of students who will vote must be selected randomly, perhaps by a random drawing of student ID numbers, so that there is no bias for age, sex, major, residence (on-campus vs. off-campus). Because most student elections do not register voters, it is difficult to be sure that only students who will vote are sampled. It may be best to ask students if they plan to vote before they are polled, but even this strategy will be subject to some bias.

9. The population is all people in this country. Blood types do not change from birth, so age is not a factor. An unbiased sample must be chosen randomly so it fairly represents the sex and racial background of people in the population.

21. This study is an observational survey because those people interviewed were simply asked questions. There are no case-control groups.

27. An experiment, either single- or double-blind, could answer this question. The treatment group would receive food with red dye and the control group would receive untainted food or food with placebo red dye. However, one may argue that such an experiment is unethical since one group would have to be administered a potentially harmful substance. The alternatives are to conduct a case-control observational study or an experiment with animals (also possibly unethical).

29. A long-term experiment could answer this question. A treatment group would take aspirin regularly and a control group would take a placebo regularly. The heart attacj incidence in the two groups would be compared after several years. Notice that this study may be unethical because one group would have to be denied a potentially beneficial substance.

Unit 2B

7. This survey is subject to selection bias because those who go to supermarkets between 10:00 a.m. and noon are not representative of all people or of all beer drinkers. For example, this process would favor unemployed or partially employed people. Many students answered that the study is biased because people don’t buy beer or drink beer at 10:00 in the morning. This doesn’t matter! People can answer a question about beer at any time of day. Nevertheless the study is biased.

19. The problem asks you to apply the eight guidelines in the unit to a statistical study!

Unit 2C

7c. Clearly there is an upward trend in the data, as the percentage of percentage of full-time women workers increases with time.

9. a. Voter turnout was lowest in 1996 (54.2% of eligible voters actually voted).

b. Unemployment was highest in 1976 (7.7%).

c. Visually there does not seem to be a correlation between voter turnout and unemployment.