3.1 Answers will vary. No, you cannot because this is your friend and not a sample.

3.3 Answers will vary.

3.5 Answers will vary.

3.7 Yes to both. Data can be available whether it resulted from an observational study or experiment.

3.9 Experiment. It is not a sample survey. The treatment is type of maize (QPM or conventional maize).

3.11 This is an experiment. Explanatory variable: apple form (juice or whole fruit); response variable: how full the subject felt.

3.13 The experimental units were 719 patients. Treatments were (1) no pills, (2) pills with no echinacea, (3) pills with echinacea with no label, (4) pills with echinacea and a label. The outcome was a measure of cold severity. The factor was whether the echinacea was given and whether it was disclosed to the subject. Because no children under age 12 were subjects, the results cannot be generalized to anyone younger than 12 years old.

3.15 Bias is probably present; whether a student is a first-year undergraduate student or a graduate student may affect the evaluations.

3.19

1. The population is 2500 faculty members at Mongolian public universities.
2. The sample is 300 faculty members at Mongolian public universities.
3. No, we cannot calculate the response rate. It would be the percent of the 300 that actually participate.

3.21 Answers will vary.

3.23 Number the list from 0 to 9 and choose three single digits from Table B.

3.25 These three proposals are clearly in increasing order of risk. Most students will likely consider that (a) qualifies as minimal risk, and most will agree that (c) goes beyond minimal risk.

3.27 It is good to plainly state the purpose of the research (“To study how people’s religious beliefs and their feelings about authority are related”). Stating the research thesis (that orthodox religious beliefs are associated with authoritarian personalities) would cause bias.

3.1 This is anecdotal. You have noticed that the can seems lacking but have no data to support the claim.

3.3

1. This is a sample survey if it is supposed to represent only those who did not get tickets; otherwise, it is anecdotal evidence and certainly doesn’t apply to all of those who tried to get tickets to the concert.
2. This is a sample survey of all those who tried to get tickets to the concert.
3. This is also a sample survey of all those who tried to get tickets to the concert. It differs from part (b) because only a sample of students were selected to participate rather than allowing participants to be self-selected.
4. It is not an experiment; there is no treatment.
5. The method used in part (a) is not very useful, especially if it was determined that the individuals were upset before being interviewed. The method in part (b) also can be problematic because those who choose to respond may have strong opinions, especially negative ones. The method in part (c) is likely the best of the three in order to get an accurate representation of students’ opinions on how the tickets were allocated.

3.5 (a)–(b) This is an experiment because patients were assigned into treatment groups. The experimental units/subjects are the patients, the treatments are (1) no pills, (2) pills with no echinacea, (3) pills with echinacea but patients did not know, (4) pills with echinacea and patients know, and the response is the severity of the cold.

3.7 The experimental unit is a mouse. The treatments are (1) no blueberries in a diet, (2) 4% blueberries in a diet, and (3) 8% blueberries in the diet. The outcome is total body bone mineral density. The control group is the group with no blueberries in the diet. Since these are mice, we would not call them subjects.

3.9

1. Experimental units: people who go to the website. Treatments: description of comfort, attractiveness, or showing discounted price. Response variable: shoe sales.
2. Comparative because we have three treatments.
3. The experiment could be biased because the treatments are confounded with time, a lurking variable.
4. One option to improve: randomly assign days or weeks.
5. Yes, a placebo (no special description or price) could give a “baseline” sales figure.

3.11 Comparison: There are three treatments. Randomization: Treatments are randomly assigned as visitors come to the website. Repetition: Would repeat the treatment on many visitors to the website.

3.13

1. The timing could be an issue. Launch all three campaigns in the same month.
2. The exercise routines are confounded with time of year. Randomly assign half of the subjects to each of the routines at each time or randomly select six months, assign the first routine to these six and the other routine to the other six.
3. Launch each promotion both online and in stores.

3.15 For example, new employees should be randomly assigned to either the current program or the new one. One possible outcome would be whether the new employee is still with the company six months later.

3.17

1. The factors are calcium dose and vitamin D dose. There are six treatments.
2. Yes, there is a placebo. The group that gets 0 mg of both calcium and vitamin D serves as the placebo group.

3.21 Design (a) is an experiment. Because the treatment is randomly assigned, the effect of other habits would be “diluted” because they would be more-or-less equally split between the two groups. Therefore, any difference in colon health between the two groups could be attributed to the treatment. Design (b) is an observational study. It is flawed because the women observed choose whether or not to take bee pollen; one might reasonably expect that people who choose to take bee pollen have other dietary or health habits that would differ from those who do not.

3.27

1. The population is all 6874 season ticket holders.
2. The sample is the 100 fans to whom the survey was sent.
3. The response rate is 62%.
4. The nonresponse rate is 38%.

3.31

1. SRS.
2. Voluntary response sample.
3. Stratified random sample.
4. Multistage random sample.

3.33

1. The population is local businesses.
2. The sample is the 70 businesses selected.
3. The nonresponse rate is 62.86%.

3.35 Using line 136 and labels 01–33, we select 08, 14, 20, 09, 24, 12, 11, and 16. Complexes chosen will vary depending on labels; alphabetically, we would choose Burberry, Crestview, Georgetown, Cambridge, Nobb Hill, Country View, Country Square, and Fairington.

3.39 Each student has a 20% chance: 5 out of 25 over-21 students and 4 of 20 under-21 students. This is not an SRS because not every group of 10 students can be chosen; the only possible samples are those with 6 older and 4 younger students.

3.41 The sample is random because the starting point is randomly selected (so every individual has an equal chance to be selected before the process begins). Once the random starting point has been selected, the rest of the sample is determined. There is no possibility of selecting students 04 and 05, which could happen in a simple random sample.

3.45

1. This design would omit households without telephones or with unlisted numbers. Such households would likely be made up of poor individuals (who cannot afford a phone), those who choose not to have phones (perhaps because they use a cell phone exclusively), and those who do not wish to have their phone numbers published.
2. Those with unlisted numbers would be included in the sampling frame when a random-digit dialer is used.

3.51 To control for changes in the mass spectrometer over time, we should alternate between control and cancer samples.

3.53 They cannot be anonymous because the interviews are conducted in person in the subject’s home. They are certainly kept confidential.

3.57

1. Answers will vary.
2. The subjects should be told what kind of questions will be asked and how long it will take.
3. Answers will vary.
4. Revealing the sponsor could bias the poll, especially if the respondent doesn’t like or agree with the sponsor. However, the sponsor should be announced once the results are made public; that way, people can know the motivation for the study, judge whether it was done appropriately, etc.

3.63

1. You need information about a random selection of his games, not just the ones he chooses to talk about.
2. These students may have chosen to sit in the front; all students should be randomly assigned to their seats.

3.65 This is an experiment because each subject is (randomly) assigned to a treatment. The explanatory variable is the price history seen by the subject (steady prices or fluctuating prices), and the response variable is the price the subject expects to pay.

3.69 The two factors are gear (three levels) and steepness of the course (number of levels not specified). Assuming that there are at least three steepness levels—which seems like the smallest reasonable choice—that means at least nine treatments. Randomization should be used to determine the order in which the treatments are applied. Note that we must allow ample recovery time between trials, and it would be best to have the rider try each treatment several times.

3.71 The latter method (CASI) will show a higher percentage of drug use because respondents will generally be more comfortable (and more assured of anonymity) about revealing illegal or embarrassing behavior to a computer than to a person, so they will be more likely to be honest.

3.73 Answers will vary.

3.75 Answers will vary.

3.79 Use a block design: Separate men and women and randomly allocate each sex among the six treatments.