Lesson 6.2.1
6-46. Hereare some more news headlines from real observational studies. Just as you didin problem 6-43, determine at least one plausible lurking variable that could explainthe cause and effect. Remember, do not argue about the link expressed in theheadline. Accept the association or link as true. Your task is to find the othervariable(s) that could be the actual cause(s).
a. “Teens with own cars more likely to crash”
[ With a car readily availablethese teens might simply be driving more and the extra time on the road is causing them to be in more crashes. ]
b. “Study connects hyperactivity, food additives”
[ A child who eats more thanaverage amounts of food additives may have parents who have a relaxedattitude about meals, schedules and overall discipline this may cause a childto appear hyperactive at school. ]
c. “The graveyard shift may be aptly named. Working nights will soon be listed asa likely cancer cause”
[ Many nighttime jobs involve working in bars,casinos, or restaurants where smoking is prevalent. Night employment isgenerally considered less desirable so people who work at night may haveless money and therefore less access to medical care. ]
d. “Daily meat diet tied to higher chance of early death”
[ This might beconnected to gender. Men as a group eat more meat and don’t live as longas women. Also, if the meat is highly processed like hotdogs, it might be theadditives that are harmful and not the meat itself. ]
6-47. Explain the difference between an association and causation. How could a researcherprove causation?
[ Association means that two or more things are related.Causation is when one event actually leads to another.]
Lesson 6.2.2
6-54. Consider the question: “Does a traditional classroom SAT preparation course improvescores more than an on-line study course?”Design an experiment that could help to answer this question. Refer to problem 6-48
for ideas. [ See sample diagram below. ]
6-55. Design experiments for the following statistical questions. If it would be unethical toconduct such an experiment, state why.
a. “Does listening to classical music during a math test improve scores?”
[ Thesimplest way to set up this study would be to get a large group of volunteersand randomly assign them to two groups who will take the same math testin different rooms. In one room classical music is played and no music isplayed in the other (control group). The scores from the groups could beaveraged and compared. An expanded version of the experiment couldhave additional groups listening to other kinds of music while taking thetest for comparison. ]
b. “Do seat belts save lives in car crashes?”
[ People would need to be randomlyassigned to random cars that will crash. One group would wear seatbeltsthe other would not and the percentage of fatalities would be measured andcompared. Clearly this is not ethical but reasonable results might beobtained in an experiment using crash dummies instead of humansubjects. ]
c. “Does vitamin C help prevent colds?”
[ A large group of volunteers would berandomly selected and exposed to a cold virus. The subjects would berandomly assigned to two treatment groups. One group gets a dailyvitamin C supplement and the other group receives a daily placebo. Thepercentage of each group that contracts a cold could be calculated andcompared. ]
Lesson 6.2.3
6-57. Answer some of the same questions from these actual newspaper headlines. If anobservational study was done, explain how. Explain why an experiment was notpossible. If the study shows an association between variables, discuss the effects of atleast two possible lurking variables. If you believe an experiment was done, state so,and outline a possible experimental design. If surveys were necessary, list a potentialsource of bias in the question(s), and a potential difficulty in getting a representativesample from the population.
a. “Study sticks it to traditional back care. Acupuncture – real and fake – getsbetter results for pain than the usual treatments.”
[ Because a placebo ismentioned this most certainly was an experiment. A group of volunteerswith back pain were randomly distributed between at least 3 test groups.One group received traditional medical care, another got acupuncturetreatments and the last got fake acupuncture. The difference in painreported by each patient was likely averaged for each group and theaverages compared. Measuring the subjects’ pain was likely done bysurvey, i.e., “Rate your back pain today from 1 to 10…” The physicalsurroundings of the interview or the compassion level of the interviewermay be a source of bias. ]
b. “MARITAL STRIFE A HEART WRECKER? Bad marriage can risk coronarydisease risk, researchers say”
[ This is an observational study. A large groupof people were given questionnaires including questions about theirmarriage and their health. This cannot be an experiment because arandomly selected group of people would have to be forced into “bad”marriages and their cardio health compared to a group forced into “notbad” marriages. Perhaps people under a great deal of stress tend to havebad marriages and tend to have heart problems. Maybe people withaddiction problems tend to have bad marriages and tend to have heartproblems. Surveys were used. Trying to get a random sample of responsesfrom the world’s married population would be nearly impossible. It islikely the respondents had something in common like an employer orhealth plan. Maybe having a bad marriage or heart condition made peoplemore likely to respond to the survey. What constitutes a “bad” marriage?Questions that start with “Have you ever …” could bias the results, e.g. …thought about a divorce … had a disagreement with your spouse … etc. ]
c. “Breastfeeding May Cut Breast Cancer Risk. Women with a family history ofbreast cancer who have ever breastfed reduce their risk of gettingpremenopausal breast cancer by nearly 60%, according to a new study.”
[ Thisis an observational study. A large group of women were givenquestionnaires including questions about breastfeeding and whether theyhave had breast cancer. This cannot be an experiment because a randomlyselected group of women would have to be forced to breastfeed theirchildren and their breast cancer rates compared to another group forced tonot breastfeed their children. Perhaps the kind of woman who breast feedsher child is less likely to drink alcohol or smoke. Maybe women who breastfeed eat healthier foods than those who do not. Surveys were used. Tryingto get a random sample of responses from the world’s population of womenwith a family history of breast cancer would be nearly impossible. It islikely the respondents had something in common like an employer orhealth plan. Maybe the type of person who takes (or has) the time to breastfeed their children, also takes the time to fill out health surveys, so thatgroup was over represented in the sample. ]
d. “Study: Oral drug better than lotion to kill lice… A new study has found that intough cases, an (new) oral medication kills the parasites more effectively than aprescription lotion applied to the scalp.”
[ The word “study” is used in theheadline but because the oral medication is new, an observational study isnot possible. This is an experiment. As an experiment it required anumber of volunteers already infected with head lice. They need to berandomly assigned to at least two groups, three if a placebo medication isincluded. One group gets the oral medication and the other gets theprescription lotion. After a period of time the number of lice on eachperson is counted and the group averages compared. ]