STAT 250 - 1.5
control group
group that does not receive the treatment or does not have the characteristic of interest
treatment group
group that receive treatment or has the characteristic of interest
placebos
harmless pill given in place of an actual treatment; often used in designed experiments to counteract the placebo effect
four key features of controlled experiments:
1. sample size must be large 2. subjects of the study must be assigned to the treatment and control group randomly 3. should be double-blind 4. should use a placebo
A doctor who believes strongly that antidepressants work better than "talk therapy" tests depressed patients by treating half of them with antidepressants and the other half with talk therapy. After six months the patients are evaluated on a scale of 1 to 5, with 5 indicating the greatest improvement. Answer parts (a) through (d) below. a. The doctor is concerned that if his most severely depressed patients do not receive the antidepressants, they will get much worse. He therefore decides that the most severe patients will be assigned to receive the antidepressants. Explain why this will affect his ability to determine which approach works best. A. If the doctor decides on the treatment, this could introduce bias. B. The doctor is letting his empathy get in the way of treating the patients objectively. This may cause personal issues for either the doctor or the patients over the course of the treatment. C. The doctor is assigning all antidepressants to only the most severely depressed people. D. The doctor is creating a second treatment group that is dependent on the severity of the patient's depression. This will further confound the results.
A
Indicate whether the following study is an observational study or a controlled experiment. A researcher is interested in the effect of music on memory. She randomly divides a group of students into three groups: those who will listen to quiet music, those who will listen to loud music, and those who will not listen to music. After the appropriate music is played (or not played), she gives all the students a memory test. A. This is a controlled experiment. She assigns students to the control and treatment groups at random in order to control for all relevant factors aside from the effect of music on memory, which is essential to conducting a controlled experiment. B. This is an observational study. The number of treatment groups is not sufficiently large to control for all relevant factors aside from the effect of music on memory; a large number of treatment groups is essential to conducting a controlled experiment. C. This is a controlled experiment. She controls for other factors prior to assigning students to the control group and two treatment groups. This is essential to conducting a controlled experiment. D. This is an observational study. The researcher observes only the results of the memory test. Without observing the subjects' performance on memory tests prior to assignment to different groups, the study is not sufficiently active to make this a controlled experiment.
A
Indicate whether the study is an observational study or a controlled experiment. Patients with multiple sclerosis are randomly assigned a new drug or placebo and are then given a test of coordination after six months. A. This is a controlled experiment because the patients were assigned drugs by those conducting the study. B. This is an observational study because the patients were assigned drugs by those conducting the study. C. This is an observational study because the patients chose whether they took the drug or the placebo. D. This is a controlled experiment because the patients chose whether they took the drug or the placebo.
A
Why is random assignment used to assign people to treatment groups and control groups in a controlled experiment? A. To make the groups as similar as possible, minimizing bias. B. To ensure that the researchers do not know which groups subjects are members of. C. To ensure that the groups have equal numbers. D. To make sure that the percentage of men and women in each group is exactly the same.
A
Which of the following is a reason we can never draw cause-and-effect conclusions from observational studies? A. Observational studies often do not involve a large enough sample to draw cause-and-effect conclusions. B. Researchers may be biased in the observations they choose to record. C. Potential confounding variables may explain the differences between groups rather than the treatment variable. D. Observational studies are not scientific in nature.
C
A doctor who believes strongly that antidepressants work better than "talk therapy" tests depressed patients by treating half of them with antidepressants and the other half with talk therapy. After six months the patients are evaluated on a scale of 1 to 5, with 5 indicating the greatest improvement. Answer parts (a) through (d) below. c. The doctor asks you whether it is acceptable for him to know which treatment each patient receives and to evaluate them himself at the end of the study to rate their improvement. Explain why this practice will affect his ability to determine which approach works best. A. If the doctor is aware of the treatment each patient receives, he may become too invested in the well-being of the patients. B. If the doctor is aware of the treatment each patient receives, that might influence his opinion about the effectiveness of the treatment. C. If the doctor is aware of the treatment each patient receives, it may negatively affect the relationship between the doctor and patient during therapy sessions. D. If the doctor is aware of the treatment each patient receives, he may bribe the patients to skew the results in his favor.
B
A doctor who believes strongly that antidepressants work better than "talk therapy" tests depressed patients by treating half of them with antidepressants and the other half with talk therapy. After six months the patients are evaluated on a scale of 1 to 5, with 5 indicating the greatest improvement. Answer parts (a) through (d) below. d. What improvements to the plan in part (c) would you recommend? A. To prevent bias, the doctor should use an evaluation procedure that is resistant to bias. B. To prevent bias, the experiment should be double-blind. Neither the patients nor the doctor evaluating the patients should know whether each patient received medication. C. To prevent bias, the evaluation should be completed by another doctor in order to make the study as objective as possible. D. To prevent bias, the sample size should be increased to reduce the effect of his bias on the study.
B
What is a placebo and what purpose does it serve in an experiment? A. A placebo is a baseline measurement that is applied to the control group. Placebos help the experimenter avoid confounding from different variables. B. A placebo is a fake treatment that looks like the treatment being tested in the experiment. Placebos blind subjects so they do not know whether or not they are receiving the treatment. C. A placebo is a combination of specific levels from all the factors that an experimental unit receives. Placebos allow the experimenter to determine the effect that different levels have on the response variable. D. A placebo is a variable whose values are controlled by the experimenter. By varying the levels of the placebo, the experimenter can try to determine the effect on the response variable.
B
A difference between two groups in an observational study that can explain why the outcomes were very different between the groups is called what? A. An outcome variable B. A causality C. A confounding variable D. A treatment variable
C
A study compared the rate of pneumonia before and after a vaccine was introduced. In the study, annual hospitalization rates were estimated from any cause using a database. Average annual rates of pneumonia-related hospitalizations before and after introduction of the vaccine were used to estimate annual declines in pneumonia-related hospitalizations. The annual rate of pneumonia-related hospitalizations among children of various age groups significantly declined relative to expected rates before introduction of the vaccine. Does this show that pneumonia vaccine caused the decrease in pneumonia that occurred? Explain. A. The study does not show that the vaccine caused the decrease in pneumonia. This is a controlled experiment. Children cannot choose to be vaccinated, so whether or not a child was vaccinated is random. Confounding factors would affect all children equally, so such biases are eliminated. B. The study shows that the vaccine caused the decrease in pneumonia. This is an observational study because the number of treatment groups isn't sufficiently large to control for all relevant factors aside from vaccination. C. The study does not show that the vaccine caused the decrease in pneumonia. This is an observational study because the children were not randomly assigned by the researchers. It is possible that confounding variables (other advances in medicine, for instance) would affect the rates of pneumonia. D. The study shows that the vaccine caused the decrease in pneumonia. This is a controlled experiment. The children not vaccinated represent the control group, and the children vaccinated represent the treatment group. Since the vaccine was administered on a sufficiently large scale, other possible biases are eliminated.
C
Of the following, which is the only method of data collection suitable for making conclusions about causal relationships? A. Observational studies B. Anecdotes C. Controlled experiments D. All three are suitable.
C
What purpose does randomization serve in an experiment? A. Randomization insures that the effect of a treatment is not due to some characteristic of a single experimental unit. B. Randomization insures that individuals do not adjust their behavior in some way to the treatment they are receiving. C. Randomization insures that the effect of factors whose levels cannot be controlled is minimized. D. Randomization fixes the value of factors whose effect on the response variable is not of interest in the experiment.
C
A study was conducted to see whether participants would ignore a sign that said, "Elevator may stick between floors. Use the stairs." Those who used the stairs were said to be compliant, and those who used the elevator were said to be noncompliant. There were three possible situations, two of which involved confederates. A confederate is a person who is secretly working with the experimenter. In the first situation, there was no confederate. In the second situation, there was a compliant confederate (one who used the stairs), and in the third situation, there was a noncompliant confederate (one who used the elevator). The subjects tended to imitate the confederates. What more do you need to know about the study to determine whether the presence or absence of a confederate causes a change in the compliance of subjects? A. Identify the absence or presence of a placebo. Without a placebo, the treatment could be confounded by the participant's expectation of the treatment. B. Identify the number of treatment groups. Without a sufficiently large number of treatment groups there is insufficient variation in the study, so we cannot infer causation. C. Identify whether there was random assignment to groups. Without random assignment there is the possibility of bias, so we cannot infer causation. D. Identify the sample size of the study. Without enough participants to observe the full range of variability in subjects we cannot control for other relevant factors, so we cannot infer causation.
C, D
A college magazine suggested that overeating reduces brain function. Is this likely to be a conclusion from observational studies or randomized experiments? Can we conclude that overeating causes a reduction in brain function? Why or why not? A. This is likely to be from controlled experiments. The magazine could have received the results from a university researcher who performed controlled experiments on overeating and neural function. It is possible to conclude causation from well-designed controlled experiments. B. This is likely to be from controlled experiments. Researchers may conduct IQ tests on people with normal diets and people who overeat to determine the relationship between overeating and brain function. We cannot conclude causation from controlled experiments because of the possibility of confounding factors. C. This is likely to be from observational studies. Surveys of people with normal diets and people who overeat could determine how overeating affects brain function. We can conclude causation from observational studies with a sufficiently large sample size. D. This is likely to be from observational studies. It would not be ethical to assign people to overeat. We cannot conclude causation from observational studies because of the possibility of confounding factors.
D
A doctor who believes strongly that antidepressants work better than "talk therapy" tests depressed patients by treating half of them with antidepressants and the other half with talk therapy. After six months the patients are evaluated on a scale of 1 to 5, with 5 indicating the greatest improvement. Answer parts (a) through (d) below. b. What advice would you give the doctor to improve his study? A. The doctor should divide the patients into groups based on how depressed they are, and give out a certain percentage of antidepressants to the least depressed group, a higher percentage to the more depressed group, and so on. B. The doctor should remain as objective as possible in order to increase the chances of successfully treating the patients. C. The doctor should remove the least depressed people from the study. D. The doctor should randomly assign the patients to the different treatments. E. The doctor must give an equal number of antidepressants to the least depressed patients and the most depressed patients in order to make the treatment sufficiently random.
D
A study concludes that the use of pesticides is associated with the development of Parkinson's disease, a neurological disease that causes people to shake. The study reported that exposure to bug killers and weed killers is "associated with" an increase of 33% to 80% in the chances of getting Parkinson's. Does this study show that pesticides cause Parkinson's disease? Why or why not? Select the correct answer below. A. The study shows that pesticides cause Parkinson's disease. This was a controlled experiment because the only way to ensure exposure to pesticides is to use low levels of pesticides on the treatment group. Assuming all key features are satisfied, controlled experiments can conclude causation. B. The study does not show that pesticides cause Parkinson's disease. The researchers did not administer a placebo along with the low levels of pesticides administered in the study. Therefore, a sufficiently robust control group was not established. C. The study shows that pesticides cause Parkinson's disease. In statistics, the term "associated with" is synonymous with "causes." Therefore, it may be concluded that pesticides cause Parkinson's disease. D. The study does not show that pesticides cause Parkinson's disease. This was an observational study because researchers could not have deliberately exposed people to pesticides. Observational studies cannot conclude causation.
D
In a television advertisement, a company called "Waist Away" claimed the workout program on their set of DVDs would help people lose weight more than any other DVD workout program. To test this claim, an independent company, called "Slim Down," selected one other DVD program. They then randomly assigned half the volunteers to the Waist Away program and the other half to the Slim Down program. Each participant was weighed before they started the program and then regularly participated in their assigned program for one month. After one month, each participant was weighed again. The percent of weight lost was recorded for each person, where negative values indicated a weight gain. What type of study was performed? A. sample B. anecdotal study C. observational study D. experiment
D
Indicate whether the study is an observational study or a controlled experiment. A group of boys is randomly divided into two groups. One group watches violent cartoons for one hour, and the other group watches cartoons without violence for one hour. The boys are then observed to see how many violent actions they take in the next two hours, and the two groups are compared.
The study is a controlled experiment.
anecdotal evidence
a story based on someone else's experience
confounding variable
a variable that has not been accounted for but which is causing a difference in the groups being studied
Two drugs were tested to see whether they helped women with breast cancer. Of 1060 women, about half were randomly assigned to drug A and the other half were assigned to drug B. After 77 months, 473 out of 539, and 426 out of 521 women assigned to drugs A and B, respectively, were alive. Complete parts (a) and (b) below.
a. The survival rate for drug A is 87.8%. The survival rate for drug B is 81.8%. The survival rate for drug A is higher than the survival rate for drug B. b. Was this a controlled experiment or an observational study? Explain why. From studies like these, can we conclude a cause-and-effect relationship between the drug type and the survival percentage? Why or why not? A. This is a controlled experiment because the researchers randomly determined the groups. Subjects are pre-screened for genetic factors that make it more likely that an experimental drug will be effective. All genetic confounding factors have been controlled for, which is essential to conducting a controlled experiment. B. This is a controlled experiment because the researchers randomly determined the groups. We can conclude a cause-and-effect relationship because the women were randomly assigned to the treatment and control groups, which controls for other variables. C. This is an observational study because the subjects determined the groups. The researchers did not assign a placebo to any women, which is essential to conducting a controlled experiment. D. This is an observational study because the subjects determined the groups. The sample size is not sufficiently large to capture the variable effects of breast cancer. This feature is essential to conducting a controlled experiment.
blinding
independent party assigns subjects to treatment groups so that the researchers do not know who is in the treatment group, preventing researchers from treating people in the treatment and control groups differently
double blind
neither researchers nor participants know whether the participants are in the treatment or control group; used in most controlled experiments
placebo effect
reacting to a treatment after being told you are receiving the treat, when in fact, you are not
controlled experiment
researchers assign subjects to a treatment group or a control group; can show causality
observational study
subjects placed into different groups by choice or someone else; can show associations but affected by confounding variables
sample sizes
the larger the better; gives opportunity to observe full range of variability in subjects
random assignment
to have treatment and control group be alike as possible; often uses technology; balances differences in groups to minimize bias
causality
what causes something to happen
outcome (response) variable
whether or not a certain outcome is seen
treatment variable
whether or not a specific treatment is used