NAME: KEY Ch. 6 Review
FREE RESPONSE:
1) What are the 3 principles of experimental design?
Control, Randomization, Replication
2) What does it mean when two variables are confounded?
The effects of one variable cannot be separated from the other variable.
3) We have 21 people that we need to assign to 3 different treatments (trt1, trt2, trt3). Write clear instructions for randomly assigning the subjects to the 3 treatments. Use the section of the table of random digits below and assign the 21 people to the 3 treatment groups.
1. Use TRD and read off 2 digits at a time
2. Let 01-21 each be a subject. Ignore 00 and 22-99
3. Read off 7 numbers, ignoring repeats, and assign those subjects to treatment 1,
Read off 7 more numbers, ignoring repeats, and assign those subjects to treatment 2.
Assign the rest of the subjects to treatment 3.
TABLE OF RANDOM DIGITS:
08|76|8 1|12|02| 34|85|9 0|92|17| 18|19|4 4|56|21| 05|07|8 6|68|13| 65|46|1 5|04|16| 99|74|2 0|86|01
Treatment 1: 08; 12; 02; 17; 18; 19; 21
Treatment 2: 05; 07; 13; 15; 04; 16; 20
Treatment 3: 01; 03; 06; 09; 10; 11; 14
4) What is the placebo effect?
When a subject shows a response to a perceived treatment even though they are not getting the treatment they believe.
5) What is the difference between a placebo and a control?
A placebo is a dummy treatment that can be used as a control. A control is either a placebo or the baseline/current treatment used to compare the new treatment to.
6) What is a lurking variable?
A variable that may have an effect on the response variable but is not accounted for in the design of the experiment.
7) What is blinding?
When the subject is unaware of which treatment he/she is receiving.
8) What is double-blinding?
Neither the subjects nor the administers of the treatments know what treatment is being used.
9) What is the difference between the explanatory variable and response variable?
The explanatory variable is the variable that is tested to see if it has an effect on the response variable. The response variable is what we measure reactions to changes in the explanatory variable.
10) What is the purpose of blocking?
We block to help control lurking variables. If it is thought that a variable might show different reactions we block before we randomize to be able to analyze the differences between treatments and blocks.
11) What are the three experimental designs we studied in class?
Completely Randomized Design
Block Design
Matched Pairs Design
12) What is the difference between an observational study and an experiment?
An observational study passively observes reactions without imposing treatments. An experiment imposes specific treatments to the subject then observes reactions to the treatment.
13) What are the benefits of an experiment over an observational study?
An experiment can show more valid cause and effect relationships. An observational study can show relationships but not necessarily cause and effect relationships.
14) An investigator wants to study the effects of two different fertilizers on plant growth (call them A and B). There are 20 plots available to test the fertilizers on. The investigator will measure the amount of growth by the plants after 3 months.
a. What is the explanatory variable?
Fertilizer
b. What are the treatments?
Fertilizer A and Fertilizer B
c. What is the response variable?
Growth of plants after 3 months
d. What are the subjects?
20 plots of plants
e. Do you think you should use a placebo group? How about a control group?
No placebo should be used
A control could be that no fertilizer is used. This would give good comparisons to see if the fertilizers are having any effect on the growth versus nothing.
f. What are some lurking variables when it comes to plant growth?
Sunlight, water, type of plants
g. Using one of the lurking variables you listed above, design a block design experiment:
15) An investigator wants to study the effectiveness of two surgical procedures to correct nearsightedness. Procedure A uses cuts from a scalpel and procedure B uses a laser. The data to be collected are the degrees of improvement in vision after the procedure is performed. There are 80 nearsighted people available for the experiment.
a. What is the explanatory variable?
Surgical procedure
b. What are the treatments?
Procedure A with a scalpel and Procedure B with a laser
c. What is the response variable?
Degrees of improvement in vision of the subjects
d. What are the individuals/subjects?
80 nearsighted people
e. Do you think you should use a placebo group? How about a control group?
No placebo should be used. No control group should be used either.
f. Could you make this a single-blind study?
No. Subjects would be aware of which treatment they would be getting.
g. Could you make this a double-blind study?
No.
h. Design a randomized comparative experiment:
i. There are two treatments. Why is it NOT ok to do a matched pairs design for this experiment?
We would not want to do separate procedures for each person’s eye. It would be an unnecessary risk.
16) A researcher wants to see if more expensive mattresses really give a better night’s sleep than the discount mattress brand. So they recruit 110 adult volunteers to participate in their study. They will have the adults sleep on the mattresses for 10 nights and then rank their overall quality of sleep (due to the mattress only).
a. What is the explanatory variable?
Mattress
b. What are the treatments?
Expensive Mattress and Discount Mattress
c. What is the response variable?
Quality of sleep
d. What are the individuals/subjects?
110 adult volunteers
e. Could there be any blinding in this study? If so explain in the context of the problem.
Yes. The subjects could be unaware of which mattress they are sleeping on.
f. Design a matched pairs experiment:
17) A new drug was tested to see if it could help decrease the appearance of wrinkles on a person’s face. The study was a randomized double-blind comparative experiment that used a placebo on a group of 80 women over the age of 60. Throughout the experiment the women’s appearances were rated by dermatologist.
a. Explain what a placebo is in the context of this experiment.
The placebo would be pill that does nothing like a sugar pill.
b. Explain why the researchers wanted to use a placebo in the context of this experiment.
The researchers wanted each woman to take some treatment so that the placebo effect would be minimized. This way we can see if there are improvements from the drug and not just from having any treatment.
c. Explain what is meant by “double-blind” in the design of this experiment.
Neither the women nor the dermatologists know what treatment the women are receiving.
d. Explain why it is important to make this experiment double-blind.
If the dermatologists don’t know what treatment is being used they would not be biased in evaluating the women’s appearances.
Multiple Choice Answers:
1. D
2. B
3. C
11. B
17. C
18. D
19. B
20. A
21. B
22. D
35. B
52. C
55. C
75. B
97. E
103. A
24. D