1. Understanding Experiments Lecture 10 Section 3.5 Mon, Sep 19, 2005

2. Examples Recall that in an experiment, the researchers control the levels of the explanatory variables. Recall that in an experiment, the researchers control the levels of the explanatory variables. Example 3.9, p. 168 – Producing Hard Gears. Example 3.9, p. 168 – Producing Hard Gears. What are the explanatory variables? What are the explanatory variables? What are the levels for each variable? What are the levels for each variable? What are the treatments? What are the treatments? How many units are there? How many units are there?

3. Design-Layout Table for 3 Factors Example 3.10, p. 170 – Three Explanatory Variables and How Many Units? Example 3.10, p. 170 – Three Explanatory Variables and How Many Units?

4. Example Think About It, p. 174. Think About It, p. 174. Read the example at the bottom of the preceding page. Read the example at the bottom of the preceding page. What is wrong with this experiment? What is wrong with this experiment? Suppose the researchers decide to divide the 100 subjects into two groups. Suppose the researchers decide to divide the 100 subjects into two groups. Group 1 receives Vitamin C. Group 1 receives Vitamin C. Group 2 does not receive Vitamin C. Group 2 does not receive Vitamin C. How should they be divided? How should they be divided?

5. Random Allocation The assignment to the groups should be random (a randomized design). The assignment to the groups should be random (a randomized design). The subjects could be numbered 1 – 100. The subjects could be numbered 1 – 100. Then use a random number generator to obtain 50 (distinct) random numbers from 1 – 100. Then use a random number generator to obtain 50 (distinct) random numbers from 1 – 100. Those 50 subjects would be assigned Group 1. Those 50 subjects would be assigned Group 1. The rest would be assigned to Group 2. The rest would be assigned to Group 2.

6. Treatment and Control Groups Treatment group – A group that receives one of the actual treatments. Treatment group – A group that receives one of the actual treatments. Control group – A group that receives no treatment, but is otherwise identical to the treatment group. Control group – A group that receives no treatment, but is otherwise identical to the treatment group.

7. Response Bias Are the subjects in the control group aware that they are not receiving Vitamin C? Are the subjects in the control group aware that they are not receiving Vitamin C? Are they aware of the purpose of the experiment? Are they aware of the purpose of the experiment? If so, will they be more likely to report cold symptoms? If so, will they be more likely to report cold symptoms? Will the treatment group be less likely to report symptoms? Will the treatment group be less likely to report symptoms? What can we do about this? What can we do about this?

8. Response Bias Response bias – The tendency of the subject to show the response that he believes the researcher is expecting. Response bias – The tendency of the subject to show the response that he believes the researcher is expecting. In this context, this type of response bias is called the placebo effect. In this context, this type of response bias is called the placebo effect.

9. Placebos Placebo – Something that looks like the real drug or treatment, but in fact has no effect. Placebo – Something that looks like the real drug or treatment, but in fact has no effect. To help eliminate the placebo effect (and other types of response bias), the researchers could give a similar tablet (a placebo), but containing no Vitamin C, to the subjects in the control group. To help eliminate the placebo effect (and other types of response bias), the researchers could give a similar tablet (a placebo), but containing no Vitamin C, to the subjects in the control group.

10. Blinded Experiments Of course, the subjects should not know whether they are receiving the placebo or the genuine treatment. Of course, the subjects should not know whether they are receiving the placebo or the genuine treatment. This is called a blinded experiment. This is called a blinded experiment. Single-blind experiment – The subjects do not know which group they are in (but the researchers do). Single-blind experiment – The subjects do not know which group they are in (but the researchers do).

11. Experimenter Bias Experimenter bias – A distortion in the observations resulting from the experimenter’s expectations. Experimenter bias – A distortion in the observations resulting from the experimenter’s expectations. This is similar to response bias except that it is caused by the experimenter, not the subject. This is similar to response bias except that it is caused by the experimenter, not the subject. How can we control for experimenter bias? How can we control for experimenter bias?

12. Double-Blind Experiments Double-blind experiment – Neither the subjects nor the researchers know which group is the control group. Double-blind experiment – Neither the subjects nor the researchers know which group is the control group.

13. Let’s Do It! Let’s do it! 3.7, p. 155 – Study Suggests Light to Back of Knees Alters Master Biological Clock. Let’s do it! 3.7, p. 155 – Study Suggests Light to Back of Knees Alters Master Biological Clock. Was there a control group? Was there a control group? Was this a blinded study? Was this a blinded study? Could there be a placebo effect? Could there be a placebo effect?

14. Let’s Do It! Let’s do it! 3.8, p. 181 – Controlling for Another Factor. Let’s do it! 3.8, p. 181 – Controlling for Another Factor.

15. Let’s Do It! 3.8 The design is The design is 5 with 1 attack 4 with > 1 attack Group 1 (Treatment) 5 with 1 attack 4 with > 1 attack. Group 2 (No treatment)

16. Let’s Do It! 3.8 The design should be The design should be 5 subjects 4 subjects One attack More than one attack TreatmentNo treatment

17. Block Design A block is a homogeneous group from the population. Essentially a block is a stratum. A block is a homogeneous group from the population. Essentially a block is a stratum. A block design is an experiment design in which the response variable is measured separately for each block. A block design is an experiment design in which the response variable is measured separately for each block.