Independent variable = dosage of the drug “Elate”

How to Design an
Experiment
MODULE 4
EXPERIMENTAL PSYCHOLOGY
GUIDED-INQUIRY LEARNING
Module 4: How to Design an Experiment
©2012, Dr. A. Geliebter & Dr. B. Rumain, Touro College & University System
Motivation
How does one set up an experiment?
Let’s say you wanted to see whether a new
drug being developed, called “Elate,” helps
patients who are depressed.
How would you set up the experiment?
Vocabulary
• Subject(s)
• independent
variable
• dependent
variable
• outcome
measure
• control variable
• subject variable
(participant
variable)
• Experimental
Group
• self-fulfilling
prophecy
• Control Group
• Independent
groups design
• blind
• double blind
• placebo
• between-subjects
design
• placebo effect
• Random
selection
• level(s) of an
independent
variable
• within-subjects
design
• random
assignment
• Latin Square
Information
The individuals participating in your experiment
are called Subjects. In order to do this experiment,
it would make sense to give some of the patients
the new drug “Elate,” whereas others would not
get it. Then we would see whether those that
received Elate showed an improvement in
depression over those that didn’t.
Subjects –
2 Groups
Yes Elate
No Elate
Information
The patients who get the new treatment are called
the Experimental Group. The ones who don’t get
the new treatment are called the Control Group
because they serve as a control against which to
measure whether the drug is actually working.
So, we can diagram it as follows:
Subjects –
2 Groups
Yes Elate =
Experimental
Group
No Elate =
Control Group
Thought Questions
1) The subjects that don’t get the Elate, the
Control Group, would you give them anything
or just nothing at all? What do you think?
You can’t give them nothing. You have to give them
a sugar-coated pill called a placebo.
2) What’s the need for a placebo?
Try to see if you can figure this out.
People get a certain psychological boost out of
knowing they are being helped. Were you ever
feeling very sick, went to the doctor, and then the
doctor told you that what you have is nothing and
that in a few days you’ll feel as good as new?
After hearing such a pronouncement, we often feel
better immediately, just because we’ve heard the
doctor say it’s nothing. The improvement people
show because they think they are being helped is
what we want to use as our baseline.
3) In an experiment, why don’t we tell the people
that are going to get a placebo that they are going
to get a placebo?
We don’t tell them because that would take away
the psychological boost they feel from knowing
they are being helped. This psychological boost
provides measurable effects in terms of their
physical well-being. We can picture it this way:
C = Elate given
Increasing
feelings of
good
health
C-B = Effect of Medication
B = Placebo given
B-A = Placebo Effect
A = No meds given = Baseline
Psychological boost due to
knowing you are being helped
C = Elate given
Increasing
feelings of
good
health
C-B = Effect of Medication
B = Placebo given
B-A = Placebo Effect
A = No meds given = Baseline
Psychological boost due to
knowing you are being helped
B - A is the gain in physical well being due to the placebo effect, i.e.,
knowing that you are being helped.
C - B is the additional gain in well-being due to the actual medication,
Elate.
If the difference C - B is significant, then Elate is effective in lifting
depression.
If C - B is small and insignificant, then patients taking Elate don't
experience a significant improvement over those taking a placebo.
So, so far we have the Experimental Group receiving Elate and the Control
Group receiving a placebo.
4) Is it possible to have more than one
Experimental Group and one Control Group?
What do you think and Why?
Yes, you might have Experimental Group I
receiving one dosage of Elate (let’s say 750 mg per
day), and Experimental Group ll receiving a
different dosage of Elate (say 1200 mg per day).
You could then compare the two groups in terms
of improvement and in terms of side effects.
2
Experimental
Groups
Experimental
Group I
= 750 mg
Experimental
Group II
= 1200 mg
You might also have two control groups, one
receiving the placebo and another one receiving a
different antidepressant medication, for example,
Zoloft. You could then compare the improvement
with each dosage of Elate to improvement with
Zoloft and determine whether one antidepressant
worked better than the other.
2 Control
Groups
Control Group I
Placebo = 0 mg
Control Group II
Zoloft
Independent and Dependent
Variables
Every experiment has at least one
independent variable and one
dependent variable.
The independent variable is the variable
you are manipulating. All other
variables we will hold constant. That we
hold all other variables constant is
called Experimental Control.
The variable we are interesting in
measuring, the outcome measure, is
called the dependent variable.
Independent and Dependent
Variables
Independent
Variable
Dependent
Variable
what we
change
what we
measure
variable we are
manipulating
outcome measure
All other variables
are held constant
Independent and Dependent
Variables
Independent Variable
= what we change =
Elate Dosage
Dependent Variable
= what we measure=
Level of Depression
In the experiment above, the independent variable is the
dosage of Elate. The dependent variable is how
depressed the subjects are.
In our example below, some subjects will get 750
mg/day of Elate. These will be Experimental Group I.
Others will get 1200mg/day of Elate. These will be
Experimental Group II.
The third group, the Control Group, will get no Elate
(0mg/day). Instead, they will get a sugar-coated pill. We
will then measure how depressed each group is.
Levels of an Independent Variable
We say there are 3 levels of the independent
variable corresponding to the three dosages. We
can diagram it as follows:
Independent variable = dosage of the drug “Elate”
If there are 3 possibilities, then there are 3 levels of
the independent variable corresponding to the 3
dosages
3 levels of the
independent
variable
(Dosage)
Experimental
Group I
Experimental
Group II
Control Group
750
mg
1200
mg
0
mg
Levels of an Independent Variable
If there are 2 possibilities, Experimental Group I
with 750 mg, and the control group (with 0 mg),
then we would say there are 2 levels of the
independent variable (corresponding to the two
dosages).
Experimental Group I
750
mg
Control Group
0
mg
2 levels of the
independent variable
(Dosage)
Levels of an Independent Variable
How do we decide how to measure
the dependent variable?
We have to decide how we are going to
measure depression.
Another way of saying this is that we have to
decide how we are going to operationalize
depression.
By “operationalize” we mean coming up with a
standard way to measure depression, so we all
have a uniform measurement tool. One reason
for this is to enable others to replicate the
study using the same measurement tool.
Thought Questions
5) How would you operationalize depression here
in this study? Remember, operationalize means to
standardize the meaning of a measure or
procedure so the study can be replicated.
One possibility is that you could use a Depression
Inventory, like the Beck Depression Inventory. The
Beck Depression Inventory is a questionnaire in
which people answer questions rating themselves
on how depressed they feel. In this case, the
dependent variable is depression as
operationalized by scores on the Depression
Inventory.
Alternatively, you could have a clinician
(psychiatrist, psychologist, or psychiatric social
worker) conduct a structured clinical interview with
each patient and make an assessment of his/her
degree of depression. In this case, depression is
operationalized by a rating given at the end of a
diagnostic interview.
You could also do both things, namely use a score
on a Depression Inventory and a rating after a
diagnostic interview. If you used both measures,
then you’d have two dependent variables or
dependent measures.
Beck Depression Inventory (BDI)
• Developed by Beck, Steer & Brown
• Assesses the intensity of depressive
symptoms
• 5-10 minutes to administer
• Highly reliable regardless of the
population tested
• 21 multiple-choice questions
Beck Depression Inventory (BDI)
Sample questions:
1.
0 - I do not feel sad.
1 - I feel sad
2 - I am sad all the time and I can't snap out of it.
3 - I am so sad and unhappy that I can't stand it.
2.
0 - I am not particularly discouraged about the future.
1 - I feel discouraged about the future.
2 - I feel I have nothing to look forward to.
3 - I feel the future is hopeless and that things cannot improve.
3.
0 - I do not feel like a failure.
1 - I feel I have failed more than the average person.
2 - As I look back on my life, all I can see is a lot of failures.
3 - I feel I am a complete failure as a person.
Beck Depression Inventory (BDI)
The remaining questions can be found at:
http://legacy.touro.edu/faculty/psychology/nsf/NewModules/
Module4_Beck_Depression_Inventory.pdf
Print out the Beck Depression Inventory and complete it.
Score your answers, and see where your total score
translates in terms of levels of depression.
Please do not share your total score or depression level with
your classmates.
Beck Depression Inventory (BDI)
• Scoring: Each answer is scored on a scale of 0 to 3 and
then all scores are added together to produce a total
score. The total score is an indicator of an individual’s
level of depression.
Total Score
Level of Depression
1-10
These ups and downs are
considered normal
11-16
Mild mood disturbance
17-20
Borderline clinical
depression
21-30
Moderate depression
31-40
Severe depression
Over 40
Extreme depression
Beck Depression
Inventory
- Patient self-rating
Could use both
measures =
2 dependent
variables
Clinical Interview
- Clinician rating
Thought Question
6) If you're going to have a clinician score
each patient on his/her level of depression,
can that clinician know what medication
that patient is receiving? Why?
No, the clinician can’t know because that
might influence his rating. For example, if a
psychiatrist knows a patient is receiving an
antidepressant, he might be biased to rate
him as less depressed than a patient
receiving a placebo.
How do we decide how to measure
the dependent variable?
A study where the patient doesn’t know
whether he’s receiving a placebo or a true
medicine is called blind.
One in which the patient and the clinician
doing the ratings don’t know which patient
is receiving what is called double blind.
Thought Question
7) How do you get people to participate in a
study where they might be getting a
placebo rather than true medication? Do
you lie to them?
No. Before an individual participates in a
study, s/he must sign an Informed Consent
Form.
8) What would you tell them?
You must say something like the following:
This is an experimental study to assess the effects of a
certain medication. Some of you will receive the
medication and some of you will receive a placebo, a
sugar-coated pill. We cannot tell you who will get what.
We do not even know whether the medication works or
what its side effects are. Your participation in this study
will help us determine whether the medication is
helpful. You are free to withdraw from the study at any
time.
If the study is being conducted at a hospital/clinic and
the patient is receiving services/treatment from the
facility, you should also state in the Informed Consent:
“If you decide to withdraw, this will not affect your
treatment or your relationship with your doctor in any
way.”
Thought Question
9) Can you have more than one independent
variable and more than one dependent variable
and how would you do this?
A study where there is more than one independent
variable and/or more than one dependent variable
is called multivariate.
The study above with two dependent variables is a
multivariate study.
Thought Questions
10) To review, what were the 2 dependent variables
in the study we were just talking about?
One dependent measure was depression as
measured by a score on the Beck Depression
Inventory.
The second dependent measure (or variable) was
depression as measured by a rating given at the
end of a diagnostic interview with a psychiatrist or
psychologist.
Multiple Independent Variables
Let’s consider now having more than one
independent variable. If one independent variable
is dosage, another might be, let’s say, amount of
exercise if you thought that exercise also had an
effect on a person’s feeling depressed.
If you also wanted to look at the effect of exercise
in addition to the effect of medication, then you’d
have 2 independent variables: (1) dosage of
medication, and (2) exercise.
Then for the “exercise” variable, you might have 2
levels of it: One level could be a “yes exercise”
condition and the other level could be a “no
exercise” condition.
2 independent variables 
Multiple Independent Variables
2 independent variables
Independent variable #1:
Dosage
Independent variable #2:
Exercise
3 levels:
750mg (experimental group #1)
1200mg (experimental group #2)
0mg (control group)
2 levels:
“Yes exercise”
“No exercise”
So, based on the above, how many groups of subjects are there?
Multiple Independent Variables
In all, there are 6 groups of subjects in this experiment.
How do we know this?
Multiple Independent Variables
We multiply the 3 levels of dosage by the 2 levels of exercise,
to get the 6 groups of subjects, or 6 conditions.
We can also see this by constructing a Latin Square, as
follows:
Independent Variable #1: Dosage
Independent
Variable #2:
Exercise
750mg
1200mg
0mg
Yes
No
Fill in what goes into each of the 6 boxes (the 6 conditions).
Multiple Independent Variables
If your Latin Square looks like the one below, then you are
correct.
Independent Variable #1: Dosage
Independent
Variable #2:
Exercise
750mg
1200mg
0mg
Yes
Group 1:
Group 2:
Group 3:
750mg and
1200mg and 0mg and
Yes Exercise Yes Exercise Yes Exercise
No
Group 4:
750mg and
No Exercise
Group 5:
1200mg and
No Exercise
Group 6:
0mg and
No Exercise