Transcript document

Research Design and
Behavioral Analysis
1
Research methods
• Correlation Research – used to determine
if there is a relationship between two
variables
– Least amount of experimental control
– Cannot be used to state cause and effect
relationships
2
Research methods (cont.)
• Experiment – provides most confidence in
determining cause and effect
relationships. Manipulate the independent
variable to see if changes occur in the
dependent variable (measure)
• Amphetamine study as an example
3
Research methods (cont)
• Quasi-experiment – similar to an
experiment except the independent
variable (IV) cannot be manipulated
– IV is usually a subject variable
– Subjects are assigned to groups for particular
reason (gender, age, ethnic group, etc.)
– Lack of random assignment reduces
confidence in cause and effect statement
– Modification of Amphetamine study as an
example
4
Research methods (cont.)
• The experimental design preferred in
psychopharmacological research – most control
and confidence in cause and effect
• Measures are usually behavioral changes that
the drug was intended to cause and physical
changes (side effects)
• Side effects not always bad – many drugs
designed for one purpose, but later found to be
useful for another because of unexpected
actions
5
Experimental designs
• All experiments begin with a control group
and one or more experimental groups. The
two principle experimental designs differ in
how subjects are assigned to the groups
6
Within-subjects design
• Only one group of subjects involved
• All subjects receive all levels of treatment. They act as
their own control and experimental group(s). Test before
giving the drug – baseline, then give the drug and
measure any changes
• Advantages
– Need fewer subjects
– You know groups are equal because they are the same
subjects
• Disadvantages
– Carry over effects – measuring independent variable to get
baseline may change subjects in some way
– Many times it requires more time – delays between
treatments
7
Between-subjects design
• Different subjects are randomly assigned to
different groups. Groups are assumed to be
equal because each subject had an equal
chance of being assigned to any group
• Advantages
– No carry-over effects
– Quick
• Disadvantages
– More variability – groups may not be equal by
chance
– More subjects required and you may not have
8
Statistical analysis
• Inferential statistics – what is the probability that
the difference between groups is the result of
chance variations?
• Within subjects design – What is the probability
that the difference in the baseline measure and
the experimental trials was the result of chance?
• Between groups – What is the probability that
the difference between the control group and
experimental group(s) was due to chance?
• Convention says to use probability of 1 or 5 out
of 100 (P<.01 or P<.05)
• With drug research using a P < .01 is preferred
9
Validity
• Internal validity – How valid is our
measurement? Are we measuring what
we say we are measuring?
• External validity – How confident are we
that we can generalize our results from our
sample to a whole population? Is our
sample truly representative of our
population? Could it be biased?
10
Important issues with validity
• 1. Within groups variability – the average of the
subjects in the experimental group performed
differently than the average of the subjects in the
control group, but perhaps a sub-group of the
subjects did not show a difference in
performance.
• Example: older subjects were very different, but
younger subjects showed no change. Your
study showed it worked for all subjects when it
only affected older subjects
11
Important issues with validity
(cont.)
• 2. Use of a drug with a population not
included in any of the samples used in the
experiment.
• Examples:
– 1. women excluded from testing hypertension
drugs because of fluctuating hormones.
These drugs used to treat hypertension in
women.
– 2. Prozac used to treat children when only
tested on adults
12
Important issues with validity
(cont.)
• Placebo – a pill or injection that has no
medicinal value, but causes changes because of
peoples expectations
• Placebo effect – if people given something that
is suppose to help them, it may have an effect
when in reality the drug had no effect. Has to be
controlled by having the control group receive a
placebo while experiment group receives
experimental drug
13
Important issues with validity
(cont.)
• Hawthorne Effect – people will sometimes
improve because they get attention and
feel that others are trying to help even if
the drug they are given has no effect itself
14
Experimenter bias
• A Person testing their drug has a strong desire
for drug testing to be successful
• They know what the desired effect is suppose to
be
• They may “see” the desired effect in the
experimental group, and not see it in the control
receiving the placebo because they know which
group is which
15
Solution to experimenter bias
• Use of multiple naïve observers who have no
expectations about the which subjects should
show a change
• Use of double-blind techniques. Subjects are
unaware if they are receiving the placebo or test
drug (single-blind), and the experimenter is
unaware of who was assigned to get the
placebo and who is receiving the test drug.
16
Three group design
• Most frequently used drug testing design
– 1. experimental groups receives test drug
– 2. control group 1 receives placebo
– 3. control group 2 receives drug with known effects
• Allows 3 different comparisons:
– 1. placebo to established drug – was test sensitive
enough to detect an effect
– 2. placebo to experimental drug – was there an
effect
– 3. established drug and experimental drug – is new
drug more effect or have fewer side effects
17
Dependent measures in
psychopharmacological research
• Changes in observed behavior – decrease in
aggressive behaviors, decreased eating, etc.
• Changes in human performance – cognitive and
motor tests
• Self-report by subjects – pain medication, “on a
scale of 1 to 10 with 1 being very little and 10
being excruciating how do you rate your level of
pain?”
18
Dependent measures in
psychopharmacological research
• Electroencephalogram (EEG)
– Measure level of arousal of the cortex –small
fast waves indicate higher levels of arousal
than large slow waves
– Measure REM sleep – deprivation of REM
sleep has similar effect to total sleep
deprivation. A side effect of some drugs is a
disruption of REM sleep
19
Dependent measures in
psychopharmacological research
• Perceptual measures:
– Sensory difference thresholds – how different
do two stimuli have to before you are able to
detect a difference – Does a drug make you
more or less sensitive to these differences?
– Critical frequency of fusion – How fast does a
light have to flicker on and off before you see
it as a continuous light
20
Dependent measures in
psychopharmacological research
(cont.)
• Cognitive tests
• Motor tests – sobriety test
• Combination – driving tests
21
Measuring performance in
nonhumans
• Drug effects on the performance on
operant conditioning and different
schedules of reinforcement
• Avoidance – escape tasks – animals learn
to avoid an unpleasant stimuli or they
learn to escape an unpleasant stimuli they
can’t avoid
– Anxiety reducing drugs decrease avoidance
behavior but not escape behavior
22
Measuring performance in
nonhumans
• Use of punishment – anything that decreases
magnitude or frequency of a behavior. Does the
drug decrease the effectiveness of punishment.
Sedatives do amphetamines do not
• Drug discrimination paradigm – teach animal to
discriminate between the effects of a placebo
and the effects of a particular drug
– Used to determine if a new drug has a similar effect
as an older drug
– Also if the two drugs have the same effect on the
actions of neurotransmitters
23
Drug development and testing
• Major issue with psychoactive drugs:
– 1. we know most mental disorders involve
biochemical imbalances
– 2. we know that psychoactive drugs operate on the
effect of neurotransmitter systems
– 3. we do not know exactly what the specific
biochemical imbalance is in most disorders so it is
difficult to know what drug will work with what
disorder
– This results in extensive testing with nonhumans
• This results in extensive testing with nonhumans
24
Nonhuman testing
• Purpose:
– Identification of exactly which
neurotransmitter system or systems the drug
effects and identifies possible side effects
– Identification of effective versus toxic dosages
based upon body weight – looking for drugs
with the largest difference between the two
• Obviously interspecies differences a
problem
25
Human testing
• Purpose:
– Identify short-term and long-term effects in
humans
– Detailed analysis of aversive effects – never
ending unless drug is discontinued
26