Transcript Lecture 4 Research D.. - HomePage Server for UT Psychology

Research Design
Operationalization
Psychological Constructs:
• Love
• Shyness
• Hunger
• Intelligence
• Anxiety
• Generosity
Operationalization
Operational Definitions:
• Amount of food consumed in one sitting
• Number of cigarettes one smokes in an hour
• Number of reasoning problems one can solve in 30 minutes
• Amount of money someone will spend for a hamburger
• Number of conversations initiated with a stranger at a party
• Percentage of income donated to charity
• Blood pressure, heart rate, respiratory rate, and sweating
• Number of friends one claims to have
• Agreeing with the statement “I would rather suffer than let
my partner suffer”
• Grade received in a psychology course
• Number of times someone is thought of during the day
Research Design
• What are the three types of research
design discussed in the book?
• Imagine that we are testing a new weight
loss drug. How would we use the three
different types of studies to investigate
weight loss and test its effectiveness?
Descriptive Studies
• Describing behavior w/out investigating
relationships between variables
– Can still be quantitative
– Usually used to “get a feel for” a subject
matter or for unusual circumstances
• Weight loss?
Correlational Studies
• Observe & measure two or more variables
to determine how well one predicts the
other
– No variable manipulation
– Cannot prove causation. If X & Y are related:
• X  Y? or Y  X?
• Third variable? Z causes X and Y?
• Weight loss?
Experiments
– Can show cause and effect relationships
between two variables.
• Independent variable
• Dependent variable
– Systematically manipulate the IV to study the
changes in the DV.
– Often done in laboratory – not always
generalizable
– Weight loss?
Elements of an experiment
•
•
•
•
•
•
•
•
•
Independent variable
Dependent variable
Extraneous variables
Operational definition
Experimental group
Control group
Placebo group
Random assignment
Single- or double-blind procedure?
Why correlation, then?
• Correlation ≠ Causation.
• Psychological studies aren’t always
amenable to experiments for practical or
ethical reasons.
– Example: Children whose parents have a lot
of books in the house are better readers and
do better in school.
• Generalizability
Mark Twain: “There are three
kinds of lies: lies, damned lies,
and statistics”
In this class, we will deal with the
latter of these three.
Do you agree?
• Men are taller than women.
• All men are taller than all women.
• Can you agree with the first statement
without agreeing with the second??
Descriptive statistics
• Used to summarize sets of data
– Make the data accessible
• Which is more useful in describing class
scores on an exam?
A. 87, 65, 90, 94, 65, 73, 81, 54, 84, 75
B. Average: 76.8
Descriptive statistics
• Correlation:
– correlation coefficients
• Measures of central tendency:
– Mode
– Median
– Mean
• Standard deviation (more variable scores = ↑ SD)
Pretend study
• Taking “Intelligencio”, a new memory
drug, will increase the chances of doing
well in PSY 301 class.
• 1 class, 75 students.
– A. 25 took intelligencio, Mean grade: 85
– B. 25 took placebo, Mean grade: 80
– C. 25 took nothing, Mean grade 79
Accidental?
• Remember: be skeptical
• What other factors could account for our
findings?
• Are they representative of all psychology
students?
Inferential statistics
• How likely is it that we found a fluke?
• Inferential statistics help us decide this.
– Depends on:
• Size of the effect (large vs. small differences)
• Within-group variance
• Number of observations – helps individual
differences cancel each other out.
Scatterplot data demonstrate
correlations:
Visualizing Various Scatterplots:
Two problems with inferences of
causation from correlational data
• 1) the directionality problem
• 2) the third variable problem
Illustrating the directionality and
third variable problems
• Neurotransmitter metabolites of
Norepinephrine and Serotonin have a
moderate negative correlation with level of
clinical depression, but what causes what?
Is depression caused by low
levels of NE and 5-HT?
• The following scatterplot empirically
characterizes the moderate negative
correlation between depression and levels
of Norepinephrine (NE) and Serotonin (5HT) metabolites
• Question to frame the discussion: Do the
data support the “biogenic amine” theory
of depression?
The directionality problem in the
metabolite scatterplot:
• Direction 1 (Pharmaceutical-Friendly):
• Insufficient neurotransmitters cause
depression (the biogenic amine theory)
• Direction 2 (favored by anti-drug groups):
• Maybe depression causes biochemical
deficiencies instead of vice versa?
The third variable look at the
biogenic amine theory:
• Maybe a third variable (or set of
variables) cause(s) both biochemical
dysregulation and depression?
• What does the correlation tell us, then?
– The variables are associated and we
need to figure out why
What are the data collection
methods of psychological science?
• Observing is an unobtrusive strategy
• Asking takes a more active approach
• Case studies examine individual lives
• Response performance measures stimulus
•
•
•
processing
Body and brain activity can be directly
measured
Research with animals provides important
data
There are ethical issues to consider
Descriptive studies
• Descriptive studies, sometimes called
observational studies because of the
manner in which the data are typically
collected, involve observing and noting
behavior in order to provide a systematic
and objective analysis
Observing is an unobstrusive
strategy
• Primate behavior is often observed in
natural settings.
• Animals are more likely to act naturally
when they are in their native habitat,
relative to animals in captivity.
Asking takes a more
active approach
Case studies examine individual
lives
• For example, many psychologists think
that immature brain development, in
combination with other factors,
contributes to violent behavior.
• Lee Boyd Malvo and Kip Kinkel were both
convicted of murder in their teens.
Response performance measures
stimulus processing
• Researchers can measure:
– Reaction time
– Response accuracy
– Stimulus judgments
Body and brain activity can be
directly measured
Positron Emission Tomography (PET)
A PET scan of blood flow in a subject’s brain under
different conditions.
Magnetic Resonance Imaging
(MRI)
MRI provides very high quality images
of intact human brains.
Research with animals provides
important data
• Throughout the history of psychological
science, many important research findings
have been obtained by studying animals.
There are ethical issues to
consider
• IRB’s
• Informed consent
• Relative risks of participation
• Research with animals
How are data analyzed
and evaluated?
• Good research requires valid, reliable,
and accurate data
• Descriptive statistics provide a summary
• Correlations describe the relationships
between variables
• Inferential statistics predict chance
results
Good research requires valid,
reliable, and accurate data
• Validity: whether the data address the research
question unambiguously
• Reliability: stability and consistency of a measure
over time
• Accuracy: the extent to which the measure is
free from error
Descriptive statistics provide a
summary
• Central Tendence
– Mean
– Median
– Mode
• Variability
– Standard deviation
Correlations describe the
relationships between variables
- Correlation is a descriptive statistic that
gives a numerical value (between +1.0
and -1.0) indicating the strength of a
relationship between 2 variables.
- Negative correlation
- Positive correlation
Inferential statistics predict
chance results
- Researchers use inferential statistics to
decide whether differences actually exist
between different sets of numbers.
Types of Error
• Random
• Systematic