PSYCH 202 Lecture Notes - Lecture 4: Influenza Vaccine, Statistical Hypothesis Testing, Biogenic Amine
Methods of Psychological Research:
Inference and Knowledge in Society and Science
Caton Roberts, Ph.D
Correlation Coefficients Summarize the Data Points in
Scatterplots
mathematical representation of relationship between
two or more variables.
Can Correlations Support Causal Statements?
Why or why not? Discuss
No, b/c of variable you haven’t thought of (third) and the directionality problem
Consider a theoretical proposition: Biochemistry
Causes Depression
The monoamine deficiency theory of depression states:
Norepinephrine (NE) and serotonin (S-HT) are the cause of clinical depression.
Do these correlational data support the theory?
Discuss in Groups:
The Directionality Problem
Not sure which is independent or dependent
The Third Variable Problem
Could be an unknown third variable that causes
Thinking through the metabolite/depression scatterplot:
Directionality Problem:
Direction 1 (favored by the drug companies): the way it’s
presented makes people want to get drugs to fix
Direction 2: Depression causes changes in chemicals
Third Variable Problem: Predisposed to depression, drug use,
stress, family conflict, exercise
Might there be any unobserved 3rd variables here,
confounding variables from the view of the causal
hypothesis?
Suggestions?
Remember the “Do Hospital’s Make People Die” example
from the tutorial?
Here, what if the apparent causation (low levels cause
depression; depression causes low levels) is better
explained by a 3rd variable like “level of bodily
movement,” where low movement causes BOTH
depression and levels of biogenic amine metabolites?
Observed correlations are said to be spurious (seems to
be casual but it’s not) when compelling confounding 3rd
variables can be identified.
How can descriptive correlational data help scientific knowledge
development?
Can help begin
Reality: Most knowledge begins with noticing correlational
associations (David Hume)
Imperial philosopher, correlation can mislead us
Multivariate correlational research: Take many variables, put them in
equations
Theory development, hypothesis testing, and experimentation:
Document Summary
Scatterplots mathematical representation of relationship between two or more variables. No, b/c of variable you haven"t thought of (third) and the directionality problem. Norepinephrine (ne) and serotonin (s-ht) are the cause of clinical depression. Could be an unknown third variable that causes. Direction 1 (favored by the drug companies): the way it"s presented makes people want to get drugs to fix. Third variable problem: predisposed to depression, drug use, stress, family conflict, exercise. Observed correlations are said to be spurious (seems to be casual but it"s not) when compelling confounding 3rd variables can be identified. Reality: most knowledge begins with noticing correlational associations (david hume) Multivariate correlational research: take many variables, put them in equations. Independent variables (iv"s): the thing you control, you change it. Therapy as treatments for major depression (adapted from rush et al. , 1977) The sample: looking for people in newspaper, most depressed didn"t try. Results: 60% on anti and cognitive were not depressed.