Transcript The Experimental Approach

-Correlation does not equal causation
-necessary but not sufficient
-Where there’s smoke there may be fire
Experimental approach attempts to
create conditions where third
variable effects and issues of
directionality can be eliminated.
But what conditions would define a
causal relationship?
Mills’ Canons of Causality
conditions that must be met if a
relationship is causal
• Where “A” is presumed cause and
“B” is presumed effect.
• Method of agreement
• Method of disagreement
• Method of concomitant variation
• Method of multiple causality
Method of Agreement
• If “A” is present then so should be “B”
Method of Disagreement
• If “A” is not present then neither should be
“B.”
Method of concomitant Variation
• If “A” is varied in intensity then “B” should
show corresponding variation
How do we set up experimental
conditions to test Mills assumptions
The Experimental approach
• Systematic manipulation (variation) of the
presumed causal variable
• Systematic measurement of the presumed
affected variable
• Under controlled conditions
Terminology
•
The presumed causal variable = the
independent variable (IV)
– The manipulated variable
– The administered variable
The presumed affected variable = the
dependent variable (DV)
– The measured variable
Identifying variables by “Titles.”
• “The effects of Stress on attention.”
• The Effects of Alcohol on social
interaction.”
• Marijuana increases attention to drug
cues.”
• Cognitive processing is diminished by
nicotine.”
• The relationship of reported drug use to
political affiliation.”
Identifying Correlational studies vs
Experimental studies. There should
be clear inference of cause or effect
in Experimental studies
•
•
•
•
The effects of peanut butter on…
Peanut butter increases…
Brain size is increased…
Typing speed is related to….
Manipulation of IV to test Mills
assumptions
• Minimally, the IV must have two Levels or
conditions
• Present/absent
or
High vs low
• IV may have more than two levels– Concomitant variation
The Experimental and Control
Conditions
• The control condition is both a point of
comparison and
• Helps control for possible unwanted
effects of extraneous (confounding)
variables.
BASIC EXPERIMENTAL DESIGNS
• Between group designs
• Within group designs
Between vs Within
• Both use groups of subjects
• Between groups test different groups at
same time. Each group exposed to different
level of IV
• Within groups use same subjects tested
repeatedly across different conditions of IV.
Design decision making
Fewer subjects
More time
The Experimental approach
• Systematic manipulation (variation) of the
presumed causal variable
• Systematic measurement of the presumed
affected variable
• Under controlled conditions
Extraneous variables
Extraneous variables may become
confounding variables
Confounding variables: An extraneous
variable that varies with the IV.
Confounding variables
• “ a variable that should be extraneous, but
becomes systematically varied along with the IV.
• Eg…”Effects of marijuana on memory.”
• MJ accidentally administered only to
experienced users but not to naïve users.
Experimental Control
• Procedures to minimize CVs
• By elimination, constancy or specific
testing of suspected confounds
• The Control group or condition
• = point of comparison and..
• A fundamental control procedure