Values range from -1 to +1

Measures linear relationships only

r² is the proportion of variance accounted for (i.e., how well X predicts Y)
 

Restriction of Range

Reduces the magnitude of r

Occurs when participants' responses do not cover the full range of one variable
    - Measuring IQ with UM students
    - Measuring weight with sumo wrestlers
    - Measuring income with movie stars
 

Mediating Variables

Variable occurring between two correlated variables

Explains how X affects Y
    * Dissonance => Arousal => Attitude Change
    * Failure => Low Self-Esteem => Prejudice
    * Proximity => Increased Contact => Attraction
 

Can Correlation Test Causality?

Can determine if one causal explanation is more plausible than others.

How?
    - Intuition
    - Longitudinal Design
    - Controlling for 3rd variables
 

Back to Overheads Page
Back to 381-002 Main Page