Correlational Research

correlation

relationship between two variables of interest

• degree/magnitude and direction

• measured, but not manipulated

correlation scatterplots

graph displaying relationship between two quantitative variables

• closer data points mirror a circle, less of a relationship

correlation coefficient (Pearson’s r)

numerical value describing the direction and strength of a linear relationship

• ranges from -1 to 1; 0 = no relationship

truncated range

occurs when one variable has little variance, artificially weakening a correlation

• certain amount of spread/variability in scores needed

• ex: SAT score/college GPA at an Ivy League school might seem to make SAT scores weak predictors of GPA, since 90% of average students not reflected

curvilinear/non-linear relationships

correlation coefficients cannot accurately reflect non-linear relationships, sometimes masking real relationships

spurious correlations

correlation that appears meaningful but is actually produced by another variable or chance

• ex: ice cream sales and shark attacks; summertime drives both

correlation advantages

• allows us to check relationships for things that cannot be manipulated or randomly assigned

• can identify potential causes, leading to later experiments

• high ecological validity

correlation disadvantages

• causal directionality issue

• reverse causality

• 3rd variables/confounds

causal directionality issue

two variables may be related, but cannot determine which caused the other

revserse causality issue

correlations cannot establish which variable is the cause and which is the outcome

• can be addressed with longitudinal designs

longitudinal design

measures same variables/people across multiple time points, helping examine temporal order