Module 6: Correlation and Experimentation
Correlation
● Correlation: a measure of the extent to which 2 variables change together
● Direction
○ Positive correlation = 2 variables act the same (↑↑ / ↓↓)
○ Negative correlation = 2 variables act differently (↑↓ / ↓↑)
● Strength
○ Strong correlation = more predictive variables
■ E.g. gal of gas pumped in increases as the $ spent on gas increases
○ Weak correlation = less predictive variables
■ E.g. car uses more gas as car color gets darker
● Correlation coefficient: a statistical measure of the relationship between 2 variables
○ Ranges from -1.0 to +1.0
○ Closer to ± 1 = stronger relationship
○ Sign depicts direction/slope
○ Number/value depicts strength
● Scatterplots: a graphed cluster of dots
○ Slope depicts direction
○ Scatter depicts strength of relationship
● Correlation ≠ causation
○ Correlation indicates the possibility of a cause-effect relationship but does not prove it
○ E.g. both ice cream sales and murder rates increase during summer, but ice cream does not
cause murder
● Illusory correlation: a perceived but nonexistent correlation
○ When we believe there’s a relationship, we are likely to notice/recall instances that confirm our
belief
○ When we notice random coincidences, we may forget that they are random and instead see
them as correlated. Thus, we can easily deceive ourselves by seeing what’s not there
Experimentation
● Samples should be representative of the population so that researchers can generalize and apply their
findings to the population
○ Representative sample: a sample of a larger group that represents it
○ Stratified sample: population is divided into subcategories and a random sample is taken from
each subcategory
● Researchers randomly assign people to a group to minimize any preexisting differences between them
○ Experimental group gets the “special treatment”
○ Control group doesn’t get the “special treatment”
● Variables
○ Independent variable: factor manipulated by experimenter
○ Dependent variable: factor that may change in response to IV; data
○ Both groups get the IV; the difference is the “degrees” of the IV
● Placebo: fake treatment
○ Placebo effect: when someone’s mental/physical health appears to improve after taking a fake
treatment
● Addressing biases
○ Single-blind procedure: when participants don’t know what group they are
○ Double-blind procedure: when neither participants nor experimenters know which group is which
● Third variable (extraneous variable)
○ 2 variables can be related but not bc they cause each other
○ Confounding variable: other items that could impact the data
● Validity is the key goal of experimental design
○ Construct validity: do the variables represent/measure what they’re supposed to?
○ Internal validity: how well constructed was the experiment to control for confounding variables?
○ Predictive validity: how well do the variables measured predict other measures of the same
construct?
○ External validity: how well do the results of the experiment translate to other
settings/participants?
Types of Research Methods
● Descriptive
○ Observe and record behavior
○ Case studies, naturalistic observations, surveys
● Correlational
○ Detect naturally occurring relationships; assess how well 1 variable predicts another
○ Collect data on 2+ variables; no manipulation
● Experimental
○ Explore cause and effect
○ Manipulate 1+ variables; use random assignment
