Statistical Terms
Interaction effect
This is an important concept that lets the researcher to establish whether or not “it depends”
Cell phone use while driving
•Effects of using a cell phone use limited to one age group only, or would it have the same effect on people of different ages?
•Independent variables
•Age of drivers, young and old
•Cell phone condition, on or off
•Dependent variables
•Accidents
•Following distance
•Breaking onset time
•Results: effect of talking on phone does not depend on age
The effect can be "generalized"
Goal of testing limits is related to external validity – can the effect be generalized, or applied across all participant or independent variables
Participant variables
Examples of participant variables: age, gender, ethnicity
Participant variable are not manipulated, just measured
Factorial designs
Study two independent variables
Behavior of a dog example..
•Depends
•Whether or not I say “sit”
•Dependent variable
•Probability dog will sit
•Independent variable
•2 independent variables: do I say “sit” or nothing
Differences in Differences, found by adding what?
Addition of an independent variable allows researchers to look for an interaction effect
Theories
Theories make statements about how variables interact with each other
Crossover interaction
Crossover interactions – “it depends” – peoples preferred food temperature depends on the type of food (think X graph)
DeWall Study on alcohol, aggression, and body weight
•Main effect
•Body weight (heavy men more aggressive than light men)
•Alcohol intake (alcohol vs. placebo)
•Significant interaction
•Result: overall difference in main effect for body weight hides the fact that body weight influences aggressive behavior when men are drunk; overall difference for alcohol hides the fact that alcohol intake makes a difference especially for heavy men.
"Especially for"
Indicates a significant interaction
Describing interactions in words – start with one level of the first independent variable and explain what’s happening with the second impendent variable, then move to the next level of the first independent variable and do the same thing
More important - main effect or interaction
When a study has both main effect and interactions the interactions are almost always the most important
A moderator
A moderator is an independent variable that changes the relationship between another independent variable and the dependent variable – a moderator results in an interaction
Strayer and Drews study: Hands free cell phones cause people to drive badly. - Younger and Older drivers studied. Example of looking for...
Interaction effect
Where in a journal should you look for the design of the study?
Methods section
Main effect
•Main effect – the overall effect of one independent variable on the dependent variable, averaging over the levels of the other independent variable
•Main effect – a simple difference, can be calculated using marginal means or an arithmetic means for each level of an independent variable, averaging over levels of the other independent variables (can use weighted averages to account for different sample sizes)
Spreading interaction
Spreading interaction – “only when” – my dog sits only when I’m holding a treat (think < )
Bartholow and Heinz Study - Alcohol intake can lead to aggressive behavior
Results: People cognitively associate alcohol cues with aggressive concepts
In reading an non-scientific article what words might indicate a factorial design study?
Words like “it depends” or “only when” indicate a factorial design
Best way to test a Theory?
Best way to study how variables interact is to combine them in a factorial design and measure whether the results are consistent with the theory