Association
Logic
Temporal Priority

Mostly, we need experimental manipulation
 

Analyses

Null Hypothesis: no difference between groups

Alternative Hypothesis: there is a difference between groups

Our goal is to reject one or the other, hopefully the null hypothesis
 

Example

Null Hypothesis: money does not influence students' course evaluations

Alternative Hypothesis: money leads to improved course evaluations

One-way design with two levels of the IV ($)
 

ANOVA

Main effect analysis

F ~ Between-groups variance
        Within-groups variance

d.f. = levels of IV - 1
 

More ANOVA

If the F-test is significant at p < .05, we reject the null hypothesis

F-test just tells us that we have a difference; it doesn't tell us in what direction

So we look to the means
 

Multi-level Designs

Let's say I manipulate the IV so one group gets $50, one group gets $20, and one gets nothing

This is still a one-way design, but it has 3 levels now

d.f. = 2 (levels - 1)
 

ANOVA with multiple levels

Same test as with 2 levels

Significant result means at least one group difference is significant, but which one?

Post-hoc tests needed
 

ANOVA Recap

d.f. = levels - 1

With 2 levels of the IV, ANOVA is the same as a t-test and we look at the means to see which group is higher on the DV

With 3 or more levels, we have to use ANOVA, and post-hoc tests tell us where significant difference lies

Back to Overheads Page
Back to 381-002 Main Page