MVDA - Repeated Measures ANOVA

# Week 6: Repeated Measures ANOVA

When do we use it? When comparing two or more (dependent) variable means for only one group (within-subjects [WS] comparisons).

We use RMA in 4 types of situations:

1. Time-series - Example: Does IQ get lower when we get older?
2. Repeated measures experiment - Example: Is task performance better when working with others than when working alone?
3. Common measuring rod - Example: Which TV program do kids like more: Pokemon, Powerpuff Girls, or Totally Spies?
4. Pairs or groups - Example: do people work harder in cohesive groups than in non-cohesive groups?

Discuss the robustness of the multivariate tests with respect to possible violations of multivariate normality in each group.

We look at the number of participants per group. If it’s larger than 15, then it’s robust to non-normality.

Is there a significant effect of condition on trait x (e.g.: anxiety)? If so, which group has the higher estimated marginal mean?

We look at the at the significance in the Multivariate tests table. If they’re below 0.05, then there is a significant effect. Example of reporting:

Yes. All multivariate tests have F(3,5) = 26.955,p=.002

The researcher uses the ready-made contrast ‘Simple’ in SPSS. Which group means are compared with which other group means in this set of contrasts?

We look at the Tests of Within-Subjects Contrasts table. Below condition, look at what each level is compared to. For example, if all of them are compared to level 1, that means each group is compared to the first group.

Check if this set of contrasts is orthogonal.

Orthogonal means that there is absolutely no overlap between contrasts. We can check this by looking at the Estimated marginal means table, and checking the lower and upper bound. If this range overlaps with another group’s, then it’s not orthogonal.

Which contrasts have a significant F test? Interpret the significant effects.

We look at the Tests of Within-Subjects Contrasts table. Then, we check which condition the significant effect (p<0.05) belongs to. We also report the mean square belonging to that condition.

Example: Level 4 vs Level 1 is significant. An example of how we can report this:

Let’s say Level 4 means 50 people and  Level 1 means 1 person. We are testing scores of anxiety and number of people in front of which participants give a presentation.

Answer: Higher anxiety in front of an audience of 50 persons (M= 8.25) than an audience of one person (M= 4.250),F(1,7) = 22.803,p= .002

Page access
Public

### Live, Study, Travel, Help or Work abroad?

Join World Supporter
Join World Supporter
Content is used in bundle
Content categories
• Public
• WorldSupporters only
• JoHo members
• Private
Statistics
 [totalcount]