# CSUN PSY 524 - Profile Analysis and Doubly Manova (23 pages)

Previewing pages 1, 2, 22, 23 of 23 page document
View Full Document

## Profile Analysis and Doubly Manova

Previewing pages 1, 2, 22, 23 of actual document.

View Full Document
View Full Document

## Profile Analysis and Doubly Manova

131 views

Pages:
23
School:
California State University, Northridge
Course:
Psy 524 - Multivariate
##### Multivariate Documents
• 8 pages

• 17 pages

• 7 pages

• 32 pages

• 32 pages

• 32 pages

• 39 pages

• 17 pages

• 24 pages

• 23 pages

• 22 pages

• 18 pages

• 28 pages

• 27 pages

• 24 pages

• 40 pages

• 42 pages

• 19 pages

• 19 pages

• 17 pages

• 28 pages

• 18 pages

• 2 pages

• 27 pages

• 33 pages

• 28 pages

• 27 pages

• 42 pages

• 18 pages

• 20 pages

• 2 pages

• 32 pages

Unformatted text preview:

Profile Analysis and Doubly Manova Comps in PA and Doubly Manova Psy 524 Andrew Ainsworth Comparisons on mains effects If the equal levels or flatness hypotheses are rejected and there are more than levels you need to break down the effect to see where the differences lie Equal levels For a significant equal levels test simply use the compute function in SPSS to create averages over all of the DVs Use this new variable as a DV in a univariate ANOVA where you can use post hoc tests or implement planned comparisons using syntax Flatness If the multivariate test for flatness is rejected than you turn to interpreting comparisons in a univariate within subjects ANOVA You can rerun the analysis removing the between subjects variables and implement post hoc tests on the within subjects variable or use syntax to use planned comparisons Testing interactions Simple Effects Simple Comparisons and Interaction Contrasts Simple effect and Simple Comparisons Interactions Whenever the parallelism hypothesis is rejected you need to pull apart the data to try and pinpoint what parts of the profile are causing the interaction Interactions Parallelism and Flatness significant equal levels not significant Simple effects would be used to compare the groups while holding each of the DVs constant Interactions Parallelism and Flatness significant equal levels not significant This is the same as doing a separate ANOVA between groups for each DV A Scheffe adjustment is recommended if doing this post hoc Fs k 1 F k 1 k n 1 K is number of groups and n is number of subjects Interactions Parallelism and Flatness significant equal levels not significant If any simple effect is significant than it should be followed by simple contrasts that can be implemented through syntax if planned or by post hoc adjustment Interactions Parallelism and Equal levels significant flatness not significant This happens rarely because if parallelism and levels are significant flatness is nonsignificant only if

View Full Document

Unlocking...