Hi David,

You're right about what each of the EVs model. The contrasts are based on linear combinations of these differences. I agree that in this construction things may not be immediately obvious, with the interaction and multiple conditions adding to the complexity.

Would it help for you if we rewrote the exact same design in a different manner? Please download again the same design_tim2.ods from the earlier email: I've updated it with a second design (designB) that uses a different construction. I'm not testing but it should yield the same results as the first one.

All the best,

Anderson

On Tue, 26 Jun 2018 at 05:16, David Hofmann <[log in to unmask]> wrote:

Absolutely no problem! 

2018-06-26 11:03 GMT+02:00 Anderson M. Winkler <[log in to unmask]>:

Hi David,

I'm away these days... I'll try to catch up with all emails next week if you don't mind...

All the best,

Anderson

On Tue, 26 Jun 2018 at 09:15, David Hofmann <[log in to unmask]> wrote:

Hey Anderson,

sorry to bother you again, but I send a reminder after 1 week by default :)..sometimes emails get lost or overseen. So ignore it when you have seen my email and answer as soon as you have time, no rush. Otherwise see below :). 

Thank you,

David

2018-06-20 18:48 GMT+02:00 David Hofmann <[log in to unmask]>:

Thank you Anderson!

I'm afraid it is not entirely clear to me how you modeled this. I may be reading this entirely wrong, but as far as I understand EV1 models the difference between A and B, EV2 models A-C and EV3 A-D asf. for the other group. And EV7 to EV14 account for the between-subjects effects.

Now, I'm not sure how the contrasts come about. Taking for example, C1, which should model the linear combination of the EV1 to EV6, i.e.: 2*EV1+1*EV2+1*EV3+2*EV4+1*EV5+1*EV6. How does this linear model A>B?

Also, what would be the 2-way interaction contrast?

Thanks for the support!

David

2018-06-20 10:05 GMT+02:00 Anderson M. Winkler <[log in to unmask]>:

Hi David,

I believe this question has appeared before. Please see this design: https://s3-us-west-2.amazonaws.com/andersonwinkler/mailinglist/design_tim2.ods.

The original thread was this: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;8d9efdc.1602

The A, B, C and D would be the 2x2 within subject conditions, and the two groups are then the between-subject final "x2" of your 2x2x2.

All the best,

Anderson

On 13 June 2018 at 19:02, David Hofmann <[log in to unmask]> wrote:

Hi all,

I'm trying to run a 2x2x2 ANOVA with one between-subject factor (Controls, patients) and two within-subject factors with 2 levels (pos. feedback/neg. feedback and observation/no observation). The number of subjects in each group is different. 

I want to calculate the 3 main effects (group, feedback, observation) as well as the 2-way (group x feedback) and 3-way interactions (group x feedback x observation). However, I'm not sure how to create a design matrix for this. 

Any help appreciated!

greetings

David

---

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1

---

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1