Dr. McLaren, 

Please forgive my dwelling on this issue but one last related question: the same stress test that I used for each alcohol and placebo day contained two runs. In order to assess the main effect of personality, sex and their interaction on each of the two rounds under a single condition (alcohol or placebo), should I also perform two separate full factorial models, one for each run or is it legit to just do one that combines the two?

Thank you again for the very clear responses and helpful answers. 

Malak 

 

On Wed, Jun 15, 2016 at 7:37 AM, Donald McLaren <[log in to unmask]> wrote:

Yes.

On Jun 15, 2016, at 2:08 AM, Malak abu shakra <[log in to unmask]> wrote:

Dear Dr. McLaren,

Thank you ! am I to understand, based on your response, that running a full factorial twice, once for alcohol and another placebo would be the valid way to go?

Malak

On Mon, Jun 13, 2016 at 1:46 PM, MCLAREN, Donald <[log in to unmask]> wrote:

No. You can't use a full factorial model for looking at the between-subject effects if you have repeated model. What you need to do is to run a model(s) without repeated measures to look at between-subject effects.

Best Regards, 

Donald McLaren, PhD

On Mon, Jun 13, 2016 at 12:49 PM, malak abu shakra <[log in to unmask]> wrote:

Dear SPMers,

I have a quick question re. how much information to model using the full factorial design: my study included one within-subject factor (2 levels, alcohol vs placebo conditions) and two between-subject factors (risk profile: 2 levels; sex, 2 levels). 

I am aware that a flexible factorial should be used to compare within-subject effects and full factorial for between-subject differences. What I'm not quite sure of is the following:

Even though I do not my full factorial model to look at within-subject differences, would it be statistically preferable that I still feed the model ALL the information obtained for subjects on both the placebo and alcohol conditions? (meaning, condition would be added as a third factor in my full factorial, with non-independent observations and equal variance). Or, would it alternatively be better if I repeat the same analysis twice, feeding it the placebo information the first time and alcohol information the second?

Thank you very much !

--

Malak Abu Shakra    Doctoral Student
Clinical Psychology
Psychology Dept.   Robert O. Pihl Lab
McGill University
1205 Dr. Penfield Avenue
Stewart Biology Building - Room W8/37
Montreal, Quebec, H3A 1B1
Canada

--

Malak Abu Shakra    Doctoral Candidate
Clinical Psychology
Psychology Dept.   Robert O. Pihl Lab
McGill University
1205 Dr. Penfield Avenue
Stewart Biology Building - Room W8/37
Montreal, Quebec, H3A 1B1
Canada

"The most erroneous stories are those we think we know best - and therefore never scrutinize or question". 

Stephen Jay Gould - 

--

Malak Abu Shakra    Doctoral Candidate
Clinical Psychology
Psychology Dept.   Robert O. Pihl Lab
McGill University
1205 Dr. Penfield Avenue
Stewart Biology Building - Room W8/37
Montreal, Quebec, H3A 1B1
Canada

"The most erroneous stories are those we think we know best - and therefore never scrutinize or question". 

Stephen Jay Gould -