Great that’s what I thought we have plotted it too and it makes sense drug down for winning and up for losing in the same regions

Thanks so much

Ciara

From: Guillaume Sescousse [mailto:[log in to unmask]]
Sent: 23 May 2012 12:44
To: Ciara Mccabe
Cc: [log in to unmask]
Subject: Re: [SPM] contrasts reversed

Hi Ciara,
this is because these two interactions are expressed with the same contrast vector:

[(Win/Drug - Win/Plac) - (Loss/Drug - Loss/Plac)] = [Win/Drug - Win/Plac - Loss/Drug + Loss/Plac)] = [1 -1 - 1 1]
[(Loss/Plac - Loss/Drug) - (Win/Plac - Win/Drug)] = [Loss/Plac - Loss/Drug - Win/Plac + Win/Drug] = [Win/Drug - Win/Plac - Loss/Drug + Loss/Plac)] = [1 -1 - 1 1]

This is why it's important to plot your interaction effect to investigate its precise form.
Best,
Guillaume

Le 23-May-12 13:15, Ciara Mccabe a écrit :

Dear SPMers

If I have a contrast drug -plac win-loss (within subjects design)

And it gives EXACTLY the same areas of activation z scores everything as

Plac-drug loss-win

This is right?

Can someone explain in layman’s terms how SPM does this?

Thanks

Ciara