Dear Anthony,
You could do statistical tests based on just the expected parameter values (mEps). The disadvantage is that you throw away the estimated variance provided by DCM for each parameter.
Let me know if you'd like any further clarification on this.
Best,
P
-----Original Message-----
From: Anthony Atkinson [mailto:[log in to unmask]]
Sent: 03 November 2015 17:20
To: [log in to unmask]; Zeidman, Peter <[log in to unmask]>
Subject: Re: DCM model evidence and free parameters
Dear Peter,
In order to compare BMA results obtained from a DCM analysis, you recommend doing t-tests based on the posterior mean (Ep) and variance (Cp). I am a bit confused about the difference between doing that and doing say, one-sample t-tests across participants for each connection/parameter using the values from mEps (e.g. the values from BMS.DCM.rfx.bma.mEPs). Could you please explain the difference? Is that latter method even a valid one?
Thanks,
Anthony
|