Email discussion lists for the UK Education and Research communities

## SPM@JISCMAIL.AC.UK

#### View:

 Message: [ First | Previous | Next | Last ] By Topic: [ First | Previous | Next | Last ] By Author: [ First | Previous | Next | Last ] Font: Proportional Font
 LISTSERV Archives SPM Home SPM April 2017

#### Options

Subject:

Re: DCM BMA

From:

Zeidman, Peter

Date:

Wed, 26 Apr 2017 08:56:14 +0000

Content-Type:

text/plain

Parts/Attachments:

 text/plain (1 lines)
 ```Dear Janina Perhaps you didn't get Will Penny's response, which I'll reproduce below: That's an equivalent z-score of z=0.5406/0.0569=9.5 So, a huge effect. spm_Ncdf returns 0 as the equivalent probability is smaller than eps, the smallest number matlab can represent. I didn't see any code below but perhaps you could replace 0 with eps and then Bonferroni could work with that. If anything remains unclear, feel free to ask again. Best Peter -----Original Message----- From: Janina Seubert [mailto:[log in to unmask]] Sent: 25 April 2017 16:28 To: [log in to unmask]; Zeidman, Peter <[log in to unmask]> Cc: Janina Seubert <[log in to unmask]> Subject: Re: DCM BMA Hi there, hope it's ok to do a very late followup on this... I am trying to use the procedure below to estimate the posterior probability of a parameter that has a mean of 0.5406 and a std of 0.0569. When I put these values into the code below 1-(spm_Ncdf(0,0.5406,0.0569^2)), spm_Ncdf becomes 0, and thus the posterior probability becomes 1. Is it really possible for the posterior probability to be 1? What does this mean conceptually, and how would you apply Bonferroni-correction to an spm_Ncdf of 0? ```