Makes sense. Thanks again. Yann On 2 Oct 2015, at 12:18 pm, MCLAREN, Donald <[log in to unmask]> wrote: > I usually limit it to 20% across the entire run. > > If there is more outliers in one condition than another, then that raises questions about the overall design and the task. Usually, that implies stimulus correlated motion as well. > > Best Regards, > Donald McLaren, PhD > > > On Thu, Oct 1, 2015 at 10:23 PM, Yann Quidé <[log in to unmask]> wrote: > Thanks Donald! > > A quick question, is there any general consensus on a maximal % of outliers detected per condition on a block design or on an event-related design? > > Thanks again. > > Cheers, > > Yann > > > On 2 Oct 2015, at 12:01 pm, MCLAREN, Donald <[log in to unmask]> wrote: > >> That is a reasonable idea and is essential what was done in my 2014 paper with our own in-house custom scripts to identify bad images and add dummy variables to the GLM for each bad time point. >> >> Best Regards, >> Donald McLaren, PhD >> >> >> On Thu, Oct 1, 2015 at 9:29 PM, Yann Quidé <[log in to unmask]> wrote: >> Dear experts, >> >> Would it be unreasonable to use the ART toolbox to identify bad images, and then use the new SPM regressors file (with the identified images + 6 movements regressors) in gPPI analyses? >> >> Thanks for your time. >> >> Cheers, >> >> Yann >> > >