Mike Lawrence

Might not make a big performance difference, but a colleague suggested these when I was getting help understanding the code...

As reported [here](https://groups.google.com/forum/#!topic/ez4r/R1sH0Xxy_Zw), ezMixed appears to simply return NA's now, attributable to something having changed in how R detects dependencies (likely exacerbated by my hack-y use of `base::eval()`). Quick fix...

Or maybe remove this default behaviour? Can't recall why I did this in the first place.

New warning message in ezANOVA in R 3.1

1

``` exp12_means9Sbj

progress_time: try to make this work when running parallel plyr jobs

1

Hadley doesn't think there will be any universal solution given the paucity of foreach's feedback, but possibly for macs using doMC and having users explicitly set the value of options('cores'),...

By default it's median, but should let user pass whatever centering function they want.

ezANOVA: add ability to change contrast types

3

Currently ezANOVA simply uses the default contrasts specified by options("contrasts").

for SE computed from ezStats

###### # Notes from Rudolf Cardinal: ###### Re Levene's test etc.: I think I've found a definitive reference, starting on p592 (with the main bit on assumptions from p607) of...