LAB NEWS                                                             

Voodoo & Type II

The Voodoo Debate Continues...

This past year (2009), I had public debate with Piotr Winkielman, one of the authors of the Voodoo paper, at the annual meeting of the Society of Experimental Social Psychologists (SESP). The session was videotaped and put online. Below are links to the videos. Because the powerpoint slides do not show up well in the footage, I've replaced the video for my parts with the slides I used. I've also included links to the original video if you'd prefer that.

Piotr Winkielman opening remarks

Matt Lieberman opening remarks [for original, click here]

Piotr Winkielman rebuttal

Matt Lieberman rebuttal [for original, click here]

The SESP webpage hosting all of the original videos is here

The original Voodoo paper is here and our reply is here

Multiple Comparisons & Type II Errors
There has been a push within the fMRI community to only consider effects significant if they survive what's called FDR correction at p<.05. Wil Cunningham and I have now published an article in SCAN detailing why this is an unduly harsh standard because it will produce countless Type II errors and is much more conservative than the thresholds typically used in behavioral research. We provide a formal defense of using combined intensity/extent threshold techniques (e.g. p<.005 for 10 contiguous voxels).

Lieberman, M. D. & Cunningham, W. (2009). Type I and Type II error concerns in fMRI research: re-balancing the scale. Social Cognitive and Affective Neuroscience, 4, 423-428. link

Our article was published as a companion piece with another article focused on the need to use FDR correction procedures.

Bennett, C. M., Wolford, G. L., & Miller, M. C. (2009). The principled control of false positives in neuroimaging. Social Cognitive and Affective Neuroscience, 4, 417-422. link