BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Circularity and selection bias in neuroscience and beyond: double 
 dipping\, publication bias\, and the decline of significant effects - Niko
  Kriegeskorte (MRC CBSU)
DTSTART:20121126T160000Z
DTEND:20121126T173000Z
UID:TALK40168@talks.cam.ac.uk
CONTACT:Mandy Carter
DESCRIPTION:A neuroscientific experiment typically generates a large amoun
 t of data\, of which only a small fraction is analyzed in detail and prese
 nted in a publication. However\, selection among noisy measurements can re
 nder circular an otherwise appropriate analysis and invalidate results. Sy
 stems neuroscience needs to adjust some widespread practices to avoid the 
 circularity that can arise from selection. In particular\, 'double dipping
 '\, the use of the same dataset for selection and selective analysis\, wil
 l give distorted descriptive statistics and invalid statistical inference 
 whenever the results statistics are not inherently independent of the sele
 ction criteria under the null hypothesis. To demonstrate the problem\, we 
 can apply widely used analyses to noise data known to not contain the expe
 rimental effects in question. Spurious effects can appear in the context o
 f both univariate activation analysis and multivariate pattern-information
  analysis. I will explain how circularity can be safely avoided in neuroim
 aging experiments. In the last 15 minutes\, I will broaden the scope beyon
 d neuroscience and discuss how selection at the level of studies and publi
 cations can introduce spurious results into the literature and create bias
 es affecting entire fields of science.
LOCATION:Seminar Room\, Department of Psychology\, Downing Site\, Cambridg
 e
END:VEVENT
END:VCALENDAR