Discussion: View Thread

  • 1.  None credible research results in the journal Science

    Posted 12-06-2014 12:24
    http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0114255

    Stir often, cook slowly until publishable.

    -- 
    Michael A. McDaniel, Ph.D.,  Professor, Department of Management, 
    Virginia Commonwealth University 301 West Main Street, PO Box 844000, 
    Richmond, VA 23284-4000
    voice: 804.827.0209     e-mail: MAMcDani@vcu.edu
    Research Professor, VCU Department of Psychology


  • 2.  None credible research results in the journal Science

    Posted 12-08-2014 14:23

    I've been going over this back and forth and I'm still confused. The multiple studies in each article are generally designed to observe the same phenomenon manifested in slightly different ways. So how can the authors treat them as independent events and simply multiply their probabilities of success to compute a cumulative probability? Am I missing something?

    The conclusion is also very puzzling: you have to show some studies do not find significant effects in order for your set of studies to be statistically credible. But when you do, your theoretical rationale is then questioned. So you're damned if you don't replicate and you're also damned if you do?