I try to make data-driven decisions in my practice, and I appreciate EBLIP attempts to take librarianship research and publication beyond "this is how we did it at my library". That said, I found David Berwick's commentary in the March 12 issue of JAMA, "The Science of Improvement," to be a very useful critique of the limitations of standard EBM methods, especially when they are used to study (essentially) social-science phenomena. Berwick proposes four changes to standard EBM research approaches: 1) embracing a wider range of scientific methodologies; 2) reconsider thresholds for action on evidence (do p values always need to be below .05 to be "significant" when we aren't talking about life-or-death interventions?); 3) rethink views about trust and bias, because interventions proceed within the context of local culture; 4) and finally, academicians must remain respectful of clinical expertise and the drive toward innovation. Complete citation: JAMA. 2008 Mar 12;299(10):1182-4. The science of improvement. Berwick DM. PMID: 18334694 It's worth a read.