How to Write a Scientific Paper Ed Bullmore You should already have started to write your first/next paper, whether you know it or not! • Ethics committee applications, job applications, grant applications, protocol presentations etc allow you to rehearse in writing: – context, background, literature review – emergent hypotheses – a sense of motivation for the study – major study design elements Introduction (1): Context, background, literature review • Make a short, simple opening statement of the context in a few, accessible sentences - avoiding over-ambitious vagueness or immediately impenetrable jargon • “Hitherto, the nature of consciousness has proved elusive.” • “AR models of residual autocorrelation will fail for 1/f noise” • Background, literature review • remember this is not a review - so be selective, play favourites • remember your paper will be peer-reviewed, by prior autors in the field, so don’t be too selective... • acknowledge history! Introduction (2) • Hypotheses • inevitable, refutable, empirically specific, statistically testable • written down a priori • Motivation • why should you bother writing this paper and why should I bother reading it? Methods Major study design elements • Sample • size, with respect to power • composition, with respect to population and stratification • Measurements • observational • experimental • Statistical models and testing • factorial structure • test statistics or outcome measures • distributions including priors • hypothesis testing, type 1 and type 2 error control Results • Use figures and tables with self-contained legends to convey your most important results “at a glance” • Let your readers see as much as possible of the data for themselves, without losing narrative coherence • use descriptive statistics/graphics as well as hypothesis tests • oragnise presentation so that logically or substantively related results are juxtaposed Discussion • It is OK to use a less constrained, more conversational style • Start positive, headlining key results in context • return to hypotheses • be thoughtful about any differences between your work and the existing literature • Do not simply rehearse results • interpretation, synthesis, predictive speculation • avoid blob-by-blob decompositions of complex function in fMRI papers • pay attention to unexpected/discrepant results • Explicitly consider the limitations of your work Title, authors, abstract: The really important stuff Title, authors, abstract: the really important stuff • The title is the only part of your paper most people will read - make it clear, self-contained, descriptive • The abstract is vitally important - without doubt the most important 200+ words in the paper • tailor it to target journal • report results • use key words for literature searching • Authors - first, second, last and corresponding • seek guidance from your supervisor How to publish a scientific paper (1) • Think about target journals early on • high impact equals tight word count • impact is not always a six letter word • if you aim low you can’t subsequently move up the food chain • if you aim high you may have to allow for turnaround time (rejection) or “second album syndrome” (success) • Obey instructions to authors • use a bibliography manager • acknowledge grant support, conflict of interest How to publish a scientific paper (2) • Dealing with reviews • anticipate revision: it is almost inevitable and generally beneficial • organise the final version of the paper and all ancillary data carefully before submission • try not to take criticism personallyor as a reflection of incompetence on the part of reviewers – their failure to understand is your lack of clarity • be respectful, exact and direct in responding to the editor • if the reviews are too negative to justify acceptance, incorporate any helpful comments and resubmit • whatever you do - do it sooner rather than later! • Dealing with proofs • Dealing with fame!