http://www.medrants.com/archives/4925 Internal medicine, American health care, and especially medical education Sat, 11 Feb 2012 15:15:48 +0000 hourly 1 http://wordpress.org/?v=3.3.1 http://www.medrants.com/archives/4925/comment-page-1#comment-529450 Ed Whitney Mon, 19 Oct 2009 19:16:28 +0000 http://www.medrants.com/?p=4925#comment-529450 When I become ruler of the universe, the first thing I plan to do is require that medical journals attach links to their online articles which will have all the raw data (stripped of identifiers) which they used in their analyses. Online journals were supposed to "revolutionize" medical publishing, but the power of the internet has not been exploited; it should have been possible to have ASCII files (at the minimum) with each study so that the reader can do additional analyses if so desired. Yes, it is nice to be able to sit at home and download literature in your stocking feet rather than trudge to the library on a snowy evening, but that is scarcely a "revolutionary" development, except in reader convenience. Often I have wanted to be able to do such simple things as run an analysis of covariance (with treatment group as a covariate) when the authors only reported change scores in their results section. Is there any reason for journals not to require authors to provide their data? When I become ruler of the universe, the first thing I plan to do is require that medical journals attach links to their online articles which will have all the raw data (stripped of identifiers) which they used in their analyses. Online journals were supposed to “revolutionize” medical publishing, but the power of the internet has not been exploited; it should have been possible to have ASCII files (at the minimum) with each study so that the reader can do additional analyses if so desired. Yes, it is nice to be able to sit at home and download literature in your stocking feet rather than trudge to the library on a snowy evening, but that is scarcely a “revolutionary” development, except in reader convenience. Often I have wanted to be able to do such simple things as run an analysis of covariance (with treatment group as a covariate) when the authors only reported change scores in their results section. Is there any reason for journals not to require authors to provide their data?

]]>