 |
 |
I enjoyed Andrew Booth's summary. I'd like to expand a bit on one topic. >What if it is a tricky statistical point? >There is no need to necessarily handle this any different >from any other learning need. Again if it severely >compromises the understanding of the paper then the >answer will have to be found before proceeding with the >article. However if it is just a question of an >unfamiliar method or technique then ask the group to take >that on trust for the moment and to check this later >after the session - point out that the statistics is one >of the easiest sections of the paper to validate, >(compare inadequate description of randomisation, for >example), and that if an inappropriate method has been >used it will often be picked up by statistical reviewers. >Two useful resources to carry with you are Last's >Dictionary of Epidemiology and Greenhalgh's How to read a >paper (especially p.73). I have a fictional story that I tell people. It's about someone who comes to my office and says he has trouble understanding a recently published paper. I look at the title "In vitro and in vivo assessment of Endothelin as a biomarker of iatrogenically induced alveolar hypoxia in neonates" and say that I understand why you would have trouble with a paper like this. Yeah, he says in return, I don't understand what this boxplot is. Dr. Booth makes an excellent point in that the statistical methods are usually easy to validate. What I stress is that you should focus not on how the data was analyzed, but on how the data was collected. The four big issues in data collection are randomization, blinding, exclusions/drop outs, and protocol deviations. You don't need a Ph.D. in Statistics to assess these issues. Also keep in mind that some of the statistical details are there only for the benefit of those who want to reproduce the research. Most of the medical professionals I know are smart enough to skim over phrases like "reverse ion phase chromatography" so they should likewise skim over phrases like "bootstrap confidence intervals using bias corrected percentiles (Efron 1982)." When a statistical method is followed by a reference as in the example above, then you can take some solace in the fact that the authors do not expect you to be familiar with this method. Also, focus on whether you understand how to interpret the results. A bootstrap is a computer intensive method for creating confidence intervals that are not dependent on assumptions like normality. Once you appreciate this fact, you don't have to worry as much. You already know to interpret confidence intervals. The bootstrap is just another tool that produces confidence intervals. You do have to know some statistical terminology, of course. Anyone reading research papers should be familiar with Type I and II errors, odds ratios, survival curves, etc. A basic appreciation of simple statistical methods is enough for nine out of ten papers. I don't want to discourage people from learning more about Statistics, of course, but neither do I want people to be intimidated by statistical jargon. I have thought that a fun prank would be to go to a poster session at a medical conference, look over each poster carefully and then say something like "Very interesting, but aren't you worried that your results would be invalidated by the presence of heteroscedascity?" And then I would slowly walk away. Steve Simon, [log in to unmask], Standard Disclaimer. STATS - Steve's Attempt to Teach Statistics: http://www.cmh.edu/stats %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%