http://www.miraibio.com/blog/2009/06/tips-for-data-analysis/ What's going on at MiraiBio Mon, 21 Jun 2010 22:02:35 +0000 hourly 1 http://wordpress.org/?v=abc http://www.miraibio.com/blog/2009/06/tips-for-data-analysis/comment-page-1/#comment-8617 aliu Thu, 10 Jun 2010 21:44:54 +0000 http://www.miraibio.com/blog/?p=344#comment-8617 Thanks for the great insight Ted. I guess we are now at "Top 12 Tips" =) Thanks for the great insight Ted. I guess we are now at “Top 12 Tips” =)

]]> http://www.miraibio.com/blog/2009/06/tips-for-data-analysis/comment-page-1/#comment-8615 Ted Mifflin Thu, 10 Jun 2010 20:56:29 +0000 http://www.miraibio.com/blog/?p=344#comment-8615 Allen's "Top Ten" certainly make sense for us when we have been running ELISAs on 4000 samples for an epidemiological study. We have gained some added insights that I offer for consideration: 1) calculate the %CV for each pair of unknown sample replicates (we run duplicates), then plot %CV vs concentration of the unknowns to see if variance is as you expect, 2) periodically run the same set (n=20-30) of 'normal' samples through your ELISA method (if doing a large study) to estimate presence of long term drift (calculate medians and distributions), and 3) graphically review the concentrations of controls over long term using a Levey-Jennings or Shuwart plots to look for shifts or drifts in results. Allen’s “Top Ten” certainly make sense for us when we have been running ELISAs on 4000 samples for an epidemiological study. We have gained some added insights that I offer for consideration: 1) calculate the %CV for each pair of unknown sample replicates (we run duplicates), then plot %CV vs concentration of the unknowns to see if variance is as you expect, 2) periodically run the same set (n=20-30) of ‘normal’ samples through your ELISA method (if doing a large study) to estimate presence of long term drift (calculate medians and distributions), and 3) graphically review the concentrations of controls over long term using a Levey-Jennings or Shuwart plots to look for shifts or drifts in results.

]]>