URL: <http://gna.org/bugs/?12456> Summary: Error analysis/propagation incorrect when using peak intensities from replicated spectra. Project: relax Submitted by: bugman Submitted on: Thursday 10/16/2008 at 09:49 Category: relax's source code Severity: 4 - Important Priority: 7 - High Status: None Privacy: Public Assigned to: bugman Originator Name: Originator Email: Open/Closed: Open Discussion Lock: Any Release: 1.3.2 Operating System: All systems _______________________________________________________ Details: This problem effects relaxation curve fitting in all relax versions. The issue is that errors are calculated for each spin and then averaged due to the low sample number (2 or 3). Rather than the variance being averaged, the standard deviations are averaged. This is statistically meaningless. For more details, see https://mail.gna.org/public/relax-users/2008-10/msg00045.html, https://mail.gna.org/public/relax-users/2008-10/msg00031.html, and related posts in these threads. _______________________________________________________ Reply to this item at: <http://gna.org/bugs/?12456> _______________________________________________ Message sent via/by Gna! http://gna.org/