I see the MIT aero thread has reared its head again; which prompts me to say this about all the "data" which gets thrown around here at ST:
Data points without stated errors and fits without stated confidence limits are junk, nothing more than anecdotal evidence. End of story.
The only circumstance under which it is acceptable to present data without error is in a *plot* where the error bars are smaller than the points, and even then that should be noted in the text. In all circumstances, the text must say something about errors.
Sometimes error analysis is a messy task and takes even longer than the original research goals, but results are worthless without knowledge of error.
Data points without stated errors and fits without stated confidence limits are junk, nothing more than anecdotal evidence. End of story.
The only circumstance under which it is acceptable to present data without error is in a *plot* where the error bars are smaller than the points, and even then that should be noted in the text. In all circumstances, the text must say something about errors.
Sometimes error analysis is a messy task and takes even longer than the original research goals, but results are worthless without knowledge of error.