NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: What precision is required in cel nav?
From: Lu Abel
Date: 2015 Jul 19, 19:41 +0000
From: Lu Abel
Date: 2015 Jul 19, 19:41 +0000
Francis:
At the risk of starting an argument that will soon outdistance my knowledge of statistics, I think there is a difference between drawing conclusions from a set of data (eg, the effectiveness or side effects of a drug) where you know nothing ahead of time about its statistics, and drawing conclusions from a set of data where you know what it should look like.
Say, for example, I take a set of sights. I can say with confidence that they ought to be going up (body ascending), going down (body descending) or be roughly the same (body at meridian transit).
Given that I am taking imprecise measurements of precise phenomena (the position of a body in the sky) it is quite reasonable to question the accuracy of a shot if it is a severe outlier from the expected results. ("Shooting the dissident" was the way my statistics prof termed it)
Lu
From: Francis Upchurch <NoReply_Upchurch@fer3.com>
To: luabel@ymail.com
Sent: Sunday, July 19, 2015 12:11 PM
Subject: [NavList] Re: What precision is required in cel nav?
Umh!John, re the statsI'm certainly no mathematician, but statistics I do know.(medical research).There are lots of problems with this. Very selective data ("cherry picking", we call it ) and fairly arbitary exclusion of "bad outliers" . This would never get past the FDA I can tell you that from 30 years experience. I very much doubt these data are worth the paper they are written on.Perhaps a professional statistician could give their opinion?Also, I assume all done on big ships? Not 30-40ft yachts? Umh.I personally do not believe a word of this. But happy to be corrected by those who know better!Best wishesFrancis
