NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: R: Averaging
From: Bill B
Date: 2004 Oct 18, 19:28 -0500
From: Bill B
Date: 2004 Oct 18, 19:28 -0500
Similar input to Federico's, but from the vantage point of marketing-research test instrument design. If, for example, I have two categories: Strongly agree Strongly disagree I am forcing respondents to one extreme or the other and they may not belong there. Let's assume I have a population that has no strong opinion on the topic. If I add a midpoint (odd number) possibility: No opinion and the vast majority choose it, then what appeared to be a polarized population becomes one with no strong opinions. But I may be comparing apples to oranges. Let's try a cel nav example (Assuming perfect sextant and operator). If I do a sight 2 minutes before LAN, 2 minutes after LAN, and average them I have a point on the midpoint of a chord (a straight line segment) from those two altitudes and below the actual LAN altitude on the arc. If I add an observation at LAN, I move altitude closer to the arc and reality. The more observations, the closer I get to approximating a position on the arc instead of on the chord. Since the sextant and operator are not perfect, the more observations I average the less affect one poor sight will have on the average. Once I get a lot of numbers in the average, I would guess (and it is only a guess) the odd vs. even would have less affect. Bill > George, > The explanation I've found for this recommendation is that this way you > can consider the central value of the odd series (i.e. the so called > median) which is definitely less affected by abnormal values than the > average value. > This is a criterion that I also remember from my statistic course at > university. > Hope this can help. > Federico > > -----Messaggio originale----- > Da: Navigation Mailing List [mailto:NAVIGATION-L@LISTSERV.WEBKAHUNA.COM] > Per conto di George Huxtable > Inviato: domenica 17 ottobre 2004 23.45 > A: NAVIGATION-L@LISTSERV.WEBKAHUNA.COM > Oggetto: Re: Averaging > > > I have read, here on Nav-l and elsewhere, the recommendation that when > averaging observations, it's best to take an odd, rather than even > number of them. > > I've no idea why. Can anyone explain? > > George. > > ================================================================ > contact George Huxtable by email at george@huxtable.u-net.com, by phone > at 01865 820222 (from outside UK, +44 1865 820222), or by mail at 1 > Sandy Lane, Southmoor, Abingdon, Oxon OX13 5HX, UK. > ================================================================