[NSRCA-discussion] Scoring Process Question

[rcmaster199 at aol.com]

Two different flights flown by two different pilots only scored one maneuver differently....the rest of the maneuvers had identical scores.  The flights are practically identical in nearly every way except this one. The pilot with the higher score in that one maneuver would win the round. 

The difference in score could be as little as 1/4 point in a potential of around 660 avg points (2 judges where one scored every maneuver the same and the other's only difference was in that maneuver). In normalized terms, that's about 3/8ths of a point in 1000. Not much of a difference but many times at large meets, that's where it comes down to. One chap goes away the Champion and the other just goes away

We have to have a yardstick and this is it

MattK


-----Original Message-----
From: glmiller3 at suddenlink.net
To: NSRCA List <nsrca-discussion at lists.nsrca.org>
Sent: Tue, 26 Jun 2007 12:50 pm
Subject: [NSRCA-discussion] Scoring Process Question



I'm going to open a can of worms here in hopes of coming up with a better system 
ut of the discussion.  Perhaps this has been discussed before and I'm not aware 
f it.  Let me preface this by saying I am not a mathematician or statistician, 
ut I have some familiarity with both subjects and the following question has 
een growing in my mind for some time. 
It seems to me that we are judging our maneuvers with limited accuracy (within 1 
oint in FAI and X.5 points in AMA classes) we are then creating the ILLUSION of 
ccuracy by multiplying that score by a K factor and then normalizing to a 1000 
oint scale.  Here is a fairly brief explanation of "Significant Digits" that 
've copied from the web which will introduce you to this thought if you haven't 
een it before:
****"SIGNIFICANT DIGITS
The number of significant digits in an answer to a calculation will depend on 
he number of significant digits in the given data, as discussed in the rules 
elow. Approximate calculations (order-of-magnitude estimates) always result in 
nswers with only one or two significant digits. 
When are Digits Significant? 
Non-zero digits are always significant. Thus, 22 has two significant digits, and 
2.3 has three significant digits. 
With zeroes, the situation is more complicated: 
Zeroes placed before other digits are not significant; 0.046 has two significant 
igits. 
eroes placed between other digits are always significant; 4009 kg has four 
ignificant digits. 
eroes placed after other digits but behind a decimal point are significant; 
.90 has three significant digits. 
eroes at the end of a number are significant only if they are behind a decimal 
oint as in (c). Otherwise, it is impossible to tell if they are significant. 
or example, in the number 8200, it is not clear if the zeroes are significant 
r not. The number of significant digits in 8200 is at least two, but could be 
hree or four. To avoid uncertainty, use scientific notation to place 
ignificant zeroes behind a decimal point: 
.200 ´  has four significant digits 
.20 ´  has three significant digits 
8.2 ´  has two significant digits
Significant Digits in Multiplication, Division, Trig. functions, etc. 
In a calculation involving multiplication, division, trigonometric functions, 
tc., the number of significant digits in an answer should equal the least 
umber of significant digits in any one of the numbers being multiplied, divided 
tc. 
Thus in evaluating sin(kx), where k = 0.097 m-1 (two significant digits) and x = 
.73 m (three significant digits), the answer should have two significant 
igits. 
Note that whole numbers have essentially an unlimited number of significant 
igits. As an example, if a hair dryer uses 1.2 kW of power, then 2 identical 
airdryers use 2.4 kW: 
1.2 kW {2 sig. dig.} X 2 {unlimited sig. dig.} = 2.4 kW {2 sig. dig.} "******
My Point is this:
I've seen many contests decided by less than 10 points on a scale of 4000 which 
as been expanded from (at most) 2 significant digits.  As a matter of 
statistics" I think that any separation of less than 100 points (two 
ignificant digits, ie,  3X00 points) is "artificial accuracy".  Unfortunately, 
 don't have any great ideas about how to improve upon the current system, I'm 
ust pointing out what I think is a scientifically valid problem with it.  
I smile when I see round scores posted to ten thousanths of a point on a scale 
hat has been expanded from two significant digit accuracy to a 1000 point 
cale.  This turns a two significant digit answer into eight significant digits!  
ie, 1234.5678)    I think that scientifically, the scores would be more 
ccurately posted as in scientific notation at   x.x  * 10 to the second power.  
ost of the contests that I've been to this year have been decided essentially 
y random statistical "noise" rather than actual scoring decisions.  

as anyone ever thought/talked about this before ?   
Let me add, that despite what I think are statistically invalid methods, in most 
ases the system seems to work pretty well.  In general the superior pilots get 
nough better scores to overcome the "noise" but it sure would be nice to come 
p with a more mathematically valid solution, IMO.
George


______________________________________________
SRCA-discussion mailing list
SRCA-discussion at lists.nsrca.org
ttp://lists.nsrca.org/mailman/listinfo/nsrca-discussion


________________________________________________________________________
AOL now offers free email to everyone.  Find out more about what's free from AOL at AOL.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.nsrca.org/pipermail/nsrca-discussion/attachments/20070627/116a947d/attachment.html