- Joined
- Aug 26, 2010
What I tried to say is that if a "true score" (the error-free value of a skater's performance) falls on the boundary of the rating levels, the measurement error places a significant role in the grade that the skater will receive.A "median with gray area" is very similar in spirit and effect to the trimmed mean.
Observed Score = True Score + Measurement Error
If the grade is in increments of 1 but the true value of a skater's performance is 0.5, the observed grade (either 0 or 1) will contain a high proportion of measurement error (0 = 0.5 - 0.5)(1 = 0.5 + 0.5)(Measurement Error = |0.5|).
If some scoring categories have a big minimum gradation than other categories (e.g., 1.00 in GOE of difficult jumps vs 0.25 in PCS), we can say that the impact of measurement errors is not randomly distributed. Luck or judge's subtle boost can play a more significant role in some elements.
I thought our discussion is to convince ourselves that the COP is good, better than 6.0 or any preferential voting system. Even if it is not good enough, maybe we can tweak a bit here (e.g., using the median), or tweak a bit there (e.g., changing the formula) and still be able to keep the spirit of COP (i.e., specified criteria). The major flaw of CoP, in my opinion, concerns judges’ focus on parts so much so that it is difficult for them to evaluate the performance as a whole at the same time. And its algorithm is basically adding up different body parts in hope of making a living thing out of them.This discussion has been going on so long that now I have forgotten exactly what perceived flaw in the CoP it is that we are trying to fix.
Don't laugh at me, but I do mean "multiply". The minimum score is 0 and maximum 10 for all four categories, with a perfect score of 200 (= 10 x 10 + 10 x 10). ∑(DxE), the sum of multiplication between difficulty and execution, is the formula used in diving for years and it hasn't caused an Olympic scandal yet. My algorithm can be seen as ∑(Parts x Whole) (i.e., skating is a product of the parts and the whole) or ∑(Difficulty x Performance) (i.e., skating is a product of the difficulty levels and the performance) or ∑(Cop x 6.0) (i.e., the judging method is a product of the CoP and the 6.0), or Elements (50%) + Presentation (50%) (i.e., first mark + second mark).^ Do you really mean to multiply the D and E scores? Couldn't the E score be negative?
Two panels of adjudicators are required: One (i.e., technical specialists) decides the type or level of each element (4T vs 3T) and assesses the skating skill levels based on prescribed criteria. The D-score is the combined base mark total of all elements (base mark is the fixed scale for each element, just like the current system except that we need to tweak the numbers). The S-score (Skating skill) is the combined score total of all subcategories (e.g., speed, edges, footwork, transition, etc.) that are rated separately on a scale. The other panel (i.e., judges) assessed the execution (E-score) and presentation (P-score) holistically as if assigning the 6.0 first mark and second mark. So, there is a panel watching for parts and another for the whole. It solves the criticism that it is impossible for the judges to focus on parts and evaluate the whole at the same time. I think all categories will be in increments of 0.1 (instead of my original 0.25) so as to accommodate the fine gradation in element base marks. And most importantly, the medians will be used to determine all ratings (level, score).
I like medians.
Last edited: