LOveLive! wrote:More difficult to overcome SDO disadvantage than before?

Please check the entire 17 necklace period record. You will see that necklace group match result is the most dominant factor in determining the OVERALL necklace finish rank. I urge you to check the Speareman's Correlation value between necklace finish, necklace group match result, and SDO. (
http://en.wikipedia.org/wiki/Spearman%2 ... oefficient" onclick="window.open(this.href);return false; ) I tried many statistical testing, and with the data at hand, there was greater chance that SDO was NOT enough to overcome necklace group match result compared to necklace group match result not being enough to overcome SDO disadvantage. Please also consider that the reason why SDO seems to play a bigger role, even though statistical testing clearly says it does not, is merely because you get to see SDO first. Also, I checked the correlation between SDO and necklace group match result and they turned out to be lower than correlation between final necklace finish rank and necklace group match result. There exist too many cases where those with very low SDO ended up with very high necklace group match result to suggest that SDO is determining factor for the necklace group match.
Check the discussions at
viewtopic.php?f=30&t=4506&start=20#p245792" onclick="window.open(this.href);return false;
Also check the following graphs for relationship between SDO and necklace group round result. The R^2 value is below 0.3, which means rather weak correlation. The correlation between necklace group round results and eventual necklace finish rank is much higher.
-------------------------
In order to save you the trouble of doing the calculations yourself, here is the Spearman's rank correlation coefficient for the 2012-2013 formula
Between SDO rank and Necklace Finish rank : 0.663509409
Between Necklace Group round result rank and Necklace Finish rank : 0.808823529
Between SDO rank and Necklace Group Round Result Rank : 0.346499358
The reason why we need to discuss the rank, and not who ended up being 1st : The simple reason is, there just isn't sufficient number of cases for this to give meaningful result. We will be using only 17 datapoint for this calculation. By including what happens at 2nd to 7th, we are increasing number of relevant datapoint by 7 fold, thus 119 datapoint. With 100+ datapoint, you can have some confidence in your analysis. With only 17, the uncertainty is too great, and what seems to be a pattern usually turns out to be mere coincidences.