Celtic Heroes

I was wondering the same thing Rob, if he drops by percentage. But...we don't seem to have a whole lot of druids on the server and we have a ton of rogues which would imply we should get massive rogue drops so that doesn't seem to apply. I suppose maybe if it takes into account every single toon created on server? I don't know.

We track all of our Aggragoth loot and we've noticed what seems to be a fairly large discrepancy in loot for rangers. With 100+ Aggs now, I feel like we have enough evidence to support it (out of those 100, I don't think we started tracking until #26 so maybe I should wait for more hehe). I didn't include main hand weapon drops or non-class specific damage rings in this tally, only skill rings and braces. I'm just wondering if any other servers that track have noticed a shortage/abundance with their classes? Here is what we have:

Druid-37
Mage-31
Rogue-30
Warrior- 28
Ranger-19

Our poor rangers have to duke it out when a brace happens to drop. Maybe he is overcompensating our druids for all the times he has fried them?

Thoughts?

First I am going to say it is probably random. Second I'm going to say you need a much bigger sample size for better accuracy. 145 items is hardly enough. 1000 would bring you closer but here are the actual figures.

37 is 25.5%
31 is 21.3%
30 is 20.6%
28 is 19.3%
19 is 13.1%

as you can see there is a 12% difference from the high to low.

with a confidence level of 99%, population size of 20000, a sample size of 145 would have a margin of error of 10.66%.

with a confidence level of 95%, same population, same sample your margin drops to 8.11%

with a confidence of 90%, same, same, you get a 6.81%, which is about what your seeing with the 12%... ±6% has a span of 12%.

99% confidence with 1000 sample size gives you 3.97% error, 90% confidence gives you 2.53% error. so at 1000 samples in a 20k pop you can be 90% certain that you will see results within 5% of each other. 99% certain you will see results within 8% of each other.

now, I used 20k population because that is what the calculator suggested...however we can create our own for the next set of points.

let's say agg spawns 3 times a week, 52 weeks a year, 14 servers. so estimated 156 spawns a year times 14 servers is 2184 spawns in one year. there are 4 drops per kill so 8736 items in one year. I don't know exactly when agg was implemented as I'm a CH v3 newb... but we will stick with 1 year.

so our population is 8736... if we can tolerate 1% uncertainty we would need 3813 samples at 90% confidence, 4575 at 95% confidence, and 5723 at 99% confidence.

if we can tolerate a 6% uncertainty (which is what you have proven with your numbers... remember ±6% is a span of 12%)...
90% confidence would be 184 samples (close to what you had), 95% confidence would be 259 samples, and 99% would be 438 samples.

if we can terate 10 percent uncertainty we would need 68 @ 90, 96 @ 95, and 163 @ 99.

what does all this mean???... your numbers fall in line with a 90% confidence level... which means the normal distribution is spot on, the system is set to random, and you must kill it tons more to see the percentage span decrease between druid and ranger.

OR...OTM loves druids and hates rangers, that could always be the case too.

Woohoooo...someone stayed awake during stats class. I slept through it I was really just curious if say, the mage heavy servers noticed more mage drops etc. Thanks for the college stats refresher!