Margin of Error calculation

[Septimus]

For this sort of thing Statistical Power would be much more important than Margin of Error. In this case, you would gain Statistical Power with increased sample size and controlling for outside factors. (Moon phase, et cetera.)

[Nystul]

where did the whole drop rate relating to moon phase come into play? I've NEVER had some kind of amazing drop rate on something and checked the clock and had it be full/new moon. Ever. It's always some odd random 37% or 82% or something. I always wondered where that even started. I know moon phase relates to fishing but I don't know how it ever got related to drop rates and how it was decided that full/new were specifically decided to be good phases. I've never seen any numbers or actual evidence for this. Just ooold (3+ years) dogma that everyone just seems to buy into.

Additionally, be wary of the difference between getting A drop and getting THE drop you want. Take nidhogg, for example. It will always drop 1 abj, and sometimes 2. TH will likely improve the odds of a second drop, however TH will not tell the mob to drop the one that you need. Don't put in TH because you're hoping to get m body rather than ebody becuase your shell needs that one more. The game doesn't know what you want for those kinds of things. I also have a gut feeling (entirely unsubstatiated and a total guess) that dring is "immune" to TH. It's a replacement for pixie earring- one of the two will drop, and since TH (again- unsubstantiated opinion and speculation on my behalf) seems to make things more likely to drop, it won't change that pixie is 99% and dring is 1%, it just says "category 2 drop increased" and then the game says "ok category 2 will drop because it's 99% earring and 1% ring, and random generator says earring drops today" and TH is a factor in saying that the slot of the loot pool will be occupied but not by what.

Another example is Kirin- TH makes it less likely that the Abj slot of the loot table is unoccupied, but it doesn't factor into the randomization of WHAT occupies that slot. Meaning "drop something" yes, "drop w. legs instead of togi" no.

Then again I could be full of shit and just talking out of my ass. Too bad we'll never really know because SE treats game mechanics like a 5th grade girls guard their secrets of who kissed who under the bleachers at the football game.

[Xaetica]

Exactly, don't worry about all those outside variables, just make sure your population includes them all by spreading your testing accross multiple days and moon faces.

If you want to know if things like moon affects drop rate, just make a note of it also when you jot down the data. There are statistical analysis you can do to determine if there is corrolation or not.

Also, since you are only gonna be killing one particular type of monster, your analysis will only be able to generalyse to that monster (if you want to be statistically honest). However this information will be more then enough to then do other smaller tests with smaller sample sizes to verify how much it can be generalyzed.

[Septimus]

You still have to control for possible factors. For example, targeting the same type of mob in the same zone controls possible variances between species and standardizes the drop tables since sometimes the same mobs will drop different items in different places. (For example, only Wespes in the Temple of Uggalepih drop Bee Larvae, not the ones in Crawler's Nest.)

You control for confounding factors because they may add a significant bias on your research, but also to limit the criticism that your research can face. If you are going to do a serious statistical analysis, take the time out to do it right from the start.

[Taj]

You still have to control for possible factors. For example, targeting the same type of mob in the same zone controls possible variances between species and standardizes the drop tables since sometimes the same mobs will drop different items in different places. (For example, only Wespes in the Temple of Uggalepih drop Bee Larvae, not the ones in Crawler's Nest.)

You control for confounding factors because they may add a significant bias on your research, but also to limit the criticism that your research can face. If you are going to do a serious statistical analysis, take the time out to do it right from the start.

If you really wanted to be serious about it, you could calculate the drop rate for each individual mob instead of the drop rate for all of the mobs in the zone. (Using something like their Character or Target ID.)

(Yes mobs have Character IDs.)

[Septimus]

You still have to control for possible factors. For example, targeting the same type of mob in the same zone controls possible variances between species and standardizes the drop tables since sometimes the same mobs will drop different items in different places. (For example, only Wespes in the Temple of Uggalepih drop Bee Larvae, not the ones in Crawler's Nest.)

You control for confounding factors because they may add a significant bias on your research, but also to limit the criticism that your research can face. If you are going to do a serious statistical analysis, take the time out to do it right from the start.

If you really wanted to be serious about it, you could calculate the drop rate for each individual mob instead of the drop rate for all of the mobs in the zone. (Using something like their Character or Target ID.)

(Yes mobs have Character IDs.)

Well, one would hope that all mobs in the zone had the same drop rate, but that is a good point. (Although most people wouldn't have the tools to check the mob ID.)

[Khamsin]

You still have to control for possible factors. For example, targeting the same type of mob in the same zone controls possible variances between species and standardizes the drop tables since sometimes the same mobs will drop different items in different places. (For example, only Wespes in the Temple of Uggalepih drop Bee Larvae, not the ones in Crawler's Nest.)

You control for confounding factors because they may add a significant bias on your research, but also to limit the criticism that your research can face. If you are going to do a serious statistical analysis, take the time out to do it right from the start.

If you really wanted to be serious about it, you could calculate the drop rate for each individual mob instead of the drop rate for all of the mobs in the zone. (Using something like their Character or Target ID.)

(Yes mobs have Character IDs.)

Well, one would hope that all mobs in the zone had the same drop rate, but that is a good point. (Although most people wouldn't have the tools to check the mob ID.)

I'd say most people do. Whether its through a third party program or not. Say you're the cleanest player ever to play FFXI and you don't even have Windower installed, let alone NPC.exe or any other tools. You could just edit your mob list .dat and tag them all with something at the end FFXII style. e.g. instead of 10 Canyon Crawlers, you have Canyon Crawler A through J. That way, you know each time you kill Canyon Crawler A, its the same one as before, which suits the purpose of this testing.

[Nystul]

You still have to control for possible factors. For example, targeting the same type of mob in the same zone controls possible variances between species and standardizes the drop tables since sometimes the same mobs will drop different items in different places. (For example, only Wespes in the Temple of Uggalepih drop Bee Larvae, not the ones in Crawler's Nest.)

You control for confounding factors because they may add a significant bias on your research, but also to limit the criticism that your research can face. If you are going to do a serious statistical analysis, take the time out to do it right from the start.

If you really wanted to be serious about it, you could calculate the drop rate for each individual mob instead of the drop rate for all of the mobs in the zone. (Using something like their Character or Target ID.)

(Yes mobs have Character IDs.)

Well, one would hope that all mobs in the zone had the same drop rate, but that is a good point. (Although most people wouldn't have the tools to check the mob ID.)

Typically they do, but there are exceptions- IE in East Saruta the crawlers or mandies or...something on that hill drop an item that the mobs elsewhere in the zone do not. Furthermore, mobs often change level in different parts of the zone (especially in the starter zones) so it's best to stick to one "spot" like...the leeches by the beach rather than every leech in the zone or something.

[Taj]

Mildly off topic, does anyone know what the hell this is ever used for?

http://euphidime.com/thottbot/img/fl_relic.png

[divisortheory]

You still have to control for possible factors. For example, targeting the same type of mob in the same zone controls possible variances between species and standardizes the drop tables since sometimes the same mobs will drop different items in different places. (For example, only Wespes in the Temple of Uggalepih drop Bee Larvae, not the ones in Crawler's Nest.)

You control for confounding factors because they may add a significant bias on your research, but also to limit the criticism that your research can face. If you are going to do a serious statistical analysis, take the time out to do it right from the start.

If you really wanted to be serious about it, you could calculate the drop rate for each individual mob instead of the drop rate for all of the mobs in the zone. (Using something like their Character or Target ID.)

(Yes mobs have Character IDs.)

If you DAT change the name of each mob to something different like Wespe 1, Wespe 2, Wespe 3, etc, DvsParse will already calculate things like this automatically, and display it in a nice tabular format. Could probably even add some more rigorous statistical things like margin of error, confidence intervals, etc.

[Septimus]

Mildly off topic, does anyone know what the hell this is ever used for?

http://euphidime.com/thottbot/img/fl_relic.png

You're the one who is supposed to know these things.

[Ashira]

*calls and drops the Statistics class for summer semester*

[Ryko]

It seems we are not alone.

[Taj]

Great, yet another thread derailed to be about Ashira.

[Ryko]

You're one to talk.

Besides, mithras are ugly.

[Taj]

http://files.myopera.com/ashira/blog/purpledalmy.jpg

How can you call that ugly? It's purple.

[Ryko]

It's got fleas.

Edit: Though that is a nice dalmy dat.

[Taj]

http://euphidime.com/frontline_ashira.png

[Ashira]

._.


Taj colored that dalmy for me; I asked him to help me figure it out and he just went and did it for me. Of course it also gives my tail a purple rinse. -.-


I don't have fleas.

[Ryko]

Yeah but how does that solve the inherent evil involved with both women and cats?