Alright there, I am testing the Militaristic Trait...

Tiny Map, Deity, 75 Regular MA vs 75 Regular Spearmen (percentage as promotion rate)

Using Militaristic Civilisations:

Test One:

Japan vs Viking

35 Veteran Produced out of 75 battles won. 46.667%

Test Two:

Japan vs America

32 Veteran Produced out of 75 battles won. 42.667%

Test Three:

Zululand vs England:

31 Veteran Produced out of 72 battles won. 43.056%



Using Non-Militaristic Civilisations:

Test Four:

India vs China:

17 Veteran Produced out of 72 battles won. 23.611%

Test Five:

France vs Celts:

18 Veteran Produced out of 74 battles won. 24.324%

Test Six:

America vs Iroquis:

18 Veteran Produced out of 74 battles won. 24.324%

My next task is to test GL generation, but won't happen until tomorrow...

.... Sounds like pretty good results right now...

Can't wait to see the GL generation.

Though... I think the military trait doesn't effect GL generation. It doesn't increase your chance by giving an elite unit a higher chance of getting one... It increase's your chance by giving you more elite units.

Is there any difference between Vet-Elite?
I only build regulars by mistake in the later game.

Any unit which can attack twice in one turn automatically gets a promotion. They must have blitz enabled. Tanks,MA's etc

But I guess in your test they only attacked once.

Originally posted by Raidne
My next task is to test GL generation, but won't happen until tomorrow...

Click to expand...

I'd be interested to see if the Heroic epic performs as advertised.

you have 1 in 16 chance of a elite producing a GL for all civs as long as it hasent made one already, then it cannot make another (but can be upgraded and then produce another i believe) heroic epic makes it 1 in 12 i think.
note: elites have 5 hit points, vets 4, regular 3, conscript 2 but all have the same A/D/M stats.

I have made a surprising discovery...

Great Leader Test, Tiny Map, Deity, 16 MA (Elite) vs 16 Warrior Opponent randomly selected

Test Result
Militaristic
Human AI Victory Great Leader Exp V Percentage
China Spain 16 0 1
Japan France 16 0 1
Aztec Iroquois 16 1 1
Mongol Rome 16 1 1
Scandinavia Egypt 16 0 1
Celts Greece 16 1 1
Germany Greece 16 0 1
Zululand Babylon 16 0 1
Rome England 16 0 1
China Mongol 16 1 1
Japan Celts 16 1 1
Aztec Scandivania 16 1 1
Mongol Egypt 16 1 1
Scandinavia Spain 16 1 1
Celts America 16 0 1
Germany Korea 16 1 1
Zululand Spain 16 0 1
Rome France 16 1 1
China Japan 16 1 1
Japan China 16 0 1
Aztec Babylon 16 0 1
Mongol Carthage 16 1 0
Scandinavia Korea 16 0 1
Total 368 12 3.26%

Non Militaristic
Greece Celts 16 1 1
India Japan 16 1 1
Iroquois Japan 16 1 1
Carthage Zululand 16 0 1
Egypt Mongol 16 1 1
Babylon Persia 16 1 1
Russia Egypt 16 1 1
America Celts 16 0 1
France India 16 1 1
Persia England 16 1 1
England Egypt 16 0 1
Spain Carthage 16 1 1
Ottoman France 16 1 1
Arabia Korea 16 1 1
Korea Ottoman 16 1 1
Greece Japan 16 1 1
India Persia 16 0 1
Iroquois Egypt 16 0 1
Carthage China 16 1 1
Egypt Japan 16 1 1
Babylon Japan 16 0 1
Russia Zululand 16 1 1
America Carthage 16 1 1
Total 368 17 4.62%

My conclusion is that the militaristic player generates less great leader from his/her elite units to offset the advantage that more elite units were generated in the battle field.

Test Defect:

1. Using 16 units in a stack to attack, eliminate the chance of generating more than 1 GL in the same turn. I would try a walk around next time.

Raidne,

All you need to do is disband the GL to be able to generate another one in the same turn.

Personally, I would use a single stack of at least 500 Elite MA's against the same number of warriors, preferably 1000, once for a militaristic civ, one for a non-militaristic civ. 368 events just doesn't seem enough to test a 1/16 chance of something.

Additionally, if you were not disbanding the first GL after it was generated (with each test) then your results are unfortunately not correct. If a GL was generated with the first attack in the stack of 16, then the remaining 15 attacks were not relevant.

Short version: Statistically, 12 events vs. 17 just isn't enough data to conclude the rates are different.

Longer version: The usual rule of thumb is that the standard deviation is about the square root of the number of events - so about 4, here. The tests differ by only about one standard deviation, which is inadequate to demonstrate a difference.

Here's another thought....does fighting against a militaristic civ give a higher chance of promtion and/or generating a GL?

For example, Rome (militaristic and commercial).

If I fight a Roman unit do my chances increase of generating a protion or GL? What if it's two militaristic civs battling it out?

It would make sense to me that if you are fighting against a militaristic civ and defeat them, you would get a better chance at a GL....the poor unwashed masses of the "peaceful" civs need a hero that defeated the barbarians afterall....

1/16 chance of getting a leader in one battle gives a chance of 0.644 of getting a leader in 16 battles. (leader not disbanded)

368=23*16 battles will generate 0.644*23=14.8 leaders

This is quite close to both 12 and 17

Assuming 1/16 chance there is :
8.0% chance for getting 12 leaders
11.6% chance for getting 17 leaders

Thanks for the comments people. I am planning to use 10 five-unit stacks so that I can have more controls. When one attacks using the "j" short-cut, it is very difficult to see how one can disband the GL. I will keep you updated.

I predict it would be too time consuming to test if Militaristic civs get leaders more or less often than non-Militaristic. The difference between 1/16 and 1/15 or 1/17 is about .004 to .003. Like Beamup was talking about, if you want your standard deviation to be somewhere around that size you need 3000+ trials (provided I didn't goof the calculation).

Doing stacks of units doesn't help if leaders aren't disbanded. For all intensive puposes a stack of 5 armors count as just one trial. The difference is that a success (1 leader) has a probability of 1 in 3.6ish for a 5 stack instead of 1/16 for a single victory. You might as well do thousands of individual battles instead of thousands of stacks of 5 since using stacks you have to do some algebra to get the measured leader probability from your measured stack probabilty.

It wouldn't, however, be infeasible to test the heroic epic. The difference between 1/12 and 1/16 is .02. To have a standard deviation around this size you only need 200 trials.

If you want to set up a low maintenance test, why not change the 'battle created unit' in the editor to something you CAN have more than one of, eg: Scouts.

You'd have to check if the one-at-any-time limit is due to the Leader itself or the way its generated but.

Its here, under general settings

This has been tested but it looks like the thread is lost due to time. Here is what I have in one of my old folders. This information is direct from Firaxis.

-----------------------------------------------------------------------------------
Odds of promotion are based on whether the civ has the Militaristic trait or not and whether the victory was against barbarians or not. Basically, chances of promotion are halved for victories over Barbarians and doubled for Militaristic civs.

non-militaristic civ vs. non-barbarians:
conscript to regular: 1 in 2
regular to veteran: 1 in 4
veteran to elite: 1 in 8

non-militaristic civ vs. barbarians:
conscript to regular: 1 in 4
regular to veteran: 1 in 8
veteran to elite: 1 in 16

militaristic civ vs. non-barbarians:
conscript to regular: 1 in 1
regular to veteran: 1 in 2
veteran to elite: 1 in 4

militaristic civ vs. barbarians:
conscript to regular: 1 in 2
regular to veteran: 1 in 4
veteran to elite: 1 in 8


__________________
Mike Breitkreutz
Programmer
FIRAXIS Games

I love you guys

In addition, there is a rule that if a unit wins 2 battle in one turn, it will always be promoted on the second win.

Originally posted by Qitai
In addition, there is a rule that if a unit wins 2 battle in one turn, it will always be promoted on the second win.

Click to expand...

Also, if it wins a third time, it gets an automatic 2nd promotion.
(@ least in the case of reg. -vet. - elite).

I think this works for battles combined between your turn and the AI's also (ie reg attacks in your turn, wins, gets attacked by AI between your turns, wins again = auto promotion)

I've seen Conscript-Regular-Veteran-Elite, and it appears to follow the usual rules. Cannot swear it wasn't just luck, since I've only seen it once or twice, but appears unlikely that it was.

It may be pointed out that an Elite's second win in a turn isn't an automatic GL!

Thanks for all the inputs, I have re-run the test with you comments in mind. Unfortunately I did not read Tacit_Exit's post before I proceeded, probably I can setup a test and try that again... But here is my result (in an excel file)

Method 3 x 5-unit stack elite MA vs 3 x 5-unit stack Warrior (allow disbanding GL)

Militaristic: 1005 battles 53 GL
Non Militaristic: 1005 battles, 47 GL

the difference is not great, I retract my previous conclusion.