- Vanilish with Mirror Shot
- Female Maractus
- Female Crustle with Rock Slide (level up)
I also checked the missing pics, based on Serebii's and Bulbapedia's data:
In the statistics department, 50% for electric versus 70% for grass is an evident difference, but just to give a statistics backup, p-value = 0.0000000 (no surprises at all)
Just for information, I performed a F-test for 2 proportions (test often used for binomials), aided by a software. An aproximation by normal distribuition can be used (so a software isn't necessary) by the formula below:
z = (P1 - P2) / sqrt [ P1 * (1 - P1) / N1 + P2 * (1 - P2) / N2 ]
where P1 and P2 are the proportions for a certain area for each pokémon type and N1 and N2 are the samples size (including all the visited areas). 'sqrt' stands for square root.
z is related to p-value:
|z| > 2.58 ==> p-value < 0.01 - the proportions are NOT the same (more than 99% sure)
|z| > 1.96 ==> p-value < 0.05 - the proportions are
probabily NOT the same
|z| > 1.64 ==> p-value < 0.10 - it's better to gather more data
|z| < 1.64 ==> p-value > 0.10 - the proportions are probabily the same
If |z| < 1.64 then a positive z means that the first sample has higher value. If z is negative, the second sample is higher.
For those who want more detail, p-value = 2 * P(Z > |z|), where P(Z > |z|) comes from normal (0,1) probability distribuition. Please keep in mind that these aren't the exact values, but work very well when the proportions are at range of ~5% to ~95%. It's more than enough for our needs here.
From now on, I will use the normal aproximation instead of the F-test.
z-values so far:
Ghost x Dark (Manor): z = +3.82 (Ghost is better)
Dark x Psychic (Manor): z = -0.45 (no statistical difference)
Fire x Fight (Mountain): z = -1.14 (no statistical difference)
Grass x Electric (Forest): z = +7.26 (Grass is better)
I also got a very interesting result when comparing Mountain and Cave for steel types (
http://www.smogon.com/forums/showpos...postcount=1172 ):
Mountain (steel) x Cave (steel): z = -1.92 (p-value = 0.055), which means that Steel benefits both Mountain and Cave (we already knew that), but there is a chance that it prefers the Cave more than the Mountain. More data is required in order to be sure though (I guess there are more posts with steel results, but I didn't have time to look for it)