I've heard a lot of ranting about fragment drops, so let's introduce a bit of statistics.
My bubble drops:
#1 - 4
#2 - 2
#3 - 13
#4 - 7
#5 - 6
Now let us introduce a little something called the chi-squared goodness-of-fit test. This guy tells us whether our observed data (above) matches the expected (20% across the board). After a bit of calculation, we get a chi-squared value of 10.42 with 4 degrees of freedom and a p-value of 0.0340.
What matters is this last number - 0.0340. This tells us that, assuming a truly random drop rate, there is a 3.40% probability of getting drops as skewed as mine (or worse). [Note: a lower number means more suspicious distribution]. In statistics, we usually take a number under 5% or 10% to be "statistically significant." This number is, which means that, statistically speaking, the districution does *not* match what it's supposed to.
N.B.
I do realize that this is a small-ish sample size, with expected values of 6.8 for each fragment, which is borderline questionable. Hence the call for more data (and me collecting more myself).
I glossed over some technical details, but the conditions (except for the aforementioned sample size issues) are satisfied, and the null hypothesis should be obvious (20% across the board).
It *has* been a while since I worked much with statistics, so if I messed up somewhere, kindly let me know instead of ranting about idiots who pretend to know more than they do.
--Zhan
Vice, Aielaman'Shan (Galaxy 18)
Moved to public forum. --Zhan