The Silph Research Group’s dedicated shiny hunters have been collecting and reporting on Minccino rewards obtained during the special Minccino Limited Research event. Our researchers completed Field Research tasks to earn 5,106 Minccino, with 527 of them shiny, for a shiny rate of approximately:
1 in 10
with a 99% confidence interval of 1 in 8.7 to 1 in 10.8. This rate is similar to the rate observed during last year’s Clamperl Limited Research event, and higher than those observed during the similar Feebas and Lotad events.
No statistically significant difference was observed between the shiny rates in the Asia-Pacific, Europe-Middle East-Africa, or the Americas regions.
Bonus: Are shiny rates the same for everyone?
It is often reported by travelers on the Road that shiny rates do not feel the same for all travelers. While not definitive, this event is an opportunity to test these claims on a rather large data set (500+ shinies) in a constrained time-frame.
We examine this topic in two ways. The first is to compare the distribution of shinies caught by each of our researchers to the theoretical distribution. While this analysis is subject to bias (if researchers do not choose their number of total encounters before they begin), it can be a simple way to look for abnormal deviations.
The distributions do not match perfectly, but there are not statistically significant differences between the distributions (chi-squared test, p-value=0.09) that would indicate anything out of the ordinary.
The second way we examine the data is by computing the probability of the exact number and distribution of shinies that we observed. We can then compare this probability to simulated data; that is, data re-created using a program that draws from a binomial distribution.
For this data set, we find that the logarithm of the probability of the observed data is -219. This value falls within the middle 95% of the simulated data (-199.3 to -221.9), and does not provide significant evidence to reject a simple binomial model. Had extra noise been present in the system (for example, if the probability was not a flat 1 in 10, but drawn randomly for each traveler), this value would have been more negative than the simulated range.
We hope that this analysis provides a small amount of peace to travelers that the playing field is level.
Until next time, keep hunting, travelers.