Statistically, if you want to check if the droprate is 50% at mp5 you need atleast 10 drops and 10 not drops in order to allow CLT deal with the non normal distribution of the samples.
But lets run the data for lulz...
Test and CI for One Proportion
Test of p = 0,5 vs p not = 0,5
Sample X N P-Value
1 1 30 0,000
That doesn't tell most peeps anything,,, So lets go through the data. X represents getting a key, N represents the number of runs, the P-Value is the chance of getting one or zero keys in 30 tries assuming the hypothesis that the probability to get a key is 50%. Generally if the P-Value is below 0.05 you discard the hypothesis.
But as we don't don't have enough data, we can't really use the test that's been used there... So we can't really attack the problem with statistics based on your small sample. Give me one where you have 10 drops and 10 non-drops and we can start talking statistics. Untill then, keep it out of it!