Apologies for the late reply. For some reason I missed the notification email.

Thank you very much for your explanation! It was very interesting.

Nevertheless, I must say that what I sent wasn't the whole dataset. It was just a sample of the first few thousands rows. The original one is too large to work in Excel and I'm using a MySQL database. In the full dataset 0 values represent just the 1.5% of the total. I checked with the organisation that produced the data and when there's a 0 it means that there was an ambulance already there, so they are not errors (at least not all of them).

The distribution is more normal than in the sample, so I'm going to stick to the median.

Thank you very much for your help.