Let's say someone qualifies 18 races on pole and due to whatever circumstances (failures, penalties,...)
he qualifies 2 races as last.
Average scores are very sensitive to extremes.
You would have
avg(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,24,24) = 3,3
So the overall picture we would get is that driver X has an average score of 3,3, so an average qualif position during the year of 3,3. But was he really that bad? No he was a great qualifier and to get an even better picture of his skills, we can calculate a median result. That result is less sensitive to extreme results and thus gives you a more accurate picture.
med(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,24,24) = 1
So that's why I have included median results, because each driver during the season is subject to issues during qualif and incidents during the start / race which could distort a bit his results.
Pretty good example would be Hamilton. He did race to pole in Barcelona but was promoted to last due to him running out of fuel. So his 24th place would lower his 2012 overal grid position, while in fact he did much better (he was 1st), so by looking at the median results, the numbers are much more accurate.
I'm not good at explaining but I hope this will do :p
If anyone could explain it better, please do
Numbers are always discussable. But it does give you a nice overview because all data you see has been calculated based on the official data, except for the start results. There I actually had to study footage and images to see how much they had advanced when entering the first corner