There are peaks - why? Is that normal? What do the peaks represent? Are they too big? Etc.
Simulate & compare actual vs. simulated (small wards/districts will likely be best excluded as usual), compare vs. 2018, 2019 p-values etc.
OK, the peaks are probably the result of the numerous small districts/wards with very discrete p-values recurring. The rest can still be interesting...
2014
%run app7_... plt.hist(df.p) plt.show()
There are peaks - why? Is that normal? What do the peaks represent? Are they too big? Etc. Simulate & compare actual vs. simulated (small wards/districts will likely be best excluded as usual), compare vs. 2018, 2019 p-values etc.