All becomes clear if you can bother reading his methodology: he uses the Google maps definition of where the station entrance is, which is not hugely realistic and ignores multiple entrances; he has an arbitrary definition of what is a 'pub' and what isn't; and the data is based on a telephone question to whoever answered the phone in the 'pub', which is not likely to be especially reliable. So virtually a complete waste of time.

Even by these criteria there are numerous mistakes, And naturally looking first at Mornington Crescent, he seems to have invented a mythical 'Beer House', rather than the Lyttelton Arms, unless such a thing has sneaked in somewhere.