However, statistical evidence is just an aggregation of anecdotal evidence.
No, sorry, completely wrong.
Statistical data comes from a well defined sample which is designed to be representative of an entire population. Anecdotes have no well defined selection criteria (my grandmother smoked until she was 100, and because I don't want to believe smoking is harmful, I remember and put great importance on this, and forget all the other people I have a connection to who were damaged by smoking) or (often where the selection criterion is 'stuff that happened to me') too little data to draw a conclusion (I've never crashed while driving drunk, so it must be safe.)
"I remember really hot days 10 years ago, hotter than now" holds very little weight. It is your experience (perhaps for every person like you there are 100 with the opposite experience), it is subject to biases of memory, and you've chosen the example retroactively to support a given position. The statistical data is thousands of thermometers measuring temperatures every hour over decades.