You can’t even do that. The number of samples vary across the years and the number of samples do not account for population size. Australia for a population of 27M has a sample size of 1000 meanwhile India with a population of 1.5B gets a sample size of 3000. At that point the data might as well be just an error. It’s like going to NYC and asking 17 people whether they like Katz’s Deli. You will not get a representative sample at all on whether NYC likes Katz’s Deli. Might as well ask the Ouija board at that point.

Do not underestimate the law of large numbers and how little your sample can be while still remaining somewhat representative.