There are a few holes glaring holes in this methodology but just one is that she has 8.7k followers in Mastodon since joining in late 2022, while she has 241k followers on twitter since 2009. It's not surprising that the users that followed her in the last 5 months are more engaged than the ones that followed her in the previous 14 years. Are the people that just aren't interested in her content or that just stopped using twitter after a 14 year span necessarily grifters/leeches/spammers?
I don't know about 1 but simply due to scale 2 is probably true, however this experiment isn't saying much.