The trouble is that when we believe something in a general sense (eg that TWAW, or that self-id poses risks to women and girls) then it seems obvious to us that 'most people' do this or that, as it chimes with our world view.
But just saying that '90% of X do Y' is not going to convince someone who takes the other view, as it will not chime with their own take on the matter. People can get stats from all over the place, and some are more reliable than others. If Stonewall put out a statement it is going to be biased towards their pro-trans perspective, whereas government statistics are far more likely to be objective. In this area in particular, interpretation is difficult, as there are only some figures in play - those about people who tick a box at a given time to say that they 'identify' into the opposite sex. There are no stats about how many people see themselves as trans, or whether those who do 'live as' a member of the opposite sex, whatever that involves.
There is no register of transpeople, so notions such as what percentage of transpeople do something are never going to be accurate. We might know that 100 transpeople have behaved in a certain way, but as we don't know how many there are, we can't know what percentage that 100 represents. Any post that quotes figures claiming that a percentage of transpeople do X Y or Z is going to be suspect for that reason alone. The poster may be quoting directly from a source, but the source simply can't be a reliable one.
It is not nitpicking to ask where generalisations come from. We all make assumptions based on our own prejudices and experiences, but it is reasonable that they are challenged when presented as facts. If I said that 90% of transpeople wore blue hats, I would be asked how I knew, and my sources would be called into question, wouldn't they? Sweeping statements about whether 90% of transwomen or transwomen prisoners do things are equally open to question.