The effect sizes here are huge and seemingly hold when controlling for the typical factors such as socioeconomic status. They appear to hold both across groups that did not seek treatment and those that did AND in follow up studies checking the rates of how feelings (e.g. ideation) evolve over time for each group. It’s ok to raise concerns about confounding variables but the hypotheses here for such a huge effect size seem limited.

The fact that people can decide they’re not trans is not particularly material. This is very uncommon.

The conspiracy that such research must be politically motivated is a contrived argument and also falsely paints a dichotomy of saying that it must either help or harm. My assumption before doing the research was that they would find no effect.