Leftism

What’s the Y axis for the middle graph?

That's the percentage of kids who've reported some kind of sexual violence.

Also only having 3 data points in such a brief window doesn’t really say much.

I disagree because it's not really just about these YRBS surveys, it's the whole pattern. When we consider how conservatives are the only ones voting in favor of child marriage, and how pundits and randos on the internet will defend teen pregnancy, even if it was just one survey that showed a difference between red and blue states that would just be confirmation of a pattern that's already pretty obvious, and we should seriously ask why their ideology leads to this kind of stuff, and how to remedy it. Even if it's just a 2% point increase, this means that hundreds of thousands of children could be saved from abuse if conservatism was less prevalent.

Finally the grouping metric of “won majority of presidential elections from 2000 to 2020” isn’t clear and isn’t necessarily reflexive of policy. A more appropriate metric might be the party of the governor or the majority parties of their chambers.

There's really no definitive metric for "red" vs. "blue" states, so while I presidential election results will obviously reflect the politics of the people in that state, I do agree that it's not a thorough measure - but this same pattern holds even when using other measures of political affiliation.

I say this because I have some additional context here, as these graphs are part of an article I'm writing about the "pedocon" theory, and I can tell you that this same pattern shows up regardless of how we measure politics or CSA. Whether it's polling on how many people identify as Republicans vs. Democrats, or liberals vs. conservatives, or left-wing vs. right-wing, this correlation is still there. Looking at governor or chambers specifically could be an interesting addition, but I fully expect the same pattern to hold.

I think the y-axis shows number of kids.

I agree with what you're saying though—3 points does not make a compelling statement. I also agree that a better metric probably exists than what was posted. I'd add on and would like to know what the error bands represent—standard error, confidence intervals, or something else?