Thread by @ErinInTheMorn

This is literally the worst social science research study I've ever seen.

It wouldn't get a passing grade in a social science research methods class.

A thread 🧵

https://twitter.com/jaypgreene/status/1536337517296902144

Lets first start with the "study design," which destroys the study's claims at the outset:

IT DOESN'T LOOK AT GENDER AFFIRMING CARE AT ALL!

Instead, it looks at states that let youth to visit doctors without parental consent... and somehow leaps to "GAC causes suicide"

They didn't even review it enough to check the labels on their charts. Their chart groups states into buckets of "has a minor access provision" and "does not have a provision" (again, for ALL care, not GAC)

But they laughably mislabeled the columns.

This confused me for a little while when I tried to research these states. Eventually I went to their source's website, and yeah. The states on the right DO have a minor access provision (to GENERAL healthcare).

This leads to an even more laughable claim...

Lets look at those states again. Remember, they are claiming that these having higher suicide rates means that GAC causes youth suicide...

Florida, Texas, Utah, Wyoming, Alabama, Arizona, North Dakota, Idaho, Indiana are all part of that bucket.

Notoriously BAD states for GAC.

Let's not even get into the fact that they failed significance on one model, so they kept adding variables and smoothing until they got significance.

Y'all this is so bad.

To even give this "research" the respect of saying "this would never pass peer review" is too much.

This wouldn't get a passing grade in a research methods class.

That right wingers are sharing it like it's some big groundbreaking study is hilarious.

Honestly I'm wondering why they bothered to waste this dude's time putting all this energy into a shit study.

Like they would have been better off just scribbling "Gender affirming care for youth bad and causes suicide" on a piece of paper in crayon and publishing that.

Oh my god one of their “researchers” is trying to actually defend their study. This is embarrassing.

https://twitter.com/jasonbedrick/status/1536396628835246080

I wonder how much this dude got paid to produce such bad research...

One of the researchers jumped in and tried to claim that they cover this by using search terms for puberty blockers as an indicator of youth trans GAC availability.

Never mind that those search terms have shot up in anti-trans states that restrict care. See Texas:

It's so bad.

In their study, this Texas datapoint would be used in SUPPORT of their claim that "gender affirming care leads to suicides."

Wait a second...

I think his google trend data is national rather than state based too....

They're basically using national search results that are time correlated to somehow claim that state level suicide rates are driven by GAC.

This is so bad. The p-hacking is obvious.

Lets pretend I was a reviewer for this "paper" (that would never even get past a review screening):

My first question would be... "why did you look at state general access provisions rather than specific accessibility of gender affirming care in each state?"

The answer is obvious:

Suicide rates at the state level are not correlated with the availability of gender affirming care in those states.

So you have to run multi-state regression model with p-hacked interaction variables that don't even align to get your finding.

They're pissed that I blocked the author.

If you can get passed peer review or if you seem reachable and are just making ill-informed conclusions, I'll engage.

If you know how to run a model like this and still put out misinfo, you're not worth my time.

That's my policy.