When I evaluate the quality of “research-based” teaching advice, I rely heavily on these websites:

First: I try to find the cited study on Google Scholar.

Second: I use AI — specifically Claude — to understand details about the study:

Who took part in the research: second graders? Dyslexic adults? People from the US or France or Iceland or Japan? Do those participants resemble my students and school?
What exactly did the participants do?
What exactly did the researchers measure?

Next: If I find the study, I check to see if other reseachers have reached roughly the same conclusion. Specifically, I explore

Scite.ai — to see how many other studies agree with, or contradict, the research I’ve been told about;

Connectedpapers.com — to get a spider-web diagram of other studies in the field (so cool); and

Elicit.com — to get a report on specific research questions.

I can find the research behind the advice, and
the participants match my students closely enough, and
the participants did something that makes sense in my classroom, and
the researchers measured something useful, and
those three websites suggest that most research comes to roughly the same conclusion, then

I feel MUCH more inclined to think seriously about following this advice in my classroom.

For the full method, check out The Goldilocks Map. More information is under the “My Books” tab.