Vernacular theories of data, like vernacular theories of politics, are incredibly important, so YES to this and so very much.
and these things are designed by academics in some places and yes i'm looking at you, UCSD.
It is the Year of Our Lord 2026 and universities still use averages of Likert scales to compare courses without even thinking about mentioning standard deviations because obviously a 4.45 is worse than a 4.50 when you have 12 responses in a class of 80. Obviously.
I hate to even mention it but NextDoor is a great window into the completely loony theories of politics that many folks walk around with and that need some serious re-shifting.
And I won't even tell you about the data side. People **think** they know how surveillance happens ("they are listening into your conversations") but they don't realize it's way more interesting and complicated than that (social network algos+IP tracing+geolocation+a few other things).
Quick crowd question: Does your institution use confidence intervals when reporting results from student evals?
Wow, Routledge is now offering 30% discount as a thank you for reviewing which is great because it brings down the cost of that $300 monograph to a very accessible $200.
And LLMs. Phew! Those are basically magical oracles. Until you explain they are just good text predictors. With some constraints.
Angry email sent. I'm now THAT faculty member.
Reviewing a paper where the authors use ", that is", over and over and I'm not sure it's the best way to frame an argument, that is.
