Thanks for all the replies! I have a class starting Jan. 12. Sounds like it's too risky for me to assign the book. I'll plan to use it in fall 2026.

@profandyfield.com Is it true that the R book is really coming out on Jan. 1? Is there somewhere we can get a table of contents?

Many thanks to MSU for issuing policies critical for my syllabus less than a week before the start of classes. After I finished the syllabus.

Wait until you read Infinite Jest. It's not that funny.

The new Cracker Barrel logo is the visual version of a 50 year old saying skibidi.

MSU Boosting Early Childhood Literacy- MSU Innovation Center PAWPrint Concepts was made by Dr. Lori Skibbe and Dr. Ryan P. Bowles to evaluate children's print concepts knowledge with a new testing method

A press release about our new measure of print concepts knowledge
innovationcenter.msu.edu/msu-boosting...

No, because it's hearsay. And not an exception to hearsay. Or it's an exception to an exception. But not an exception to an exception to an exception.

Already on my syllabus for next year!

This analysis was supposed to be easy. F*** you, data.

White and Nerdy

Great to see a student in my program get recognized for her expertise.

Where is the groan button in bluesky?

It finally happened. I have a student in my grad stats course whose advisor was once my student in the same grad stats course.

Language people: my trivia team had a rhyming question where the answer was Towel Howl. We had a debate about whether those words actually rhyme. Thoughts?

Delete everything. If it's important enough, they'll email again.

Note: Despite the brilliance of this strategy, I have never been able to implement it.

Lick hurt. Don't like hurt.

A Likert scale is always some form of Disagree to Agree rating scale. It is a type of rating scale, not synonymous with rating scale.

This is a hill I will not die on, because I have given up and have nothing left but to scream into the Bluesky void.

Underappreciated fact:

„When the entire population is measured there is no need to perform a hypothesis test. After all, there is no pop. to generalize to. When data from the entire pop. has been collected the pop. effect size is known, and there is no confidence interval to compute.“

A standard error is a standard deviation of a sampling distribution. What is a sampling distribution? That is a question for another weekend's mediation.

Don't be a hardass. At least one of your students is struggling with something major right now. No matter when now is.

How much different do two samples need to be for factorial in invariance to be meaningful validity evidence? For example, randomly splitting a sample in two and finding factorial invariance is definitely not meaningful validity evidence.

As long as the old friend is a companion during eating, not a side dish to go with the noodles. English can be a little vague.

I taught a lecture on IRT today, and I have to say that even after 25 years in the field, I don't really get it. In the absence of a causal theory of the response process, what can IRT really say about measurement and validity?

Not in Python, but I've tried several R packages and keep coming back to Winsteps.

I'd count myself in this group.

My prediction is fewer international students in higher tier programs, more students from high reputation U.S. colleges, and no change for low/mid reputation colleges.

This color scheme wins worst graph for me

Excellent threat. A comparison of the statistical properties of RMMs and FMMs is irrelevant. All that matters is the nature of the latent variable (although I would not necessarily say the formative approach involves a latent variable).

Embretson's construct representation was about whether the construct causes the item response. Did sadness cause the score? Whether the manipulation caused the level of sadness is a separate question, and there has to be a latent trait in between the manipulation and the score.