Hello, all!
My name is Katie Onkst; I enjoyed meeting you all this morning. I am happy to be serving as Dr. Vandenberg's TA this week and will be posting periodically throughout the course.
To get us started, I wanted to share three primary things that stood out to me from the morning session:
-
Measurement invariance isn’t optional. Even when we have reliable and valid measures under CTT, assuming those measures work the same across groups or time without testing them can lead to serious interpretational confounding.
-
Interpretational confounding can quietly undermine your whole model. If the meaning of a latent variable shifts between groups or over time, you may think you’re comparing apples to apples when you’re not. That risk multiplies if you skip testing your measurement model first.
-
The sequence of ME/I testing builds a foundation for meaningful comparisons. Each step, from configural to metric to scalar invariance, adds a layer of confidence that our constructs are functioning consistently. Skipping steps or rushing the process can jeopardize the validity of later group comparisons.
Here's to a great week!