Teaching Rasch measurement, Rasch analysis

Congratulations! Teaching Rasch methodology can be rewarding, but also has its challenges.

 

You will probably find that your class is imbued with "Classical" ideas. CTT usually starts with data and statistics. Avoid arguing the wrongs and rights of CTT for as long as possible ....

 

So avoid CTT pitsfalls by starting from Rasch first principles with lots of examples and classroom interactions.

 

1.What do we want to measure? Can we express it along a line (latent variable) from less (easy, beginner) to more (hard, expert)?

2.Can we convert this progression into dichotomous items? (relevant items = content validity, item hierarchy = construct validity)

3.How do we expect people of different abilities to respond to these items? (ability ordering = predictive validity)

4.What about people with abilities exactly at, then close to, an item's difficulty?  This introduces the idea of probability.

5.We can tell when a person is centered on the dichotomous items when the person's raw score is about 50%

6.Similarly we can tell when an item is close to the mean of the person abilities (thetas) when the frequency (p-value, probability) of success is close to 50%

7.So, based on these probabilities, we can put items and persons on the same "map" of the latent variable based on probabilities: "conjoint measurement". (In my experience, people are amazed at this. They have been taught item analysis and person reporting as separate topics.)

8.Draw a picture of probability (% success) vs. latent variable: logistic ogive -> logits unit of measurement

 

9.So with these first principles firmly in mind, introduce the Rasch model. Perhaps starting with one of the "Rasch model derived from" in RMT - such as https://www.rasch.org/rmt/rmt62c.htm

10.And now you are off and running ... time to introduce software to do the hard work!

11. If people need to see what is going on "under the hood", there are the Excel worksheets at www.rasch.org/moulton.htm