To continue from today’s class, here’s what we’ll be discussing next time:
– Estimating the direction and the magnitude of the discrimination parameters.
– How to tell when your data don’t fit the model.
– When does ideal-point modeling make a difference? Comparing ideal-point estimates to simple averages of survey responses.
P.S. Unlike the previous post, this time I really am referring to the class we had this morning.
Looks interesting:
-In an ideal point context my impression was that people generally do not care that much about the discrimination parameter (other than maybe to look for problematic situations); I thought the interest in discimination parameters falls more in the realm of item-response theory.
-On model fit: Andrew, personally I like the work you have done in the PA article with Bafumi and Park. I have applied it myself but I still wonder to what extent it is useful in practice. Sure, it may be very useful to understand how well your model does, but at times I want to convey this to the reader. You can explain this in lucid proze and with nice graphs, but really, what size of excess error rate is acceptable? (to use one of the things you worked on)
-I am curious on the “When does it make a difference.” Reading Simon Jackman, I thought you want to do it because you want to obtain the joit probability distribution of all parameters (especially the ideal points) so you can analyze quantities of interest (e.g. ranks) while taking uncertainty into account. But this line suggests that you are approaching it from another way.