Running dwnominate on the Swedish parliament

[potmo]

Running dwnominate on the Swedish parliament

[wmay]

So it sounds like you had three sessions of data. For less than 5 sessions, if I recall correctly, only a constant model is fit. This is hardcoded in the fortran source code. For such a small number of sessions it's not clear that a more complicated model is a good idea.

Oh, you know what I should do-- I should just add a warning to let people know when that happens.

Anyway, yes, you can get higher resolution coordinates. One approach would be to split the votes into more sessions. Another approach is interpolation-- the coordinates are just a polynomial fit to the session numbers (usually constant or linear), so interpolating is straightforward and valid and may or may not be valid, depending on the application. If you want a more flexible trajectory type, you'd have to look at some alternative methods like MQ scores for example

[potmo]

Thanks! Thats great! I'll try splitting the data into shorter session and I'll look into Martin-Quinn scores too. Seems like there is a C++ reference implementation for it.