Ordinal Time-Series Dependent Variable

#1
I am attempting to run a regression with an ordinal DV that is panel/time series data. I only have SPSS and with the ordinal DV, it seems I should be running the PLUM (Polytomous Universal Model). Will this work with time series data?

My dataset is based on election results over 20 years, covering 40 countries (roughly 4-5 election results per country). The DV is an ordinal measure (0 to 15) of the quality of the election management.
 

noetsi

Fortran must die
#2
I don't know SPSS at all when it comes to time series so I won't comment on that. Generally when you run time series you want to run it with some form of autoregressive error component [regression with autoregressive error is a good example of this - some call this dynamic regression]. I have not seen this done with non-interval dv so I do not know if that works or not.
 

noetsi

Fortran must die
#4
I suggest looking up either AR or ARDL in the context of regression time series. The use of the distributed lag portion of these is really complicated IMHO....