spatial Markov models

[Stephen Clark]

Hello, 

I am using the spatial Markov models provided by pysal / giddy. For convenience I have created the .gal file in GeoDa.

My data has 6 time points (years) and within each year I have converted the raw scores into quintiles. So each year will have roughly 20% in each quintile. I HAVE NOT CALCULATED THEM ON A POOLED BASIS. This is because an area's position relative to its peers in that year is the best way to capture the dynamics.

Here is my problem, I am confused by the effect of the fixed and discrete parameters in giddy.markov.Spatial_Markov(). I am using these values at the moment:

sm = giddy.markov.Spatial_Markov(c, w, fixed=False, discrete=True, k=5, m=5)

fixed is false because "If false, quantiles are taken each time period over n."

discrete is True because "If true, categorical spatial lags which are most common          categories of neighboring observations serve as the conditioning and fixed is ignored;"

What I don't understand is why is fixed ignored if discrete is True?

The quintiles I provide are just fine, I don't want them changing. For my data is fixed=False and discrete=True correct?