Dear listers,

My question is about comparing two value lists mapped onto the same

point pattern:

We have calculated a kind of economical potential estimate from data

collected for 576 towns from gravitation modell using linear distances

between towns first, then using the minimum time required to reach one

town from the other via road network. So, we have two values for the

same 576 points which differ from each other with some magnitude and in

spatial distribution, too. We would like to spatially compare two data

to find areas where the two methods gives the most different results.

Our idea was to calculate a (simple) linear regression for the two

values, and then map the residual values, looking for the continuous sub

areas where residuals are autocorrelated. We are not interested in

estimation, but in finding the anomaleous places, where one modell

locally significally over- or underestimates the other one.

I am interested in your opinion about this approach, I would be thankful

to get ideas about probably better (standard?) methods for comparing two

mapped value sets.

Jozsef Fabian