Loading ...
Sorry, an error occurred while loading the content.

GEOSTATS: Significance tests

Expand Messages
  • Chris Lloyd
    Hello, I have a question which relates to previous messages I can recall vaguely on AI-Geostats. I am working on segmentation of various spatial data sets. I
    Message 1 of 1 , Sep 29, 2000
    • 0 Attachment
      Hello,
       
      I have a question which relates to previous messages I can recall vaguely on AI-Geostats. I am working on segmentation of various spatial data sets. I want to compare one observation with another and on some basis decide they are from the same or different populations, which should be simple enough. In the past I have used standard t-tests but, of course, the results are questionable if a Gaussian distribution and independence of samples cannot be assumed. So the problem is this: what approach can I use that is similar in purpose to the standard t-test but that makes (much) weaker assumptions about independence (my data are strongly spatially autocorrelated) and normality (the data are positive skew and transformation would be very problematic in this case). None of the standard non-parametric tests seem suitable. I suspect that I am asking the impossible, but any suggestions for appropriate tests will be gratefully received. I will provide a full summary of any answers to the list.
       
      Best regards,
       
      Chris Lloyd
       
      --
      Chris Lloyd,
      Lecturer in Geography (GIS),
      School of Geography,          
      Queen's University Belfast,
      Belfast,
      BT7 1NN
      Northern Ireland, UK
       
      Email: c.lloyd@...
      Tel: +44 (0)28 9027 3478
      Fax: +44 (0)28 9032 1280
       
    Your message has been successfully submitted and would be delivered to recipients shortly.