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

[ai-geostats] need help with a negative binomial spatial model

Expand Messages
  • Matthew Z.
    Dear list members, I am new to spatial analysis and appreciate your help. I have heard that there are problems with using discrete outcomes in spatial
    Message 1 of 1 , Feb 16, 2005
      Dear list members,

      I am new to spatial analysis and appreciate your help. I have heard that
      there are problems with using discrete outcomes in spatial regression, but I
      do not remember the exact errors/problems that arise. Negative binomial is the
      becoming the reviewer's standard for homicide research; if I fight them to use
      2SLS, I'd like to be coherent in my argument. Could you please give me a brief
      rundown of the errors I create when I attempt a negative binomial model and/or
      suggest a few sources for reference.

      This is the procedure I would ideally use:
      1. negative binomial model of homicide counts, with population as an
      exposure variable;
      2. save predicted count from the negative binomial model;
      3. multiply predicted count by a weights matrix to make spatial lag;
      4. add the spatial lag to the first negative binomial model.

      Why is a two-stage least squares model appropriate, but a two-stage
      negative binomial model not? -- I have highly significant findings for
      spatial lag using 2SLS, but lose significance when I run a neg. binomial

      Thank you very much,
      Matthew
    Your message has been successfully submitted and would be delivered to recipients shortly.