I want to test the relationship between categorical variables - fire severity (very low, low, moderate, and high) and cold-air pool (no, maybe, yes). My question is, Does cold-air pool status influence fire severity?

However, both variables are spatially auto-correlated. Fire severity is spatial auto-correlation beneath 100 m and cold-air pools are spatial auto-correlation beneath 1500 m. Due to the small area, I cannot sample beyond the spatial auto-correlation and thus my data violates the chi-square test's assumption of independence. I've looked into the Cochran's Q test, but it doesn't seem quite right either.

Is there a test that incorporates spatial auto-correlation for comparing two categorical variables?

  • $\begingroup$ I wanted to follow up on how I solved this problem. I ultimately used the continuous version of fire severity (RdNBR). I originally didnot want to use this method because the remotely sensed results are not ground-truth and normalized between fires. While I used the continuous RdNBR variable, I tried to account for my concerns and added a random effect term for each fire to account for remote sensing anomalies between fires. $\endgroup$ – Kate Wilkin Nov 11 '16 at 14:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.