Abstract
An area of increasing interest to agricultural and ecological researchers is the analysis of spatially correlated non-normal data. A generalized linear model(GLM) accounting for spatial covariance was presented by Gotway and Stroup (1997). Their method included approximate inference based on asymptotic distributions. A simulation study was conducted to assess the small sample behavior of their proposed estimates and test statistics. This study suggests that the spatial GLM yields unbiased estimates of treatment means and differences for binomial data, that the spatial GLM improves precision, as measured by MSE, and that the approximate F-statistic is acceptable for hypothesis testing.
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Recommended Citation
Collins, Roger G.; Stroup, Walter W.; and Kachman, Stephen D.
(1997).
"HOW GOOD ARE SPATIAL GLM'S? A SIMULATION STUDY,"
Conference on Applied Statistics in Agriculture.
https://doi.org/10.4148/2475-7772.1304
HOW GOOD ARE SPATIAL GLM'S? A SIMULATION STUDY
An area of increasing interest to agricultural and ecological researchers is the analysis of spatially correlated non-normal data. A generalized linear model(GLM) accounting for spatial covariance was presented by Gotway and Stroup (1997). Their method included approximate inference based on asymptotic distributions. A simulation study was conducted to assess the small sample behavior of their proposed estimates and test statistics. This study suggests that the spatial GLM yields unbiased estimates of treatment means and differences for binomial data, that the spatial GLM improves precision, as measured by MSE, and that the approximate F-statistic is acceptable for hypothesis testing.
