Table 3.
Western Colorado SAR GLM results with an inverse distance weighting matrix. Parameter estimates are in units of WN cases per capita per kg/m2. The SAR framework reduces model residual spatial autocorrelation to non-significant levels, as assessed by comparing each model residual Moran I statistic to a bootstrapped distribution of Moran I statistics for random permutations of those model residuals. The expected Moran I statistic value for this sample size is −0.0011. The best-fit model is shown in bold.
| Month | Estimate | T Statistic | Significance | Deviance | Moran I Stat. | Moran Prob |
|---|---|---|---|---|---|---|
| March | −0.0079 (0.0045) | −1.76 | 0.0788 | 4.07e4 | 0.0001 | 0.633 |
| April | −0.0122 (0.0044) | −2.79 | 0.0053 | 4.00e4 | 0.0001 | 0.624 |
| May | −0.0191 (0.0067) | −2.87 | 0.0042 | 3.98e4 | 0.0001 | 0.622 |
| June | −0.0464 (0.0122) | −3.80 | 0.0002 | 3.91e4 | 0.0000 | 0.617 |
| July | −0.0443 (0.0124) | −3.56 | 0.0004 | 3.94e4 | 0.0000 | 0.626 |
| August | −0.0423 (0.0116) | −3.66 | 0.0003 | 3.93e4 | 0.0000 | 0.223 |