TABLE 6.
Analysis of sawmill data including all sampled sites using the unbalanced repeated-measures model with structured covariance matricesa
| Source | dF | χ2 | P |
|---|---|---|---|
| Site type | 3 | 38.82 | <0.0001 |
| Sampler | 1 | 40.61 | <0.0001 |
| Replicate | 2 | 4.59 | 0.101 |
| Total | 193 |
a
Terms are as defined for Table 4. In this analysis, 193 measurement values representing 49 measurement sets were included. Neither wood type nor any of the interactions were significant in the full model, so they were removed in this reduced model. Regression equation: Endotoxin concentration = 5.91 + 0.5841 · sawing − 0.00272 · sorting − 0.172 · planing + 0.669 · filter + 0.0893 · second replicate − 0.0014 · third replicate. Reference variables are debarking for site type, AGI for sampler, and first replicate for replicate. In this analysis Akaike's information criterion was −214.5.