TABLE 4.

Endotoxin concentration data from the sawmills studied using analysis of variance for repeated measures^a

Sourceb	dFc	SSd	MSe	F	P
Grand mean	1	2,683.3	2,683.3	1,392.7	<0.0001
Between groups
Site type	3	20.91	6.97	3.62	0.02
Sampler	1	43.63	43.63	22.64	<0.0001
Residual	39	75.14	1.93
Within groups
Replicate	1	0.52	0.52	1.07	0.31
Rep∗ site type	3	0.68	0.68	1.39	0.26
Rep∗ sampler	1	0.006	0.006	0.01	0.91
Residual	39	18.97	0.49
Total	88	2,844.4

^aShown are the results of a reduced model that includes all sites with simultaneous AGI and filter data run in duplicate or triplicate. For sites with triplicate determinations only samples 1 and 2 were included. Wood type (P = 0.78) and the interaction between site type and sampler (P = 0.46) were not significant and were eliminated from the full model. 

^bGrand mean, comparison of data after controlling for all listed parameters and interactions; site type, debarking, sawing, sorting, or planing; sampler, AGI or filter; Residual, error term; replicate, two randomly selected replicate samples with each sampler were used in this analysis; Rep∗, interaction terms between the replicates and the site type and sampler. 

^cdF, degrees of freedom. 

^dSS, sum of squares. 

^eMS, mean square.