Table 3.

Random effect model (Model 0) of log-transformed data (ppm) and mixed-effects models including the fixed effects: job group, design of process area, season, wind speed, and sampling duration. Model 2 includes sampling year in addition to the other fixed effects.

	Model 0	Model 1	P-value
	β (SE)	β (SE)
Intercept	−5.49 (0.22)	−3.71 (0.44)
Job group
Laboratory technicians		ref.
Mechanics		−0.41 (0.26)	0.116
Process operators		0.01 (0.26)	0.971
Industrial cleaners		0.84 (0.38)	0.029
Other		0.09 (0.30)	0.761
Design of process area
Open		ref.
Partially restricted		0.98 (0.37)	0.008
Restricted		0.72 (0.38)	0.060
Season
Winter		ref.
Summer		0.31 (0.16)	0.056
Wind speed
Light air (0–3.9 m s−1)		0.15 (0.24)	0.519
Breeze (4.0–11.9 m s−1)		ref.
Gale (12.0–20.0 m s−1)		−0.74 (0.19)	0.008
Sampling duration
Minutes (continuous)		−0.004 (0.0003)	0.000
Between-installation variance (bpS2)	0.90 (0.34)	0.27 (0.13)
Within-installation variance (wpS2)	4.89 (0.27)	3.93 (0.22)
Total variance (bpS2 + wpS2)a	5.79	4.2
% Variance explained by fixed effectsb		28

β, intercept. Total variance _{(random effects)} – Total variance _{(fixed effects)} * 100/Total variance _{(random effects).}

^ª Total variance = bpS² + bwS2.

^b % reduction in variance from random effect model to the respective mixed-effects models.