Table 6.
Effect of changing the analyte-specific paint concentrationsa on predicted mean BZCs of each measured polyisocyanateb in downdraft booths
| Paint concentration (mg l−1) | Predicted mean BZC (μg m−3) |
|||
| HDI | Uretidone | Biuret | Isocyanurate | |
| 10 | 5.33 | 3.5 | 3.4 | 2021 |
| 25 | 5.50 | 5.7 | 5.4 | 2021 |
| 50 | 5.79 | 8.2 | 7.7 | 2021 |
| 100 | 6.42 | 11.9 | 11.0 | 2022 |
| 250 | 8.76 | 19.4 | 17.7 | 2023 |
| 500 | 14.7 | 28.0 | 25.2 | 2026 |
| 1000 | 41.4 | 40.4 | 36.0 | 2030 |
| 2500 | 65.7 | 57.6 | 2044 | |
| 5000 | 95.0 | 82.2 | 2067 | |
| 10 000 | 137 | 117 | 2115 | |
| 25 000 | 2264 | |||
| 50 000 | 2537 | |||
| 100 000 | 3184 | |||
| 250 000 | 6296 | |||
a
Predictions were not made for paint concentrations exceeding the maximum measured paint concentrations of individual monomeric and polymeric HDI.
b
Linear mixed models (Table 5) specific to downdraft booths were used to predict mean BZCs of the logged exposure data (μln). Median values (Table 2) were used for all other covariates in the models. Marginal R2 statistics and total variance estimates (σln2) for the analytes measured in downdraft booths were used to compute arithmetic means (μx) with the following formula: 
.