Table 2

Source model characteristics Characteristics of the two models and their factors of the principal factor analysis and their associations with the two PM2.5-measures: Model A includes all components shown in Table 1, Model B excludes Ca, Cr, Se and W.
Model characteristics
 Model A


Model B

Number of exposure variables included
 13


9

Total variance explained
 62.3%


68.9%

Correlation with PM2.5Mass (R2)
 0.73


0.63

Correlation with PM2.5Lightscatter (R2)
 0.71


0.52

Factor characteristics
 Factor 1 "crustal"
 Factor 2 "steel wear"
 Factor 3 "gasoline"
 Factor 4 "speed-change"
 Factor 1 "road surface"
 Factor 2 "gasoline"
 Factor 3 "speed-change"
Sum of squares of factor loadings
 2.48
 2.11
 1.82
 1.67
 3.01
 1.68
 1.51
Proportion of total variance
 19.1%
 16.3%
 14.0%
 12.9%
 33.5%
 18.6%
 16.8%
Slopea of correlation with PM2.5Mass
 2.95
p = 0.0006
 4.11
p < 0.0001
 2.16
p = 0.003
 4.50
p < 0.0001
 4.74
p < 0.0001
 3.03
p = 0.002
 3.21
p = 0.002








Slopea of correlation with PM2.5Lightscatter
 0.68
p = 0.6 (n.s.)
 3.58
p = 0.01
 3.74
p = 0.004
 11.32
p < 0.0001
 2.48
p = 0.15 (n.s.)
 4.03
p = 0.026
 9.27
p < 0.0001








a) Slope in μg/m3 per 1 SD change in the exposure factor score

Riediker et al. Particle and Fibre Toxicology 2004 1:2   doi:10.1186/1743-8977-1-2