Table 3.
Estimated change in mean performance (in s) for Trail Making A according to job title, adjusted for various demographic and exposure variables. All models use robust standard errors to account for non-constant variance; comparisons between models show significance of those additional terms used to adjust performance beyond job title. and age.
| Model 1 (M1) | Model 2 (M2) | Model 3 (M3) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Variable | Est. | SE | p | Est. | SE | p | Est. | SE | p |
| Occupation | |||||||||
| Eng. – Appl. | −19.7 | 10.2 | 0.06 | −18.9 | 8.5 | 0.03 | −18.2 | 8.6 | 0.04 |
| Tech. – Appl. | −16.3 | 8.2 | 0.05 | −15.3 | 8.8 | 0.09 | −15.3 | 8.5 | 0.08 |
| Eng. – Tech. | −3.5 | 7.2 | 0.64 | −3.6 | 4.5 | 0.43 | −2.9 | 5 | 0.56 |
| Age per 10 yrs | 10.1 | 3.6 | 0.01 | 8.6 | 4.0 | 0.04 | 8.6 | 4.1 | 0.04 |
| Education per 1 yr | −0.2 | 2.4 | 0.92 | −0.5 | 2.4 | 0.84 | |||
| Work for MOA per 1 yr | 0.4 | 0.2 | 0.05 | 0.4 | 0.2 | 0.05 | |||
| GMa TCPy per 15 ng/mg creat. | −0.6 | 0.5 | 0.21 | ||||||
| Relativeb AChEper 10 percentage point decrease | 0.6 | 4.4 | 0.88 | ||||||
| Relativeb BuChE per 10 percentage point decrease | 1.6 | 1.6 | 0.34 | ||||||
| [Bayesian Information | 1252.7 | 1261.2 | 1274.3 | ||||||
| Criterion (lower is better) | |||||||||
| Model Comparisons | |||||||||
| M2 vs M1 (simultaneous test of education, work) | F(2, 118) = 2.03, p = 0.14 | ||||||||
| M3 vs M2 (simultaneous test of all 3 exposures) | F(3, 115) = 0.60, p = 0.62 | ||||||||
| M3 vs M1 (simultaneous test of all effects other than age, occupation) | F(5, 115) = 1.12, p = 0.35 | ||||||||
a
Geometric mean from Session 2 & 3.
b
Geometric mean activity from Session 2 & 3, divided by activity at Session 1.