Table 2.
Morning waking cortisol by 5-HTTLPR, gender and day of samplinga
| Variable | Coefficient | s.e. | z | P | 95% CI |
|---|---|---|---|---|---|
| 5-HTTLPR | 0.11 | 0.03 | 3.36 | 0.001 | 0.05 to 0.18 |
| Females | 0.24 | 0.05 | 4.66 | 0.0001 | 0.13 to 0.34 |
| Day of sampling | -0.03 | 0.01 | -2.64 | 0.008 | -0.05 to 0.1 |
| Adversity number at entry | 0.02 | 0.03 | 0.5 | 0.62 | -0.044 to 0.074 |
| Constant | 0.63 | 0.1 | 5.99 | <0.0001 | 0.42 to 0.84 |
| Variable components | |||||
| Between participants | 0.44 | 0.02 | 0.40 to 0.48 | ||
| Within participants | 0.51 | 0.01 | 0.49 to 0.53 | ||
| Intraclass coefficient | 0.42 | 0.03 | 0.37 to 0.48 |
a. This model was obtained from 1551 observations of cortisol assayed from 393 individuals with an average of 3.9 (range 2-4) samples per participant. The log likelihood was -1427.7653 and Akaike’s information coefficient (AIC) was 2869.432, which gave the best fit compared with analyses that dropped sample day (AIC = 2874.462). Inclusion of any two-way interaction did not improve the fit (e.g. gender × 5-HTTLPR alone, AIC = 2871.305; and together with gender × adversity number, AIC = 2871.142).