Table 4. Univariate series analysis for predicting time to REM sleep.
| Variable | Mean (h) | Time to REM sleep
estimate bc (SE) |
p-value | |
|---|---|---|---|---|
| μ 0 | 1.27 | 1.09 (0.03) | < 2e-16 | |
| β cym30 | 3.73 | 0.54 (0.04) | < 2e-16 | |
| β cym60 | 4.22 | 0.62 (0.04) | < 2e-16 | |
| β mel3 | — | — | — | |
| β cym30 mel3 | 3.55 | 0.51 (0.04) | < 2e-16 | |
| β cym30 mel6 | 3.43 | 0.49 (0.06) | 8.1e-15 | |
| β cym60 mel3 | 4.33 | 0.64 (0.08) | 1.8e-14 | |
| β cym60 mel6 | 5.16 | 0.76 (0.08) | 4.9e-16 | |
| β cym60 tem15 | 4.27 | 0.63 (0.07) | < 2e-16 | |
| β cym60 tem30 | 4.43 | 0.64 (0.06) | < 2e-16 | |
| β Night | — | — | — | |
| β Weekday | ||||
| β Friday | 1.05 | -0.08 (0.04) | 0.0458 | |
| β Saturday | 1.01 | -0.09 (0.04) | 0.0174 | |
| bc: exponent = 0.375 | ||||
| Adjusted R2: 0.6823 | ||||
| p-value: < 2.2e-16 | ||||
Adjusted mean in hours (h), transformed mean estimate bc, standard error (SE) and p-value (Pr > |t-value|). R2: R-squared; bc: Box-Cox transformed variable raised to exponent given in final model. Back-transformation to original units was performed (after adjustments relative to intercept) by taking the nth (exponent) root of estimate.