Table 3.
Generalized estimation equations analyzing the effect of day of the week on response rate
| Parameter | Model 1 b (SE) |
Model 2 b (SE) |
Model 3 b (SE) |
Model 4 b (SE) |
Model 5 b (SE) |
Model 6 b (SE) |
Model 7 b (SE) |
Model 8 b (SE) |
Model 9 b (SE) |
|---|---|---|---|---|---|---|---|---|---|
| Intercept | – 0.86 (.02)** | – 0.86 (.02)** | – 0.86 (.01)** | – 0.86 (.02)** | – 0.86 (.02)** | – 0.86 (.02)** | – 0.88 (.02)** | – 0.88 (.02)** | – 0.88 (.02)** |
| Study | |||||||||
| Study 1 (ref.)a | 1.07 (.04)** | ||||||||
| Study 2 | 2.47 (.10)** | 2.46 (.10)** | 2.46 (.10)** | 2.46 (.10)** | 2.46 (.10)** | 2.47 (.10)** | 2.75 (.10)** | 2.75 (.10)** | 2.75 (.10)** |
| Study 3 | 1.00 (.04)** | 0.99 (.04)** | 0.99 (.04)** | 0.99 (.04)** | 0.99 (.05)** | 0.99 (.05)** | 1.20 (.04)** | 1.20 (.04)** | 1.20 (.04)** |
| Study 4 | – 0.20 (.01)** | – 0.20 (.01)** | – 0.20 (.01)** | – 0.20 (.01)** | – 0.20 (.01)** | – 0.20 (.01)** | – 0.25 (.01)** | – 0.25 (.01)** | – 0.25 (.01)** |
| Study 5 | – 0.45 (.01)** | – 0.45 (.01)** | – 0.45 (.01)** | – 0.45 (.01)** | – 0.45 (.01)** | – 0.45 (.01)** | – 0.51 (01)** | – 0.51 (01)** | – 0.51 (01)** |
| Day of the weekc | |||||||||
| Monday | 0.08 (.03)** | ||||||||
| Tuesday | 0.05 (.03) | ||||||||
| Wednesday | – 0.04 (.03) | ||||||||
| Thursday | – 0.05 (.03) | ||||||||
| Friday | – 0.09 (.04)* | ||||||||
| Saturday | 0.03 (.04) | ||||||||
| Sunday (ref.)a | 0.01 (.03) | ||||||||
| Weekend | – 0.07 (.03)* | – 0.09 (.05) | – 0.07 (.03)* | – 0.07 (.03)* | – 0.07 (.03)* | – 0.07 (.03)* | – 0.07 (.03)* | ||
| Day of workweek (linear trend) | – 0.04 (.01)** | – 0.07 (.06) | – 0.04 (.01)** | – 0.04 (.01)** | – 0.04 (.01)** | – 0.04 (.01)** | – 0.04 (.01)** | ||
| Day of workweek (squared trend) | 0.01 (.01) | ||||||||
| Holiday | – 0.00 (0.07) | – 0.00 (.07) | |||||||
| Holiday * Weekend | 0.12 (.16) | ||||||||
| Holiday * workweek (linear) | 0.06 (.07) | ||||||||
| Employment | – 0.03 (.02) | – 0.03 (.02) | – 0.03 (.02) | ||||||
| Education | 0.12 (.02)** | 0.12 (.02)** | 0.12 (.02)** | ||||||
| Gender | – 0.03 (.02) | – 0.03 (.02) | – 0.03 (.02) | ||||||
| Age | 0.35 (.02)** | 0.35 (.02)** | 0.35 (.02)** | ||||||
| Not employed * weekend | – .01 (.04) | ||||||||
| Not employed * day of workweek (linear trend) | .02 (.01) | .02 (.01)* | |||||||
| Quasi-likelihood | – 16864.00 | – 16855.00 | – 16856.03 | – 16855.92 | – 16856.03 | – 16855.60 | – 16451.1 | – 16448.5 | – 16448.6 |
| QIC | 33738.00 | 33733.00 | 33726.03 | 33727.81 | 33728.03 | 33731.16 | 32929.8 | 32928.7 | 32926.8 |
Note. N = 12,876 using 29,592 observations for Models 1 to 6, and N = 12,845 using 29,526 observations in models 7 to 9 due to missing data in employment variable. All models use an independent working correlation structure. Effects represent weighted effects (i.e., deviations from overall sample mean) for categorical variables. Continuous variables were centered. Employment status (0 = working, 1 = not working), Gender (0 = male, 1 = female), day of the week (0 = no, 1 = yes, reference category is Monday), weekend (0 = weekday, 1 = weekend), and holiday (0 = no holiday, 1 = holiday)
aThe values for the reference categories were calculated in additional models and added to the table (Nieuwenhuis et al., 2017). Since weighted effects represent deviations from the overall sample mean, the values of other effects do not depend upon choice of the reference category
*p < .05. **p < .01