Table 2.
Summary model results
| Model 1 | Model 2 | Model 3 | Model 4 | |
|---|---|---|---|---|
| Estimate (95% C.I.) | Estimate (95% C.I.) | Estimate (95% C.I.) | Estimate (95% C.I.) | |
| Constant | 0.921 (0.848, 0.995) | −0.578 (−0.862, −0.294) | 0.099 (−0.085, 0.283) | |
| Fixed Effects | ||||
| Age | −0.026 (−0.048, −0.004) | −0.015 (−0.042, 0.012) | −0.054 (−0.081, −-0.027) | |
| Age2 | 0.006 (0.004, 0.008) | 0.006 (0.004, 0.008) | 0.006 (0.004, −0.008) | |
| Age3 | −0.00012 (−0.00016, −0.00008) | −0.00011 (−0.00015, −0.00008) | −0.00011 (−0.00015, −0.00008) | |
| Hours TV on in the home | 0.067 (0.055, 0.079) | 0.041 (0.027, 0.055) | 0.020 (0.004, 0.036) | |
| Mother TV-time | 0.205 (0.187, 0.223) | 0.192 (0.168, 0.216) | 0.135 (0.104, 0.167) | |
| Mother TV attitude a | 0.426 (0.314, 0.538) | 0.249 (0.131, 0.367) | 0.277 (0.167, 0.387) | |
| Ethnicity (ref: White British) | ||||
| Pakistani | 0.224 (0.129, 0.319) | 0.219 (0.072, 0.366) | 0.103 (−0.060, 0.266) | |
| Other | 0.155 (0.015, −0.295) | 0.069 (−0.154, 0.292) | −0.007 (−0.262, 0.248) | |
| Country Birth (ref: UK) | ||||
| Pakistan | 0.247 (0.147, 0.346) | 0.149 (0.008, 0.290) | −0.025 (−0.194, 0.144) | |
| Other | 0.148 (−0.006, 0.302) | 0.170 (−0.083, 0.423) | 0.137 (−0.153, 0.427) | |
| In most deprived IMD quintile | 0.166 (0.071, 0.260) | 0.053 (−0.065, 0.171) | ||
| No previous children | 0.148 (0.056, 0.240) | 0.072 (−0.042, 0.186) | 0.098 (0.002, 0.194) | |
| Child BMI at 24 months (z score) | −0.062 (−0.107, −0.018) | 0.008 (−0.043, 0.059) | ||
| Mother psychological stress (12 months) scored 4 or more. | 0.06 (0.011, 0.109) | 0.148 (0.042, 0.254) | 0.133 (0.035, 0.231) | |
| Interactions | ||||
| Age*Mother TV-time | 0.004 (0.002, 0.006) | |||
| Age*TV on in the home | 0.001 (0.001, 0.001) | |||
| Age*Pakistani ethnicity | 0.012 (0.004, 0.020) | |||
| Age* Other ethnicity | 0.005 (−0.009, 0.019) | |||
| Age*Pakistan born | 0.014 (0.004, 0.024) | |||
| Age* Other non UK born | 0.009 (−0.007, 0.026) | |||
| Random effects | ||||
| Uj(1) Random slope | 0.385 | 0.254 | 0.254 | 0.228 |
| Uj(2) Random intercept | 0.001 | 0.0004 | 0.0004 | 0.0005 |
| eij Between occasion | 1.11 | 1.047 | 1.047 | 1.041 |
| Total variance | 1.496 | 1.3014 | 1.3014 | 1.269 |
| Variance between children | 26% | 20% | 20% | 18% |
| Variance within children | 74% | 80% | 80% | 82% |
Model 1: average trajectories of child TV viewing by age with a random effect for age along with age squared and age cubed to allow the modelling of non-linear trends
Model 2: reporting the coefficients from a series of univariate models considering each covariate independently
Model 3: as model 1 but with the inclusion of all covariates together in a multivariate model
Model 4: final model derived, as model 1 with all significant covariates and interaction terms in a multivariate model. This is the model used to estimate predicted child TV viewing
Variance between children is the percentage of variance in the outcome attributable to differences between children; variance within children is the percentage of variance attributable to change over time within children
All covariates are time invariant apart from mother TV time which was measured at each survey wave
BMI Body Mass Index, IMD Index of Multiple Deprivation
aMother TV attitude: not agree that it is important that child not watch too much TV