Table 4. Results from longitudinal single mediation models (regression coefficients (B) and 95% confidence intervals (95%CI)) examining potential intrapersonal mediators of leisure time physical activity using residual change scores.
| Independent variable | Potential mediator | c-path B (95%CI)1,7 |
a-path B (95%CI)2,7 |
b-path B (95%CI)3,7 |
c'-path B (95%CI)4,7 |
Mediated effect a*b (95%CI)5 |
Proportion mediated (a*b)/c 6 |
|---|---|---|---|---|---|---|---|
| Family social support | Enjoyment | -4.06 (-16.68; 8.57) | 0.28 (-.05; 0.60) | -3.00 (-1.25; 7.24) | -4.88 (-17.11; 7.35) | 0.82 (-0.86; 2.51) | - |
| Outcome Expectations | -3.90 (-16.00; 8.20) | 0.21 (0.04; 0.37) | 6.39 (-4.01; 16.79) | -5.21 (-17.98; 7.55) | 1.31 (-1.11; 3.73) | - | |
| Self-Efficacy | -3.29 (-15.50; 8.92) | 0.15 (-0.08; 0.39) | 0.39 (-9.42; 10.20) | -3.35 (-15.36; 8.67) | 0.06 (-1.82; 1.94) | - | |
| Friend social support | Enjoyment | -7.91 (-21.98; 6.16) | 0.28 (-0.01; 0.57) | 3.33 (-0.93; 7.59) | -8.83 (-22.27; 4.61) | 0.92 (-0.74; 2.58) | - |
| Outcome Expectations | -8.42 (-22.32; 5.47) | 0.09 (-0.10; 0.28) | 6.61 (-3.28; 16.49) | -9.03 (-23.346; 5.19) | 0.60 (-1.06; 2.26) | - | |
| Self-Efficacy | -8.02 (-22.11; 6.07) | 0.30 (0.08; 0.52) | 1.16 (-8.54; 10.86) | -8.37 (-21.95; 5.21) | 0.35 (-2.60; 3.30) | - |
Note. Values represented in bold are significant associations
1 c-path (total effect): is the association of changes in social support from family or friends between baseline and 3 year follow-up and changes in leisure time physical activity between 3 year and 5 year follow-up analysed with a linear regression analyses using residual change scores
2 a-path: the association between changes in social support from family or friends between baseline and 3 year follow-up and changes in the mediator between baseline and 3 year follow-up analysed with a linear regression analyses using residual change scores
3 b-path: the association of changes in the mediator between baseline and 3 year follow-up and changes in leisure time physical activity between 3 year and 5 year follow-up analysed with a linear regression analyses using residual change scores
4 c’-path (direct effect) is the association of changes in social support from family or friends between baseline and 3 year follow-up and changes in leisure time physical activity between t 3 year and 5 year follow-up adjusted for changes in the mediator between baseline and 3 year follow-up. This association analysed with a linear regression analyses using residual change scores
5 Mediated effect (a*b) is the indirect effect of the independent variable on the outcome variable through the potential mediator
6 Proportion mediated effect (a*b/c)
7 All analyses are models adjusted for educational level (high, medium, low), living with children (yes, no), living area (rural vs urban) and robust standard errors were used to account for potential clustering within the neighbourhoods