Table 2.
H0: Model 1 fits better than full model | H0: Model 2 fits better than full model | H0: Model 3 fits better than full model | |||||
---|---|---|---|---|---|---|---|
Subgroup | Effect mediated by: | x 2 | df | x 2 | df | x 2 | df |
Younger1 boys | Paternal BMI | 22.17*** | 1 | 14.83*** | 1 | 43.99*** | 2 |
Maternal BMI | 16.10*** | 1 | 10.53** | 1 | 31.30*** | 2 | |
Older2 boys | Paternal BMI | 10.32** | 1 | 3.89* | 1 | 16.79*** | 2 |
Maternal BMI | 13.09** | 1 | 3.38 | 1 | 19.80*** | 2 | |
Younger1 girls | Paternal BMI | - | - | - | |||
Maternal BMI | - | - | - | ||||
Older2 girls | Paternal BMI | - | - | - | |||
Maternal BMI | 7.07* | 1 | 2.91 | 1 | 11.54** | 2 |
Abbreviations: BMI, body mass index; x2, chi-square; df, degrees of freedom.
1Younger, ca 3–8 years.
2Older, ca 11–16 years.
***= p < .001, ** = p < .005, * = p < .05.
Chi-square difference test for path model fit.
Model 1: path from paternal/maternal education to paternal/maternal BMI (b1) was fixed to 0.
Model 2: path from paternal/maternal education to child’s overweight (b3) was fixed to 0.
Model 3: both paths b1 and b3 were fixed to 0.