*Latent Structural Equation Models*

*Model 1*: Both model 1a (preschool) and model 1b (primary school) fit the data well (Table 3).


**Table 3.** Data fit of the calculated models.

Note: CFI = comparative fit index; RMSEA = root mean square error of approximation; MIMIC = structural equation model with covariate age; MGM = multigroup model; SS = subsample.

Table 4 shows that the associations of the latent constructs of BMC and social integration with general HRQoL differed between preschool and primary school children. A positive small to moderate association between children's BMCs and their social integration (assessed by the teachers) was found in both preschool and primary school. In preschool, no correlations were found between general HRQoL (assessed by the parents) and BMCs or between general HRQoL and social integration.


**Table 4.** Intercorrelations between variables in model 1 (subsample 1).

Note: (1) object movement, (2) self-movement, (3) social integration, (4) general HRQoL. \*\* *p* < 0.01, \*\*\* *p* < 0.001.

In primary school, positive significant associations were found between general HRQoL and self-movement (*r* = 0.14, *p* = 0.005), as well as general HRQoL and social integration (*r* = 0.13, *p* < 0.001), whereas no correlations were found between general HRQoL and object movement. The results show that children who were better with selfmovement and children who were better socially integrated obtained higher values for general HRQoL. The correlations with age show that older children had better BMCs and seemed to be better socially integrated than younger children. No correlation with age was found for general HRQoL.

*Model 2:* Taking model 1 as a starting point, the correlations between the latent factors, as well as with age as a covariate, were calculated for both genders separately in a multigroup model (Table 5). This model achieved a good fit for preschool (model 2a, Table 3) and primary school (model 2b, Table 3). Deviation estimates for boys and girls appeared to be comparable in preschool (boys: χ <sup>2</sup> = 106.924, *n* = 484; girls: χ <sup>2</sup> = 94.146, *n* = 459) and within primary school (boys: χ <sup>2</sup> = 62.191, *n* = 450; girls: χ <sup>2</sup> = 117.932, *n* = 430).


**Table 5.** Intercorrelations between the variables in model 2 (subsample 1).

Note: (1) object movement, (2) self-movement, (3) social integration, (4) general HRQoL. \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001. Girls below the diagonal, boys above the diagonal

In model 2, the results were similar for boys and girls. Boys and girls with better BMCs were rated higher for their social integration by their teachers in both preschool and primary school. Regarding general HRQoL, there were no associations with BMCs or social integration for either boys or girls in preschool. For primary school, significant relationships with general HRQoL, as assessed by the parents, were only found for selfmovement among girls (*r* = 0.18, *p* = 0.007) and social integration among boys (*r* = 0.16, *p* = 0.004). The finding that age was positively associated with BMCs and social integration was evident for girls and boys.

*Model 3:* The structural equation model with object movement, self-movement, social integration, and the subscale physical well-being (t-value of subscale sum score) achieved a good model fit (Table 3). A high correlation between object movement and self-movement was found (*r* = 0.79, *p* < 0.001). As already became clear from model 1 and model 2, older children showed higher BMCs and were assessed as being better socially integrated. Moreover social integration was significantly correlated with object movement (*r* = 0.29, *p* < 0.001) and self-movement (*r* = 0.40, *p* < 0.001). No significant correlation was found between social integration and physical well-being. The BMCs of the children were positively

correlated with both object movement (*r* = 0.20, *p* = 0.004) and self-movement (*r* = 0.29, *p* < 0.001) (Table 6).


**Table 6.** Intercorrelations between the variables in model 3 (subsample 2).

Note: (1) object movement, (2) self-movement, (3) social integration, (4) physical well-being. \*\* *p* < 0.01, \*\*\* *p* < 0.001.

*Model 4:* Taking model 3 as a starting point, the correlations between the factors, as well as with age as a covariate, were calculated for both genders separately. This multigroup model achieved a good fit (Table 3). Separate deviation estimates were χ <sup>2</sup> = 67.209, *n* = 198 for boys and χ <sup>2</sup> = 108.436, *n* = 186 for girls. For both genders, the children's social integration, as assessed by their teachers, was moderately related to BMCs. Correlations with BMCs could also be found for physical well-being. Both boys and girls showed high correlations between self-movement and physical well-being (*r* = 0.35/0.34, *p* < 0.001). In the competency domain of object movement, a significant correlation was only observed for boys (*r* = 0.21, *p* = 0.009). No significant correlation was found between physical well-being and social integration (Table 7).

**Table 7.** Intercorrelations between the variables in model 4 (subsample 2).


(1) Object movement, (2) self-movement, (3) social integration, (4) physical well-being. \*\* *p* < 0.01, \*\*\* *p* < 0.001.
