Page 182 - Second language development of newly arrived migrant kindergarteners - Frederike Groothoff

P. 182

```
182 Chapter 8 older pupils. We however do refrain from plotting the model in a graph because this would be difficult to interpret. A larger sample is necessary to confirm the effect. A likelihood ratio test did not show that the main effect of Regard for Student Perspectives contributed significantly to the fit of the model to the observed data, nor did we find an interaction between Age and Regard for Student Perspectives for MLR. This means that we could not show that the fact that teachers at Mainstream schools were more likely to take into account student perspectives compared to teacher at the DL2-schools had any significant impact on the pupils’ development of lexical richness. Macrostructure In the next three sub-sections the macrostructural elements of the narrative ability will be modelled and related to learning environmental aspects based on observed teacher behavior. First the growth model of Story Structure (SS), then Structural Complexity (SC), and finally Internal State Terms (IST) is presented. Story Structure (SS) From the comparison between the consecutive models for SS score (Table 8.8) it is apparent that a model with a fixed linear component – allowing for differences in Age – fit the data better than a model with only an intercept (Δχ2 (SSb) = 55.90; df = 1; p < .001). The variance within individuals depends on the age of participants (ΔΧ2 (SSc) = 11.98; df = 2; p = .002). However, the variance between individuals (SSd) did not. Finally, adding Age2 to the model improved the fit significantly (ΔΧ2 (SSe) = 10.17; df = 1; p = .001). Hence, in the final model (SSDEF) a fixed effect of Age1 and Age2, as well as a variance within pupils component which depends on Age needs to be included, and with this model we continued the analysis. Table 8.8: Fit of Different Polynomials (-2LL) for Changes in Story Structure (114 cases) as well as the Comparison of Consecutive Models. Model -2LL SSa: β0ijcons a 587.10 SSb: SSa + β1Age1ij 520.20 SSc: SSb + e1ijAge1ij 508.22 SSd: SSc + u10jAge1ij 506.53 SSe: SSd + β2Age2ij 496.36 SSDEF: SSa + β1iAge1ij + 503.37 β2Age2ij Comparison Models SSa vs SSb SSb vs SSc SSc vs SSd SSd vs SSe ΔΧ2 Δdf p 55.90 1 <.001 11.98 2 .002 1.69 2 .43ns 10.17 1 .001 a In addition to the intercept, variance components for differences individuals are estimated. within and between
```