Page 198 - Second language development of newly arrived migrant kindergarteners - Frederike Groothoff

198 Chapter 8 Figure 8.14 shows that the effect of Age on SS is significant. This is a quadratic relation; the increase in SS development is larger for younger pupils. With a constant variance within and between variance we cannot say that there are differences in the precision of our estimates between older and younger pupils, nor can we say that the variance between older pupils is different than the variance between younger pupils. The analysis was continued with the new General Development Model, SSDEF. The addition of the percentage of time spent in Balanced Language Situations improved the General Development Model of the SS score (ΔΧ2 (SSVIII) = 4.37; df = 1; p = .04; see Table 8.20) the higher the percentage of time spent in Balanced Language Situations, the higher the SS.. No interaction between Age and percentage of time spent in Balanced Language Situations was found (SSIX). Table 8.20: Fit of Different Polynomials (-2LL) for Changes in Story Structure (162 cases) with the Addition of Balanced Language Situations (BalLanSit) as Explanatory V ariable. Comparison -2LL Models ΔΧ2 Δdf p 4.37 1 .04 3.24 1 .07ns Model SSDEF: β0ijcons + β1Age1ij + 743.78 β2Age2ij SSVIII: SSDEF + β3BalLanSitij SSIX: SSVIII + β4Age*BalLanSitij 739.41 SSDEF vs SSVIII 736.16 SSVIII vs SSIX The difference in SS when the percentage of time spent in Balanced Language Situations was taken into account is graphically presented in Figure 8.15. There was a fixed main effect of the percentage of time spent in Balanced Language Situations, but there was no interaction effect between Age and percentage of time spent in Balanced Language Situations (see Table 6.10 in Appendix 6 for the estimated parameters). Pupils with higher percentages of time spent in Balanced Language Situations had higher SS scores.