Page 79 - Second language development of newly arrived migrant kindergarteners - Frederike Groothoff

Development of receptive vocabulary 79 in fit between consecutive models are presented in the tables with the Likelihood Ratio Tests and expressed by -2 log likelihoods. In the subsequent sections and chapters, the General Development Model will be extended to include the child characteristics and the characteristics of the school learning environment. The explanatory variables in this study are added separately one by one, and not in combination, because this way we can make our models more reliable. Due to our sample size with a small number of participants and a limited number of observations per participant we need to be extra careful. 4.4 Results The results of the receptive vocabulary development are presented as follows: first, the developmental data from PPVT is presented to answer the question: How does the receptive vocabulary (in Dutch) of newly arrived migrant kindergarteners develop during the first two-and-a-half years of schooling in the Netherlands in relation to school type? The basis of the analysis is the General Development Model, which includes only Age at which receptive vocabulary was measured as an explanatory variable. Second, the variables Exposure to Dutch at School and Educational Facility are introduced into the model. To describe the development of the receptive vocabulary scores, several models were fitted. Models will be built but no tables with mean scores will be presented since the participants differ in too many aspects. That is, an overview of descriptive statistics will not be informative, whereas the models will be. In the tables that are presented, the increase in fit in relation to the previous model is shown. From the comparison between the consecutive models (see Table 4.3) 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 (PPVT1) = 125.87; df = 1; p < .001). The variance within individuals depends on the age of participants (ΔΧ2 (PPVT2) = 24.54; df = 2; p < .001). The variance between individuals is a (linear) function of age as well (ΔΧ2 (PPVT3) = 4.00; df = 1; p = .045). Finally, adding Age2 to the model improved the fit significantly (ΔΧ2 (PPVT4) = 8.18; df = 1; p = .004). Hence, in the final model (PPVT4) a fixed effect of Age1 and Age,2 as well as variance within and between pupils components which depends on Age needs to be included, and with this model we continued the analysis.