The results of both of the OECD-Programme for International Student Assessment (PISA) studies have called educational researchers´ attention to the educational success of immigrant students in the participating countries. Especially in Germany, students who immigrate together with their parents (first-generation immigrant students) perform much worse than native students and children of immigrants born in Germany (second-generation students) perform even worse than first-generation students. Research indicates that the differences in mean competencies in reading, mathematics and science between the aforementioned student groups correlate with the students´ social, cultural and economic background and their language abilities. But since PISA is designed as a sequence of cross-sectional studies it is very difficult to make causal inferences. Because of this the German PISA research consortium has conducted a supplement study to PISA 2003. This supplement study has been called PISA-I-Plus and has assessed the performances of the same 6,020 students in mathematics and in science in their ninth and tenth grade. Therefore, it was possible to analyze the immigrant students´ performance gains and to relate their competencies at ninth and tenth grade to social, cultural and economic factors and language usage. The results for science show that first-generation students learn the most, whereas second-generation students lag behind native students. In contrast, there were no significant differences between the groups' mean gain scores in mathematics. Multi-level-analyses show that social, cultural and economic factors have a significant influence on performances until the end of ninth grade, but lost much of their predictive power at the end of tenth grade. They also show that negative effects of a high percentage of immigrant students in school seems to be mediated by a low level of general cognitive abilities at the school level.

 

Both studies of the Programme of International Student Assessment (PISA) have shown lower mean competencies of fifteen-year old immigrant students in the OECD (e.g. OECD, 2006). Especially in Germany immigrant students´ educational success is considerably lower than for native students. The situation in Germany is – with the solitary exception of New Zealand – unique insofar as children of immigrants born in Germany (second-generation students) tend to perform lower in reading, mathematics and science than students who immigrate together with their parents (first-generation students). These performance differences correlate with differences in social, cultural and economic factors (e.g. family income, parental education, amount of cultural possessions). But since the studies of PISA are cross-sectional it is difficult to make causal inferences from these correlations. As a national supplement to PISA 2003 the German PISA research consortium conducted a repeated measurement study called PISA-I-Plus to – among other things – examine the conditions of immigrant students´ competency status and development in mathematics and science in grade 9 and grade 10.

 

 

The aims of PISA-I-Plus included the examination of the status, the development and the conditions of immigrant students´ competencies in mathematics and science in grade 9 and grade 10 in Germany. Three research questions were analyzed in relation to immigrant students´ performances:

 

In PISA-I-Plus the same 6,020 students´ competencies in mathematics and science were assessed in two consecutive years at the end of the German lower secondary school years (Sekundarstufe I). The assessment instruments were

-         a mathematics test consisting of 77 items in 2003 and 98 items in 2004,

-         a science test consisting of 63 items in 2003 and 28 items in 2004;

-         a general ability test consisting of 25 items;

-         a student questionnaire consisting of several scales assessing social, cultural and economic factors, language usage etc.

The data collected with these instruments were scaled using Item Response Theory models, e.g. the Rasch model, and marginal maximum likelihood parameter estimation technique. For the achievement data conditional Plausibe Values were computed, for the questionaire data Weighted Likelihood Estimates were computed. The analyses related to the research questions included methods such as the analysis of variance and multi-level modelling.

 

 

The results for the first research question show that the mean competency gain scores for the groups of first-generation students and second-generation students were essentially the same as for the group of native students in mathematics (24 to 28 points on the PISA scale; d = 0.31 to d = 0.36; differences between groups not significant at p = .05), although native students in general perform higher than immigrant students. The greatest mean difference in performance status was 55 points (d = -0.72; p < .01) at the end of grade 9. In science native students also perform higher than immigrant students. The greatest mean difference was 74 points (d = -0.89, p < .01). But in this domain the performance gains were significantly different between the student groups (p < .01): First-generation students learned the most (32 points on the PISA scale, d = 0.41) and second-generation students the least (12 points on the PISA scale, d = 0.16).

The second and the third research question were addressed by multi-level analyses. In the first series of analyses the performances in mathematics or science at the end of grade 9 were regressed on general cognitive abilities, social, cultural and economic indicators (e.g. ISEI) and indicators of language usage at the individual level. At the school level the mean level of general cognitive abilities, the mean and the standard deviation of the ISEI and the percentage of immigrant students in school were taken into account. These analyses confirm that lower values in social, cultural and economic factors are associated with lower competencies and that the performance differences between immigrant students and native students decline under their control. The results also show that a high percentage of immigrant students in school has not per se a negative influence on individual performance but seems to be mediated by a lower level of general cognitive abilities which is confounded with lower socio-economical status and immigrant status.

The conditions of performance gains were analyzed by a second series of multi-level analyses. At the individual level the performance in mathematics or science at the end of grade 10 were regressed on the performance in the same domain at the end of grade 9 in addition to the variables mentioned above. These analyses show that social, cultural and economic factors lost much of their predictive power for the competency gains. In addition, the effect of the percentage of immigrant students in school on competency gains seems to be also mediated by a lower level of general cognitive abilities in school.

 

 

The empirical findings emphasize the poor educational success of second-generation students in Germany which seems to get even worse at the end of the lower secondary school years. But they also confirm the more successful integration of the first-generation students in Germany. Social, cultural and economic factors really seem to be one cause of the poorer performances of these students, but cannot account for all differences. They even lost much of their predictive power for performance gains at the end of the lower secondary school years. The results also indicate that a high percentage of immigrant students has not per se a negative influence on individual development of competencies, because its effect seems to be mediated by confounding factors. Both results seem to signal that educational programs for immigrant students will not fail because of determining socio-economic differences.