BOOK OF ABSTRACTS AND EXTENDED SUMMARIES The Book of Abstracts and Extended Summaries is organised by type of presentation - paper, symposium etc. It is taken directly from the submission software, and thus does not include changes in titles, names or affiliations which have been made in the programme itself. To search this file, using the pdf search facility in the top mid-right of the screen. It should say Find and you can enter the name or a title and click on Find Next in Current PDF. PAPER PRESENTATION Bias factors related to math test performance of Ethiopian students in Israel Michal Levi- Keren, Tel Aviv University, Kibbutzim College of Education, Israel The present study examines bias factors which might grant a relative advantage to native Israelis over groups of immigrants in a mathematics achievement test administered to fifth grade students. The bias factor analysis process, aided by quantitative and qualitative data gathering methods, comprised three phases: (1) Using a psychometric procedure called DIF (Differential Item Functioning) to identify differential functioning in test items. DIF was examined in two differentiated groups of immigrants from the FSU and Ethiopia, which were compared to a group of native Israelis with the same ability level in the construct which the test aimed at measuring. (2) Identifying the sources of difficulty leading to DIF, while distinguishing between difficulties that are relevant to the construct measured in the test ("impact factors"), as opposed to those which are not ("bias factors"). (3) Determining whether the item incorporates bias factors. Phase 1 was based on the analysis of the results of a mathematics achievement test administered to a national, representative sample of native Israeli students and immigrant students from the FSU and from Ethiopia. Phases 2 & 3 were made through the judgment of subject-matter experts of those items found to be having DIF, as well as through interviews with immigrant students and "culture experts", who contributed a cultural perspective to the answers of the interviewees. The paper focuses on the results related to Ethiopian students, which support Vygotsky's socio-cultural approach, emphasizing the influence of culture on the formation of high mental functions, including mathematical thinking. The practical implications of these results are discussed. In recent years there has been a significant increase in the number of immigrants in many European countries, as well as in Canada, the US and Australia. Israel has absorbed immigrants from the day of its foundation in 1948, including, since the end of the 1980s, large immigration waves from the Former Soviet Union (FSU) and Ethiopia. Mass immigration, in Israel as abroad, tends to include large numbers of children, who, as they join their new country's education system, are forced to cope with many difficulties. In Israel, as elsewhere, follow-up research has been conducted to accompany these integration processes. Studies concerning the academic achievements of immigrant children paint a complex, multi-dimensional picture. While some indicate that immigrant students' achievements level out and , after years, may even come to exceed those of native students, there are other studies that reveal persistent academic achievement. In Israel, it was reported that the academic achievements of immigrants from the FSU and Ethiopia are lower than those of native Israelis, even long after immigration. Many studies have tried to uncover the sources of immigrant students' academic difficulties. They pointed at a complex array of student characteristics, on both the individual and group level. These may be demographic: relating to the number of years students have spent in their new country, to the socio-economic status of their families, to their school's socioeconomic composition, etc; other characteristics are culture-dependent and therefore relate to language, ways of thinking and values typical of the students' culture of origin. However, since the findings of these studies are not mutually consistent, it is impossible to generalize about the respective impact of these characteristics. Another important component contributing to immigrant students' academic difficulties, however, has to do with learning conditions, including the use of modes of assessment which were occasionally found to be incompatible with the students' learning needs. The assessment measures used supplied only meager information regarding the knowledge the students acquired in their countries of origin. Even the use of testing accommodations, specially developed to improve assessment validity and thus to help the immigrant students skirt these problems, was not always proven to be effective. The main innovation of this study lies in the attempt to explain immigrant students' difficulties with math tests by focusing on identification and conceptualization of the sources of difficulties they experienced with test items. This approach comprised three phases: (1) Using psychometric procedures to detect test items which are differentially difficult or easy, for two differentiated groups of immigrants from the FSU and Ethiopia which were compared to a group of native Israelis with the same ability level in the construct which the test aimed at measuring.. The measured phenomenon is called DIF: Differential Item Functioning. (2) Identifying the possible sources of DIF, namely, identifying the attributes of the items which caused the difficulty, bearing in mind that there are two kinds of difficulty: those which can be related to the construct being measured (math ability), and therefore reflect true performance differences (impact), as opposed to those which are not, and which are called bias factors .It is the latter type of factors which might distort test results and threaten the validity of the decisions taken on the basis of the test scores. (3) Determining whether the possible source of DIF is relevant to the construct being measured. Three research questions were formulated, reflecting the three above-mentioned phases: (1) Which items function differentially (i.e., which items are DIF items), regarding the Ethiopian and FSU immigrant students who were tested in math? (2) What are the characteristics of the DIF items that were found to favor the achievement of native born Israelis? (3) Which of those characteristics are bias factors and which reflect true performance differences? This study is based on a secondary data analysis of findings available from a comprehensive research conducted by Levin et al., in 2002[1]. In this paper, a research design combining quantitative and qualitative methods was used. The identification of DIF (question 1) was based on a statistical technique used to detect DIF items, called Delta-plot. For this purpose a stratified national sample of fifth graders was used, consisting of 530 native Israelis and 559 immigrants from the FSU and Ethiopia, who were tested in math, as part of the earlier mentioned research project. The identification of DIF sources (question 2) and the decision regarding their interpretation (question 3) were made through the judgment of subject-matter experts of those items found to be having DIF, as well as through interviews with immigrant students and "culture experts", who contributed a cultural perspective to the answers of the interviewees. The study results support Vygotsky's socio-cultural approach, which emphasizes the impact of culture on thinking processes, including mathematical thinking. The salient bias factors among immigrant students from the FSU originate in their unfamiliarity with certain test items, which proved to be alien to their ways of thinking; and also from a lack of sufficient attention to those item-related instructions which appeared in a different typographical format. Immigrant students from Ethiopia, however, tended to have difficulties in manifesting test taking skills differing from those relevant to their indigenous culture. Additionally, the answers they supplied to certain items, in which a concrete situation was presented, were rooted in their peculiar life experience, instead of being anchored in the items' data. This paper focuses on bias factors related to Ethiopian students, and discusses the theoretical as well as the practical implications of these results (such as providing testing accommodations), that may assist in generating further research aimed at the development of a fairer assessment culture, appropriate to Israel's current multi-cultural reality. [1] PAPER PRESENTATION Many questions, few answer(s)? A non response analysis of students’ evaluations of teaching. Bert Brockx, University of Antwerp, Belgium; Pieter Spooren, University of Antwerp, Belgium; Dimitri Mortelmans, University of Antwerp, Belgium Student evaluations of teaching (SET) are used at many universities to assess teaching quality. The validity and reliability is often discussed in research papers. Several studies that examine different aspects of SET report that a significant number of students do not return their evaluations. This can cause problems concerning the reliability and validity of SET. We do not know of any study that has examined the non-response issue in SET research. In the present paper, an exploratory study on student and course characteristics that might influence a student's decision (not) to participate in course evaluations is presented using multilevel binomial regression models. We find that some student characteristics significantly influence the fact that students complete their SET questionnaires (i.e. course grade, overall grade, gender, and first year of higher education). These findings might have implications for the interpretation of SET results and the use of SET in general. For example, the results show that good students (in terms of having received good grades) are more likely to complete evaluation questionnaires than students who received grades. All stakeholders should take this into account when drawing conclusions about teaching quality based on SET results. Introduction Student evaluations of teaching (SET) are seen as the most common means to assess teaching quality at universities (Apodaca & Grad, 2005). A massive amount of research papers have been published on SET (Richardson, 2005). A frequent problem with teacher evaluations is the fact that a significant number of students do not respond to the survey. Layne, Decristoforo & McGinty (1999), for instance, showed that response rates for were 47,8% for online course evaluations and 60,8% for in-class surveys. These results were confirmed by Dommeyer, Baum, Hanna & Chapman (2004). Due to rather low response rates it might be possible that a selection bias occurs. This could for example mean that students who perform well tend to respond more easily than students who receive low grades. If these students would also have different opinions about the teacher's performance, the selection that occurred could bias the results. It might mean that the results do not reflect the opinion of the complete population but of a selective sample of students. In this paper, we focus on the topic of non response within SET and examine whether course or student characteristics influence a student's decision to participate or not. Data The data for the present study were gathered at the University of Antwerp during the fall semester of 2008-2009 academic year by means of the SET -37 course evaluation questionnaire (Mortelmans & Spooren, 2009). 1224 online evaluations were sent to 598 students who were subscribed in 44 courses. As evaluations were administrated through an online system every student was given the possibility to evaluate the courses, opposed to an in-class surveys which requires full class attendance. The response rate was similar to the one reported by Layne, et al. (1999), namely 41,3%. Method The collected data have a cross classified hierarchical structure. This means that some students evaluated multiple courses by completing several evaluations. The most appropriate analysis method to analyze this data would be a cross classified multilevel binomial logistic regression analysis. However calculating a model with 49 teachers, 609 students, 1238 evaluations and several explanatory variables leads to estimation problems. Therefore, two separate multilevel logit models were estimated, with at the first level the questionnaires and at the second level the students (model1) and, in a second step, the courses (model2). The dichotomous dependent variable (‘response') was coded ‘1' if students participated in the evaluation of a course. Several explanatory variables were added to the model. At the questionnaire level the student's course grade was added, and whether the evaluation was about an elective course or not. At the student level, the overall grade, the student's gender and whether or not it was the student's first year of higher education were added. In the second model we added the course characteristics namely the class size and the required workload. Results and conclusion The results show that mostly the student characteristics influence the student's decision to participate in a SET survey. Table 1 shows that female students are more likely to fill in a questionnaire than their male counterparts, also good students (in terms of grades) tend to respond more likely than students who receive low grades. On the other hand, a negative effect is found for students who are in the first year of their university career. The course characteristics show no significant effects, a questionnaire of a course with high or medium workload is not more or less likely to be responded to, than one that required low workload. The same counts for courses with high, low or medium class sizes. The results of this study might have implications for the use of SET. All stakeholders should take into account that the student's performance and gender play an important role in the decision to respond to SET questionnaires. They thus should bear in mind that the sample of students who respond to the evaluations might not be a completely representative sample of the complete population. References Apodaca, P., & Grad, H. (2005). The Dimensionality of Student Ratings of Teaching: Integration of Uni- and Multidimensional Models. Studies in Higher Education, 30(6), 723-748. Dommeyer, C. J., Baum, P., Hanna, R. W., & Chapman, K. S. (Writer) (2004). Gathering faculty teaching evaluations by in-class and online surveys: their effects on response rates and evaluations [Article], Assessment & Evaluation in Higher Education: Routledge. Layne, B. H., Decristoforo, J. R., & McGinty, D. (1999). Electronic versus traditional student ratings of instruction. Research in Higher Education, 40, 221-232. Mortelmans, D., & Spooren, P. (2009). A revalidation of the SET37 questionnaire for student evaluations of teaching. Educational Studies, 35(5), 547 - 552. Richardson, J. T. E. (2005). Instruments for Obtaining Student Feedback: A Review of the Literature. Assessment and Evaluation in Higher Education, 30(4), 387-415. PAPER PRESENTATION The relationship between item type, students’ characteristics and media-effect in CBA Gyongyver Molnar, University of Szeged, Hungary; Krisztina R. Toth, German Institute for International Educational Research, Germany; Beno Csapo, University of Szeged, Hungary Paper-based assessment reached its limits, however the shift from paper-pencil (PP) to computer-based assessment (CBA) poses new questions. Validity issues regarding media-effect studies belong to one of the key areas. Several studies have been conducted with older students to measure validity (Csapo, Molnar & R. Toth, 2009), however, only a few focused on early childhood education (Choi and Tinkler, 2002). The purpose of this paper is (1) to study the media- effects in a cross-curricular competency field by first grade (age 6-7) students; (2) categorize item formats and students according to their media-dependence. Due to the young age of the target population, the instrument of the study consisted of figural, nonverbal items. The same inductive reasoning test was administered in PP and in CB mode (N=5156 and 313, respectively). The online data collection was carried out with the TAO platform. Student-level differences indicated media-effect, students' achievement was higher in PP (45.33%) than in CB format (32.6%; t=6.11, pThe highest media-effect was noticeable at items where the answer and distractors contained more figures. Students' socio-economic factors, gender and frequencies of computer-usage did not influence their CB test scores. Theoretical framework Technology has fundamentally changed the quality of education, including educational assessment. Technology-based assessment (TBA) opens new areas, raises new issues, offers new assessment methods (see Csapo, Latour, Bennett, Ainley, & Law, 2010), changes the whole assessment process and poses several new questions (see Scheuermann, & Bjßrnsson, 2009). It is no longer doubt that TBA will replace traditional paper-pencil (PP) testing. The transition from PP to TBA in educational context requires a step-by step procedure (Csapo, Molnar, & R. Toth, 2009); the first is the adequate control of media effect during testing. Several media effect studies were carried out in the past few years; however, only a few focused on testing very young students in technology-based environment. Aims Administering technology-based tests by very young students poses several challenges; therefore, special attention must be paid to control instrument validity and reliability. The purpose of this paper is to study the media effects in a cross curricular competency field by first grade (age 6-7) students, to control media effect and make detailed comparisons of test results delivered by different media. In this paper we (1) compare pupils' achievement in PP and computer-based (CB) environment; (2) categorise the item formats according to their media-dependence; and (3) characterise students who achieve better or worse in CB tests. Methodology The sample for the PP study was drawn from 1st grade students (age 6-7; N=5156), representative for the school population of Hungary; while the sample for the CB study was drawn from 1st and 2nd grade students (age 6-8; N=313). The instrument of the study was an inductive reasoning test, developed directly for young learners. It consisted of 37 figural, non-verbal items. The same test was used in PP and CB format. Special attention was paid to the paper and screen layout of the test, which were kept as similar as possible. The online data collection was carried out with the TAO (Testing Assistéê par Ordinateur) platform via Internet by using computers available at schools. The detailed item analyses were performed by using several means of classical test theory and IRT. Findings and conclusions The reliability index of the inductive reasoning test did not differ significantly in PP and in CB environment (Cronbach- a=.88 and .85, respectively). The comparison of PP and CB test scores indicates significant differences between the achievements based on the media. The average student achievement was higher in PP mode (M=45.33%, sd=20.07%) than in CB format (32.66%, SD=18.17%; t=6.11, p Regardless of the item type applied the delivery media caused significant achievement differences. However, depending on the used item format the amount of differences are changing. The highest media effect was noticeable at items where the answer and distractors contained more figures. The influential factor of the delivery media proved to be the strongest (t=8.24, p Regarding gender analyses, there were no achievement differences between the achievement of boys and girls in CB test results. Similar result was found at subtest level as well. Students' socio-economic factors (e.g. number of books, number of PCs, number of mobile phones) did not influence their CB test scores; similarly the common usage of computer and/or internet did not result in higher CB test results. The delivery media had a significant impact on first graders' achievement regardless of pupils' different background variable the measured cognitive structure, context and item format. The results suggest that if the goal was to develop equivalent summative assessment for the two media, studying the particular differences between the two media using different research condition and research design may support a developmental process towards the improvement of the validity of online assessment. Further research is needed to study the effect of media in other domains, to identify the differences of the cognitive processes relevant in the two media and for controlling the effects of other variables that were not the goal of these studies. Educational and scientific importance Paper-based assessment reached its limits. Existing models of assessment typically fail to measure the skills, knowledge and competencies that are increasingly important in the 21st century, and direct feedback is not available after the actual testing. Technology helps to solve this problem. Within a few years large scale international educational assessment programs plan to introduce computer-based assessments of student achievements, therefore, students are expected to be familiar with CB testing. This fact may facilitate the early use of CB testing at schools. However, most cases studies do not focus on pupils' CB testing in an educational context. In connection with media effect control, the detailed analyses of a comparison of PP and CB testing establish scientific bases to improve the efficiency, effectiveness and validity of computerized tests. References Csapo, B., Latour, T., Bennett, R., Ainley, J. éês Law, N. (2010). Technological Issues of Computer-Based Assessment of 21st Century Skills. Draft White Papter 3. Cisco, Intel, Microsoft. The University of Melbourne, Melbourne. Csapo, B., Molnar, Gy., & R. Toth, K. (2009). Comparing paper-and-pencil and online assessment of reasoning skills. A pilot study for introducing electronic testing in large-scale assessment in Hungary. In Scheuermann, F. & Bjßrnsson, J. (Eds.), The transition to computer-based assessment. New approaches to skills assessment and implications for large- scale testing. Luxemburg: Office for Official Publications of the European Communities. pp. 113-118. Scheuermann, F. & Bjßrnsson, J. (2009, Eds.). The transition to computer-based assessment. New approaches to skills assessment and implications for large-scale testing. Luxemburg: Office for Official Publications of the European Communities. PAPER PRESENTATION Test follow-up: developing productive synergies between summative and formative assessment David Carless, University of Hong Kong, Hong Kong The main aim of this paper is to explore how teachers derive formative instructional power from internal school tests. It analyzes potentials and constraints in seeking to enhance student learning capacities from test processes which carry both summative and formative purposes. The paper uses qualitative case study data from six teachers in three elementary schools in Hong Kong involving 31 classroom observations and 26 interviews with teachers. The observations were mainly of test follow-up lessons and interviews probed issues arising from them and other themes related to the interface between testing and student learning. The findings document two particular test follow-up strategies: student self-evaluation of performance on a test; and peer co-operative processes. A common thread across both of these is that they involve active student participation in the post-test learning process. The significance of the study lies in its focus on building productive synergies between summative and formative assessment. Test follow-up can be used to show that test results can look forward to future learning, as well as summarizing previous performance. Aims Managing the relationship between summative and formative assessment is a major challenge for the enhancement of the learning aspects of assessment. The power and influence of summative assessment often undermines the learning potential of formative assessment. The development of productive synergies between these two functions of assessment represents a potentially powerful way forward. Within this theme, the particular strategy discussed in this paper is post-test follow-up as a means to develop ongoing student learning capacities. The main aims of the paper are: To explore how elementary school English teachers in three selected Hong Kong schools carry out learning-oriented follow-up to internal school tests; To contribute to the development of assessment theory by exploring how summative assessment can be used formatively to support student learning; To analyze the role of context in formative assessment and suggest some features of contextually grounded assessment practices. Theoretical framework: Formative use of test data The framework for the paper is based on work probing the relationship between formative and summative assessment. Formative assessment is defined as the elicitation and interpretation of evidence, so as to enhance instruction and improve student learning. The formative assessment strategy discussed in this paper is the formative use of a test designed principally for summative purposes (Black et al., 2003), hereafter test follow-up. Test follow-up uses information from tests to advance student learning, so has the potential to contribute to a positive relationship between summative and formative assessment. For test follow-up to be exploited optimally it requires the right kinds of tests: those that are aligned with curriculum aims; those that are instructionally sensitive (Popham, 2008); and those that promote mastery as well as performance (cf. Dweck, 2000). There are a number of challenges for the implementation of test follow-up: frequent classroom testing risks making judgments about student learning rather than suggesting means to improve (Black et al., 2003); it can result in micro- teaching of ways to gain additional marks rather than developing student learning power (Stobart, 2008); and teachers are generally more effective at drawing inferences about student levels of understanding than they are in planning future instructional steps (Heritage et al., 2009). Methods The research questions guiding the study are: What test follow-up strategies do the teachers carry out? What is the rationale, strengths and limitations of these test follow-up strategies? The research involved multi-site qualitative case studies of teachers implementing test follow-up. Case studies are a particularly useful research strategy to probe the context-dependent nature of classroom implementation of learning activities. The case studies were carried out in English as a foreign language classes in elementary schools in Hong Kong. The unit of analysis in the study was the teacher. Six teachers from three schools were involved over a three-year period. Data sources comprised classroom observations and interviews. Observation is the most direct way of understanding classrooms and so a series of lesson observations were conducted before and after tests. A customized observation schedule was developed to facilitate the collection of relevant information, including: the relationship between lesson content and items in tests; student learning progress and difficulties; and identifiable test follow-up practices. Semi-structured interviews were conducted with teachers to identify their views about assessment and its relationship with classroom practice. Baseline interviews with participant teachers served to understand assessment practices in the school. Post- lesson interviews probed teacher perceptions of relevant issues occurring in observed lessons. The data-set comprises 31 lesson observations and 26 interviews with teachers. Data were analyzed inductively using standard qualitative data analysis procedures, including coding, categorizing and respondent validation. Findings Classroom observations revealed various test follow-up strategies. Those which carried potential for enhancing student learning mainly involved students as active participants in test follow-up. Peer co-operation was a key element and was exploited in various ways: students working in pairs or small groups to work out answers to challenging test questions or develop correction sheets; and students sharing test preparation or test-taking strategies. Another potentially useful strategy was engaging students in self-evaluation through written reflections on test performance: what they did well, what they were less successful in and how they wanted to improve. Whilst some students wrote relatively trivial comments, such processes had potential to remind students that a test was not just an end-point but part of ongoing learning. Teachers' potential to implement test follow-up effectively related to a complex interplay of multiple factors, including: their background, training and experience; beliefs and understandings related to testing and formative assessment; the extent of satisfaction with existing practices; and the pedagogic priorities in their own school context. Underpinning these elements were various contextual and cultural factors, pertaining to the Confucian-heritage setting of Hong Kong: the expected roles of teachers and students; the dominance of examination-oriented education; and collectivist notions encouraging positive co-operation with peers. Significance The theoretical significance of the work relates to the development of productive synergies between summative and formative assessment. By acknowledging the centrality of summative assessment in teachers' and students' lives, but deriving much needed formative impetus, test follow-up can be used to show that test results can look forward to future learning, as well as summarizing previous performance. The study also casts light on the interplay between assessment and educational context via the notion of contextually grounded formative assessment practices which acknowledge the realities of the host setting. Such an orientation honors the existing beliefs of teachers and the associated classroom practices, grounded as they are in a particular socio-cultural setting. It takes a pragmatic view of what is feasible in a specific context, such as Hong Kong, in view of the dominance of examination- oriented education. This starting-point can act as a seed for further development. Finally, the study also carries implications for practice. The findings suggest two main interlinked keys to test follow-up: a) the importance of active student involvement; and b) engaging students in reflecting on their work and performance, including peer learning and self-evaluation. References Black, P., Harrison, C., Lee, C., Marshall, B., and Wiliam, D. (2003). Assessment for learning: Putting it into practice. Maidenhead: Open University Press. Dweck, C.S. (2000). Self-theories: Their role in motivation, personality, and development. Lillington, NC: Taylor & Francis. Heritage, M. et al. (2009). From evidence to action: a seamless process in formative assessment. Educational Measurement: Issues and Practice, 28 (3), 24-31. Popham, W.J. (2008). Transformative assessment. ASCD: Alexandria. Stobart, G. (2008) Testing times: The uses and abuses of assessment. London: Routledge. PAPER PRESENTATION What do we know about the digital divide in students with an immigrant background? Oliver Walter, Germany; Martin Senkbeil, Leibniz Institute for Science Education, Germany In this paper we analyze the ‘digital divide', i.e. disparities in the access to digital media, frequency of computer use, and computer literacy, with regard to students with an immigrant background in Germany. Since ICT competencies have an increasing relevance to young people's opportunities on the labour market, especially for students with an immigrant background, we studied the amount of differences on all three dimensions of the digital divide in a sample of 8,874 ninth-graders in Germany. In doing so we distinguished between students of different immigrant generations and different ethnic origin. Our results show increasing levels of computer competency with ascending immigrant generation. With respect to ethnic origin we found deficits for some (e.g., students from Turkey), but not for all migrant groups (e.g., students from Poland). After identifying these disparities, we related them to indicators of structure and process characteristics of the students' family, e.g. socioeconomic status. We found that disparities in computer skills between different groups of immigrants and natives can largely be traced back to differences in access to digital media and to family characteristics. Interestingly, there were different effects of these factors for different immigrant groups. Our results suggest that schools should be aware of the different patterns of the digital divide with respect to immigrant generation and ethnic origin and should try to compensate for them by making an adequate educational offer. Aims Many studies have been conducted on the ‘digital divide' with regard to social inequalities since the early 1990s. But they often focused only on access to digital technologies and neglected other dimensions as frequency of computer use and computer literacy. Only some research has focused on migrants, especially in Europe. In addition, the existing research on migrants and the digital divide rarely distinguish between different migrant groups (e.g., according to immigrant generation and ethnic origin) and take little notice of differences due to family structure (e.g. socio- economic status, parental education) and family process characteristics (e.g. family language, family communication) into account. Therefore, our concern was to study access, use and knowledge of computers and the internet of different groups of immigrant students in Germany and to relate our findings to structure and process characteristics of their families. In particular, we were interested in the following questions: (1) What is the digital divide like between natives and students of different immigrant generations? (2) What is the digital divide like between natives and students of different ethnic origins? (3) Do disparities in computer literacy stem from differences in the access to digital media and from family structure and process characteristics? Methodology For answering our research questions we used data of 8,874 9th graders participating in PISA 2006 in Germany. Immigrant background was defined by the students' and parents' country of birth. We distinguished the following groups with respect to immigrant generation: (a) students without an immigrant background: The student and both parents were born in Germany (N = 6,950). (b) 2.5th generation: The student and one parent were born in Germany (N = 491). (c) 2ndgeneration: The student was born in Germany, both parents were born abroad (N = 731). (d) 1st generation: The student and both parents were born abroad (N = 687). Additionally, we used a second classification differentiating between students whose parents were born in the former Soviet Union (N = 441), Turkey (N = 424), Poland (N = 230) and other countries (N = 801). Computer literacy was measured by a paper-pencil-test consisting of 14 items (a = .78). The other dimensions of the digital divide and the family background of the students were assessed by questionnaire. Findings Table 1 shows the results concerning research questions 1 and 2. With respect to computer access there are only some minor differences to students without an immigrant background, especially for students from the 1st and 2.5th immigrant generation. More pronounced disparities can be found for internet access. There, our results show that 1st generation students have substantially lesser access and that the disparities decreases over generations. In contrast to access we did not find significant lesser differences in frequency of computer use at home and at school. Instead 2nd generation students use computers a bit more often than natives at home. If we look at computer use outside both school and home we also see that the computer is even more often used by students of every immigrant generation than by natives. Even so, 2nd and 1st generation students show distinct deficits in computer proficiency, on average. In contrast to the results for immigrant generation there are greater disparities in access to digital media between groups of different ethnic origin. With the exception of students whose parents were from Poland, immigrant students have significantly lesser access to digital media than natives. Especially a substantial proportion of students from the former Soviet Union and from Turkey do not have access to the internet. The findings for frequency of computer use are similar to the ones we found for students from different immigrant generations: There are no differences in the frequency of use at home and at school, but significantly more students whose parents came from Turkey and other countries use computers outside school and home than natives. In accordance with findings reported above and with the exception of Polish students, students with another than a German origin have lower computer proficiencies, on average. With respect to research question 3, we found that differences in computer proficiency between groups of immigrant and native students can mostly be traced back to access to digital media, family structure and family process characteristics, although these factors only explain less than 10 percent of the total variance (see table 2 for results with respect to ethnic origin). More detailed analyses point to different mediating mechanisms regarding the development of knowledge gaps between natives and different immigrant groups. Whereas access conditions as well as family structure and process characteristics contribute significantly to explain the disparities for first-generation students, only structure and process characteristics do so for 2nd generation students. The results are similar if we categorize students depending on their parents' country of origin: Family structure and process characteristics are important for Turks, access conditions and family process characteristics are important for those from the former Soviet Union. Theoretical and educational significance Our results have some practical implications for the promotion of computer literacy through school activities. Schools have a responsibility to teach students how to competently use a computer and the internet so as to guarantee each student equal opportunities in reaching their future vocation or profession. Schools could, for example, specifically advise parents with an immigrant background, depending on their origin and their children's specific deficits in dimensions of the digital divide, to invest in a computer and internet access at home, or if this is not possible, to provide this through access at school, a library or an internet cafe. Schools should ensure that Turks and second- generation immigrants receive the necessary computer skills. Although these students claim to use a computer frequently they seem not to be able of acquiring sufficient skills on their own. As the parents of Turkish students often lack the necessary education and cognitive skills, schools must take on the role of a mediating agent when it comes to conveying computer-oriented skills. PAPER PRESENTATION Multilevel analysis on quadratic relationship between socioeconomic status and math performance Ningning Zhao, UGent, Belgium The purpose of the present study is to explore, in a large sample of Chinese primary school students, the relationship between family socioeconomic status and mathematics performance on the base of a multi-level analysis. A rather weak relationship is found between socioeconomic status and performance in the Chinese context. The relationship follows a quadratic curve, implying that students from a disadvantaged family and higher socioeconomic background have a higher probability to attain higher mathematics scores. This can be explained on the base of Chinese cultural beliefs about exams and social class mobility. Moreover, the average socioeconomic status at the school level seems to moderate in the relation between individual SES and academic performance. This suggests that individuals from a disadvantaged family will benefit more from the school with a higher average higher family socioeconomic status than other students who are enrolled in schools with a lower average family socioeconomic status. 1. Introduction The relationship between socioeconomic status (SES) background and academic performance has received ample attention since the publication of the "Coleman Report" in 1966 (Coleman, Campbell, Hobson, McParland, Mood, Weinfeld and York, 1966). A consistent finding is that students with a high family SES perform better than students with a lower SES. As shown in the Third International Mathematics and Science Study (TIMSS), the influence of family socioeconomic status on educational performance appears to vary systematically depending on the economical development level of a region (Schiller, Khmelkov and Wang, 2002). Whether the relationship between SES and mathematics performance is different in other countries (developed or developing, various cultural value) is still an open question. This brings us to the research aims of the present study: (a) to construct a comprehensive SES index based on input from previous studies; (b) to explore the general relationship between SES and mathematics in P. R. China, considering different developmental levels of the region; and (c) to analyze the extent to which aggregated SES at the school level influences student mathematics performance, regardless of students different individual SES levels. Multilevel analysis was applied to study the impact of variables at the school, class and student level on mathematics performance. 2. Theoretical background 2.1 Measurement of socioeconomic status SES which was approached in the 1980s stress family income, father's educational level, mother's educational level, and father's occupational status or occupation type (White, 1982). In later studies, additional variables were added to the index; e.g., home resources (Sirin, 2005), home atmosphere or context, number of books in the household, and other resources related to the learning (Caldas and Bankston, 1997; OECD, 2003). 2.2. Varying impact of SES on mathematics performance 2.2.1 Family SES and academic performance Recently, the meta-analysis of both White (1982) and Sirin (2005) reveals that the direct relation between socioeconomic status and performance might be less strong. White's meta-analysis claimed that the meta-analysis reveals that the average correlation is .299 while Sirin's meta-analysis claimed a correlation value of.343. 2.2.2 SES and performance: the moderating effect of school aggregated SES variables Previous studies show that higher levels of an aggregated school SES is related to an increase in student performance and in students with a different level of family SES (Perry and McConney, 2010). 3. Method 3.1 Sampling These twenty schools are located in five Chinese regions reflecting different development levels; and are located in a rural or urban setting. Total school enrolment ranged from 318 to 897 students (M=547.95, SD=140.19). Sampling strata were based on the location of the school in a specific region. 3.2 Variables 3.2.1 Dependent variables Mathematics performance level. All the items and cases were calibrated on a continuum scale ranging from grade one to six by Item Response Theory (IRT) programme of BiLog (See Authors, revising). Reliability of the scales (Cronbach's a) was reported to be high: for grade one to grade six respectively .94, .96, .95, .94, .94 and .93. Reported means are: -1.24, -.89, .05, .18, .69, .83. 3.2.2 Independent variables 3.2.2.1 Individual learner's level SES In a first step, an exploratory factor analysis (EFA) was carried out by applying the WLSM method that is able to deal with eight items for the family information as categorical data (Mplus5.1). The results suggest a two-factor structure in the SES variable. A first factor grouped SES variables focusing on parents' occupation status; a second factor groups SES variables in relation to family wealth. 3.2.2.2 School level SES Two aggregated SES indexes were calculated at the school level: (1) the average parent's socioeconomic status of the learners attending this school (SCFSES_J) and the average level of wealth of the children in the school (SCFSES_W). 4. Results 4.1 Multilevel analysis of the relationship between SES variables and mathematics performance 4.1.1 Weak relationship between SES variables and mathematics performance These models only account for 0.41% of the variance in mathematics performance at the individual learner level (.486 vs. .484) (compare model 0 to model 6 in Table 4). This implies that the SES variables under study are not strong predictors of mathematics performance in primary school after controlling for school level variables. 4.1.2 The U-shaped relationship between SES variables and mathematics performance Another interesting finding from the study is the U-shaped relationship between SES variables and performance in the primary school. This implies that the students with lower SES in China do not always attain lower performance levels, while students with a higher SES level do not always attain higher performance in primary schools. The performance of the students depends on the corresponding family SES level of the student. 4.2 SES and mathematics performance: the moderator effect of school level aggregated SES indexes 4.2.1 Stronger Effect of School level of FSES_J on mathematics performance In sum, when means of SES at school level are entered into the models, about 22.02% of the total variance in mathematics performance can be explained (.227 vs. .177 see model 8 in Table 4) at the school level. 4.2.2 The moderator effect of school SES on mathematics performance Figure 2 shows how the mathematics performance of learners with a different level of SES, based on their parents job level (recoded in three categories), varies according to the aggregated SES variable at school level. The slope of the disadvantaged group with lower individual family SES (FSES_J_R = 1) is less steep as compared to the middle group (FSES_J_R = 2) and the advantaged group (FSES_J_R = 3). This implies that the moderating effect is stronger for learners with a higher individual SES. In general, the higher the school aggregated SES, the higher performance of learners with a higher family SES. 5. Discussion and conclusions A variety of rationales can be presented to explain these specific findings based on data from the Chinese context. Although the cultural values and political management policies can have a positive impact, it remains nevertheless clear that disadvantaged students run a higher risk for encountering learning difficulties. PAPER PRESENTATION Assessment of adult’s mathematical competence and the use of mathematics in work and daily life Timo Ehmke, University of Lueneburg, Germany; Christoph Duchhardt, Leibniz Institute for Science and Mathematics Education, University of Kiel, Germany; Eva Knopp, Leibniz Institute for Science and Mathematics Education, University of Kiel, Germany; Irene Neumann, Leibniz Institute for Science and Mathematics Education, University of Kiel, Germany The Programme for International Student Assessment (PISA) has drawn attention to the role of mathematical literacy as an important prerequisite for lifelong learning and active participation in society and culture. Many everyday situations require mathematical abilities for sound judgements or important decisions. The aim of this study is to find out how often adults use mathematics in their work and daily life and how this is connected to their mathematical competence. The sample comprises N = 461 adults from Germany (age: M = 44.2 years, SD = 12.2 years). The testing took place by individual sessions in the adults' personal home environment. A Latent-Class-Analysis identified three classes of adults which differ in their profiles of using mathematics in work and daily life. Regression analyses on the adults' mathematical competence showed statistically significant effects for age (beta = -0.19), socio-economic status (beta = 0.26), educational attainment (beta = 0.18) and the interaction between gender and educational attainment (beta = 0.57). Adding the variable "use of mathematics" to the regression model resulted in an additional effect mathematics user profile classes. The findings showed that the use of mathematics in work and daily life is a strong predictor for mathematical competence. This effect is stable even when characteristics of the social background are controlled. Further implications for the life-long learning in mathematics with regard to limitations of the study (sample bias and cross sectional design) will be discussed. The Programme for International Student Assessment (PISA; OECD, 2003) has drawn attention to the role of mathematical literacy as an important prerequisite for lifelong learning and active participation in society and culture. Many everyday situations require mathematical abilities for sound judgements or important decisions. Mathematical literacy thereby emphasizes the knowledgeable use of mathematics as a tool for solving problems in real-world situations. The aim of this study is to find out how often adults use mathematics in their work and daily life and how this is connected to their mathematical competence. More concrete we follow two research questions: (1) Which profiles can be identified concerning the use of different mathematical concepts in work and daily life? (2) Of what kind is the relationship between the adults' use of mathematics in work and daily life, their social status and their mathematical competence? The sample comprises N = 461 adults from Germany (age: M = 44.2 years, SD = 12.2 years; gender: 55 percent women and 45 percent men). It was taken within a pilot study of the National Educational Panel Study (NEPS) in Germany (Blossfeld, Schneider & Doll, 2009). The testing took place by individual sessions in the adults' personal home environment. Test administrators who had been trained in advance conducted the test according to standardized guidelines. Each session took 90 minutes, and the participating adults were rewarded with a sum of 20 Euros. The test time was divided into 60 minutes for a paper-and-pencil mathematics test and 30 minutes for a questionnaire. The