During the last decades International Large Scale Assessments (ILSAs) have made great contributions to the understanding of how student background measured in terms of socio economic status (SES) contributes to student performance (see e.g. Broer, Bai, & Fonseca, 2019), and thereby extended the theoretical work in the area of social inequality in education (Lareau, 2011).

This research has to a large extent focused on the size of the relationship and how macro level features like level of homogeneity and centralization of the education system affects this relationship (Broer et al., 2019). More direct investigations of factors in the home that contribute to the relationship have been carried out on elements such as parents ability to help students with their homework (Bergem, Kaarstein, & Nilsen, 2016) or home environment support such as access to a desk, students own room and internet connection (Mullis, Martin, Foy, & Hooper, 2016). Elements that are more abstract includes how SES influences the forming of student’s language and hence their possibility to meet the language demands in school (Walzebug, 2014). Thus, previous research has had a focus primarily on the structural elements constituting the SES – student performance relationship, although it seems difficult to point at what exactly in the home that controls SES relationship with student performance (Allerup, Kristensen, & Torre, 2019).

This presentation takes on an opposite approach to the analysis of the relationship between SES and student performance. It focuses on how the student’s ability to solve different math problems relates to socioeconomic background and hence turns from focusing on what characteristics in the home that influences the relationship to which difficulties the student face when meeting different elements of the curriculum.

Within item response theory (IRT) differential item functioning (DIF) is usually used to investigate how different items in an assessment functions for different groups of students to make sure assessment is unbiased (Angoff, 1993). Besides DIF’s contribution to item development and testing, DIF has been used to investigate differences in performance due to students’ group belonging. These investigations have mostly concentrated on differences across cultures and gender. However, analyses of DIF in ILSAs have also been used to study differences between students due to diversities in student characteristics such as language, curriculum, gender and culture (e.g. Bundsgaard, 2019; Grover & Ercikan, 2017; Huang, Wilson, & Wang, 2016; Le, 2009; Oliveri, Ercikan, & Zumbo, 2013). Less emphasize seems to have been put on DIF with relation to students SES (see e.g. Grover & Ercikan, 2017; Walzebug, 2014), and none of these studies have investigated the content of the items showing DIF or made comparisons hereof between countries.

Using data from the mathematics assessment in the Trends in International Mathematics and Science Study (TIMSS 2019), this presentation will investigate the extension of DIF across different levels of students’ SES at grade 4 with a focus on Danish 4^th grade students, comparing the results to other Scandinavian countries. The presentation examines whether similar characteristics apply to the math problems that shows DIF for different levels of SES, and compares whether there are differences or similarities in the items that students with different social background find difficult when comparing to Norwegian and Swedish students.

Hence, the objective of the presentation is twofold. First to investigate to which degree DIF exists between different groups of students based on socioeconomic measures; and second, to investigate whether similarities exists in DIF for these groups of students across the Scandinavian countries.

The concept of differential item functioning (DIF) was developed as an alternative to the notion of item bias to avoid the negative implications of a test item functioning differently for different groups of test takers (Angoff, 1993) and can take different forms and strategies to detect DIF (Hanson, 1998). DIF analyses have been used to inform test construction in ILSAs like TIMSS, e.g. to avoid favoring one country over another or avoid gender bias in assessments (Zwitser, Glaser, & Maris, 2017). The presentation will model student responses from selected countries to the mathematics items from TIMSS 2019 as a generalized partial credit model and tests for DIF between different levels of SES using the mirt package in R (Chalmers, 2012). Items showing DIF for different levels of SES will be scrutinized for similarities in relation to content, including curriculum area, cognitive domain and language use in the item. However, as the project is in its initial phase, a final methodology for the study is still under development. The project will draw on experiences from the existing literature regarding DIF (including the works mentioned above) in developing and selecting a final model for DIF analyses. As seen in previous studies of DIF (e.g. Bundsgaard, 2019), a number of countries will be included to be able to make comparisons across education systems with differences and similarities. Comparisons of analyses will focus on Danish results and compare for similarities and differences with Norwegian and Swedish results. These countries are selected as they have some similarities in relation to the composition of SES among students and to some degree holds parallels in relation to their education systems and didactic traditions. Further, as different levels of language demands have previously been shown to be a part of the explanation for SES related DIF (Walzebug, 2014), comparing countries with languages from the same language family and a large pool of shared vocabulary is expected to ease the interpretation of similarities and differences related to language use in test items.

Initial analyses indicate that DIF can be found between different levels of SES to a degree that is higher than what must be expected by chance alone. Hence, the aim of the presentation is to make two contributions. First, it will provide a systematic overview of the extent to which DIF is an issue in relation to students’ socioeconomic status in the mathematics items of the TIMSS study. This has implication for both item development in ILSA studies and for the use of ILSA data for analyses of student performance across different SES groups. Second, it will provide relevant knowledge on the relation between students’ background and performance, investigating whether specific types of assignments or certain subject domains seems to be more prone to contribute to the strong and well documented SES-student attainment relationship, or whether the relationship is more general, and hence not related to subject content. This will have didactical implications for how students’ social background should be taken into account when developing a curriculum, as well as it is expected to affect how teachers should approach students from different social backgrounds when teaching different content. Together, these two points are expected to contribute to the general study of social inequality in education.

Allerup, P., Kristensen, R. M., & Torre, A. (2019). Deconstruction of the negative social heritage? – a search for variables confounding the simple relation between socioeconomic status and student achievement. Paper presented at the 8th IEA International Research Conference. Aarhus University. Angoff, W. H. (1993). Perspectives on differential item functioning methodology. In P. W. Holland & H. Wainer (Eds.), Differential item functioning (pp. 3–24). Hillsdale, NJ: Lawrence Erlbaum. Bergem, O. K., Kaarstein, H., & Nilsen, T. (2016). Vi kan lykkes i realfag: Resultater og analyser fra TIMSS 2015. Oslo: Universitetsforlaget. Broer, M., Bai, Y., & Fonseca, F. (2019). Socioeconomic Inequality and Educational Outcomes: Evidence from Twenty Years of TIMSS (Vol. 5): Springer. Bundsgaard, J. (2019). DIF as a pedagogical tool: analysis of item characteristics in ICILS to understand what students are struggling with. Large-scale Assessments in Education, 7(1), 9. doi:10.1186/s40536-019-0077-2 Chalmers, R. P. (2012). mirt: A multidimensional item response theory package for the R environment. Journal of statistical Software, 48(6), 1-29. Grover, R. K., & Ercikan, K. (2017). For Which Boys and Which Girls Are Reading Assessment Items Biased Against? Detection of Differential Item Functioning in Heterogeneous Gender Populations. Applied Measurement in Education, 30(3), 178-195. doi:10.1080/08957347.2017.1316276 Hanson, B. A. (1998). Uniform DIF and DIF Defined by Difference in Item Response Functions. Journal of Educational and Behavioral Statistics, 23(3), 244-253. doi:10.3102/10769986023003244 Huang, X., Wilson, M., & Wang, L. (2016). Exploring plausible causes of differential item functioning in the PISA science assessment: language, curriculum or culture. Educational Psychology, 36(2), 378-390. doi:10.1080/01443410.2014.946890 Lareau, A. (2011). Unequal childhoods: Class, race, and family life (2nd ed.). Berkeley: University of California Press. Le, L. T. (2009). Investigating Gender Differential Item Functioning Across Countries and Test Languages for PISA Science Items. International Journal of Testing, 9(2), 122-133. doi:10.1080/15305050902880769 Mullis, I. V., Martin, M. O., Foy, P., & Hooper, M. (2016). TIMSS 2015 International Results in Mathematics. Chestnut Hill, MA: TIMSS & PIRLS International Study Center, Boston College. Oliveri, M. E., Ercikan, K., & Zumbo, B. (2013). Analysis of Sources of Latent Class Differential Item Functioning in International Assessments. International Journal of Testing, 13(3), 272-293. doi:10.1080/15305058.2012.738266 Walzebug, A. (2014). Is there a language-based social disadvantage in solving mathematical items? Learning, Culture and Social Interaction, 3(2), 159-169. doi:10.1016/j.lcsi.2014.03.002 Zwitser, R. J., Glaser, S. S. F., & Maris, G. (2017). Monitoring Countries in a Changing World: A New Look at DIF in International Surveys. Psychometrika, 82(1), 210-232. doi:10.1007/s11336-016-9543-8