Comparative Study on Indicators of Success for Non-Dominant Student Groups on the 8th Grade 2015 TIMSS Assessment

• In Event: Publication mentoring workshop session
  In Special Session: Early Childhood Edu / Assessment & Outcomes

Proposal

At the time of the conference, I will be a recent graduate. This study is the final part of my dissertation. The central purpose of this research study is to build new mathematical models of success for immigrant students and emergent bilinguals using the 2015 8th grade TIMSS data and identified indicators of success. The main research questions are:
1. What is the relationship between immigrant status and/or language use and performance on large-scale mathematics assessments?
2. How do different countries compare in terms of mathematics achievement scores for immigrants and/or emergent bilinguals?
3. In which ways do indicators of success affect immigrant students’ and emergent bilinguals’ performance on large-scale assessments?
The conceptual framework of this study builds upon some ideas about the achievement gap in mathematics. I chose to evaluate achievement within groups, but across countries, to alleviate concerns about normalizing each country’s dominant groups’ performance. Any study that compares the achievement of one group of students against another risks normalizing the dominant groups’ performance. This then results in viewing the oppressed group through a deficit lens in comparison (Gutierrez, 2008). Additionally, large-scale assessments, such as TIMSS, are intended to show how well students perform in certain subject areas, but are frequently used to evaluate the differences between groups, rather than within groups. This study does not intend to understand the nuances of the differences between groups, however it does aim to understand some of the more general indicators that affect mathematics achievement, which may be tied to race and social structure. In this sense, this data will not be used for “gap gazing” (Guitiérrez & Dixon-Román, 2011).
I am also building upon the literature within mathematics education that reflects on multilingualism, emergent bilinguals, and immigrant students in mathematics. There is some conflation between these ideas as many students whom are immigrants are also emergent bilinguals. As mathematics classes become more multilingual, the focus on language acquisition, rather than academic content, may play a role in the long-term achievement of immigrant students (Battey, Llamas-Flores, Burke, Guerra, Kang, & Kim, 2013). Generally, teachers of immigrant students hold a deficit view of their students’ abilities in mathematics classes (Civil, 2010). It would be amiss to assume that language was a hurdle and the only defining characteristic of emergent bilinguals and immigrant students. They “have a variety of unique linguistic resources at their disposal that they utilize productively to learn in their second language” (Langer-Osuna, Moschkovich, Norén, Powell, & Vazquez, 2016, p. 164). With the increasing reliance of discussion and justification in mathematics classrooms, emergent bilinguals can be active participants and contribute to classroom activities when the teacher allows them to use their first language for academic purposes. This positions them as knowledgeable resources in the classroom (Turner, Dominguez, Maldonado, & Empson, 2013).
To expand on this research, this study aims to find other, more specific variables that affect mathematics achievement on the TIMSS 2015 assessment in different educational contexts for immigrant students, emergent bilinguals, and for students who identify as both. This is important as most previous studies have focused solely on language acquisition rather than mathematical ability.
I used pre-existing TIMSS data and chose to focus on five countries for analysis: the United States, Canada, Italy, the United Arab Emirates, and Botswana. The first three were chosen as they had similar percentages of students in each of the target groups (immigrants and emergent bilinguals). The United Arab Emirates was chosen due to the high percentage of immigrant students, and Botswana for its multilingual context. Once countries were selected, I performed a factor analysis to build latent variables that could be used for analysis across all chosen countries. I then used those latent variables to perform hierarchical linear modelling using HLM software to build models.
At this time, the final analysis of the models is still incomplete. As this is the last part of my dissertation, I plan to finish within the next few weeks. As a preliminary analysis, I anticipate the models to be quite different for each country, as I expect each latent variable to impact assessment scores of students in different ways and these variables do not necessarily account for the educational context of each country.

Author

• Amanda E. Lowry, Rutgers University