A cross-cultural analysis of the dimensions of mathematics-related affect: assessing the psychometric properties and the relationship with achievement

Emmanuel Adu-Tutu Bofah

    Research output: ThesisDoctoral ThesisCollection of Articles

    Abstract

    The aim of the present set of studies in this dissertation was to examine the psychometric properties of measures of mathematics-related affect developed and normalized in one culture for use in another, how these properties transcend cross-culturally in an African context, and the methodological challenges associated with the process. Further aims were to examine the relationships between these constructs on a cross-cultural level, and to explore any associations be-tween students background variables and mathematics achievement. With these aims in mind, we conducted four original empirical studies based on different types of structural equation modeling. Studies I and II explored the problems of importing an instrument from one culture into another, and the associated methodological challenges. More specifically, Study I gives a detailed account of the processes involved in applying structural equation modeling to validate mathematics-related affective measures developed in one culture (Finland) for use in another (Ghana). Reliability estimates and confirmatory factor analyses indicated that the Ghanaian data set did not fit the original hypothesized model (seven-factor structure). A series of factor and confirmatory factor analyses indicated a four-factor structure for the Ghanaian sample. Study II examined the possible causes of the differences in the factor structures from a cross-cultural perspective. The results indicate that measurement artifacts, cultural differences, and construct validity and adaptability were possible causes of the observed differences in factor structure between the Ghanaian and the theoretical model. In conclusion, it is suggested that re-searchers should be aware of construct importation and adaptation, and of the fact that measurement errors, question order, negatively worded item, translation, and content overlap may influence the reliability and validity of survey measures. Moreover, it is necessary to consider cultural variation and the methodological approaches involved in the theoretical settings in order to make any meaningful comparative assessment. Researchers focusing on cross-cultural mathematics-related affect are recommended to acquire the theoretical and practical knowledge necessary to address these issues using appropriate tools such as structural equation modeling. Study III investigated the psychometric properties (factor structure, reliabilities, method effect, and measurement invariance country and gender) of the mathematics-related affective constructs used in the 2011 Trends in International Mathematics and Science Study (TIMSS 2011) across the five participating African countries. It also examined the relationship between these mathematically related affective constructs, as well as the associations amongst the constructs, and between the students background variables and mathematics achievement cross-culturally. The results empirically support the multidimensionality of the construct, and the measures were largely invariant across the five educational systems and gender. There was also some evidence to suggest that negatively worded items in a construct can attenuate the reliability and validity of the measures: a series of confirmatory factor analyses (CFAs) revealed the need to control for the method effects associated with such items. It seems that responses to negatively worded items differ systematically across countries and are systematically linked to students achievement. The analyses also covered the relationships between the student affect, achievement, and background variables such as parental education, gender and students educational aspirations. Lower mathematics achievement was associated with students from countries reporting higher motivational belief. On the association between students mathematics-related affect, achievement, parental education, educational aspirations and gender, it seems that there are culture-specific as well as cross-cultural universal outcomes. For instance, the relationship linking parental involvement, teacher responsiveness and mathematics achievement turned out to be culture-specific whereas, boys rated their mathematics competence more highly than girls did. Parental education, gender and long-term educational aspirations also influenced student achievement and motivation, but to different extents in different countries. Study IV, based on non-recursive structural equation models, tested theoretical and methodological models of the reciprocal relationship between mathematics-related affect (e.g., self-concept) and achievement using the TIMSS 2011 cross-sectional data set. The results in different countries support the existence of a unidirectional influence of affect on achievement and of achievement on affect, and a direct feedback-loop relationship between affect and achievement. According to the evidence, the reciprocal determinism between affect and achievement is dependent on the national context. Moreover, there was a cross-cultural universal pattern among males: reporting a stronger mathematics self-concept and higher long-term educational aspirations significantly predicted higher mathematics achievement. The effects of socioeconomic status on mathematics achievement and self-concept among the students were dependent on the national context.
    Original languageEnglish
    Place of PublicationHelsinki
    Publisher
    Print ISBNs978-951-51-2055-7
    Electronic ISBNs978-951-51-2056-4
    Publication statusPublished - 2016
    MoE publication typeG5 Doctoral dissertation (article)

    Fields of Science

    • 516 Educational sciences

    Cite this