作者: Herbert W Marsh , Adel Salah Abduljabbar , Maher M Abu-Hilal , Alexandre JS Morin , Faisal Abdelfattah
DOI: 10.1037/A0029907
关键词: Psychometrics 、 Trends in International Mathematics and Science Study 、 Psychology 、 Test validity 、 Discriminant validity 、 Social psychology 、 Measurement invariance 、 Mathematics education 、 Convergent validity 、 Academic achievement 、 Science education
摘要: For the international Trends in International Mathematics and Science Study (TIMSS2007) math science motivation scales (self-concept, positive affect, value), we evaluated psychometric properties (factor structure, method effects, gender differences, convergent discriminant validity) 4 Arab-speaking countries (Saudi Arabia, Jordan, Oman, Egypt) English-speaking Anglo-Saxon (United States, England, Australia, Scotland). In this article, also highlight methodological weaknesses TIMSS approach to these measures. We found reasonable support for within-group invariance across domains between-group full factor loading partial item intercept invariance. However, structure is complicated by strong negative-item effects correlated unique characteristics associated with use of items parallel wording. Correlations involving factors were reasonably similar countries, supporting both validity relation achievement, plans take more coursework science, long-term educational aspirations. differences largely favor girls Arab (with single-sex education systems) relative Anglo (and norms). The juxtapositions latent mean achievement are perplexing; students from had substantially higher than but lower all 8 factors.