Abstract
This substantive-methodological synergy demonstrates evolving multilevel latent-variable models for cross-cultural data. Using Trends in International Mathematics and Science Study (TIMSS) 2007 data for U.S. and Saudi Arabian eighth grade students, we evaluate the psychometric properties (measurement invariance, method effects, and gender differences) of math self-concept, positive affect, coursework aspirations, and achievement. Extending the studies of the "paradoxical cross-cultural self-concept effect" largely based on U.S.-Asian comparisons, country-level differences strongly favored the United States for achievement test scores, but favored Saudi Arabia for self-concept and aspirations. Latent mean gender differences, of particular interest because of Saudi Arabia's single-sex school system, interacted with country for all constructs. The largest interaction was for achievement test scores; there were no significant gender differences for U.S. students (in coed schools), but in single-sex Saudi schools, Saudi girls performed substantially better than Saudi boys. Consistently with previous (mostly Western) research, but not previously evaluated with TIMSS, in each of the four (2 gender x 2 country) groups all three outcomes (self-concept, affect, and aspiration) were positively influenced by individual student achievement but negatively influenced by class-average achievement (the Big-Fish-Little-Pond Effect: BFLPE). BFLPEs were similar in size for boys and girls in coeducational (United States) and in single-sex (Saudi) classrooms.