How Central Exams Affect Educational Achievement: International Evidence from TIMSS and TIMSS-Repeat.

This study asserts that central examinations need not necessarily be those administered by the government; they can be examinations centralized at some administrative level, but in any case, they are external to the school. This study also assessed the impact of central examinations on the functioning of the education system and students academic performance using data from the Third International Mathematics and Science Study (renamed Trends in International Mathematics and Science Study) (TIMSS; 1994-1995) and the TIMSS-Repeat (1998-1999). Data on nationally representative samples of middle-school students are available from 39 countries for the TIMSS and 38 for the TIMSS-Repeat, with 23 countries participating in both studies. In all, about 450,000 students were represented. The international evidence of both TIMSS studies confirms that central examinations are a powerful accountability device. Student performance in mathematics and science is substantially higher in school systems with central examinations than those without central examinations, and this is true for students from all performance quartiles and family backgrounds. Parents, administrator, schools, teachers, and students all appear to respond favorably to the changed incentive environment created by centralexams by behaving more favorably to students educational achievement. Parental involvement grows more informed and effective, and opportunistic behavior of schools and teachers is curbed. An appendix discusses construction of the TIMSS Repeat database. (Contains 4 figures, 9 tables, and 32 references.) (SLD)