Around the world, education is overwhelmingly conducted in the public sector and one of the central debates among governments is how to improve the quality of public schools (Kremer & Sarychev, 2000). Politicians and policymakers have proposed incorporating the private sector and market rules into the educational arena as a way to improve the quality of public education. Indeed, the introduction of market-based reforms to public education systems has not only been popular in the United States, but has also gained ground across Latin America, and especially in Chile. Proponents of market-based reforms in education argue that in the traditional system, principals and teachers do not face strong enough consequences for the poor performance of schools or their students. They believe that a competitive and test-based accountability system is more likely to improve social mobility and equalize the quality of education among families of different socio-economic backgrounds (Hursh, 2007; West, 1997)

The assumption is that implementation of standardized test will provide students, teachers and administrators an incentive to work harder as well as help identify struggling students and schools (Jacob, 2005).  However, accountability system often set objectives for improvement that are “simply impossible to meet by legitimate means” (Madaus & Clarke, 2001). In this context, teacher or schools administration are tempted to make inappropriate decisions when there is overstated emphasis on a single indicator and their job, salary or future is define by students’ performance in a standardized test (Nichols & Berliner, 2007). This research studies the impact of using a standardized test, as a single indicator, to measure the quality of education in the Chilean case. It asks whether or not the use of standardized test creates incentives for schools to use non-academic strategies, such as the exclusion of low-achievement students from the test-taking pool, in order to improve their performance on standardized tests.

I use a logistic regression model, with the dependent variable being (1) if the student does not take the standardized test, and (0) otherwise and after controlling for student’s annual attendance rate, family, teachers and school characteristics. The preliminary results of this analysis shows that students with lower grades a classroom have a higher likelihood to be absent during at least one of the days of the application of a standardized test; these results are statistically significant at .05 alpha level. This finding supports the information reported by teachers, academics and policy-markers about how schools manipulate the composition of students in the test-taking pool in order to maximize ratings in Chile, and it is coherent with similar findings reported by other scholars in the U.S. (Brian & Levitt, 2003; Cullen & Reback, 2006).