In the USA, in the last decade, standards have been adapted for each grade level. These standards are annually checked by state-wide tests. The results of these tests often determine the school's funding and even the school's future existence. Due to this importance, a large amount of time is spent on teaching to the tests.Most teachers believe that this testing approach is detrimental to student education. This belief seems to be empirically supported by the fact that so far, the testing approach has not led to spectacular improvements promised by its proponents. While this empirical evidence is reasonably convincing, the teacher community has not yet fully succeeded in clearly explaining their position to the general public -because the opposing argument (of the need for accountability) also seems to be reasonably convincing.In this paper, we show that the situation becomes much clearer if we take uncertainty into account -and that, hopefully, a proper use of uncertainty can help in resolving this situation.What is "teaching to the test"? In the last few decades, in the US school education, state-wide tests have been developed for testing the mathematical knowledge of students at the end of each grade. Student performance on these state tests (and on similar tests in other disciplines) has become the most important criterion of how the performance of schools and teachers are gauged:• Funding of individual schools is largely determined by the test results. • In some cases, schools are disbanded and teachers are fired if the test results are unsatisfactory several years in a row. Because of the importance of the test results, schools are understandably paying a large amount of attention to making sure that the students pass these tests. In other words, instead of spending most of the time teaching the material -as it was in the past -teachers now spend a significant amount of time teaching "to the test".