Does lowering the bar help? Results from a natural experiment in high-stakes testing in Dutch primary education

In many countries, high-stakes tests play an important role in the allocation of pupils to prestigious tracks or schools in secondary education or students to prestigious programs or colleges in tertiary education. It is not clear what would happen if the standards for these tests were systematically raised or lowered. Would that affect the subsequent educational career? This paper exploits a unique natural experiment in the Netherlands using the market entrance of two new suppliers of high-stakes tests in primary education. In the first year of introduction, these new tests were not yet properly calibrated: For one test the standards were too low, while for the other test they were too high, compared to the standards of the traditional test that continued to be the main supplier. We use high-quality register data and a within-schools-across-cohorts design to model the short- and long-term outcomes (i.e., change in teacher advice and actual track three years later) for the students that were affected by the new tests. We find evidence for short-term effects, but no evidence for long-term effects. This implies that the Dutch educational system is sufficiently flexible to allocate pupils to the appropriate track, even if a high-stakes test advice does not recommend the right track. At the same time, it also implies that lowering the bar is not a simple way to increase the share of students going to prestigious tracks