The COVID-19 pandemic resulted in significant disruption in schooling worldwide. Global test score data is used to estimate learning losses by modeling the effect of school closures on achievement by predicting the deviation of the most recent results from a linear trend using data from all rounds of PISA. Mathematics scores declined an average of 14 percent of a standard deviation, roughly equal to seven months of learning. Losses are greater for students in schools that faced relatively longer closures, for boys, immigrants, and disadvantaged students. Educational losses may translate into significant national income losses over time.
Over the last two decades, the Polish education system has been reformed several times, with the comprehensive structural reform in 1999, curriculum and evaluation reform in 2007, and early education reform introduced gradually until 2014. Student outcomes, as documented by PISA, but also other international assessments, largely improved over the last 20 years. Poland moved from below the OECD average to a group of top-performing countries in Europe. This chapter describes the reforms and research on their effects. It also discusses how it was possible to find political support for the reversal of changes that seemed to be highly successful. It provides three lessons from the Polish experience. First, the evidence should be widely disseminated among all stakeholders to sustain reforms. Second, the sole reliance on international studies is not sufficient. Additional investment into secondary analyses and national studies is necessary to develop evidence for better-informed political discussions. Third, some positive changes are more difficult to reverse. In Poland, increased school autonomy, but also external examinations, broader access to preschool and higher education, are among the changes that the new government could not alter.
This paper examines how including latent variables can benefit propensity score matching. Latent variables can be estimated from the observed manifest variables and used in matching. This paper demonstrates the benefits of such an approach by comparing it with a method where the manifest variables are directly used in matching. Estimating the propensity score on the manifest variables introduces a measurement error that can be limited with estimating the propensity score on the estimated latent variable. We use Monte Carlo simulations to test how the proposed approach behaves under distinct circumstances found in practice, and then apply it to real data. Using the estimated latent variable in the propensity score matching limits the measurement error bias of the treatment effects’ estimates and increases their precision. The benefits are larger for small samples and with better information about the latent variable available.