Does Private Schooling Affect Non-Cognitive Skills? International Evidence Based on Test and Survey Effort on PISA
*Names in bold indicate Presenter
Recent research has also shown the possibility of building proxy measures for these important non-cognitive skills by measuring student effort in achievement tests and surveys. Borghans and Schils (2012) used the fact that students participating in PISA were randomly given different test booklets to quantify the rate of decline in performance in the test as the test progresses, independent of question difficulty. The authors found that cross-country differences in motivation explained 19 percent of the variance in PISA 2006 scores between countries. Boe, May and Boruch (2002) examined the role of student effort on the test on cross-country comparisons using TIMSS. They used item response rates on a student survey given as part of the TIMSS as their measure of student effort and found that more than 50 percent of the cross-country variation in test scores was explained by survey item non-response. Finally, Zamarro, Hitt & Mendez (2016) combined measures of decline in performance in the PISA test with measures of student effort in the PISA survey (i.e. item non-response and careless answering) and found that they explained between 32 and 38 percent of the observed variation in test scores across countries in 2009.
We use PISA data from over 300,000 individual students within 44 countries in 2009 and a historical natural experiment to estimate the causal impact of private schooling on student non-cognitive skills. Since nations with larger shares of Catholics in 1900 tend to have larger shares of private schooling today, we use the Catholic share of the population in 1900 as an exogenous instrument to predict whether a given child is in a private school in 2009. This technique has been successfully used by previous scholars to identify the causal effects of private schooling on PISA scores (West & Woessmann, 2010). After using the instrumental variable in addition to controlling for student, family, and country-level characteristics, we find evidence that private schooling increases student effort on PISA tests, as measured by smaller rates of test-decline, while decreasing effort on the student surveys, as measured by careless answer patterns and non-response rates.