The Pedagogy of Science and Environment: Experimental Evidence from Peru

Abstract

In today's knowledge-based societies, understanding basic scientific concepts and the capacity to structure and solve scientific questions is more critical than ever. Accordingly, in this article we test an innovative methodology for teaching science and environment in public primary schools where traditional (teacher-centred) teaching was replaced with student-centred activities using LEGO kits. We document positive and significant improvements of 0.18 standard deviations in standardised test scores. Such positive results are mainly concentrated within boys that were located above the median of baseline academic performance.

Acknowledgements

The authors wish to thank the Japan Special Fund for Poverty Reduction at the Inter-American Development Bank (IDB) for financing this study.

Notes

1. For instance, between 1998 and 2009 enrolment in pre-school education increased from 53.4 per cent to 66.3 per cent; enrolment in primary education increased from 90.6 per cent to 94.4 per cent; enrolment in secondary education increased from 59 per cent to 76.5 per cent (source: own calculations using the Peruvian Ministry of Education statistics portal: http:////escale.minedu.gob.pe/).

2. Source: own calculations using the database from the 2009 national evaluation: http:////umc.minedu.gob.pe

3. Minimum wage in Peru during the pilot was S/550 per month.

4. Notice that we express the function in a linear specification for ease of exposition. However, the unknown production function may take several types of non-linear forms.

5. Notice, however, that there is an extensive non-experimental literature using simulation techniques with the objective of estimating education production functions and the relative incidence of different cognitive and non-cognitive inputs (Cunha & Heckman, 2007).

6. Notice, however, that if the choosing of two sections within schools with three sections or more was not really random and, furthermore, this had not been balanced between treated and control schools then our research design would be biased. However, as we show in Table 1, schools are balanced between treated and control groups regardless of the number of sections that the school has.

7. Validation of the design of the exams for science and the environment, reading comprehension and mathematics had been tested previously in two schools in Lima that are not part of the sample. Baseline tests had been applied to third grade pupils at the beginning of the school year and the end line at the end of the school year. While the pilot programme did not include specific activities to improve reading and mathematics skills, it was considered desirable to also evaluate these areas. These exams covered the learning expectations on each subject at each point in time. These were standard exams that covered the areas that were supposed to be learned according to the national curriculum. Recall that our methodology only changed the method of teaching; learning targets were the same under the traditional and the new experimental methodologies being tested.

8. Notice that all test scores are expressed in standard deviations with respect to the control group. That is, the control group test scores have mean zero and standard deviation of unity.

9. At baseline, the total number of students was 2802; however, from this number were excluded: pupils pertaining to the inclusion programme (a programme that integrates students with special education needs in regular classrooms), those that moved to the fourth grade during the school year, and those that did not finish all the tests; leaving 2771 effective observations. At the end line the total number of students tested was 2663. Excluding pupils of the inclusion programme, those that had moved to the fourth grade and those that did not finish all the tests the figure falls to 2552. Of the latter group only 2373 pupils were evaluated at baseline.

10. The exam results were standardised such that the control group had a zero mean and a standard deviation of unity. Therefore, differences between treated and control groups are expressed in standard deviations with respect to the control group.

11. Again, exam results were standardised such that the control group had a zero mean and a standard deviation of unity. Therefore, differences between treated and control groups are expressed in standard deviations with respect to the control group.

12. We also did this exercise for rural schools; however, no impacts were found.