Putting inputs to work in elementary schools: What can be done in the Philippines?

Economic Development and Cultural Change

Volumn 45, Issue 4, 1997, Pages 857-879

  Tan, Jee Peng ^a   Lane, Julia ^b   Coustère, Paul ^c  

^a WORLD BANK GROUP   (United States)

^b AMERICAN UNIVERSITY   (United States)

^c UNIVERSITÉ DE BOURGOGNE   (France)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 34248105460     PISSN: 00130079     EISSN: None     Source Type: Journal    
DOI: 10.1086/452310     Document Type: Article

References (46)

1. See, e.g., Bruce Fuller and Prem Clarke, "Raising School Effects While Ignoring Culture? Local Conditions and the Influence of Classroom Tools, Rules, and Pedagogy," Review of Education Research 64 (Spring 1994): 119-57; Eduardo Velez, Enersto Schiefelbein, and Jorge Valenzuela, "Factors Affecting Achievement in Primary Education," HRO Working Paper no. HROWP2 (World Bank, Human Resources Development and Operations Policy, Washington, D.C., 1993); Marlaine E. Lockheed, Adriaan M. Verspoor, and associates, Improving Primary Education in Developing Countries (Washington, D.C.: Oxford University Press, for World Bank, 1991); and Jaap Scheerens, Effective Schooling: Research, Theory and Practice (London: Cassell, 1992).
2. See, e.g., Bruce Fuller and Prem Clarke, "Raising School Effects While Ignoring Culture? Local Conditions and the Influence of Classroom Tools, Rules, and Pedagogy," Review of Education Research 64 (Spring 1994): 119-57; Eduardo Velez, Enersto Schiefelbein, and Jorge Valenzuela, "Factors Affecting Achievement in Primary Education," HRO Working Paper no. HROWP2 (World Bank, Human Resources Development and Operations Policy, Washington, D.C., 1993); Marlaine E. Lockheed, Adriaan M. Verspoor, and associates, Improving Primary Education in Developing Countries (Washington, D.C.: Oxford University Press, for World Bank, 1991); and Jaap Scheerens, Effective Schooling: Research, Theory and Practice (London: Cassell, 1992).
3. See, e.g., Bruce Fuller and Prem Clarke, "Raising School Effects While Ignoring Culture? Local Conditions and the Influence of Classroom Tools, Rules, and Pedagogy," Review of Education Research 64 (Spring 1994): 119-57; Eduardo Velez, Enersto Schiefelbein, and Jorge Valenzuela, "Factors Affecting Achievement in Primary Education," HRO Working Paper no. HROWP2 (World Bank, Human Resources Development and Operations Policy, Washington, D.C., 1993); Marlaine E. Lockheed, Adriaan M. Verspoor, and associates, Improving Primary Education in Developing Countries (Washington, D.C.: Oxford University Press, for World Bank, 1991); and Jaap Scheerens, Effective Schooling: Research, Theory and Practice (London: Cassell, 1992).
4. See, e.g., Bruce Fuller and Prem Clarke, "Raising School Effects While Ignoring Culture? Local Conditions and the Influence of Classroom Tools, Rules, and Pedagogy," Review of Education Research 64 (Spring 1994): 119-57; Eduardo Velez, Enersto Schiefelbein, and Jorge Valenzuela, "Factors Affecting Achievement in Primary Education," HRO Working Paper no. HROWP2 (World Bank, Human Resources Development and Operations Policy, Washington, D.C., 1993); Marlaine E. Lockheed, Adriaan M. Verspoor, and associates, Improving Primary Education in Developing Countries (Washington, D.C.: Oxford University Press, for World Bank, 1991); and Jaap Scheerens, Effective Schooling: Research, Theory and Practice (London: Cassell, 1992).
5. Unesco, World Education Report (Paris: Unesco, 1993).
6. Armand V. Fabella, "The Educational Strategy for Philippines 2000" (paper presented at the Oxford Roundtable on Educational Policy, Oxford University, September 30, 1993).
7. Marcelina M. Miguel, Variables That Affect Student Achievement in the Philippines (Manila: Bureau of Elementary Education, Department of Education, Culture and Sports, 1993).
8. International Association for the Evaluation of Educational Achievement (IEA), Science Achievement in Seventeen Countries: A Preliminary Report (London: Pergamon, 1988).
9. See Fabella.
10. See Eric A. Hanushek, "The Economics of Schooling: Production and Efficiency in Public Schools," Journal of Economic Literature 24 (September 1986): 1141-77; and Scheerens, for extensive summaries.
11. The tests contain 25-50 items each and were created by the Bureau of Elementary Education on the basis of the syllabus for the grade.
12. The marked decrease in the grade 1 sample occurred because, due to logistical difficulties, the mathematics test was not administered in several schools.
13. The separate treatment is appropriate for several reasons. One reason is that, whereas nearly all teachers teach all subjects in grade 1, teachers in grades 4 and 6 tend to specialize in specific subjects. As a result, the regression equation requires a slightly different specification for grade 1 pupils than for the other two grades. Another reason is that the effect of policy variables on student learning may differ across grades. Finally, the relation between incoming and outgoing test scores tightens with grade level, suggesting that there are fewer policy interventions that will improve achievement in the later grades. Thus, from a policy perspective, the empirical results are especially important for the grade 1 sample.
14. A full definition and data description is available from us on request.
15. To facilitate interpretation of the regression results reported below, all test scores are rescaled as standardized deviations from the sample mean. A score of 0.50, e.g., signifies that a pupil tested at 0.5 SD above the sample mean.
16. Family wealth is notoriously difficult to measure. In the data set, the teacher asked the pupil about the family's ownership of a list of tangible household assets. The pupil is likely to know the answers with some precision, thereby reducing measurement error. Following Ralph Harbison and Eric A. Hanushek, Educational Performance of the Poor: Lessons from Rural Northeast Brazil (Washington, D.C.: Oxford University Press for World Bank, 1992), we use the method of principal components to create an index of family wealth based on the pupil's family income (as estimated in broad categories by the teacher), as well as the family's ownership, as reported by the pupil, of five of the assets for which there is sufficient variance: flush toilets, electric lighting, in-house piped water supply, radio, and television.
17. Some apparently appealing candidates for inclusion in the list of variables were not used for lack of variance. For example, the sex of the teacher was not used because nearly all the teachers in the sample were women. Similarly, other interesting aspects of the learning context, such as the use of guide books by teachers, whether or not the teachers or school head enjoyed the work, whether or not the school depended on community contributions, and so on, were considered and not used because the responses were relatively homogeneous across teachers and schools.
18. The questionnaire simply asked teachers to indicate the number of times during the past 8 years they attended training in different subject areas at the national, regional, divisional, district, and school levels.
19. Angus Deaton, "The Analysis of Household Surveys: Microeconometric Analysis for Development Policy" (Princeton University, Research Program in Development Studies, September 1993 draft).
20. In the education literature the problem is commonly described as arising from the hierarchical structure of the data. One approach for addressing it involves the use of multilevel models to specify and estimate the structure of the error term as an input toward estimating the regression model. See Anthony S. Bryk and Stephen W. Raudenbush, Hierarchical Linear Models: Applications and Data Analysis Methods (Newbury Park, Calif.: Sage, 1992) for a full discussion.
21. Joseph Altonji and Thomas Dunn, "The Effects of School and Family Characteristics on the Return to Education," Working Paper no. 5072 (National Bureau of Economic Research, Washington, D.C., 1995).
22. See, e.g., M. Aitkin and N. Longford, "Statistical Modelling in School Effectiveness Studies," Journal of the Royal Statistical Society 149, A (1986): 1-43; and Harvey Goldstein, Multilevel Models in Educational and Social Research (New York: Oxford University Press, 1987).
23. See, e.g., M. Aitkin and N. Longford, "Statistical Modelling in School Effectiveness Studies," Journal of the Royal Statistical Society 149, A (1986): 1-43; and Harvey Goldstein, Multilevel Models in Educational and Social Research (New York: Oxford University Press, 1987).
24. Cheng Hsiao, Analysis of Panel Data (Cambridge: Cambridge University Press, 1986), p. 135.
25. For a good discussion, see Russell Davidson and James G. Mackinnon, Estimation and Inference in Econometrics (New York: Oxford University Press, 1992).
26. Our referees pointed out that the decision to attend preschool and the event of grade repetition are endogenous variables and perhaps should also be instrumented. As these events occurred prior to entering the grade under analysis, we assume that they are uncorrelated with the disturbance term, and therefore we did not apply the technique of instrumental variables to them.
27. The full set of regression results is available from us on request.
28. We use teachers' ratings of their pupils' activeness as a control variable in the regression models, assuming that the ratings are consistent across classes. This assumption is admittedly subject to error, since each teacher may rate on a scale that depends on the behavior of pupils in her or his own classroom. However, because the scale is a standard three-part Likert scale, the assumption is likely to be valid in reality. A different issue is that this variable correlates with pupils' incoming test scores; i.e., teachers know who the high performers are and rate them as such. To the extent that this is true, the coefficient estimates will not be biased but their standard errors will increase.
29. Results for scores in English also indicate that preschool programs have a favorable influence on this subject. The coefficient estimate is unreliable, however, given that the randomness in the workbook availability variable could not be addressed satisfactorily using the instrumental variables approach.
30. For presentational simplicity, table 1 does not report the results corresponding to the nonlinear specification of the variable. These are available from us on request.
31. The data do not permit a close examination of the reasons for the finding that parental education affects performance in Filipino but not in mathematics. It may be that parents find it easier to help their children with language skills, since this help can be offered in the course of everyday conversation.
32. The inadequacy of classroom furniture may reflect generally poor physical classroom conditions.
33. One possible explanation is that the mathematics test in grade 1 covers the most basic of numeracy skills, so that teacher's subject knowledge matters little in affecting student performance.
34. The questionnaire also canvassed teachers on their attitudes toward other statements, but the responses show limited variance in the sample.
35. For simplicity, the table does not include the results relating to school location; these are available from us on request. School location in our model serves as a control variable to capture a variety of community influences, including the possibility that schools differ in their emphasis on the teaching of various subjects. For example, a school in a non-Tagalog speaking area might wish to emphasize Pilipino in the early grades. Although interesting, the issue is not central to this article, and we do not explore it in detail here.
36. The survey obtained information on many aspects of school heads' management practices or attitudes. Except for the facets listed in the table, the responses lacked sufficient variance to be used meaningfully in the regression.
37. The difference in effects across subjects is curious and invites speculation about the likely causes. It is possible that more scope for experimentation exists in the teaching of Pilipino than it does for mathematics, but the data do not allow us to confirm the validity of this explanation.
38. The amount received by each school was calculated by multiplying the number of copies of each item received with its price and then summing the total.
39. 2 value of 0.14.
40. See Scheerens (n. 1 above).
41. We do not use class and school dummy variables in the reported regressions, since we want to allow for variables that include variation at the class and school levels. The exercise here is intended to estimate the aggregate importance of the class and school effects that are explicitly specified in our regressions and those that are omitted.
42. We omit reporting the results for English test scores simply to make the presentation consistent with previous tables. Note that because textbook and workbook availability in grade 1 is irrelevant in determining initial test scores, the random regressor problem associated with this variable does not apply here.
43. As discussed above, the regression variable "inadequate furniture" can also be interpreted as signaling generally poor physical classroom conditions. As such, spending money to improve these conditions may also be cost-effective if the unit costs involved are comparable to those in table 6.
44. Stephen P. Heyneman, Dean T. Jamison, and Xenia Montenegro, "Textbooks in the Philippines: Evaluation of the Pedagogical Impact of a Nationwide Investment," Educational Evaluation and Policy Analysis 6, no. 2 (1984): 139-50.
45. Marlaine E. Lockheed and Eric A. Hanushek, "Improving Educational Efficiency in Developing Countries: What Do We Know?" Compare 18, no. 1 (1988): 21-38.
46. See, e.g., the review by Hanushek (n. 7 above).