1 minute read

10 Mediation analysis

Table 10: Mediation analysis

Inputs Inputs & Management (1) (2) (3) (4) (5) Treatment 0.19∗∗∗ 0.09∗∗ 0.11∗∗ 0.01 0.01 (0.03) (0.04) (0.05) (0.05) (0.06) Writing materials -0.00 -0.00 -0.00 (0.00) (0.00) (0.00) PTR -0.00 -0.00 -0.00 -0.00 (0.00) (0.00) (0.00) (0.00) Teachers’ age -0.01∗∗∗ -0.02∗∗∗ -0.01∗∗∗ -0.01∗∗∗ (0.00) (0.00) (0.00) (0.00) % exp. in private schools -0.00 -0.00 -0.00 (0.00) (0.00) (0.00) Textbooks -0.00 -0.00 (0.00) (0.00) Teachers’ experience 0.01∗∗ 0.01 (0.01) (0.01) Teachers’ test score 0.06 0.07 (0.05) (0.05) Certified teachers 0.00 0.00 (0.00) (0.00) Teacher attendance 0.00∗∗∗ 0.00∗∗∗ (0.00) (0.00) % of time spent on management -0.04 (0.08) Index of good practices (PCA) 0.07∗∗∗ (0.02) Student attendance 0.12 (0.10) Instruction (Classroom obs) -0.00 (0.00) Hrs/week 0.01∗ (0.00) No. of obs. 3,498 3,498 3,498 3,498 3,498 R2 0.53 0.53 0.53 0.53 0.54 Mediators None Lasso All Lasso All

Column 1 replicates the results from Table 3 and columns 2 and 3 include only raw inputs. Columns 4 and 5 include raw inputs and the use of these inputs. Column 2 and column 4 only include mediators selected via Lasso, and columns 3 and 5 include all the mediators. Standard errors are clustered at the school level. The sample is the original treatment and control allocation. ∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01

Note that in Section 3 we estimated equation (4) for several mediators. Combining those results with the results from Table 10, we show in Figure 5 the mediation effect (β4 × θ5) for the intermediate outcomes selected by Lasso, as well as the direct effect (β5). The left panel uses only raw inputs as mediators, while the right panel also includes changes in the use of inputs.

Approximately half of the overall increase (37.4%–53.5%) in learning appears to have been due to

39

This article is from: