| Evaluation | Effect | Average Age | Observations | Experimental Evaluation | Country | Intervention |
|---|---|---|---|---|---|---|
| Cullen 2006 | -2 | Grade 8 | N/A | Yes | United States | Students obtained access to preferred schools, such as magnet schools |
| Adukia 2014 | 5 | Grade 7 | 53,338 | No | India | The Total Sanitation Campaign provided construction of school latrines |
| Rosati 2007 | 2 | Grade 9 | 48,957 | No | Mexico | Computers, equipment, training, materials, incentives |
| Cristia 2014 | 4 | Grade 9 | 33,583 | No | Peru | Schools received hardware, software, training and internet access |
| Zhang 2009 | 6 | Grade 10 | 9,424 | Yes | China | Students obtained access to magnet schools |
| Ome 2012 | 1 | Grade 11 | 9,283 | No | Colombia | Introduction of examination for teachers to get hired or promoted |
| Estrada 2015 | -1 | Grade 8 | 8,365 | No | Mexico | Federal teacher hiring program based on national standardized test |
| Pires 2015 | 3 | 15 | 2,330 | No | Chile | The reform increasing the number of schooling hours by 30% |
| McEwan 2015 | -10 | Grade 7 | 1,386 | No | Honduras | Flexible teacher contracts and alternative recruitment practices |
| Khandker 2009 | 11 | 14 | 1,237 | No | Bangladesh | Improvement of roads, construction of culverts and small bridges |
| Fairlie 2013 | 1 | Grade 8 | 1,123 | Yes | United States | Computers to schoolchildren for home use |
| Glewwe 2009 | 9 | Grade 8 | 887 | Yes | Kenya | Schools received official government textbooks and teachers guides |
| Kemple 2000 | 11 | Grade 10 | 345 | Yes | United States | Academic and vocational training and career development |
Graph of Effects
This graph shows the effects of [Number of Evaluations] evaluations of [Program Type] programs. The first evaluation, [Evaluation name 1], estimated an effect of [Estimated effect Eval. 1] points (graphed with the top grey circle). Because this is an estimate, there is uncertainty about what the true effect is. But we know that the true effect will probably lie between a low value of [True effect lie between – Low value, Eval. 1] and a high value of [True effect lie between – High value, Eval. 1] (graphed with the top horizontal line). We can combine the estimates from all the evaluations and compute an average effect of [Estimated effect, Average] (graphed with the black circle). There is also uncertainty about where the true average effect is. But we know that the true average effect probably lies between [True effect lie between – Low value, Average] and [True effect lie between – High value, Average] points (graphed with the bottom horizontal line).
| Evaluation | Effects | Low Value | High Value | Weight in Average Effect (%) |
|---|---|---|---|---|
| Adukia 2014 | 5 | 3 | 7 | 10 |
| Cristia 2014 | 4 | -28 | 36 | 0 |
| Cullen 2006 | -2 | -4 | 0 | 10 |
| Estrada 2015 | -1 | -6 | 4 | 6 |
| Fairlie 2013 | 1 | -2 | 4 | 9 |
| Glewwe 2009 | 9 | 3 | 15 | 5 |
| Kemple 2000 | 11 | 4 | 18 | 4 |
| Khandker 2009 | 11 | 6 | 16 | 7 |
| McEwan 2015 | -10 | -18 | -2 | 3 |
| Ome 2012 | 1 | 0 | 2 | 12 |
| Pires 2015 | 3 | 1 | 5 | 10 |
| Rosati 2007 | 2 | 1 | 3 | 12 |
| Zhang 2009 | 6 | 4 | 8 | 11 |
| Overall |