**CALCULATING THE LEARNING LEVEL**

Each country is classified into a group range of 10% based on an evaluation of country-available performance data. Data on learning outcomes was derived from a large number of sources. The main sources are described below, followed by a list briefly describing the main data source for each country. Detailed data can be found at http://www.acasus.com/who-learns/circle-map/

This analysis is not perfect. The underlying data is of variable accuracy; each test has its own margin of error and may not always reflect the performance of the system more broadly. The multiple sources are difficult to compare. Occasionally, different sources for the same country in the same year yield widely diverging results. The methodology and the final results bear testament to the lack of reliable data on learning levels in many countries, itself a major obstacle to improvement. However, the overall spectrum of performance is extremely wide, and even with these challenges, it is possible to develop a reasonably accurate estimate of where each country falls on that spectrum.

Countries with a population of less than one million were excluded from the analysis.

**PISA**

PISA is a large international assessment run by the OECD. The tests involve students from more than 60 countries.

The tests are conducted at age 15. Evidence from a range of countries shows a strong correlation between outcomes at age 12 and outcomes at age 15 at a country level, so the PISA tests are a reasonable instrument for estimating comparative levels of achievement at the end of primary schooling.

The outcome levels show in the global learning data set were based on ‘the proportion of children on track to reach level 2 in the PISA examination at the end of primary schooling.’ The number of children who read at level 2 or above in the PISA assessment is assumed to be equal to the number of children ‘on track to reach level 2 by age 15’ at the end of primary school.

In general, data from the 2012 PISA assessments was used, except in a small number of cases where only data from earlier PISA assessments was available.

In a few countries where a large number of children drop out of school before the age of 15, adjustments were made to allow for this dropout.

**TIMSS**

Where data from PISA was not available, TIMSS (Trends in International Mathematics and Science Study) was used as an alternative data source. TIMSS tests children in mathematics and science.

In general, data from the 2011 assessment was used, however, for five countries for which no 2011 data or alternative testing data was available, 2007 data was used.

Twenty-seven countries participated in both the PISA assessments and the TIMSS assessment. Results were well correlated, with the best correlation existing between Grade 8 Science scores in TIMSS and age 15 math scores in PISA (r-squared=0.86). The relationship between TIMSS and PISA results for the 27 countries participating in both was used to estimate a PISA equivalent score for countries participating in TIMSS but not PISA.

**SACMEQ **

SACMEQ (Southern and Eastern Africa Consortium for Monitoring Educational Quality) is a test run across 15 countries in Southern and Eastern Africa on a periodic basis.

One country participates in both PISA and SACMEQ. For SACMEQ, we compared the results of Mauritius on SACMEQ and PISA. 53.3% of Mauritian students achieved level 2 or above in PISA. 53.4% achieved level 6 or above in SACMEQ. So for reading we assumed that the same proportion of students would score level 2 or above in PISA as scored Level 6 or above on SACMEQ. For Mathematics, applying the same logic, we assumed that all students at level 6 and 57% of students at level 5 on SACMEQ would score at or above level 2 on PISA.

To check the methodology, we compared PISA scores estimated based on SACMEQ with PISA scores estimated based on TIMSS for countries participating in both TIMSS and SACMEQ.

**SERCE**

For six countries, the most recently available data comes from the SERCE assessment. This assessment was conducted in 1997.

A further nine countries participated in both SERCE and PISA. For these countries, we looked at both their improvement between 1997 and the present day and their performance in the most recent PISA assessments.

There was a strong correlation between results in the SERCE test and the PISA tests. The proportion of children scoring at level two in PISA was comparable to the number of children scoring at level 3 and above, plus 49% of the children scoring at level 2, in the SERCE assessment.

This relationship was used to estimate results for the remaining six countries.

http://www.unesco.org/new/en/santiago/education/education-assessment-llece/second-regional-comparative-and-explanatory-study-serce/

**PASEC**

PASEC is a test of Math and French run across several countries in Western Africa.

No country participates in both PASEC and SACMEQ, TIMSS, or PISA, making comparison with other tests difficult. Average primary completion rates for the group of PASEC countries were compared to average primary completion rates in the other African countries for which learning data was available. This information was used to calibrate the mean and average of performance for the PASEC group with other countries for which data was available. Based on this, learning levels were estimated. The PASEC methodology, relying on comparison of completion rates, is considered the least reliable of the group.

http://www.confemen.org/le-pasec/rapports-et-documents-pasec/les-rapports-du-pasec/

**Other tests**

In a small number of countries, other tests were used. These are described in the individual country notes.

**Completion**

For a small number of countries with no learning level data, but with very low primary completion rates, learning levels were assumed to be in the lowest category based on comparisons with other countries with comparable completion rates and in country reports.

**Other data sources**

Data on enrolment and completion was sourced from databases maintained by UNESCO and the World Bank. Occasionally, other sources were used to fill gaps in the data, as described in the source information.

**Individual country summary notes**

Country | Group | Description |

Afghanistan | No data was available. | |

Albania | 4 | The proportion of children performing at reading level 2 on the 2012 PISA tests is used to estimate learning levels among children who complete primary school. The estimate is adjusted for the relatively small target population in the Albanian sample. |

Algeria | 2 | Seven percent of children do not complete primary school. 2007 TIMSS data is used to estimate learning levels among those who do complete primary school. |

Angola | 0 | Completion rates from primary school are very low. Based on comparisons with countries with similar completion rates for which learning data is available, the proportion of children leaving primary school with a good level of learning is estimated as below 10%. |

Argentina | 4 | The proportion of children performing at reading level 2 on the 2012 PISA tests (46%) is used as a proxy for learning levels among those who complete primary school. |

Armenia | 3 | Seventeen percent of children do not reach the last grade of primary school. 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Australia | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (86%) is used as a proxy for learning levels among those who complete primary school. |

Austria | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (81%) is used as a proxy for learning levels among those who complete primary school. |

Azerbaijan | 2 | Seven percent of children do not reach the last grade of primary school. The proportion of children performing at reading level 2 on the 2009 PISA tests (27%) is used as a proxy for learning levels among those who do complete primary school. |

Bahrain | 4 | 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Bangladesh | 1 | Data from national assessments suggest that only 13% of grade 5 students are able to do simple mathematical equations and 31% are able to read a newspaper headline; however early grade reading tests suggest a slightly higher level of fluency than India |

Belarus | No data was available. | |

Belgium | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (84%) is used as a proxy for learning levels among those who complete primary school. |

Benin | 0 | In the PASEC tests, 26% of children read at level 3. Based on this, the proportion of children completing school with a good level of learning is estimated as less than 10%. |

Bolivia | 3 | Little recent data on learning for Bolivia is available. In 1997, Bolivia participated in LLECE, and performed at a similar level to neighbouring Paraguay. This estimate is based on that comparison. |

Bosnia and Herzegovina | 5 | Seventeen percent of children do not reach the last grade of primary school. 2007 TIMSS data is used to estimate learning levels among those who complete primary school. |

Botswana | 3 | The estimate is based on the proportion of students who reach the last grade of primary school (end of G6) and achieve level 6 on the SACMEQ reading test. |

Brazil | 4 | The proportion of children performing at reading level 2 on the 2012 PISA tests is used to estimate learning levels among children who complete primary school. The estimate is adjusted for the relatively small target population in the Albanian sample. |

Bulgaria | 6 | The proportion of children performing at reading level 2 on the 2012 PISA tests (61%) is used as a proxy for learning levels among those who complete primary school. |

Burkina Faso | 0 | In the PASEC tests, 38% of children read at level 3. Only 39% of children graduate from primary school. Based on this, the proportion of children completing school with a good level of learning is estimated as between 10% and 20%. |

Burundi | 0 | In the PASEC tests, 53% of children read at level 3. Only 36% of children graduate from primary school. Based on this, the proportion of children completing school with a good level of learning is estimated as less than 10%. |

Cambodia | No data was available. | |

Cameroon | 1 | In the PASEC tests, 71% of children read at level 3. Only 48% of children graduate from primary school. Based on this, the proportion of children completing school with a good level of learning is estimated as between 10% and 20%. |

Canada | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (89%) is used as a proxy for learning levels among those who complete primary school. |

Central African Republic | 0 | Completion rates from primary school are very low. Based on comparisons with countries with similar completion rates for which learning data is available, the proportion of children leaving primary school with a good level of learning is estimated as below 10%. |

Chad | 0 | In the PASEC tests, 27% of children read at level 3. Only 26% of children complete primary school. Based on this, the proportion of children completing school with a good level of learning is estimated as less than 10%. |

Chile | 6 | The proportion of children performing at reading level 2 on the 2012 PISA tests (67%) is used as a proxy for learning levels among those who complete primary school. |

China | 8 | The proportion of children performing at reading level 2 is estimated based on results for the 12 Chinese provinces in which the data has been collected. |

China, Hong Kong | 9 | The proportion of children performing at reading level 2 on the 2012 PISA tests (93%) is used as a proxy for learning levels among those who complete primary school. |

Colombia | 4 | The proportion of children performing at reading level 2 on the 2012 PISA tests is used to estimate learning levels among children who complete primary school. The estimate is adjusted for the relatively small target population in the Albanian sample. |

Congo | 0 | In the PASEC tests, 37% of children read at level 3. Only 47% of children graduate from primary school. Based on this, the proportion of children completing school with a good level of learning is estimated as between 10% and 20%. |

Costa Rica | 6 | |

Côte d’Ivoire | 0 | In the PASEC tests, 37% of children read at level 3. Only 46% of children graduate from primary school. Based on this, the proportion of children completing school with a good level of learning is estimated as between 10% and 20%. |

Croatia | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (81%) is used as a proxy for learning levels among those who complete primary school. |

Cuba | 8 | Estimate based on the scores of children in the SERCE Reading tests. |

Cyprus | 6 | The proportion of children performing at reading level 2 on the 2012 PISA tests (67%) is used as a proxy for learning levels among those who complete primary school. |

Czech Republic | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (83%) is used as a proxy for learning levels among those who complete primary school. |

D.R. Congo | 0 | EGRA reading tests suggest that 70% of grade 4 and 5 students do not understand basic reading passages. In countries with similar results, fewer than 10% of children are on track to complete primary with good learning levels. |

Denmark | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (85%) is used as a proxy for learning levels among those who complete primary school. |

Dominican Republic | 1 | Estimate based on the scores of children in the SERCE Reading tests. |

Ecuador | 3 | Estimate based on the scores of children in the SERCE Reading tests. |

Egypt | 3 | 2007 TIMSS data is used to estimate learning levels among those who complete primary school. |

El Salvador | 3 | 2007 TIMSS data is used to estimate learning levels among those who complete primary school. |

Eritrea | 0 | Completion rates from primary school are very low. Based on comparisons with countries with similar completion rates for which learning data is available, the proportion of children leaving primary school with a good level of learning is estimated as below 10%. |

Estonia | 9 | The proportion of children performing at reading level 2 on the 2012 PISA tests (91%) is used as a proxy for learning levels among those who complete primary school. |

Ethiopia | 0 | Estimate based on the 2012 Ethiopian National Learning Assessment (ENLA). |

Finland | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (89%) is used as a proxy for learning levels among those who complete primary school. |

France | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (81%) is used as a proxy for learning levels among those who complete primary school. |

Gabon | 2 | In the PASEC tests, 84% of children read at level 3. Based on this, the proportion of children completing school with a good level of learning is estimated as between 20% and 30%. |

Gambia | 0 | Independent tests show that 54% of children cannot read a single word by the end of grade 2. Based on this and comparison with similar countries, the proportion of children reaching a good basic level of learning is estimated as below 10%. |

Georgia | 3 | The proportion of children performing at reading level 2 on the 2009 PISA tests (38%) is used as a proxy for learning levels among those who complete primary school. |

Germany | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (86%) is used as a proxy for learning levels among those who complete primary school. |

Ghana | 1 | 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Greece | 7 | The proportion of children performing at reading level 2 on the 2012 PISA tests (77%) is used as a proxy for learning levels among those who complete primary school. |

Guatemala | 2 | Estimate based on the scores of children in the SERCE Reading tests. |

Guinea | 0 | |

Guinea-Bissau | No data was available. | |

Haiti | 0 | In an EGRA assessment, only 10% of grade 3 students could complete basic reading comprehension in creole. In countries with similar results, fewer than 10% of children are on track to complete primary with good learning levels. |

Honduras | 2 | 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Hungary | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (80%) is used as a proxy for learning levels among those who complete primary school. |

India | 1 | The proportion of children performing at reading level 2 on the 2009 PISA tests in the two states which participated in PISA (14%) is used as a proxy for learning levels among those who complete primary school. |

Indonesia | 3 | Twelve percent of children do not reach the last grade of primary school. The proportion of children performing at reading level 2 on the 2012 PISA tests (45%) is used as a proxy for learning levels among those who complete primary school. |

Iran | 5 | 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Iraq | 3 | In early grade reading and numeracy tests, students in Iraq had comparable performance to students in Egypt. The same estimates of learning levels are used. |

Ireland | 9 | The proportion of children performing at reading level 2 on the 2012 PISA tests (90%) is used as a proxy for learning levels among those who complete primary school. |

Israel | 7 | The proportion of children performing at reading level 2 on the 2012 PISA tests (76%) is used as a proxy for learning levels among those who complete primary school. |

Italy | 8 | |

Jamaica | 3 | Jamaica does not participate in any standardized international assessments. National assessments are run at grades 3 ,4 and 6. These generally find that 30-40% of students meet curriculum expectations. A similar proportion achieve the target grades I,II or III at secondary level. |

Japan | 9 | The proportion of children performing at reading level 2 on the 2012 PISA tests (90%) is used as a proxy for learning levels among those who complete primary school. |

Jordan | 4 | Seven percent of children do not reach the last grade of primary school. The proportion of children performing at reading level 2 on the 2012 PISA tests (49%) is used as a proxy for learning levels among those who do complete primary school. |

Kazakhstan | 4 | The proportion of children performing at reading level 2 on the 2012 PISA tests (43%) is used as a proxy for learning levels among those who complete primary school. |

Kenya | 2 | The estimate is based on the proportion of students who reach the last grade of primary school (end of G5) and achieve level 6 on the SACMEQ reading test. |

Korea D.P.R. | No data was available. | |

Korea, R. | 9 | The proportion of children performing at reading level 2 on the 2012 PISA tests (92%) is used as a proxy for learning levels among those who complete primary school. |

Kuwait | 1 | 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Kyrgyzstan | 1 | The proportion of children performing at reading level 2 on the 2009 PISA tests (17%) is used as a proxy for learning levels among those who complete primary school. |

Laos | No data was available. | |

Latvia | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (82%) is used as a proxy for learning levels among those who complete primary school. |

Lebanon | 3 | Seventeen percent of children do not complete primary school. 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Lesotho | 0 | The estimate is based on the proportion of students who reach the last grade of primary school (end of G6) and achieve level 6 on the SACMEQ reading test. |

Liberia | 1 | This estimate is based on reading levels measured through the EGRA assessment. |

Libya | No data was available. | |

Lithuania | 7 | The proportion of children performing at reading level 2 on the 2012 PISA tests (79%) is used as a proxy for learning levels among those who complete primary school. |

Macedonia T.F.Y.R. | 3 | Six percent of children do not reach the last grade of primary school. 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Madagascar | 1 | In the PASEC tests, 36% of children read at level 3. 73% of children enter the last grade of primary. Based on this, the proportion of children completing school with a good level of learning is estimated as between 10% and 20%. |

Malawi | 0 | The estimate is based on the proportion of students who reach the last grade of primary school (end of G5) and achieve level 6 on the SACMEQ reading test. |

Malaysia | 4 | The proportion of children performing at reading level 2 on the 2012 PISA tests (47%) is used as a proxy for learning levels among those who complete primary school. |

Mali | 0 | |

Mauritania | 0 | In a 2006 PASEC assessment, scores were significantly lower than in any other PASEC country |

Mauritius | 5 | The proportion of children performing at reading level 2 on the 2009 PISA tests (53%) is used as a proxy for learning levels among those who complete primary school. |

Mexico | 5 | |

Moldova | 3 | Eight percent of children do not reach the last grade of primary school. The proportion of children performing at reading level 2 on the 2009 PISA tests (43%) is used as a proxy for learning levels among those who do complete primary school. |

Mongolia | 4 | 2007 TIMSS data is used to estimate learning levels among those who complete primary school. |

Morocco | 1 | 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Mozambique | 0 | The estimate is based on the proportion of students who reach the last grade of primary school (end of G6) and achieve level 6 on the SACMEQ reading test. |

Myanmar | No data was available. | |

Namibia | 1 | |

Nepal | 0 | Reading assessments show that 80% of children cannot read a single word by the middle of grade 2. Based on comparison with other countries, the proportion of children leaving primary school with good levels of learning is estimated as less than 10%. |

Netherlands | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (86%) is used as a proxy for learning levels among those who complete primary school. |

New Zealand | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (84%) is used as a proxy for learning levels among those who complete primary school. |

Nicaragua | 3 | Estimate based on the scores of children in the SERCE Reading tests. |

Niger | 0 | |

Nigeria | 0 | Estimate based on a DFID-funded learning assessment conducted in 2010, which found that 58% of Nigerian students in Grade 4 could not respond correctly to grade 1 and grade 2 test items. |

Norway | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (84%) is used as a proxy for learning levels among those who complete primary school. |

Oman | 4 | 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Pakistan | 1 | The ASER survey, administered in both India and Pakistan, suggests that overall literacy levels at grade 5 are broadly similar, once out-of-school children are accounted for |

Palestinian Territory | 4 | Seven percent of children do not reach the last grade of primary school. 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Panama | 3 | The proportion of children performing at reading level 2 on the 2009 PISA tests (35%) is used as a proxy for learning levels among those who complete primary school. |

Papua New Guinea | 0 | Achievement estimated at below 10%, based on the proportion of students who complete primary and the proportion of primary school leavers who can read a sentence. |

Paraguay | 3 | Estimate based on the scores of children in the SERCE Reading tests. |

Peru | 4 | The proportion of children performing at reading level 2 on the 2012 PISA tests (40%) is used as a proxy for learning levels among those who complete primary school. |

Philippines | 2 | Nine percent of children do not reach the last grade of primary school. 2003 TIMSS results are used to estimate learning levels among those who complete primary school. |

Poland | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (89%) is used as a proxy for learning levels among those who complete primary school. |

Portugal | 8 | |

Qatar | 4 | The proportion of children performing at reading level 2 on the 2012 PISA tests (43%) is used as a proxy for learning levels among those who complete primary school. |

Romania | 6 | The proportion of children performing at reading level 2 on the 2012 PISA tests (63%) is used as a proxy for learning levels among those who do complete primary school. |

Russian Federation | 7 | The proportion of children performing at reading level 2 on the 2012 PISA tests (78%) is used as a proxy for learning levels among those who complete primary school. |

Rwanda | 1 | Independent assessment data show that 47% of grade 6 students are able to answer basic questions on a grade 2/3 level text. Based on this, and comparisons to other countries with similar learning levels, the proportion of children leaving primary school with a good basic level of learning is estimated as 10-20%. |

Saudi Arabia | 4 | Seven percent of children do not reach the last grade of primary school. 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Senegal | 0 | In the PASEC tests, 45% of children read at level 3. However, only 35% of children enter the last grade of primary. Based on this, the proportion of children completing school with a good level of learning is estimated as less than 10%. |

Serbia | 6 | The proportion of children performing at reading level 2 on the 2012 PISA tests (67%) is used as a proxy for learning levels among those who complete primary school. |

Sierra Leone | 0 | |

Singapore | 9 | The proportion of children performing at reading level 2 on the 2012 PISA tests (90%) is used as a proxy for learning levels among those who complete primary school. |

Slovakia | 7 | The proportion of children performing at reading level 2 on the 2012 PISA tests (72%) is used as a proxy for learning levels among those who complete primary school. |

Slovenia | 7 | The proportion of children performing at reading level 2 on the 2012 PISA tests (79%) is used as a proxy for learning levels among those who complete primary school. |

Somalia | 0 | Learning data for Somalia is not available, but based on enrolment estimates and field evidence, the number of students completing primary school with a good learning level is estimated as below 10%. |

South Africa | 2 | We use the proportion of students who reach the last grade of primary school (end of G6) and achieve level 6 on the SACMEQ reading test. |

Spain | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (82%) is used as a proxy for learning levels among those who complete primary school. |

Sri Lanka | 5 | It is difficult to compare learning levels in Sri Lanka internationally, and this estimate should be read with caution. The National Assessment of Achievement finds that 80-85% of students achieve basic proficiency in reading and math by end of grade 4, and around 50% achieve mastery. 37% of students taking the GCE O-level exam, historically linked to British standards, pass. 99% of children complete primary school and 91% of students survive to grade 9, remarkably high levels for a developing country. Based on this, the number of students completing primary with a good level of learning is estimated as in the range of 50-60%. |

Sudan | 0 | |

Swaziland | 2 | |

Sweden | 7 | The proportion of children performing at reading level 2 on the 2012 PISA tests (77%) is used as a proxy for learning levels among those who complete primary school. |

Switzerland | 8 | |

Syria | 4 | 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Taiwan | 8 | |

Tajikistan | No data was available. | |

Tanzania | 4 | The estimate is based on the proportion of students who complete primary school and achieve level 6 on the SACMEQ reading test |

Thailand | 6 | Eight percent of children do not reach the last grade of primary school. The proportion of children performing at reading level 2 on the 2009 PISA tests (67%) is used as a proxy for learning levels among those who do complete primary school. |

Timor-Leste | 0 | EGRA reading tests show 50-60% of students do not understand basic reading passages. In countries with similar results, fewer than 10% of children are on track to complete primary with good learning levels. |

Togo | 0 | In a 2010 PASEC assessment, average scores in Togo were among the lowest of PASEC countries. |

Trinidad and Tobago | 5 | Nine percent of children do not reach the last grade of primary school. The proportion of children performing at reading level 2 on the 2009 PISA tests (55%) is used as a proxy for learning levels among those who do complete primary school. |

Tunisia | 4 | Ten percent of children do not reach the last grade of primary school. The proportion of children performing at reading level 2 on the 2012 PISA tests (51%) is used as a proxy for learning levels among those who do complete primary school. |

Turkey | 7 | |

Turkmenistan | No data was available. | |

Uganda | 0 | |

Ukraine | 7 | 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

United Arab Emirates | 6 | The proportion of children performing at reading level 2 on the 2012 PISA tests (65%) is used as a proxy for learning levels among those who complete primary school. |

United Kingdom | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (83%) is used as a proxy for learning levels among those who complete primary school. |

United States of America | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests (83%) is used as a proxy for learning levels among those who complete primary school. |

Uruguay | 5 | The proportion of children performing at reading level 2 on the 2012 PISA tests (53%) is used as a proxy for learning levels among those who complete primary school. |

Uzbekistan | 2 | Seven percent of children do not reach the last grade of primary school. We use an national assessment from 2006 which suggested that only 30% of students met the expectations of the national curriculum. This assessment could not be calibrated to any external benchmark. |

Venezuela | 5 | The proportion of children performing at reading level 2 on the 2009 PISA tests is used as a proxy for learning levels among those who do complete primary school (40%). Only one state participated in PISA, Miranda, which has the highest HDI of any state in Venezuela, so this is likely a high estimate |

Viet Nam | 8 | The proportion of children performing at reading level 2 on the 2012 PISA tests is used as a proxy for learning levels among those who complete primary school. This is adjusted for the small target population sampled in PISA. |

Yemen | 0 | 2011 TIMSS data is used to estimate learning levels among those who complete primary school. |

Zambia | 0 | |

Zimbabwe | 1 | |

Rest of World | Assumed equal to the average of the countries for which data is available |