Gender and Education for All THE LEAP TO EQUALITY

Appendix 2

The Education for All Development Index (EDI) and prospects for gender parity As explained in Chapter 2, if an Education for All Development Index is to measure overall progress towards EFA, its constituents should ideally reflect all six Dakar goals. In practice, however, this is difficult, since not all the goals have a clear definition or target. For example, Goal 3 – learning and life skills programmes – is not yet conducive to quantitative measurement. For rather different reasons, early childhood care and education (Goal 1) cannot easily be incorporated at present because the data are insufficiently standardized across countries, and they are, in any case, available for only a small minority of states. Accordingly, for the time being, the EFA Development Index (EDI) only incorporates indicators for the four goals of universal primary education, adult literacy, gender parity and the quality of education. One indicator is included as a proxy measure for each of these four EDI components1. This is in accordance with the principle of considering each goal to be equally important and, thus, of giving the same weight to each of the index constituents. The EDI value for a particular country is the arithmetical mean of the observed values for each of its different constituents. As each of its constituents is a percentage, its value can vary from 0 to 100% (or, when expressed as a ratio, from 0 to 1). The closer a country’s EDI value is to the maximum, the nearer the country is to the goal and the greater the extent of its EFA achievement. The following are the EDI constituents and related indicators: - Universal primary education: net enrolment ratio in primary education. - Adult literacy: literacy rate of the age group 15 years and over. - Quality of education: survival rate to Grade 5 of primary education. - Gender: gender-related EFA index; this is the simple average value of the GPIs in primary education, in secondary education and in adult literacy. Choice of indicators as proxy measures of EDI constituents In selecting indicators, the issue of data availability must be taken into account. On the other hand, this should not be at the expense of the relevance of the indicator as a measure of the index component. A balance between these considerations is needed. Thus, among a range of indicators which might be used to proxy the different aspects of a given component such as educational quality, both the most relevant and that for which the data coverage is acceptable should be chosen. Universal primary education (UPE) The indicator selected for UPE is the net enrolment ratio (NER), which reflects the percentage of school-age children who are enrolled in school. The value varies from 0% to 100%. An NER of 100% means that all eligible children are enrolled in school. Additionally, if a country maintains that level over time, it implies that all the children enrolled are also completing their studies. Adult literacy The adult literacy rate is used as a proxy to measure progress against EFA Goal 4. As discussed in Chapter 2, the existing data on literacy are not entirely satisfactory. Efforts to provide a new data series will, however, take some years to materialise, and the literacy estimates used are presently the best available on an international basis. As regards relevance, it should be noted that the indicator for adult literacy is a statement about the stock of human capital. As such it is slow to change, and it could be argued that it is not a good ‘leading indicator’ of progress towards improvement in literacy levels on a year-by-year basis. Quality of education There is considerable debate about the concept of quality and how it should be measured. Several proxies are generally used to measure quality that are all far from satisfactory. They include the pupil/teacher ratio – although its impact on students’ performance is ambiguous, and its distribution is as important as the national average value; the repetition rate, which can be a reasonable proxy of quality but policies of automatic promotion undermine its value as an indicator in a number of countries; the percentage of trained teachers, which is problematic because national definitions vary considerably and data availability is limited; public current expenditure variables, which also suffer from limited data coverage and provide only a rough proxy of quality; learning outcome measures, which would constitute the most appropriate proxy of the quality of education, but again the lack of comparable data across countries makes their inclusion impossible at present. Among these and other ‘candidates’ for proxy measures of quality, the survival rate to Grade 5 was selected. This is linked to UPE as a rough proxy of completion, and it has much higher data coverage compared with the other candidates. There is a strong link between survival within the primary cycle and educational achievement. The survival rate to Grade 5 – often taken as the threshold for acquisition of literacy – also captures aspects of grade repetition, promotion policy and early drop-out. Gender This fourth EDI component is measured by a composite index, the gender-related EFA index (GEI). Ideally, the GEI should reflect the whole spirit of the gender-related EFA goal that calls for ‘eliminating gender disparities in primary and secondary education by 2005, and achieving gender equality in education by 2015, with a focus on ensuring girls’ full and equal access to and achievement in basic education of good quality’. Two sub-goals are distinguished: gender parity (achieving equal participation of girls and boys in primary and secondary education) and gender equality (ensuring educational equality between boys and girls). The first sub-goal is measured by the GPI of gross enrolments at primary and secondary levels separately. Measuring and monitoring the broader aspects of equality in education is difficult, as Chapters 2 to 4 demonstrate. Essentially, outcome measures disaggregated by sex are needed for a range of educational levels. These are not available on an internationally comparable basis. As a step in that direction, the GEI includes the gender parity of adult literacy. Thus, it is calculated as a simple average value of GPIs in primary education, secondary education and adult literacy. For this reason, the second aspect of the EFA gender goal is not fully reflected in the GEI. However, this is a priority area and a challenge for future reports. Calculating the GEI As indicated above, the gender-related EFA index assesses a country’s relative achievement in gender parity in participation in primary and secondary education, as well as gender parity in adult literacy. Note that the range for GPI, when expressed as the female/male enrolment ratio or literacy rate, can exceed unity when more girls are enrolled than boys. For purposes of the index, however, in those cases where the GPI is higher than 1, the usual F/M formula has been inverted to M/F. This solves mathematically the problem of the inclusion of the gender-specific EFA index in EDI (where all components have a theoretical limit of 1, or 100%), while keeping the indicator’s capacity to show gender disparity. Figure A2.1 shows how the ‘transformed GPI’ has been created in order to highlight gender disparities neutrally – whether to the disadvantage of females or males. Once all three GPI values have been converted into the ‘transformed’ GPI (from 0 to 1), the composite gender related index is calculated as a simple average of the GPIs in primary education, in secondary education, and in adult literacy, each being equally weighted. To illustrate the calculation we will use data for South Africa in 2000, where the GPIs in primary, secondary and adult literacy were 0.95, 1.09 and 0.98 respectively. Calculating the EDI Once the gender-related EFA index (GEI) has been calculated, determining the EDI is straightforward. It is the arithmetical mean value of its four constituents – NER, adult literacy rate, GEI and survival rate to Grade 5. The EDI value varies from 0 to 1. The closer to 1, the nearer a country is to EFA achievement. A country with an index of 0.5 may be considered as being halfway towards its goals. As a simple average the EDI may, of course, hide important variations between its constituents. In other words, by giving the same weight to each EFA constituent some of those which are more advanced may be outweighed by less-developed ones, as shown in Box A2.1. On the other hand, if all the EFA goals are equally important, a country would not have achieved EFA if only some of its aspects had been concentrated upon. The objective of a synthetic indicator such as the EDI is to inform the policy debate on the prominence of all EFA goals, and to highlight the synergy between them. To illustrate the EDI’s calculation, South Africa is again taken as an example. For the remaining EDI components – NER, adult literacy rate and survival rate to grade 5 – values for this country in 2000 were 0.89, 0.85 and 0.64 respectively. Data sources Almost all the data used to calculate the EDI for both 1990 and 2000 are drawn from the UNESCO Institute for Statistics (UIS) database, with the exception of some survival rates to Grade 5, which were missing for some countries. These were obtained from the Education for All 2000 Assessment country reports. Only those countries with a complete set of indicators required to calculate the EDI are included in this analysis. This means that only ninety-four countries are currently included, and that a comprehensive global overview and monitoring of progress towards the EFA goals cannot yet be given. Gender parity prospects: methodology Chapter 2, Table 2.25 indicates the extent to which the gender goals would be achieved around the world based on an extrapolation of past trends. Achievement of the gender goal is defined as having a GPI value of between 0.97 and 1.03. This tolerance of up to 3% of inequality between the reported enrolment ratios of boys and girls is to allow for measurement error in international statistical series, and does not imply a judgement about the acceptability of any particular level of disparity. The results reported in Table 2.25 have been derived by applying a linear extrapolation of the change in GPI at primary and secondary levels in the period 1990 to 2000. This method is simple, but is subject to a number of limitations. First, as the GPI is a ratio between the separate enrolment rates for girls and boys, it would have been best to calculate, and extrapolate, its constituent elements. Owing to the introduction of ISCED 1997, however, and to consequent changes in the defined length of each branch of schooling over the 1990–2000 decade, this method proved not to be possible. Accordingly, the more approximate method of projecting the GPI itself has been used. Secondly, there are well-known pitfalls in interpreting the results of point-to-point projections. In particular, they are unable to take account of non-linear rates of expansion or contraction that may have held between the two chosen years. However, in the absence of an international series of GPI data for the intervening years, a more sophisticated extrapolation method could not be used. In cases where the extrapolated value of GPI, calculated as above, falls outside the range 0.97–1.03 in 2005 or 2015, the country is classified as being at risk of not achieving gender parity in those years. This includes all those countries that moved away from the goal over the decade 1990–2000. These comprise thirteen countries at primary level and twenty-eight countries at secondary level. If they were moving away from the goal over that period, any method for extrapolating the past would clearly indicate that they were at risk of not achieving the goal in the future. On the other hand, this implies that there are a number of countries in the ‘at risk’ group that were very close to goal-achievement in 2000, but where their past rates of growth were either negative or so insignificant that parity would not be delivered were such experience to be continued. The nine such cases are mentioned in chapter 2. In cases where a simple extrapolation of past rates of growth would cause the goal to be exceeded (i.e. where previous inequality in favour of boys would thus become an inequality in favour of girls, or vice versa), it is assumed that no ‘overshooting’ will occur and that parity will be maintained. In practice, however, it should be noted that there have been a number of countries where such crossovers have occurred – particularly at secondary level – and where a male bias in enrolments has been replaced by a female bias. Table A2.3 summarizes the methods used in dealing with each possible pairing of values for GPI at primary and secondary levels in 1990 and 2000. Table A2.4 compares the outcomes of this methodology – for primary education – with the outcomes that would result if the methodology used in the EFA Report 2002 had been used. The columns show the 2002 quadrant framework categories, while the rows show the categories used by the new methodology. It can be seen that there is no difference between the two methods in the number of countries considered to have already achieved parity, because each uses the same definition for goal achievement. For all the remaining countries the new methodology gives more precision about the likelihood of different countries achieving the goals. There are sixteen countries in Quadrant I (close and moving towards the goal), all of which would have been considered to have high chances of achieving the goal. However, the new methodology shows that nine of them are likely to achieve it by 2005 and five by 2015. Additionally, it also suggests that two countries with a very low rate of progress would not be able to reach the goal. With the exception of Cuba and Paraguay, all the countries considered at risk in one methodology are also shown to be at risk in the other. Finally, some countries that the quadrant framework would have classified as having low chances of achieving the goal (Quadrant III), would achieve it according to the present method (three of them by 2005 and seven by 2015) Thus, the results from using each of these methods are reasonably consistent. Most differences between the EFA Report 2002 and this report in the extrapolated likelihood of particular countries reaching the gender goals are caused by changes in data and, more particularly, by the incorporation of secondary level parity in the treatment of the gender goal for this report. 1. However, as explained below, the gender component of the EDI is itself a composite index comprising measures of gender parity in primary education, secondary education and adult literacy.