OOSCI Operational Manual: Data sources and profiles of children
Process of assessing data sources, calculating indicators and analysing disaggregated data to produce the main profiles of out-of-school children and children at risk of dropping out
Researchers of an OOSCI national study must consider multiple complementary data sources, because the limitations inherent in each imply that no single source will be sufficient to provide a complete profile of out-of-school children and children at risk of dropout.
The data sources and indicators section of the OOSCI Operational Manual includes the steps required to source, appraise, and prepare the data needed to conduct an OOSCI study, identify gaps and limitations, and produce the basic quantitative profile of out-of-school children and children at risk of dropout (RODO) in the country, including the calculation of numbers and rates for each dimension of exclusion.
Step 1: Review data sources, quality, gaps and limitations
Every OOSCI study should contain a brief section that outlines the quantitative and qualitative data sources examined, the rationale for using those retained and a discussion of the data limitations and advice on the interpretation of indicators.
Out-of-school children study teams are encouraged to access, download and use the Data Inventory and Quality Assessment Tool. This tool is designed to produce a comprehensive overview of data sources available in the context of an out-of-school children study, assess the reliability of each, and identify important differences between them that may lead to different estimates of the number of out-of-school children and RODO.
Include as many columns as required to cover all sources of population data, enrolment and attendance data, and other data on out-of-school children and RODO, collected during the last five years (or more, if a comparison of trends over time is desired). Examples for household survey data and administrative data are included in the Data Inventory tool.
Include information on data collection systems and sources that are both national in coverage, or sub-national but provide information on out-of-school children for a specific geographic region of the country (for example, a province or state) or for a specific population group or minority.
While this step is the first in the process of developing an out-of-school children study, the tools it covers may be of use beyond this chapter and should be revisited as appropriate at later stages. For instance, once the initial quantitative data analysis covered in this chapter is complete, the findings may highlight the need to consider further data sources, or point to further research, for instance to develop profiles or identify barriers. If so, new sources discovered, or surveys carried out can be added and reviewed at later stages.
In performing the exercise, reference may be made to metadata, questionnaires, codebooks and existing analytical reports, to gain a better understanding of the data. All such reference material should be retained and listed.
View Section 3.1 of the OOSCI Operational Manual for further guidance on common data sources and issues, creating a data inventory, assessing data quality, and documenting data gaps and limitations.
Step 2: Calculate the 7DE indicators
The Out-of-School Children Initiative focuses on children out of school and those at risk of dropping out, over a wide age range. To help distinguish distinct groups of children for analysis and policy support, it uses a dimensions of exclusion framework, where each group of children is represented by a particular dimension. In line with the SDG 4 commitment to achieve universal primary and secondary education, the OOSCI dimensions of exclusion model has been expanded. It is now called the Seven Dimensions of Exclusion (7DE) and includes two dimensions relating to youth of upper secondary age.
Each dimension of exclusion represents a distinct group of children that can be quantified and analysed using statistical methods to identify the particular characteristics (or profiles) of the children most likely to be excluded.
After determining which data sources to use for analysis, indicators can be calculated, and the relevant data tables can be generated. Study teams are encouraged to use the 7DE Calculation Tool, a bilingual (English-French) Excel-based tool available soon, that has been specifically developed for this purpose.
The section on 7DE in Chapter 1 of the OOSCI study should present the key indicators for the rate and number of children in each of the dimensions. Teams are encouraged to select the tables according to the most relevant data in their context, and determine which tables belong in the main text and annexes. The section can be relatively short, providing a brief overview of the latest 7DE values, with a brief analysis of national or regional trends.
View Section 3.2 of the OOSCI Operational Manual for more detail on the Seven Dimensions of Exclusion and using the 7DE Calculation Tool, including greater clarity in terms of indicator definitions for the 7DE and an explanation of its underlying approach. For example Stata statistical software code to calculate education indicators, see Operational Manual Annexes G, H and I.
Profiles of children
The profiles of children section of the OOSCI Operational Manual describes the process of drafting Chapter 2 of the study. This includes the steps required to produce the main profiles of out-of-school children and children at risk of dropping out of school.
A ‘profile’ is a group of children in one or more of the 7DE with certain shared characteristics. Profiles presented in the OOSCI study should be created for relatively large groups of out-of-school children (scale of exclusion) or for groups in which the out-of-school rate is relatively high (severity of exclusion). The profiles chapter should highlight results from relevant indicators, disaggregated analyses and qualitative data.
Identifying the profiles of children, adolescents and youth most likely to be out of school, or at risk of dropping out can involve determining:
- If specific groups face higher out-of-school/risk of dropout rates, such as children with disabilities, a given ethnic minority or internally displaced children;
- The most common characteristics of out-of-school children and children at risk of dropping out, such as household wealth, work status or area of residence;
- The educational experiences of out-of-school children, whether certain groups are more likely to have dropped out (and at what age and grade), enter in the future, or unlikely to enter at all; or
- If specific locations, such as regions or districts, out-of-school/risk of dropout rates that are considerably higher than the national average.
Step 3: Conduct disaggregated data analysis
Disaggregated analysis is the comparison of indicator values for different characteristics or groups, to determine for whom and where the numbers or rates are the highest. Disaggregated data analysis is crucial to determine the key profiles of children, adolescents and youth out of school and at risk of dropping out. This is the analysis of population subgroups, which is important for understanding the individual, household, school, or community characteristics of children in the 7DE.
The purpose of disaggregated analysis is to identify groups of children that experience higher rates of school exclusion and risk of exclusion, to later identify the specific barriers they face and develop solutions to reach them.
Household survey data, dedicated surveys and administrative data can all be used in disaggregated analysis to develop key profiles of out-of-school children and children at risk of dropping out. However, because household surveys collect data on children in and out of school, with a range of individual and household characteristics, it is particularly valuable for the profiles analysis described here.
Tailoring the disaggregation of data to the most critical and relevant issues for national policies will help the relevance of findings for policy recommendations. For example, if the education system is decentralized and decision-making occurs at the province or district level, it is important to try to provide robust estimates at the sub-national level.
View Section 4.1 of the OOSCI Operational Manual for further detail. For guidance on data for specific groups of children, see Operational Manual Annexes C, D, E, and F.
Step 4: Analyse the flow of children in and out of the education system
Analysts are also advised to present complete profiles of children who left school early by identifying at what level and grade they left school. Step 4 enhances the findings in Steps 1, 2 and 3 by considering interaction with the education system over time to understand school exclusion. This step looks at the constriction in flows of children through the education system and identifies points in time, or critical milestones, where children are ‘lost’ from the education system. It builds on the analysis of out-of-school children by school exposure described earlier.
Here are some common points in time – points of constriction – that may create or worsen educational exclusion:
- (Non-) or late entry into school;
- Repetition, which may be more common in Grade 1 or in grades coinciding with national examinations;
- Promotion between grades; and
- Transition between levels of education, particularly from basic to upper secondary, which often faces greater supply constraints and has higher expectations of learning.
There are two primary methods to identify exclusion points: current trend analysis and retrospective or pathway analysis. Disaggregated analysis of the results of either approach listed above may show that different groups of children face different exclusion points. For example, children whose mother tongue is not the language of instruction may face much higher repetition rates in Grade 1, or rural children may have lower transition rates to upper secondary education than urban children, due to lack of nearby schools. Such dynamic analysis provides insights into the particular moments in the schooling system that merit further analysis and attention.
View Section 4.2 of the OOSCI Operational Manual for more detailed information on current trend analysis and pathway analysis.
Step 5: Cumulative risk analysis and other multivariate analyses
The data to develop profiles of children in the 7DE can be analysed using multivariate regression models. Such models are used to identify the strongest determinants of being out of school or dropping out, among the range of individual, household, community and school characteristics.
Cumulative risk analysis (CRA) refers to the usage of simple line graph to show how the probability of being out of school changes as risk factors (such as disadvantaged background characteristics) cumulate. It looks at the ‘added’ impact of disadvantaged background characteristics. Based on data availability, a standardized CRA considers four risk factors of being a girl, living in rural area, coming from poorer family, and having a less educated mother. CRA can be conducted for Dimensions 2, 3 and 6. The Stata code to calculate CRA can be found in Annex I of the OOSCI Operational Manual.
The added value of cumulative risk analysis is to shift the focus from the correlation between being out of school and various risk factors and background characteristics (as done in profiles analyses described earlier) to the causal inference between these characteristics and being out of school by decomposing the joint influence of various risk factors.
Correlation does not equal causality. CRA analysis moves beyond a simple tabulation of out-of-school rates and individual background characteristics. For example, rural children may be more likely to be out of school, but we cannot conclude that living in rural areas leads to being out of school. Rural areas may have more poor households (larger economic constraint) and lower levels of adult literacy (lower capacity to supervise children’s learning). These factors are all common drivers of being out of school. Due to the correlation between location, household wealth and adult literacy, a simple rural/urban tabulation might hide the fact that a rich rural child with educated parents could have the same possibility of being out of school as his urban peer. In contrast, a CRA decomposes the various influences to tell a more accurate story about how much the rural status affects school attendance, disregarding wealth and parental education. As such, the CRA analysis helps analysts identify more detailed profiles of children who face the highest risks of being out of school.
View Section 4.3 of the OOSCI Operational Manual for more detail on calculating the CRA for profiles analysis.
Step 6: Identify key profiles of out-of-school children and children at risk of dropping out
The OOSCI profiles chapter should present a synthesis of the most important characteristics and information about the children in the 7DE. This will form the basis for the barriers analysis and the focus for policies and strategies.
For each dimension of exclusion, the chapter should clearly explain: who is most likely to be out of school or at risk of dropping out, where they live and what kind of school exposure they have. It should also note major data gaps or highlight where there may be ‘invisible’ out-of-school children that the barriers analysis should also consider.
In addition, the profiles chapter should include a summary table of the main characteristics or risk factors that cut across the dimensions of exclusion.
View Section 4.4 of the OOSCI Operational Manual for a summary table example and further information.