Thrive by Five Index and ECD Baseline Audit 2021
South Africa, 2021
Get Microdata
Reference ID
zaf-dbe-dd-tb5i-ecda-2021-v4
Producer(s)
DataDrive2030, Department of Basic Education
Collections
Metadata
Created on
Oct 13, 2023
Last modified
Jun 16, 2026
Page views
202670
Downloads
2137
Identification
Survey ID number
zaf-dbe-dd-tb5i-ecda-2021-v4
Title
Thrive by Five Index and ECD Baseline Audit 2021
Abbreviation or Acronym
TBFI 2021
Country
| Name | Country code |
|---|---|
| South Africa | zaf |
Study type
Child Assessment Survey
Abstract
The cumulative effect of South Africa’s investments in early childhood development services – good health care and nutrition, parenting support, high quality early learning programmes (ELPs), and child safety and protection – can be determined by measuring the proportion of children who are developmentally ‘on track’ by the age of 5 years. No national data currently exists to track this important indicator. The South African Thrive by Five Index (formerly the SA Early Years Index) intends to address this gap. The Index will monitor trends over time in the proportion of preschool children who are on track for age in key areas of development. The aim is to use the data to strengthen collective efforts to ensure that more children receive the full suite of nurturing care and services they need to be On Track, and to support those children whose development is Falling Behind. The Index was initiated by First National Bank and Innovation Edge, in collaboration with the Department of Basic Education (DBE), and supported by the United States Agency for International Development (USAID) and ECD Measure. The sample consists of a nationally representative selection of children aged 50-59 months who are enrolled in various types of ELPs. Data was collected electronically using tablet devices, in-person. For each child, the following developmental domains were assessed:
1. Early learning outcomes (measured using the ELOM 4&5 direct assessment tool),
2. Social and Emotional Functioning (measured using the ELOM Social and Emotional Functioning Rating Scales)
3. Height for Age (measured using stadiometers)
Within each cluster of Early Learning Programmes (ELPs) surveyed, 1 ELP was selected to "audit". This involved interviews with the principal and one ECD practitioner responsible for a class of 4-5 year old children, an assessment of the overall environment and infrastructure plus a 2 hour observation of the quality of the early learning programme. This dataset is referred to as the ECD baseline audit. The ECD audit data is merged with the Thrive By Five assessment data in the current dataset.
1. Early learning outcomes (measured using the ELOM 4&5 direct assessment tool),
2. Social and Emotional Functioning (measured using the ELOM Social and Emotional Functioning Rating Scales)
3. Height for Age (measured using stadiometers)
Within each cluster of Early Learning Programmes (ELPs) surveyed, 1 ELP was selected to "audit". This involved interviews with the principal and one ECD practitioner responsible for a class of 4-5 year old children, an assessment of the overall environment and infrastructure plus a 2 hour observation of the quality of the early learning programme. This dataset is referred to as the ECD baseline audit. The ECD audit data is merged with the Thrive By Five assessment data in the current dataset.
Kind of Data
Other
Unit of Analysis
Individuals
Version
Version Description
v4: Edited, anonymised data for public distribution
Version Date
2023
Version Notes
Version 3 of the Thrive by Five Index (2021) data was uploaded on 2 April 2023. Version 3 of the data is the Thrive by Five Index data merged with the ECD Baseline Audit data, as in version 1. Changes were made to Version 1 to produce Version 2 of the data, but that version was superceded by Version 3 before it could be published.
Changes in version 3 are the inclusion of a weight variable and the corrections to an item, domain and total ELOM score.
An error was discovered in the digital tablet scoring of Emergent Numeracy and Maths (ENM) item 10 (addition trial), where the programme recorded correct answers as incorrect and some incorrect answers as correct. Due to this programming error, ENM total scores were incorrectly calculated and have now been corrected. The effect of the correction has resulted in a slight increase in the score for the ENM total scores, as well as a slight adjustment of the ELOM 4&5 Total Score. The error in the tablet form has been corrected and no further errors have been detected. It is important to note that only children whose scores were close to one of the ELOM standard score bands would be likely to be misclassified. Data files in the public domain dataset hosted by DataFirst have been corrected.
Version 4 has corrections to the weights in the data the additional variable 'quintile_natemis' which is described below (from page 38 of the Thrive by Five Technical Report).
Description of the 'quintile_natemis' Variable
"The quintile rank system is a valuable proxy for the wealth status of the children assessed and was used for stratification during the sampling process, as well as for disaggregation in the analysis. There are three approaches that can be used in assigning a quintile status to an ELP:
Using the quintile status of the primary school that was used when constructing the sample (quintile_original);
Using the quintile status of the closest primary school in the school sample used to identify clusters (quintile_sample); or
Using the quintile status of the closest school in the DBE 2021 Masterlist data (quintile_natemis).
The research team decided to use the quintile status of the primary schools that was used to construct the sample (quintile_original) for both the construction of the weights, as well as for disaggregation.For the construction of weights, the quintile_original variable is most appropriate because it determined the probability of an ELP having been sampled. For disaggregation, the variable quintile_original was deemed the most conservative choice of classification to use, since it will not introduce any additional measurement error that cannot be accounted for (for example, the closest school being in a typical quintile 5 area, while the ELP is in a neighbouring Q3 area)."
Changes in version 3 are the inclusion of a weight variable and the corrections to an item, domain and total ELOM score.
An error was discovered in the digital tablet scoring of Emergent Numeracy and Maths (ENM) item 10 (addition trial), where the programme recorded correct answers as incorrect and some incorrect answers as correct. Due to this programming error, ENM total scores were incorrectly calculated and have now been corrected. The effect of the correction has resulted in a slight increase in the score for the ENM total scores, as well as a slight adjustment of the ELOM 4&5 Total Score. The error in the tablet form has been corrected and no further errors have been detected. It is important to note that only children whose scores were close to one of the ELOM standard score bands would be likely to be misclassified. Data files in the public domain dataset hosted by DataFirst have been corrected.
Version 4 has corrections to the weights in the data the additional variable 'quintile_natemis' which is described below (from page 38 of the Thrive by Five Technical Report).
Description of the 'quintile_natemis' Variable
"The quintile rank system is a valuable proxy for the wealth status of the children assessed and was used for stratification during the sampling process, as well as for disaggregation in the analysis. There are three approaches that can be used in assigning a quintile status to an ELP:
Using the quintile status of the primary school that was used when constructing the sample (quintile_original);
Using the quintile status of the closest primary school in the school sample used to identify clusters (quintile_sample); or
Using the quintile status of the closest school in the DBE 2021 Masterlist data (quintile_natemis).
The research team decided to use the quintile status of the primary schools that was used to construct the sample (quintile_original) for both the construction of the weights, as well as for disaggregation.For the construction of weights, the quintile_original variable is most appropriate because it determined the probability of an ELP having been sampled. For disaggregation, the variable quintile_original was deemed the most conservative choice of classification to use, since it will not introduce any additional measurement error that cannot be accounted for (for example, the closest school being in a typical quintile 5 area, while the ELP is in a neighbouring Q3 area)."
Scope
Notes
The sample consists of a nationally representative selection of children aged 50-59 months who are enrolled in various types of ELPs. For each child, the following developmental domains were assessed:
- Early learning outcomes (measured using the ELOM 4&5 direct assessment tool), including:
- Gross Motor Development: This domain assesses the abilities required to control the large muscles of the body. Ideally, we would want to see children showing good control and coordination in their large movements.
- Fine Motor Coordination and Visual Motor Integration: This domain assesses the abilities required to control the small muscles of the body and to coordinate fine motor movements with visual information that is perceived by the eyes. This allows coordinated movements such as copying shapes.
- Emergent Numeracy and Mathematics: This domain assesses the ability to understand number concepts, symbols, shapes, and size.
- Cognition and Executive Functioning: This domain assesses the ability to think critically, solve problems, form concepts, attend to instructions, and control impulses.
- Emergent Literacy and Language: This domain assesses the ability to communicate effectively, use language and demonstrate listening comprehension.
- Social and Emotional Functioning (measured using the ELOM Social and Emotional Functioning Rating Scales): This domain covers self-care, social relations with peers and adults, and emotional readiness for school.
- Height for Age (measured using stadiometers): This domain measures whether a child is at risk of stunting. Stunting is a condition that arises from prolonged under-nutrition and it affects physical and brain development.
The child outcomes data is complemented by baseline assessments (audit) of selected Early Learning Programmes with respect to the type of early learning services provided, staff qualifications and training, resources and funding, quality of the learning environment, as well as access to government support services. Programme quality was measured using the ELOM Programme Quality Assessment Tool and interviews were conducted with principals and practitioners.
Coverage
Geographic Coverage
The lowest level of geographic representivity of the data is Ward-level. The data is also available at the level of Local Municipality.
Producers and sponsors
Primary investigators
| Name | Affiliation |
|---|---|
| DataDrive2030 | |
| Department of Basic Education | Government of South Africa |
Funding Agency/Sponsor
| Name | Abbreviation | Role |
|---|---|---|
| DataDrive2030 | DD | Funding Agency |
| ECD Measure | ECDM | Funding Agency |
| First National Bank Foundation | FNBF | Funding Agency |
| Innovation Edge | IE | Funding Agency |
| US Agency for International Development | USAID | Funding Agency |
Sampling
Sampling Procedure
Thrive by Five Index
The Index has child outcomes data on over 5000 children sampled from 1200+ early learning programmes (ELPs) / ECD sites nationally. In the absence of a complete sampling frame for either children or ELPs, it was decided to cluster ELPs via primary schools for random sampling. In each province, 48 schools (432 nationwide) were randomly selected to be used as clusters. As many ELPs were then identified as possible within a 5-10km radius around each school, or, where this impractical because there were not enough ELPs in the vicinity of the schools, as the ward in which the school is located. The school sample in each province is stratified by socio-economic quintiles (1 to 5). Child assessments (ELOM 4&5, ELOM socio-emotional rating and height for age) were conducted with an average of 4 children at each of 3 randomly selected ELPs per cluster.
ECD Baseline Audit
Within each cluster of 3 ELPs, 1 ELP was selected to "audit". This involved interviews with the principal and a practitioner (the practitioner working with the 4 year old children), an assessment of the overall environment and infrastructure plus a 2 hour observation of the quality of the early learning programme. This data is referred to as the ECD baseline audit. The Index and Baseline data are combined in the final publicly-available version of the dataset.
The Index has child outcomes data on over 5000 children sampled from 1200+ early learning programmes (ELPs) / ECD sites nationally. In the absence of a complete sampling frame for either children or ELPs, it was decided to cluster ELPs via primary schools for random sampling. In each province, 48 schools (432 nationwide) were randomly selected to be used as clusters. As many ELPs were then identified as possible within a 5-10km radius around each school, or, where this impractical because there were not enough ELPs in the vicinity of the schools, as the ward in which the school is located. The school sample in each province is stratified by socio-economic quintiles (1 to 5). Child assessments (ELOM 4&5, ELOM socio-emotional rating and height for age) were conducted with an average of 4 children at each of 3 randomly selected ELPs per cluster.
ECD Baseline Audit
Within each cluster of 3 ELPs, 1 ELP was selected to "audit". This involved interviews with the principal and a practitioner (the practitioner working with the 4 year old children), an assessment of the overall environment and infrastructure plus a 2 hour observation of the quality of the early learning programme. This data is referred to as the ECD baseline audit. The Index and Baseline data are combined in the final publicly-available version of the dataset.
Weighting
Use of the quintile status of the primary schools for the construction of the weights, and for disaggregation.
The research team decided to use the quintile status of the primary schools that was used to construct the sample (quintile_original) for both the construction of the weights, as well as for disaggregation.
For the construction of weights, the quintile_original variable is most appropriate because it determined the probability of an ELP having been sampled. For disaggregation, the variable quintile_original was deemed the most conservative choice of classification to use, since it will not introduce any additional measurement error that cannot be accounted for (for example, the closest school being in a typical quintile 5 area, while the ELP is in a neighbouring Q3 area)."
The research team decided to use the quintile status of the primary schools that was used to construct the sample (quintile_original) for both the construction of the weights, as well as for disaggregation.
For the construction of weights, the quintile_original variable is most appropriate because it determined the probability of an ELP having been sampled. For disaggregation, the variable quintile_original was deemed the most conservative choice of classification to use, since it will not introduce any additional measurement error that cannot be accounted for (for example, the closest school being in a typical quintile 5 area, while the ELP is in a neighbouring Q3 area)."
Survey instrument
Questionnaires
Instruments used to collect data were ELOM 4&5 direct assessments of young children; ELOM Social-Emotional rating scales (interview with child’s teacher), ELOM learning programme quality assessment (classroom observation tool), CAPI Questionnaire for Principals, CAPI Questionnaire for Practitioners, and a Facility Observation Questionnaire. Stadiometers were used to measure height.
Data collection
Dates of Data Collection
| Start | End | Cycle |
|---|---|---|
| 2021-09 | 2021-11 | Round 1 |
Data Collection Mode
- Other
Data Collectors
| Name |
|---|
| Ikapadata |
access_policy
Access conditions
Public access data for use under a Creative Commons CC-BY (Attribution-only) License
Citation requirements
DataDrive and Department of Basic Education. Thrive by Five Index and ECD Baseline Audit 2021 [dataset]. Version 4. Cape Town and Pretoria: DataDrive and DBE [producers], 2021. Cape Town: DataFirst [distributor], 2022. DOI: https://doi.org/10.25828/d0zx-6k11
Metadata production
Producers
| Name | Affiliation | Role |
|---|---|---|
| DataFirst | University of Cape Town | Metadata producer |
Date of Metadata Production
2023-11-01
Metadata version
DDI Document version
Version 6