NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Back to History Current Version
Cluster Randomised Trial of the effect and Timing and Duration of Early Childhood Interventions in Odisha, India
Last registered on August 21, 2018


Trial Information
General Information
Cluster Randomised Trial of the effect and Timing and Duration of Early Childhood Interventions in Odisha, India
Initial registration date
July 29, 2018
Last updated
August 21, 2018 9:39 AM EDT
Primary Investigator
Other Primary Investigator(s)
PI Affiliation
Pratham Education Foundation
PI Affiliation
Institute for Fiscal Studies
PI Affiliation
Inter American Development Bank
PI Affiliation
Institute for Fiscal Studies
PI Affiliation
University of Pennsylvania
PI Affiliation
University College London
PI Affiliation
Yale University
Additional Trial Information
On going
Start date
End date
Secondary IDs
The importance of early life has been highlighted in three series on Early Childhood Development (ECD) published in The Lancet between 2007 and 2016 and in many other publications. One of the key conclusions of this burgeoning literature is that stimulation interventions can be life-changing with long-run persistent effects leading to tangible improvements in quality of life through adulthood. There have been an increasing number of studies, including randomized trials, of interventions (typically based at home) during the first life-cycle stage, from pregnancy to 2 years or interventions (often center-based) during the second life-cycle stage (3 to 5). Ultimately, the aim of such interventions is to break the intergenerational transmission of poverty and to offer equal opportunity to all. However, there are three critical questions outstanding:
(i) how to design scalable ECD interventions that can reach broad segments of poor populations in an effective and sustainable way and lead to longer-run impacts,
(ii) at what stage should one intervene in a child’s development and for how long, and
(iii) can such approaches be adopted effectively in the contexts of rural South Asia, wherein live almost half of the world’s children under 5 years of age who are predicted to not reach their developmental potential.
To answer the questions above about optimal timing requires however a randomized design at both the earliest and the next stage. In this study we implement the second phase of such a design in 192 villages in rural Odisha (India), with randomized curricular improvements for the pre-school stage (ages 3-5 years) overlaid on a completed randomized intervention to improve parental stimulation during the first 1-3 years of life. With such a design, some children will be exposed to our interventions in both the first 1-3 years of life and when 3-5 years old, some children will be exposed in one or the other of these two age ranges, and some will not be exposed at all, but will of course have access to whatever is currently provided in these communities by the government or the private sector. Odisha provides an excellent setting for the study as it represents a prototypical context of low levels of stimulation, high malnutrition, poor developmental outcomes and extreme levels of poverty. Only with such a design can we confidently learn answers to the critical questions about timing that are raised above.
External Link(s)
Registration Citation
Attanasio, Orazio et al. 2018. "Cluster Randomised Trial of the effect and Timing and Duration of Early Childhood Interventions in Odisha, India." AEA RCT Registry. August 21. https://doi.org/10.1257/rct.3161-2.0.
Former Citation
Attanasio, Orazio et al. 2018. "Cluster Randomised Trial of the effect and Timing and Duration of Early Childhood Interventions in Odisha, India." AEA RCT Registry. August 21. http://www.socialscienceregistry.org/trials/3161/history/33342.
Sponsors & Partners

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information
Experimental Details
Currently pre-school services are provided through the Anganwadi program, started by the Indian government in 1975 as part of the ICDS program to combat child hunger and malnutrition. A typical Anganwadi center provides supplementary nutrition, non-formal pre-school education, nutrition and health education, immunization, health check-up and referral services of which latter three services are provided in convergence with public health systems. The non-formal Pre-School Education component of the ICDS for 3-to-6 year-old children in the Anganwadis is directed towards school readiness and development of positive attitudes towards education and is intended to contribute to the universalization of primary education, by providing to the child the necessary preparation for primary schooling and offering substitute care to younger siblings, thus freeing the older ones - especially girls - to attend school. However, the pre-school aspect of the program, is very weak and has a particularly low awareness in the wider community.

Our treatment is an enhanced preschool program (EPP) organized in the Anganwadi centers and accompanied by parenting classes embedded in currently existing parental meeting at the center. This includes the implementation of the enhanced curriculum, along with all of the materials required to implement it, and the assistance of a Pratham Mentor who will attend each center 2 days per week for 18 months. Our starting point for the curriculum is the existing one used by the ICDS, which the research team, including CECED and Pratham, has enhanced further drawing on existing expertise and knowledge of those working with young children in India.
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
Key Final Outcomes
1. Children’s child cognitive development
2. Children’s school readiness

Primary Outcomes (explanation)
1. We will use the Wechsler Preschool and Primary Scale of Intelligence (WPPSI, Wechsler, 1967; Wechsler, 2002) to assess cognition and language. The WPPSI is widely-used internationally to assess intellectual function and contains the following subscales: Verbal Comprehension Index, Visual Spatial Index, Working Memory Index, Fluid Reasoning Index, and Processing Speed Index. We will adapted for the context and choose the subscales that work best.
2. Daberon-II Screening for School Readiness test (Danzer et al., 1991) and/or the School Readiness Instrument (which was designed and standardized by The World Bank in India and the new version of this tool will be standardized by CECED at the end of 2018)

Secondary Outcomes
Secondary Outcomes (end points)
Key Intermediate Outcomes
1. Children's language development, executive function and behaviour.
2. Preschool quality
3. Child anthropometric outcomes
4. Child rearing practices: level of stimulation in the home
5. Parental knowledge of child rearing practices
Secondary Outcomes (explanation)
1. We will assess behavioral problems and prosocial behavior by mothers’ report with the Strength and Difficulties Questionnaire (Goodman 1997). Language development and executive function will be measured using the relevant subscales of the WPPSI.
2. We will measure preschool quality using a new instrument that we will develop (Preschool Assessment Tool) that will be based on the Early Childhood Education Quality Assessment Scale (ECEQAS, which is an observation rating scale developed by CECED at Ambedkar University, Delhi along the lines of Early Childhood Education Rating Scale (ECERS)), the CLASS (Classroom Assessment Scoring System) Pre-K and the TIPPS ( The Teacher Instructional Practices and Processes System) developed by Global TIES for Children, NYU University
3. Anthropometric measures will be conducted to assess height-for-age, weight-for-age, and weight-for-height z scores and thus the extent of growth faltering and growth recovery using standard procedures.
4. The presence of toys and learning materials in the house will be assessed together with parental involvement with the child, the child’s routines and organisation of the child’s time inside and outside the family house. This will be assessed using the Family Care Indicators, developed by UNICEF, and possibly selected subscales of the Home Observation for the Measurement of the Environment (HOME).
5. We will collect information on the mother’s knowledge of nutrition and stimulation, and her beliefs regarding the importance of these for children’s development. To test knowledge, we will rely on an adapted selection of items from the Knowledge of Infant Development (KIDI) as well as another instrument developed as part of the 0-3 intervention to elicit maternal beliefs on the effect of parental investment on child outcomes.
Experimental Design
Experimental Design
The evaluation will take place using a cluster-randomized controlled trial and will be based on two further rounds of data collection. The initial phase of the trial has three data collection rounds: a baseline survey (completed and report is available), one at midline (December 2016) and one at endline (December 2017). In this second phase, we will build on this data collection with two further rounds: one in June 2019 (midline of the new project) and one in December 2019. The end product will be a unique data set tracking child development from age 1-5 with children having received different intervention combinations in a randomized fashion.

Overall there are 192 communities in our experimental population. Each current intervention arm (three in total) includes 48 communities and we also have 48 control communities. The allocation of a community to an intervention or control arm was random.

In the proposed project extension, we intend to randomize half the communities of each current experimental arm to the new intervention described above. This means that we will include 96 communities in the new intervention, while the remaining 96 will not be included. Randomization will ensure unbiased comparisons with the control group and across alternative arms. All children will be evaluated based on the same measurements. Hence a comparison of children who only received the first part of the intervention with those who received only the second part will reveal the effect of starting early versus later – a key question of this research; a comparison of those always in the intervention with those only later will quantify the extent to which the two components reinforce each other and so on.

Importantly the data include detailed household and parental characteristics, which will allow us to understand the heterogeneity of outcomes across different groups. Specifically, we will be considering how the educational level of the mother affects child outcomes and we will focus on how parental investments in time and materials (which we measure) are affected by the interventions and in turn reinforce them.
Experimental Design Details
Not available
Randomization Method
Randomization will be done in office by a computer using the statistical software stata, version 14.1.
Randomization Unit
Randomization will be carried out at the village level.
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
192 villages (clusters), 96 in treatment and 96 in control.
Sample size: planned number of observations
The total sample sizes will be 1,427 families with children who were a part of the initial phase of the trial, distributed among the two treatment arms.
Sample size (or number of clusters) by treatment arms
96 communities in treatment and control for the new trial, creating 8 arms of 48 clusters each.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Overall there are 192 communities in our experimental population with a total of 1427 children and their mothers. Each original intervention arm (three in total: Stimulation by home visits, stimulation by playgroups, nutritional education) includes 48 communities and we also have 48 control communities. The allocation of a community to an intervention or control arm was random. In the proposed pre-school intervention, we will randomize half the communities of each current experimental arm to the new enhanced quality Anganwadi intervention. This means that we will include 96 communities in the new intervention, while the remaining 96 will not be included. By using data from the baseline we found the intracluster correlation for a number of measures (ASQ and anthropometrics) to be no more than 0.1, which is the number we assume below. The detectable effects below are all in standard deviation units of any outcome of interest. The calculations are based on a simple comparison of means. In practice the power is likely to improve (leading to even smaller detectable effects), because we will use baseline characteristics as controls to increase precision. The sample is large enough to detect effects of policy-meaningful size with significance below 5% and power ≥ 80%: It is unlikely that policy makers would want to implement a program with smaller effects than these. By sub treatment arm the minimum detectable effect is as follows; overall new intervention (96(T) and 96 (C)) 0.19 of a SD; receiving both the Anganwadi intervention and early stimulation versus no stimulation intervention and no Anganwadi intervention (48(T) and 48 (C)) 0.27 of a SD; receiving the Anganwadi intervention for those who received no early stimulation treatment versus no early stimulation and no Anganwadi intervention 48(T) and 48 (C) 0.27 of a SD ; receiving the Anganwadi intervention on those not treated by early Stimulation compared to those who only got Early Stimulation 48(T) and 48 (C) 0.27 of a SD ; receiving the Anganwadi intervention and Early Stimulation versus those who only got early Stimulation. 48(T) and 48 (C) 0.27 of a SD. The “no early stimulation” group includes the pure control group and the group that received nutritional education in the first phase of the project. The nutritional education arm of the early intervention had no statistically significant effect on any of the primary outcomes of the trial. The “early stimulation group” includes those who received early stimulation in the first phase of the program either through weekly home visits or through the weekly playgroups. However, it is also possible to estimate finer effects, where for example we distinguish between types of stimulation (home visiting versus group). In this case the minimum detectable effects are between 0.33 and 0.38 of a standard deviation of the outcome, depending on the specific comparison.
IRB Name
Institute for Financial Management and Research (IFMR)
IRB Approval Date
IRB Approval Number
IRB00007107; FWA00014616; IORG0005894
IRB Name
Pratham Ethics Committee
IRB Approval Date
IRB Approval Number
IRB Name
Yale Human Subjects Committee
IRB Approval Date
IRB Approval Number
IRB Protocol ID: 1112009492 Submission ID: MOD00006266
IRB Name
University College London Research Ethics Committtee
IRB Approval Date
IRB Approval Number