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.
Evaluating integrated schooling models in Tamil Nadu
Initial registration date
April 08, 2020
April 09, 2020 7:28 PM EDT
Other Primary Investigator(s)
UC San Diego
Stockholm School of Economics
Additional Trial Information
This randomized control trial will measure the effectiveness of the Tamil Nadu Government’s policy intervention to improve early childhood education standards in government schools and shift student enrollment from private to public schools. The policy will implement LKG/UKG (lower and upper kindergarten) classes in Anganwadi Centers (AWCs) that are co-located in middle schools by assigning surplus government teachers to teach the LKG/UKG programs. By changing the provision of public education in these AWCs/schools, the intervention changes the menu of educational options in the entire market. Thus, the experiment and sampling are designed to measure market-level treatment effects along two main dimensions: 1) Learning outcomes of students at the end of kindergarten and through early primary grades, and 2) changes in enrolment shares across public/private/out-of-schooling sectors at the pre-primary and primary levels.
Muralidharan, Karthik, Mauricio Romero and Abhijeet Singh. 2020. "Evaluating integrated schooling models in Tamil Nadu." AEA RCT Registry. April 09.
The Government of Tamil Nadu (GoTN) notified, through its order of 11 Dec 2018, a pilot program of introducing LKG/UKG (lower and upper kindergarten) classes in Anganwadi centers (AWCs) which are co-located with middle schools. These classes would be conducted by teachers from the school. The aim of this policy is to provide high-quality early childhood education to children and to attract parents who would otherwise send their children to private lower- and upper-kindergarten (LKG/UKG) classes and save them the financial burden of paying for private schooling.
This policy capitalizes on the fact that Tamil Nadu, unlike many other states, has spare teacher capacity which can be used to improve the school readiness for incoming cohorts into Grade 1. Focusing in the first instance on AWCs that are located within the premises of middle schools allows for sharing manpower between the school and the AWCs, an optimization that is typically made very hard because the two facilities are overseen and administered by different ministries both in state governments and at the national level. If successful, this pilot may serve as a template for a much broader system-level integration of services across the ICDS and the school system
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Enrollment decisions (and enrollment shares)
Primary Outcomes (explanation)
Learning outcomes will be measured for all children in the village. We will conduct one-on-one testing at children's home.
Enrollment will be measured using data collected by us, as well as administrative data.
Secondary Outcomes (end points)
Household expenditure on education
Secondary Outcomes (explanation)
The Government of Tamil Nadu has identified a list of 2,381 potential AWCs which are eligible for inclusion in this pilot program. In four districts --- Kancheepuram, Tiruchirapalli, Karur and Virudhunagar --- for the purpose of evaluation of the program, the Government has agreed to a randomized rollout of the program. These districts have been randomly selected for a prior JPAL study on improving health and nutrition outcomes in Anganwadi centers and conducting this evaluation in the same districts will enable valuable additional comparisons for policy.
In four districts, a treatment group of 108 AWCs and a control group of 89 AWCs based on random allocation --- stratified at the block level --- has been selected. The study universe is restricted to rural areas where there are at least 2 AWCs per block which are co-located with middle schools.
Experimental Design Details
Randomization done in office by a computer
AWC center. Since at most one AWC in a village meets the eligibility criteria of being co-located with a middle school, thus randomization in this set of eligible AWCs is equivalent to village-level randomization.
Was the treatment clustered?
Sample size: planned number of clusters
197 AWC centers/villages
Sample size: planned number of observations
Roughly 20,000 households at baseline
We tested every child between 2-5 in these households (approx 22,000 children)
We surveyed every ECE facility in each village
Sample size (or number of clusters) by treatment arms
Treatment group ---108 AWCs
Control group --- 89 AWCs
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Assuming (conservatively), that we have information on 30 children per village at endline, an intra-village correlation of 0.05 (it's 0.025 at baseline), a correlation of 10% between the baseline and endline outcome, a size of 5% and a power of 90%, the minimum detectable effect is 0.125 SD.
INSTITUTIONAL REVIEW BOARDS (IRBs)
Institute for Financial Management and Research (IFMR)
IRB Approval Date
IRB Approval Number