The effect of affordable daycare on health and economic well-being over the life course in India: A cluster-randomized impact evaluation study

Last registered on December 09, 2020


Trial Information

General Information

The effect of affordable daycare on health and economic well-being over the life course in India: A cluster-randomized impact evaluation study
Initial registration date
July 15, 2015

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
July 15, 2015, 12:54 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated
December 09, 2020, 1:43 PM EST

Last updated is the most recent time when changes to the trial's registration were published.



Primary Investigator

McGill University

Other Primary Investigator(s)

PI Affiliation
McGill University
PI Affiliation
Institute for Financial Management and Research
PI Affiliation
Institute for Financial Management and Research

Additional Trial Information

Start date
End date
Secondary IDs
There are structural barriers to women's economic participation and empowerment in India. Among these barriers is the lack of affordable and reliable daycare services. The responsibility of providing childcare is imposed primarily on women and contributes to gender inequalities experienced by women over the life-course. The provision of affordable and reliable daycare services is a potentially important policy lever for reducing gender inequality and improving health and socioeconomic well-being over the life-course. Access to daycare might reduce barriers to labor force entry and generate economic opportunities for women, improve education for girls caring for younger siblings, and promote nutrition and learning among children. However, empirical evidence concerning the effects of daycare programs over the life-course in low-and-middle-income countries is scarce. We propose a cluster-randomized impact evaluation for estimating the effect of a community-based daycare program on health and economic well-being over the life-course among women and children living in rural Rajasthan, India. This interdisciplinary research initiative will address an important research gap and has the potential to inform policies for improving the daycare system in India in ways that promote inclusive economic growth.
External Link(s)

Registration Citation

Agarwal, Parul et al. 2020. "The effect of affordable daycare on health and economic well-being over the life course in India: A cluster-randomized impact evaluation study." AEA RCT Registry. December 09.
Former Citation
Agarwal, Parul et al. 2020. "The effect of affordable daycare on health and economic well-being over the life course in India: A cluster-randomized impact evaluation study." AEA RCT Registry. December 09.
Sponsors & Partners

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

Request Information
Experimental Details


The proposed intervention is the introduction of full time, affordable daycare centers, or balwadis, that will be implemented by Seva Mandir, in 80 hamlets in the Udaipur district of Rajasthan. These hamlets will serve as the treatment group for the randomized control trial. Each of the balwadis will provide childcare, nutritious food and supplements, basic medicines, and preschool education to children 1-6 years old. Balwadis are operated by local women, called Sanchalikas, who meet with mothers on a quarterly basis to discuss their child's progress. They also promote immunization of children by maintaining immunization records and following-up with parents and government nurses.

In December 2014 and early January 2015, potential study hamlets were selected. These hamlets were selected from areas where Seva Mandir works and operates programs besides the balwadi program. We identified 160 eligible hamlets that satisfied five criteria, specifically: (1) no readily accessible balwadi or government-run aganwadi nearby (within 1.5 kilometers) to reduce the potential for contamination effects; (2) minimum number of children in the appropriate age range (>24) in the village to ensure adequate demand; (3) existing building suitable for a balwadi; (4) a qualified Sanchalika, living in the study hamlet or nearby, to operate the balwadi; and (5) adequate demand from the village council (Panchayat) for a new balwadi, as expressed by an application for a balwadi. Seva Mandir provided a listing of 160 hamlets that satisfied these criteria to the study investigators, after which a household census was conducted in each hamlet to confirm the eligibility of the hamlet, enumerate the population, and identify potential households for inclusion.

In January 2015, we conducted a household census in each of the 160 hamlets to confirm the eligibility of the hamlet, enumerate the population, and identify potential respondents for inclusion. Eligible respondents were mothers (biological or guardian) with a child between one and six years of age. The total number of eligible respondents was similar to our desired sample size. Therefore, we did not probabilistically sample individuals among eligible respondents and all eligible respondents were selected for interview.

Within each eligible household, a woman with a child less than 6 years old was randomly selected as the primary respondent using a Kish procedure. Outcomes will be measured via in-person survey at baseline, just before the balwadis are introduced, at midline, approximately halfway through the intervention period, and at endline, approximately 18-24 months post the start of the intervention.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
We will measure the impact of our intervention on self-reported empowerment (self-esteem, role in decision-making, freedom of mobility, and control of material resources), measured via in-person survey. Other outcomes include labor market attachment (employment status, occupation, hours worked), economic status (personal and household income), time use, and mental health (depressive symptoms). Additionally, we will measure nutritional status (stunting, wasting, underweight) among children less than six years of age, as well as educational attainment and literacy among school-aged children.
Primary Outcomes (explanation)
Regarding the measurement of empowerment, we adopted the conceptual approach of Kabeer, which incorporates both the personal and political dimensions of empowerment, and developed indicators that captured "women's sense of self‐worth and identity, their willingness to question their own subordinate status, their control over their own lives and their voice and influence within the family". We constructed indicators of empowerment, adapted from the Indian NFHS4 whenever possible to facilitate comparability, which encompassed four domains: (1) decision making within the family and control over income (e.g., Who decides how the money you earn will be used: mainly you, mainly your husband, or you and your husband jointly?); (2) freedom of movement in the public domain (e.g., Are you usually permitted to go to the following places on your own, only if someone accompanies you, or not at all?); (3) participation in community and public life (e.g., Are you a member of any type of association, group or club which holds regular meetings?); (4) and views and attitudes on critical gender issues (e.g., Please tell me if you agree or disagree with each statement: A married woman should be allowed to work outside the home if she wants to). We tested the face validity of the questionnaire by consulting individuals with local expertise and added, deleted, and revised existing indicators. We pilot tested the questionnaire in sample of approximately 200 women. The questionnaire was modified accordingly based on the results of the pilot study. A data reduction technique (i.e., factor analysis) will be used to provide a summary score. We plan to test the reliability of empowerment measures during the second survey wave.

Use of time was measured using a structured questionnaire, adapted from a study by Beaman et al. (2012), that asked respondents whether they spent any time in the past 24 hours on specific activities (e.g., gathering fuel or firewood), how much time they spent on each activity, and whether this amount reflected the usual amount of time spent on the activity. The questionnaire also asks whether respondents were paid in cash or in-kind for the activities they engaged in.

The survey asks about employment experiences, including whether respondents work, their occupation, the type of work, the quantity of work, whether they are paid for their work in cash or in-kind, and what they do with their children while working. We will ask about household income received in the past 12 months from various categories (e.g., agricultural income, business income, rents, remittances, government payments). Household wealth will be measured using a series of questions about ownership of specific assets (e.g., telephone, bicycle, radio), environmental conditions (e.g., type of water source, sanitation facilities), and housing characteristics, (e.g., number of rooms, materials used for housing construction). Additionally, we will ask respondents about savings accounts held by household members, including for each account the type of account, its purpose, the total value, and whether the respondent can use the account to make purchases.

With respect to anthropometry, the equipment used will be calibrated daily before home or hospital visits. Two anthropometrists will record data independently and compare values. The average of these two values is taken, and any large discrepancies can be resolved by a first repeat measurement. Length is measured using a length board (sometimes called an infantometer) placed on a flat, stable surface such as a table. To measure height, a height board (sometimes called a stadiometer) will be mounted at a right angle between a level floor and against a straight, vertical surface such as a wall or pillar. If a child is less than 2 years old (or if age is unknown, less than 85 cm), recumbent length will be measured. If the child is aged 2 years or older (or if age is unknown, greater than 85 cm) and able to stand, standing height will be measured. The child's shoes, socks, and hair ornaments will be removed. Braids will be undone if they will interfere with the measurement of length/height. The Shorr height board is will be used to measure recumbent length. Adult and child heights will be measured using the Harpenden Portable Stadiometer (range 65-206 cm). For weight measurement tared weighing will be performed for children less than 2 years old. If the child is 2 years or older the will be weighed alone (if the child can stand still). The SECA 814 digital scale (up to 200kg) will be used to measure child weight. To measure arm circumference, any clothing covering the child's arms will be removed. The midpoint of the child's upper left arm will be calculated and marked with pen on the child’s arm. The child's arm will be straightened and the tape used to measure the circumference. While measuring, the tape should be flat on the child's arm. The measurement should be recorded to the nearest 0.1 cm. Circumferences will be measured using a metal tape (range 0-200 cm). The suggested instrument for measuring mid-upper arm circumference (MUAC) is the MUAC Tape (UNICEF Item No. 145600 Arm circumference insertion tape/pack of 50).

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
The intervention, i.e. the introduction of balwadis, will be randomly assigned to 80 of the 160 hamlets. The remaining 80 hamlets will serve as comparison group. The survey, which measures the study outcomes, will be administered at the baseline (before the intervention starts), at the mid-line (one year after the intervention starts), and at the end-line (when the intervention/program ends).
Experimental Design Details
Randomization Method
Randomization was done in office by a computer (using Stata software's random number generator)
Randomization Unit
We randomly assigned the 160 hamlets to treatment or control stratified by block (Badgaon, Girwa, Jhadol, Herwara, Kotra) because we wanted to prevent random variations in the distributions of blocks across treatment groups from confounding effects (e.g., if women in treated hamlets were more likely to reside in blocks with more economic opportunities). A random integer was assigned to each of the five blocks and each of the 160 hamlets using a random number generator.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
160 village hamlets
Sample size: planned number of observations
3200 individuals
Sample size (or number of clusters) by treatment arms
80 treated hamlets; 80 control hamlets
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
In the context of clustered treatment assignment the minimum detectable effect (MDE, Bloom 1984) will vary as a function of the extent of clustering. We calculated a range of MDEs under different clustering scenarios for our primary binary outcome, labor force participation, conditional on the number of clusters (j) the number of subjects per cluster (n) and the degree of clustering (estimated by the intra-cluster correlation coefficient [ICC]). We focus on estimating the MDE for a pre-specified level of power, based on our initial design of j=160 clusters (80 treated, 80 comparison) and n=20 subjects per cluster. Plugging in values for type II error (kappa=0.20), type I error (alpha=0.05), a balanced 50% allocation fraction (P=0.5), the standard deviation of the outcome in the control population (labour force participation rate of 30%, so SD=.3*(1-.3)=.21), and sample size (N=3200), this gives a MDE of 0.03, meaning that we could detect a treatment effect as small as 3 percentage points (33% or 27% in the treated vs. 30% in the control). Cluster randomization will increase the MDE because it reduces precision, and this is a function of the degree of clustering, as measured by the intra-cluster correlation (ICC). Effectively, this deflates the precision so we will either need to be satisfied with a larger MDE or a bigger sample to estimate the same MDE. If we revise the scenario above to take into account the clustered design, fixing the clusters again to 80 treated and 80 untreated, increasing the ICC increases the MDE. For k=160 clusters and n=20 individuals per cluster, the MDE for various ICCs are calculated as: ICC=0.00, MDE=0.05 ICC=0.05, MDE=0.07 ICC=0.10, MDE=0.08 ICC=0.20, MDE=0.10 For degrees of clustering typical in social science surveys (between 0.01 and 0.05, Bloom 2005), the present design of 160 clusters and 20 individuals per cluster still provides us with MDEs that seem feasible and relevant (5 to 7 percentage point differences for binary outcomes with baseline proportions of around 0.3).

Institutional Review Boards (IRBs)

IRB Name
McGill Faculty of Medicine Institutional Review Board
IRB Approval Date
IRB Approval Number
FWA 00004545
IRB Name
IFMR Human Subjects Committee
IRB Approval Date
IRB Approval Number
IRB00007107; FWA00014616; IORG0005894
Analysis Plan

Analysis Plan Documents

Affordable Daycare to Empower Indian Women: Pre-Analysis Plan

MD5: f49a2664339f158d5d56c1ace32d02b9

SHA1: 95ebc5ac37ebe20884d53387e0d506624f867bad

Uploaded At: February 22, 2017


Post Trial Information

Study Withdrawal

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

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials