Reducing the burden of unpaid care work among women farmers in Tanzania: A clustered RCT

The intervention will consist of Unpaid Care Work modules integrated into the existing FFBS curriculum. The Unpaid Care Work modules and activities are currently being designed in response to findings from the baseline data collection. Once the modules are complete, the implementer CARE Tanzania will train frontline staff randomly assigned to the treatment group to deliver these modules to their groups in addition to the standard FFBS curriculum.
Activities
• Monthly dialogue sessions with producer group members, facilitated by frontline staff, focusing on unpaid care work and its impact on household members. Sessions will run year-round, independent of agricultural seasons to ensure continuity.
• Collaboration with a local partner to develop and provide low-cost labor-saving devices to reduce unpaid care work burdens. Since most producer group members are also VSLA members, the project will encourage them to use their savings and loans to purchase these devices.

Compliance, spillover and differential participation and attrition
• No major compliance issues are anticipated, as CARE has successfully used the dialogue-based approach in similar community programs. CARE will use a monitoring tool to track the outcomes of the dialogue sessions relating to shifting care work norms and practices.
• Spillover effects are expected to be minimal, as activities will be delivered within group settings by frontline staff assigned to either treatment or control, reducing cross-group exposure.
• Since the Unpaid Care Work Modules build on the existing FFBS infrastructure, treatment and control groups will face similar participation dynamics. Endline survey participation is also not expected to differ significantly between treatment and control households, given the shared FFBS framework.

2025-03-03

2026-07-31

The primary experimental outcomes of the RCT are
a. Total hours unpaid care work time performed by women
b. Daily workload of women
c. Gender gap in time allocation to unpaid care work among men and women in dual-adult households
d. Gender gap in daily workload among men and women in dual-adult households

Using an adapted 24-hour recall Pro-WEAI survey, respondents recall and report all activities from wake-up to sleep in 15-minute intervals. Activities are classified into four categories:
● Income generation: Farm work, employment, self-employment, and paid care work.
● Unpaid care work: Tasks such as water and fuel collection, meal preparation, laundry, cleaning, childcare, caring for the elderly/disabled, community support, commuting, shopping and other errands, and household management.
● Leisure: Personal care, social and religious engagements, entertainment, leisure travel, rest, and schooling.
● Sleep: The period from bedtime to wake-up.
During each primary activity, respondents also note any concurrent childcare responsibilities as a secondary activity. We calculate the time spent on each of the above categories, plus the:
● Daily workload: total time spent on income generation, unpaid care work, plus half of the time allocated to secondary childcare.
● Gender gap in dual-adult households: the difference in time allocation between men and women in dual-adult households, where a positive value indicates women carry a larger share.

The secondary outcomes are:
● Gender norms and decision-making power surrounding unpaid care work
● Household investment in labor-saving devices and outsourcing of care work
● Household income
● Control over household income
● Time allocation to paid and educational activities for all household members
● Intra-household conflict in dual-adult households
● VSLA investment in labor-saving devices and outsourced care work
● Gender gap in unpaid care for FFBS groups whose VSLA funded childcare systems

The secondary outcomes are constructed as follows:
Gender norms and decision-making power will be assessed using data from both men and women respondents to the household survey. We will use:
● Household Gender Roles in Chores and Childcare, a series of two validated Likert scales, that measures perceived support for sharing responsibilities in household chores and childcare
● Relative Autonomy Index, which assesses individual motivations behind participating in household chores, employment decisions, and major purchases. Positive scores indicate autonomous decision-making, whereas negative scores suggest decisions are constrained by social norms.
Household investment in labor-saving devices and outsourced care work will be assessed using household survey data on the ownership, quantity purchased in the past year, and price for 23 device types (e.g., kitchen appliances, farming equipment). This list of items was derived from =the Tanzania Household Budget Survey and used to calculate:
● Annual expenditure on labor-saving devices: Sum of (quantity of items purchased in the past 12 month × average price) for all 23 devices
● Outsourced care work: total expenditure on outsourced household services (chores, childcare, elder care, delivery) in the past month.
Household income will be assessed using household survey data.
● Annual agricultural income: Total income from agriculture and livestock over the past year, based on crop types, yields, prices, and livestock revenue.
● Annual FFBS income: Income from sunflower and soybean production in the last year, reflecting FFBS program focus.
● Weekly income from all sources: Combined income from agriculture, wages, self-employment, and informal work, aggregated from household roster data.
Control over income will be assessed using household survey data. Respondents reported the proportion of household income they control in three ways: (i) sole control, (ii) joint control, and (iii) no control, adding up to 100%.
Time allocation to paid and educational activities: Will be assessed using household survey data. We will collect the hours spent on paid work and educational activities over the past week by household members, categorized by age and gender. With this measure, we explore potential spillover effects when changes occur in women’s unpaid care work. For instance, if mothers reduce unpaid care duties, daughters might assume more responsibilities.
Intra-household conflict in dual-adult households: Frequency of disputes over work, chores, finances, leisure, and childcare is rated from 0 (never) to 4 (very frequently).
VSLA investments in labor-saving devices and outsourced care work: Variable defined using transaction records from the Chomoka app, a smartphone tool for VSLA bookkeeping. If available, we will use these records to understand members’ savings usage and allocation toward community-funded labor-saving devices or childcare. We would then check whether FFBS groups in the treatment show greater VSLA investment in these areas than standard, control FFBS groups.
Gender gap in unpaid care for groups whose VSLA funded childcare systems: the average difference in unpaid care work hours between women and men within dual-adult households that participate in groups with VSLA-funded childcare systems.

Our study utilizes a cluster randomized controlled trial (cRCT) to evaluate the impact of adding Care Work Modules to the existing FFBS curriculum. Paraprofessionals trained to lead FFBS groups are randomly assigned to either the intervention arm, which receives the additional Care Work Modules, or the control arm, which continues with the standard curriculum.
Randomization was conducted at the level of the group facilitator (called ‘paraprofessional’) rather than by FFBS group. This approach was selected to prevent contamination since a single paraprofessional can train multiple FFBS groups, and mixing intervention and control groups under one facilitator could bias results.

Not available

Using the statistical software Stata, we executed 1,000 rounds of randomization and group selection to enhance study power and ensure comparability between the treatment and control arms. We then selected the iteration that yielded the fewest average groups per paraprofessional and the greatest equilibrium across wards.

The unit of randomization is the paraprofessional. After assigning all 60 paraprofessionals to treatment or control, we randomly selected FFBS groups led by each for inclusion in the study. As noted above, we executed 1,000 rounds of randomization and group selection to enhance study power and ensure comparability between the treatment and control arms. We initially sampled 60 first-choice groups per arm, plus 20 backups per arm, anticipating potential list inaccuracies.

Within each selected FFBS group, we then randomly sampled 10 households per group, plus replacement households. If groups were large, up to 25 replacement households were sampled to address potential field issues. To avoid double-sampling spouses—since household composition was not in the sampling frame—we exclusively sampled women respondents, determining in the field whether each household was dual-adult or female-headed.

Yes

60 paraprofessionals

Our target sample was initially 1200 households (960 dual-adult households and 240 female-headed households). Because we sought to recruit both men and women from dual-adult households, this would reflect a total of 2,160 individual survey respondents. At baseline, we recruited 1,186 households (946 dual-adult households and 240 female-headed households), reflecting a total of 2,132 individual respondents.

Our target sample was 600 treatment households and 600 control households. At baseline, we recruited 586 control and 600 treatment households.

For investigating whether households receiving additional Care Work Modules experience greater reductions in total unpaid care work among women, the study uses a sample of 1,186 households (600 treatment, 586 control). At baseline, women in these households spend on average 5.36 hours per day on unpaid care work (SD = 2.75), with an ICC of 0.044. At a 5% significance level, the study has 80% power to detect a reduction of about 36 minutes in the daily time women spend on unpaid care work. For examining whether dual-adult households receiving additional Care Work Modules report a more equal allocation of time dedicated to unpaid care work between men and women, the study focuses on the 946 dual-adult households (480 treatment, 466 control). At baseline, in these households the female spouse spends on average 3.91 hours per day more than the male spouse on unpaid care work (SD = 3.48), with an ICC of 0.039. At a 5% significance level, the study has 80% power to detect a reduction of approximately 48 minutes in this gender gap for households receiving the intervention.