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.
Food Without Fire: Environmental and Nutritional Impacts from a Solar Stove Field Experiment
Last registered on March 31, 2019


Trial Information
General Information
Food Without Fire: Environmental and Nutritional Impacts from a Solar Stove Field Experiment
Initial registration date
March 26, 2019
Last updated
March 31, 2019 11:08 PM EDT
Primary Investigator
University of Arizona
Other Primary Investigator(s)
PI Affiliation
Additional Trial Information
Start date
End date
Secondary IDs
Much of the population in rural sub-Saharan African relies of firewood or charcoal to prepare food. Population pressure is speeding the rate of deforestation, raising the monetary and opportunity costs of cooking meals. We use a field experiment in Zambia to investigate the impact of solar cook stoves on the money households spend on charcoal and the time allocated to collecting firewood. Additionally, we examine changes
in diet that result from the reduction in the cost of meal preparation.
External Link(s)
Registration Citation
Estrada Carmona, Natalia and Jeffrey Michler. 2019. "Food Without Fire: Environmental and Nutritional Impacts from a Solar Stove Field Experiment." AEA RCT Registry. March 31. https://doi.org/10.1257/rct.4054-1.0.
Former Citation
Estrada Carmona, Natalia, Jeffrey Michler and Jeffrey Michler. 2019. "Food Without Fire: Environmental and Nutritional Impacts from a Solar Stove Field Experiment." AEA RCT Registry. March 31. http://www.socialscienceregistry.org/trials/4054/history/44389.
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
In rural sub-Saharan African (SSA), many households rely on firewood and/or charcoal as the main sources of energy for cooking. Increasing population pressure in many parts of SSA is having a large impact on deforestation and, by extension, human health. As the forested landscape around human habitats is reduced (e.g. in Zambia, 0.2% of the land was deforested in a yearly basis between 2001-2014 - Hansen et al. 2013), the cost of procuring energy for cooking increases. The opportunity cost of finding firewood increases as household members must venture further and further from habited areas to find wood. This in turn increases the monetary cost of purchasing charcoal. As the price of energy inputs to preparing a meal rise the ability of households to incorporate healthy cooking habits falls. With high energy costs, households are less likely to boil water and milk, increasing susceptibility to water and milk borne diseases. Additionally, dietary diversity decreases as households refrain from preparing slow-cooking foods, like legumes, and instead focus on a small number of quick cooking foods such as corn meal.

In order to investigate the link between rising fuel costs for cooking and poor nutritional outcomes we will conduct a solar stove field experiment in the Barotse region of western Zambia. We will examine the impact of solar cookers which are randomly assigned to households across several communities. We will document the effects of the cookers on both the costs of preparing food and the types of food that the household prepares. Households will record the ingredients used in preparing each dish for each meal on each day plus the fuel used to prepare each dish over the six weeks of the experiment. This will provide us with detailed information on food and fuel inputs that each household used in preparing an expected 126 meals (3 meals a day * 7 days * 6 weeks) consumed over the study period.

Barotseland, the area of our intervention, has seen extensive work in the context of the CGIAR Research Program – (CRP) Aquatic Agricultural Systems (AAS) where

Bioversity International and World Fish to promoted nutritional information as well as crop diversification strategies for resilient and productive farming systems. Households in the region have self-selected into nutritional cooking clubs, participation in farm plot demonstrations, or chosen to abstain from participation in these development activities. We will stratify our randomization across these four groups in order to help us understand which information, from which activity might have a larger impact on shifting people towards more diverse diets, particularly when there is access to a low cost source of energy for cooking.

The research questions are as follows:
1. To what extent the provisioning of solar stoves reduce the amount of fuel (firewood, charcoal, dung) used by the household to cook meals?
2. Does the provisioning of solar stoves change the composition of the diet (measured by household dietary diversity, dietary species richness, count of the number of dishes, and count of the number of meals skipped) eaten by the household?
3. To what extent the provisioning of solar stoves increase the amount of liquid (water or milk) that is boiled by the household?
4. Does the provisioning of solar stoves increase the amount of legumes consumed by the household?
5. Does the use of solar stoves to cook some dishes affect the composition of other dishes cooked by the household using traditional stoves?
6. Are there heterogeneous effects based on self-selection into participation in cooking demonstrations, participation in agricultural demonstrations, or both?

Five main assumptions are behind our theory of change. The first assumption is that deforestation on the Barotse floodplain has increased the cost of traditional cooking fuel (firewood and charcoal) to the point where the price of cooking fuel is a binding constraint on the household’s decision regarding what meals to prepare. The second assumption is that households with a solar stove will reduce their use of traditional cooking fuel, thus reducing their costs of meal preparation. The third assumption is that, with a reduction in fuel costs, households will prepare more healthy meals. This includes preparing meals that include more ingredients, preparing meals that result in greater dietary diversity, boiling more liquids, and cooking more legumes. The fourth assumption is that, conditional on a reduction in fuel costs, households will change the composition of dishes cooked using traditional fuel as they re-optimize over their consumption decisions. The fifth assumption is that these effects will differ for households that had previously self-selected into participating in nutrition and/or farming demonstrations compared to households that self-selected to not participate.
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
1) Fuel use
2) Composition of diet
3) Boiling of liquids
4) Cooking of legumes
Primary Outcomes (explanation)
1) Fuel use - either firewood, charcoal, dung or all three measured as:
The time spent collecting fuel in each week
The time spent collecting fuel over all six weeks.
The money spent purchasing fuel in each week.
The money spent purchasing fuel over all six.
The amount of fuel that was collected in each week.
The amount of fuel that was collected over all six weeks.

2) Composition of diet - measured in four ways:
Dietary diversity. The household dietary diversity score (HDDS) will be calculated following FAO guidelines: http://www.fao.org/3/a-i1983e.pdf
Species richness. The household dietary species richness (SR) will be calculated following Lachat et al. (2018): https://www.pnas.org/content/pnas/115/1/127.full.pdf
Number of dishes
Number of meals skipped

3) Boiling of liquids - either milk, water, or both measured as:
The number of times liquid was boiled in a given day.
The number of times liquid was boiled in a given week.
The number of times liquid was boiled over all six weeks.
The volume of liquid boiled in a given day.
The volume of liquid boiled in a given week.
The volume of liquid boiled over all six weeks.

4) Cooking of legumes - measured as:
The number of times legumes were cooked in a given day.
The number of times legumes were cooked in a given week.
The number of times legumes were cooked over all six weeks.
The volume of legumes cooked in a given day.
The volume of legumes cooked in a given week.
The volume of legumes cooked over all six weeks.
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
The Barotse Floodplain System (BFS), Western Province in Zambia, provides diverse ecosystem services important for local and downstream communities (Schuyt, 2005). The livelihoods and migratory patterns of local communities, the Lozi people, are adapted to the natural flow of the unregulated Upper Zambezi (Tweddle, 2010). The livelihoods of the Lozi depend on fishing, cattle or farming, or a combination of those activities. The floodplain is experiencing severe declines in fish catch rates, fish species, population size and fish diversity (Tweddle, 2010; Tweddle et al., 2015). Shifting cultivation and burning are widely practiced in the area, with large impacts on forest composition, deforestation and regeneration (Wolski, 1998, Tambara, et al., 2012). The communities in Barotse shared their concern about the depletion of their natural resources, particularly regarding fish and forest (Kwashimbisa and Puskur, 2014).

Western Province has one of the highest poverty rates in the country, low agricultural productivity and the region is highly vulnerable to internal and external shocks (Flint 2008; Rajaratnam et al. 2015). Poor sandy soils (Kalahari sands), limited access to agricultural inputs (e.g. manure), equipment, and knowledge about improved management techniques using organic matter (Baidu-Forson et al. 2014) prohibit many rural people from increasing production and productivity. The region experiences a period of four to five months with limited access to food (hunger season) (Castine et al. 2013, Baidu-Forson et al. 2014, Rajaratnam et al. 2015).

The BFS is a pilot Nutrition-Sensitive Landscape (NSL) embedded in CGIAR Research Programs Aquatic Agriculture Systems ( ended in January 2016) and Agriculture for Nutrition and Health. Bioversity International lead the characterization of diets, food availability (Pascualino 2014); agrobiodiversity (Baidu-Forson et al. 2014), farming systems (Del Rio 2014) and ecosystem services (Del Rio et al., 2018). Similarly, in collaboration with local partners’, three main activities has been promoted and supported to increase knowledge on nutrition, healthy cooking habits, diversify diets and crops named cooking demonstrations-nutrition clubs and learning plots across the eleven villages. During those activities, legumes have been promoted as good and cheap source of proteins while just improving soil fertility and reducing soil erosion as cover crops. However, legumes occupy a minor part of the Lozi diet.

Among the eleven villages in the BFS pilot, we will first randomly select three villages in which to conduct our experiment. Each BFS village has at least two community facilitators (women and men) and one Induna or traditional leader. We will introduce our activities with the community facilitators and Indunas to incorporate their opinions and ideas. They will be in charge of inviting the community to our activities or first events. The invitation for the solar cookers will be open to everyone within the village who is interested.

In each community we will have an introductory day-long event. During the morning we will: 1) start an open discussion with participants about the objectives, commitment and expected results from the solar cookers project; 2) conduct a test and use some of the cookers to make our communal lunch highlighting safety management and precaution measures with a hands-on experience.

During the afternoon session, we will invite those participants interested in the project and in volunteering to have and use the solar stove during six weeks. The commitment is to properly manage the solar stove, record its daily usage for six weeks in the assigned form and record charcoal and firewood consumption during the same period. Households who volunteer to participate will be part of a raffle of the solar stoves at the end of the six week experiment, conditional on their satisfactory completion of their cooking and fuel log. This is to incentivize members of the control group, who do not receive a stove, to record their data through the six weeks.
Experimental Design Details
Randomization Method
We will stratify our randomization of the stoves based on a household’s previous self-selection into one of four groups: 1) cooking demonstrations (nutrition clubs), 2) farmer demonstrations (learning plots), 3) both activities, and 4) no activities. Within each strata and within each village, stoves will be assigned via a random draw (without replacement) of participant names from a bowl.
Randomization Unit
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
4 clusters based on pre-experiment self-selection into either 1) learning plots, 2) nutritional clubs, 3) both, or 4) none.
Sample size: planned number of observations
158 households
Sample size (or number of clusters) by treatment arms
59 households
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB Name
Office for the Protection of Research Subjects [OPRS], University of Illinois at Urbana-Champaign
IRB Approval Date
IRB Approval Number
Details not available
Analysis Plan
Analysis Plan Documents
Pre-Analysis Plan

MD5: f9c7608fe27b4eae103e261ea938779d

SHA1: 913716781d629e8b387a764365367722161c6c9d

Uploaded At: March 25, 2019

Post Trial Information
Study Withdrawal
Is the intervention completed?
Is data collection complete?
Data Publication
Data Publication
Is public data available?
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)