Do Spouses Share Information? Exploring Intra-Household Information Sharing and Learning Using a Lab in the Field?

Last registered on May 02, 2022

Pre-Trial

Trial Information

General Information

Title
Do Spouses Share Information? Exploring Intra-Household Information Sharing and Learning Using a Lab in the Field?
RCT ID
AEARCTR-0009295
Initial registration date
April 22, 2022

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
April 28, 2022, 5:58 PM EDT

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

Last updated
May 02, 2022, 6:18 AM EDT

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

Locations

Region

Primary Investigator

Affiliation
University of Georgia

Other Primary Investigator(s)

PI Affiliation
University of Georgia

Additional Trial Information

Status
In development
Start date
2022-04-14
End date
2022-06-14
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Exploring Intra-Household Information Sharing Using a Lab in the Field

1 Overview and significance
There is a growing consensus among development economists that the adoption of modern farming practices
and technologies improves the welfare of smallholder farmers and sparks the growth and development of the
agriculture sector in developing countries. By adopting modern farming practices and technologies, farmers
can increase profits (Bezu et al., 2014; Takahashi et al., 2019), improve food security (Shiferaw et al., 2014),
and increase the quality and diversity of both their children’s diets and their own (Kim et al., 2019; Bezu
et al., 2014). Despite these welfare improving outcomes, technology adoption rates remain consistently low
and issues such as chronic malnutrition remain persistent in most of the developing world. Lambrecht et al.
(2014) states technology adoption occurs in three stages; awareness, tryout, and continued use. Access to
and the quality of information plays a vital role in an individual’s decision to adopt new technology.
In many developing countries, information is first introduced to farmers through extension agents, local
traders, or marketers. Extension agents rarely meet the demand for extension services while subsistence households do not interact frequently with local traders and marketers reducing farmers’ access to information.
Recent studies have focused on increasing smallholder farmers’ access to information by increasing extension
access (Kondylis et al., 2016), using peer to peer (P2P) training (BenYishay & Mobarak, 2019; Fafchamps,
2019), and leveraging social networks to improve diffusion (Banerjee et al., 2014; Beaman et al., 2018).
Although these studies show increased rates of information sharing and technology adoption, many of these
studies find that women were less likely to obtain information than men (Beaman & Dillon, 2018), men were
less likely to listen to female trained farmers (BenYishay et al., 2020), and information sourced through a
female social network had less impact on demand for agricultural technologies (Magnan et al., 2015).
Although women play a key role in development, technology adoption among women continues to lag
behind men. This lag is often attributed to the unique barriers women face such as time constraints, credit
constraints, information constraints, and gender norms. These barriers to information are found within the
household where husbands have more awareness and knowledge of agricultural technology (Fisher et al., 2019)
and credit opportunities (Fletschner & Mesbah, 2011) than their wives. Studies on information diffusion
may overestimate the impact of their intervention by assuming household level access to information or
household level knowledge of new technology. Information diffusion designs, such as peer-to-peer farmer
training, that do not include within household behavior may underperform or lead to larger gender gaps in
access to information. Recent work determining spousal roles in adopting technology has found mixed results.
Women may have little impact on the household’s decision to adopt new technologies if the information enters
the house through her (Gulati et al., 2019), however, Magnan et al. (2020) find that the adoption decision
for improved maize varieties are correlated to risk preferences of both the husband’s and wives showing that
intrahousehold decision making is complicated, heterogeneous, and how the information enters the household
impacts which technology is adopted. Despite these effects information has within the household, few studies
have attempted to directly test for information sharing in the household.
This study uses a laboratory in the field experiment to observe information sharing within the household.
This paper contributes to two broad areas of research. First, this paper builds on the existing literature
on household behavior. Several papers have tested the assumptions of many household models. Our paper
contributes to this literature by experimentally observing information sharing by both husbands and wives
as well as how the control of the income within the household impacts information sharing. Second, our
paper contributes to the information dissemination literature. Many papers have explored ways to improve
information dissemination across households. BenYishay & Mobarak (2019) uses performance based incentives to improve peer-to-peer information sharing. Others such as Banerjee et al. (2014) use social network
characteristics to improve information dissemination. Many of these papers focus on farmer to farmer or
household to household information dissemination. Our paper contributes to this literature by expanding on
the findings in Beaman & Dillon (2018) and focusing on the barriers to information women face specifically
within their household.

2 Empirical strategy

We use a lab in the field with a 2x2 treatment design to exogenously vary the training of one spouse in each
household and the control each spouse has over the income the household receives from the experiment. We
ask couples from groundnut-producing households in Northern Ghana to attend two meetings. Within each
meeting, participants are asked to solve the 4-disk version of the Tower of Hanoi for a cash prize. Households
will be assigned to one of four treatment groups, husband first- individual payment, wife first- individual
payment, husband first- joint payment, and wife first- joint payment. These four treatment groups allow us
to observe information sharing between husbands and wives through their performance in this task as well
as the impact income control has on information-sharing behavior across genders.
The primary goal of using a lab in the field is to create a unique environment to observe information
sharing exogenous to the many factors that impact information sharing. We expect the results of the study
to reflect real-world practices. We collect information about groundnut practices and income from both
spouses. We use this information to create outcome variables that are correlated with information sharing.
We then compare household information sharing in the lab to information sharing in practice.

3 Data, Methods, and Expected Results

Our study involves 480 households among 12 villages in Northern Ghana. Both the husband and wife
of each household are interviewed separately. The interview consists of 5 parts: household demographics, information sources, groundnut production, spouse’s groundnut production, and groundnut technology
knowledge. Participants perform the task using an android based tablet that collects the number of moves
made, what move was made, and the time until completion which are uploaded directly from the tablet to a
cloud-based database. We interview the untrained spouse at the second meeting about the information they
received from their spouse or other members outside of the household. We expect to experimentally find
results similar to Fletschner & Mesbah (2011) where men often do not share information with their wives.
Further, we expect to expand on those results and observe information sharing from women to their husbands.
We expect that when income is shared women are more likely to share information, however, information
will be under-supplied in individual payments where the benefits of sharing may not outweigh the costs.

4 Potential for discussion

We will start by discussing the policy-relevant question of unintended consequences of information dissemination strategies that focus on the household level. Women face unique barriers to information-making
technology adoption that may reduce food insecurity, improve food safety, and increase women’s empowerment less likely. The focus of the discussion will be on improving future information dissemination strategies
to be more inclusive of women and shedding light on the barriers to information women face within their own
households. Furthermore, information dissemination strategies may need to account for household income
structure.
External Link(s)

Registration Citation

Citation
Magnan, Nicholas and Sean Posey. 2022. "Do Spouses Share Information? Exploring Intra-Household Information Sharing and Learning Using a Lab in the Field?." AEA RCT Registry. May 02. https://doi.org/10.1257/rct.9295-1.1
Sponsors & Partners

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

Request Information
Experimental Details

Interventions

Intervention(s)
Households will be assigned to one of four treatment groups, husband first- individual payment, wife first- individual payment, husband first- joint payment, and wife first- joint payment. Both the husband and wife of each household are interviewed separately. The interview consists of 5 parts: household demographics, information sources, groundnut production, spouse’s groundnut production, and groundnut technology knowledge. One spouse in each household will be trained to optimally solve the four disk version of the Tower of Hanoi. The treatment status of the household will determine which spouse will be trained. The other participant in the household will rely on their spouse for training. Participants will perform the task one day apart to allow for information sharing at home. Participants perform the task using an android based tablet that collect the number of moves made, what move was made, and time until completion which are uploaded directly from the tablet to a cloud-based database. We interview the untrained spouse at the second meeting about information they received from their spouse or other members outside of the household.
Intervention Start Date
2022-05-02
Intervention End Date
2022-06-14

Primary Outcomes

Primary Outcomes (end points)
The outcome variables is the inverse of the total number of moves needed to finish, total number of correct moves made before an error
Primary Outcomes (explanation)
total number of moves and total number of correct moves will be recorded by the android tablet as the individual performs the task.

Secondary Outcomes

Secondary Outcomes (end points)
Knowledge of spouse's groundnut practices, knowledge of current groundnut technology and practices, knowledge of spouses income from groundnuts.
Secondary Outcomes (explanation)
Knowledge of spouse's groundnut practices is an index that increases as individuals accurately identify their spouse's groundnut practices.
Knowledge of current groundnut technology is an index that increases if an individual states they are aware of the technology and if they correctly identify the benefits of the technology.
Knowledge of spouses income from groundnuts is the gap between the reported groundnut harvest by the spouse and the estimated groundnut harvest of the participant

Experimental Design

Experimental Design
Our study involves 480 households among 12 villages from Northern Ghana, the largest groundnut producing region in Ghana. Groundnuts are produced by both men and women in the household despite being referred to as a \woman's crop". Both the head of household and their spouse produce groundnuts, allowing us to compare information sharing relative to groundnut production. The husband and wife of each household will be interviewed separately. The interview consists of 6 parts: household demographics, information sources, groundnut production, spouse's groundnut production, groundnut technology and disease knowledge, and information sharing within the household. Each household will be randomly assigned into one of four treatment groups. The treatment groups can be broken down into two categories, spouse and payment. The spouse category consists of two treatments, the husband first treatment (HF) and the wife first treatment (WF). These treatment groups determines which spouse in the household will be trained and which member of the household will rely on their spouse to train them. In the husband (wife) rst treatment, the husband (wife) will be trained by the research team on how to solve the puzzle in the minimum number of moves. They will be able to share this information with their spouse after the rst meeting. The payment category consists of the individual payment treatment and the joint payment treatment. All payment will be given at the end of the second meeting. Participants will receive an envelope with their individual identi cation code. Participants in the individual payment treatment will receive an envelope of cash corresponding to their individual performance in the task. This treatment simulates a management scheme where individuals manage their own plots and have control over the income from those plots. Unlike the individual payment treatment, the joint payment replicates the income ownership of a joint management scheme. Instead of each participant within the household receiving their prize money separately, the household's earnings are combined and split evenly between both spouses. In the rst meeting after the husband
(wife) finishes the task they will be informed of how much they have earned and informed that they will receive half of the total household's prize money after the second meeting. We will use the household's responses in the survey to identify households as either joint or individual producers and stratify by production scheme into payment treatment For example, if 200 households report having a joint production scheme, we will randomly assign 100 of those households in the joint payment scheme and the other 100 households in the
individual production scheme. We will do the same to the households that report an individual production scheme.
Households will be invited to attend two meetings. Each spouse will be asked to attend one meeting. In the fi rst meeting participants will be asked to complete the 4 disk version of the Tower of Hanoi puzzle, a novel game that requires no experience to play, it is easy to teach with a simple objection and few rules, there is a unique way to solve the game and any information obtained about this game should not hinder a players performance. The 4 disk version of the puzzle can be completed in no less than 15 moves. The puzzle is attempted using an android based tablet where the number of moves, time, and each move they make will be recorded. Participants will be between the ages of 18 and 65 years of age. This is for four reasons, that population is more relevant to this topic, farmers 65 and below often adopt technology more frequently (Adesina & Zinnah, 1993), have a higher willingness to pay for extension services (Mwaura et al., 2010), older farmers may be less capable of using a touch screen device causing the data to be misleading, and due to the COVID-19 pandemic, it is deemed safer for participants over the age of 65 to not participate in in-person studies. The rst meeting will consist of 20 participants, each participant will be given two practice rounds where they will be allowed to make as many moves as possible to ensure they understand the game and are comfortable using an Android tablet.
Once each participant in the meeting has completed both practice rounds, the participants will break up into groups of 4 where they will be trained to complete the game in only 15 moves. The training will consist of an enumerator slowly showing each move and explaining why that move must be made to complete the game in only 15 moves. The goal of the training is to not only help them memorize the order in which they need to move, but to help them understand why that order is necessary. Finally, after the training, the participants will be asked to complete their third and nal round where they will earn a cash prize based on their performance. The cash prizes are shown in Figure 3. To prevent censoring at 75 moves, we o er an additional 1GhC for individuals who complete the task. Any participant who chooses to forfeit a round, the tablet will record this and they will be allowed to move forward on the next round.

The cash prizes are broken down into bins to reduce the complexity of how much money each participant wins and to avoid any confusion by the participant while playing. Traditionally using a non-linear payment system can lead to censoring, players stop after a certain point where they no longer believe they can make it to the next payment system and receive no bene t from moving forward, or strategic playing that may be counter to the researchers intended goals. Neither of these issues are a concern in this experiment. Any player that can strategically make a certain number of moves to maximize their pay to effort level has a dominant strategy to complete the game in 15 moves. We don't expect any outcomes where participants strategically play the game that results in bunching or a result that doesn't align with dominant strategy of the game. A cost per move payment system may be overly confusing to a participant without adding any additional benefi ts.

Each meeting is one day apart to allow for information to be shared between spouses, but to minimize information obtained outside the household. At the second meeting, the spouse from each household who has not been trained by the research team will be asked questions to determine if information was shared, how information was shared, and to what extent did the participant from the rst meeting train their spouse. After these questions are answered, the meeting will follow directly as the rst meeting without the training. After all the participants have fi nished the task and the interview, they will receive their prize money. If the participant is in the joint payment group they will receive their portion of the household's winning, but if they are in the individual payment group they will only receive their winnings.
Experimental Design Details
Our study involves 480 households among 12 villages from Northern Ghana, the largest groundnut producing region in Ghana. Groundnuts are produced by both men and women in the household despite being referred to as a \woman's crop". Both the head of household and their spouse produce groundnuts, allowing us to compare information sharing relative to groundnut production. The husband and wife of each household will be interviewed separately. The interview consists of 6 parts: household demographics, information sources, groundnut production, spouse's groundnut production, groundnut technology and disease knowledge, and information sharing within the household. Each household will be randomly assigned into one of four treatment groups. The treatment groups can be broken down into two categories, spouse and payment. The spouse category consists of two treatments, the husband first treatment (HF) and the wife first treatment (WF). These treatment groups determines which spouse in the household will be trained and which member of the household will rely on their spouse to train them. In the husband (wife) rst treatment, the husband (wife) will be trained by the research team on how to solve the puzzle in the minimum number of moves. They will be able to share this information with their spouse after the rst meeting. The payment category consists of the individual payment treatment and the joint payment treatment. All payment will be given at the end of the second meeting. Participants will receive an envelope with their individual identi cation code. Participants in the individual payment treatment will receive an envelope of cash corresponding to their individual performance in the task. This treatment simulates a management scheme where individuals manage their own plots and have control over the income from those plots. Unlike the individual payment treatment, the joint payment replicates the income ownership of a joint management scheme. Instead of each participant within the household receiving their prize money separately, the household's earnings are combined and split evenly between both spouses. In the rst meeting after the husband
(wife) finishes the task they will be informed of how much they have earned and informed that they will receive half of the total household's prize money after the second meeting. We will use the household's responses in the survey to identify households as either joint or individual producers and stratify by production scheme into payment treatment For example, if 200 households report having a joint production scheme, we will randomly assign 100 of those households in the joint payment scheme and the other 100 households in the
individual production scheme. We will do the same to the households that report an individual production scheme.
Households will be invited to attend two meetings. Each spouse will be asked to attend one meeting. In the fi rst meeting participants will be asked to complete the 4 disk version of the Tower of Hanoi puzzle, a novel game that requires no experience to play, it is easy to teach with a simple objection and few rules, there is a unique way to solve the game and any information obtained about this game should not hinder a players performance. The 4 disk version of the puzzle can be completed in no less than 15 moves. The puzzle is attempted using an android based tablet where the number of moves, time, and each move they make will be recorded. Participants will be between the ages of 18 and 65 years of age. This is for four reasons, that population is more relevant to this topic, farmers 65 and below often adopt technology more frequently (Adesina & Zinnah, 1993), have a higher willingness to pay for extension services (Mwaura et al., 2010), older farmers may be less capable of using a touch screen device causing the data to be misleading, and due to the COVID-19 pandemic, it is deemed safer for participants over the age of 65 to not participate in in-person studies. The rst meeting will consist of 20 participants, each participant will be given two practice rounds where they will be allowed to make as many moves as possible to ensure they understand the game and are comfortable using an Android tablet.
Once each participant in the meeting has completed both practice rounds, the participants will break up into groups of 4 where they will be trained to complete the game in only 15 moves. The training will consist of an enumerator slowly showing each move and explaining why that move must be made to complete the game in only 15 moves. The goal of the training is to not only help them memorize the order in which they need to move, but to help them understand why that order is necessary. Finally, after the training, the participants will be asked to complete their third and nal round where they will earn a cash prize based on their performance. The cash prizes are shown in Figure 3. To prevent censoring at 75 moves, we o er an additional 1GhC for individuals who complete the task. Any participant who chooses to forfeit a round, the tablet will record this and they will be allowed to move forward on the next round.

The cash prizes are broken down into bins to reduce the complexity of how much money each participant wins and to avoid any confusion by the participant while playing. Traditionally using a non-linear payment system can lead to censoring, players stop after a certain point where they no longer believe they can make it to the next payment system and receive no bene t from moving forward, or strategic playing that may be counter to the researchers intended goals. Neither of these issues are a concern in this experiment. Any player that can strategically make a certain number of moves to maximize their pay to effort level has a dominant strategy to complete the game in 15 moves. We don't expect any outcomes where participants strategically play the game that results in bunching or a result that doesn't align with dominant strategy of the game. A cost per move payment system may be overly confusing to a participant without adding any additional benefi ts.

Each meeting is one day apart to allow for information to be shared between spouses, but to minimize information obtained outside the household. At the second meeting, the spouse from each household who has not been trained by the research team will be asked questions to determine if information was shared, how information was shared, and to what extent did the participant from the rst meeting train their spouse. After these questions are answered, the meeting will follow directly as the rst meeting without the training. After all the participants have fi nished the task and the interview, they will receive their prize money. If the participant is in the joint payment group they will receive their portion of the household's winning, but if they are in the individual payment group they will only receive their winnings.
Randomization Method
Randomization will be done using a randomized number generator in Stata with a set seed for replicability
Randomization Unit
Randomization will be done at the household level
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
12 communities
Sample size: planned number of observations
40 households per community
Sample size (or number of clusters) by treatment arms
480 households and a husband and wife from each household for a total of 960 participants
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
There are 6 hypotheses for the main outcomes. For the first and third hypotheses we expect a MDES of 3.51 moves (.18 of a standard deviation). The second and fourth hypotheses we expect a MDES of 4.66 moves (.17 standard deviations) . For the fifth hypothesis our power calculations report a MDES of 2.27 moves (.26 standard deviations). Finally for our sixth hypothesis, we calculated a MDES of 2.68 moves (.26 standard deviations)
IRB

Institutional Review Boards (IRBs)

IRB Name
University of Georgia
IRB Approval Date
2021-06-30
IRB Approval Number
PROJECT00002326
Analysis Plan

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

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

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

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials