Download StudyGeneral

Study Overview

Title:
Confidence, Experimentation, and Belief Updating
Study ID:
RIDIE-STUDY-ID-594a08a839f14
Initial Registration Date:
06/21/2017
Last Update Date:
04/13/2017
Study Status:
Ongoing
Location(s):
Kenya
Abstract:
We have developed an app that combines crop modeling, rainfall data and soil health information, within which farmers can experiment with agricultural inputs on a field that is calibrated to their own production context. The app provides farmers an opportunity to learn about the complex returns to (and interactions between) different fertilizer combinations and soil characteristics – without having to bear the upfront cost of purchasing the inputs. We elicit subjective expectations over fertilizer and incentive-compatible selections of optimal inputs before and after farmers interact with the game as well as risk preferences, confidence, and self-efficacy and aspirations. We will evaluate how farmers’ beliefs and input choices change in response to interaction with the app and test whether individuals’ risk preferences, confidence, and plot characteristics affect game play and belief updating.
Categories:
Agriculture and Rural Development
Additional Keywords:
Secondary ID Number(s):

Principal Investigator(s)

Name of First PI:
Emilia Tjernström
Affiliation:
University of Wisconsin-Madison
Name of Second PI:
Travis Lybbert
Affiliation:
University of California-Davis

Study Sponsor

Name:
Global Center for Food Systems Innovation
Study Sponsor Location:
United States
Funding Proposal:

Research Partner

Name of Partner Institution:
Tegemeo Institute of Agricultural Policy and Development, Egerton University
Type of Organization:
Research institute/University
Website:
http://www.tegemeo.org/
Location:
Kenya
Intervention

Intervention Overview

Intervention:
The intervention is to give Kenyan farmers supervised access to an app that enables farmers to experiment with agricultural inputs over a series of seasons that has been calibrated based on soil samples from that farmers’ field. Mahindi Master, an Android application created in Unity, animates the results from the simulations, allowing farmers to make in-game choices over fertilizer types, rates and different rainfall levels.  The application shows crop growth over fictional seasons with animations varying with the fertilizer and weather selections, and then displays expected yield ranges. At the end of each season, farmers can adjust the fertilizer -- and in later rounds, the rainfall levels -- and simulate a new season with these parameters.  This project targets two constraints on farmers' ability to learn about new inputs: (i) temporal variability from stochastic weather shocks that make signal extraction harder, and (ii) heterogeneity in soil quality, which makes it harder for farmers to use the experiences of others as a proxy for their own.
Multiple Treatment Arms Evaluated?
No

Implementing Agency

Name of Organization:
Tegemeo Institute of Agricultural Policy and Development, Egerton University, Kenya.
Type of Organization:
Research Institution/University

Program Funder

Name of Organization:
Global Center for Food Systems Innovation
Type of Organization:
Research Institution/University

Intervention Timing

Intervention or Program Started at time of Registration?
Yes
Start Date:
09/16/2016
End Date:
04/03/2017
Evaluation Method

Evaluation Method Overview

Primary (or First) Evaluation Method:
Regression with controls
Other (not Listed) Method:
Additional Evaluation Method (If Any):
Other (not Listed) Method:

Method Details

Details of Evaluation Approach:
We will evaluate the effect of the intervention on farmers' beliefs as well as differential effects of farmer characteristics on updating using a regression with controls. To measure the magnitude of updating, we will regress the change in measured beliefs on a constant and controls. We will also run a regression of the post value on the pre value and controls. To evaluate the effects of farmer characteristics on game play, we will regress the game outcome on the farmer characteristic (e.g. measured confidence) and controls.
Outcomes (Endpoints):
The outcomes of interest include the following: Beliefs and input choices: (1) changes to the farmers' fertilizer order following game play, (2) changes to farmers beliefs about fertilizer and optimal inputs Game play: (1) number of rounds played, (2) the type of weather simulations chosen, (3) whether farmers choose to change their final order, (4) number of times the final order is changed, (4) share of rounds played with unfamiliar inputs Aspirations: (1) changes in reported aspirations, (2) changes in reported self-efficacy
Measurement:
We will construct a variable of farmers' beliefs using information elicited through subjective expectations of yields under different fertilizer types and levels. We will calculate the mean and distribution of beliefs which we estimate by fitting a log-normal approximation to the responses and estimate the parameters of individual distributions using least squares. We will then use the first and second moments from this and estimate using other distributions as robustness check.
Unit of Analysis:
The main unit of analysis is individual (farmer).
Hypotheses:
Hypotheses related to beliefs: (1) We expect that farmers with less experience and/or more diffuse priors will update their beliefs more after playing the game. (2) We expect that farmers who update their beliefs about the returns to fertilizer and/or lime after playing the game will be more likely to change their fertilizer “order” after playing the game. (3) We expect farmers whose ex ante predicted returns to lime (low pH on their field) will allocate a greater share of their budget to lime after playing the game than those with low expected returns to lime. Hypotheses related to characteristics: (1) We expect more confident farmers to be less likely to want to modify their final order. (2) The number of rounds that farmers play in the game (pre-final round) is expected to be decreasing in the participant’s measured confidence. (3) We expect farmers who play fewer rounds of the game to be less likely to update their beliefs and fertilizer orders.
Unit of Intervention or Assignment:
We chose Western Kenya, south of Lake Victoria, as a geographic region, and households within villages were randomly chosen, proportional to size
Number of Clusters in Sample:
Number of Individuals in Sample:
The expected number of observations is 200.
Size of Treatment, Control, or Comparison Subsamples:
All farmers receive the intervention, but we are conducting a quasi-experiment with farmers with low pH fields (approx. #)

Supplementary Files

Analysis Plan:
Pre-AnalysisPlan.pdf
Other Documents:
Data

Outcomes Data

Description:
Data for this project were collected from 200 farmers in 19 villages. At this location, farmers were be asked survey questions, and given the opportunity to play the app-based farming game. Enumerators facilitated the initial learning of how to navigate the game. Farmers were asked survey questions, including a confidence elicitation. After the survey, risk preferences were elicited as well as other questions.
Data Already Collected?
Yes
Survey Name:
Data Previously Used?
No
Data Access:
Not restricted - access with no requirements or minimal requirements (e.g. web registration)
Data Obtained by the Study Researchers?
Yes
Data Approval Process:
Approval Status:

Treatment Assignment Data

Participation or Assignment Information:
Yes
Description:
Data Obtained by the Study Researchers?
Data Previously Used?
Data Access:
Data Obtained by the Study Researchers?
Data Approval Process:
Approval Status:

Data Analysis

Data Analysis Status:
No

Study Materials

Upload Study Materials:

Registration Category

Registration Category:
Prospective, Category 3: Data for measuring impacts have been obtained/collected by the research team but analysis for this evaluation has not started
Completion

Completion Overview

Intervention Completion Date:
Data Collection Completion Date:
Unit of Analysis:
Clusters in Final Sample:
Total Observations in Final Sample:
Size of Treatment, Control, or Comparison Subsamples:

Findings

Preliminary Report:
Preliminary Report URL:
Summary of Findings:
Paper:
Paper Summary:
Paper Citation:

Data Availability

Data Availability (Primary Data):
Date of Data Availability:
Data URL or Contact:
Access procedure:

Other Materials

Survey:
Survey Instrument Links or Contact:
Program Files:
Program Files Links or Contact:
External Link:
External Link Description:
Description of Changes:

Study Stopped

Date:
Reason: