Upload
leon-lewis
View
220
Download
2
Tags:
Embed Size (px)
Citation preview
AGENDA
Introduction Problem Current Solutions
Participatory Design Process Overview of Pilot Study Evaluation
Qualitative Results Quantitative Harvest Results
Lessons Learned Future Work
In developing countries: High illiteracy rates
752 million people Limited electricity
1.5 billion people Agriculture is major source
of livelihood—often use inefficient practices
PROBLEM
CURRENT METHODS
Traditional Agriculture extension
High transaction costs Small fraction reached Low adoption levels
ICT4D Radio broadcast Digital Green Grameen’s CKW
Farmers need locally generated agriculture
guidance that they can listen to
repeatedly and when they need it.
Iterative Design 2 years of research Numerous trips to Ghana Explored many form factors
Current Model Users can:
Listen to local experts Record their own messages Copy messages between devices
PARTICIPATORY DESIGN PROCESS
1. Content
Created by local experts in agriculture, health, and education.
2. Launch
Delivered 21 devices to a rural village (~1000 residents, 90% illiteracy, no electricity).
3. Allocation and Training
Completed by a local leadership committee.
4. Continual Support
Visited biweekly for feedback/support.
5. Evaluation
Interviewed about device, implementation, and harvest results.
OVERVIEW OF PILOT STUDY
This research has been funded in part by the Seattle International Foundation
QUALITATIVE STUDY
Methodology>40 in-person interviews Implementation Device
Physical Software / Content
QUALITATIVE RESULTS
Training and Usability Audio instructions alone were effective for
some, others required training Misunderstandings about program Requests for more training
General use Proper handling
Allocation Devices were very valuable Committee was protective, feared breakage. Unequal access: schooling, gender, regions Many requests for more devices
I would change the device distribution to
make it more fair. So if there are thirty devices, then women get 15 and men get 15 to prevent
conflicts.
QUALITATIVE RESULTS (cont.)
Common Requests Lights for night use More pronounced buttons for blind Embedded radio Solar or rechargeable power Reduce to pocket-size
Durability and Maintenance Rough handling of microphone jack Some software issues One device disassembled but still
functioned
QUALITATIVE RESULTS (cont.)
Behavior Change Farmers reported:
Telling peers about what they had learned.
Learning and applying new methods.
Seeing improved results.
MEET SUGLO
It has a lot of benefits to me. It taught me that we should start clearing our farm lands before the farming season begins,
start by March and finish between May and June. That we can just cultivate the land and plant the crops or plant them in
beds and/or lanes; that those methods increase the amount of crops per land area compared to mounds which waste land
and take up a lot of space. Beds also help accumulate water, prevent erosion and keep the soil within the farm moist. The beds actually make a big difference in terms of keeping the
soil moist. Mounds are too high from the ground and they dry up very fast and our crops suffer during insufficient rain fall.
Now we can still smile during short periods of draught because planting in beds keeps the soil moist for a little while.
Since I heard that from this thing (talkbook),
I tried it this year, and I am a woman but people exclaim whenever they see my crops in the farm and I just keep
my mouth shut because I know the harvest is going to be good. With the small amount of rain that we get, the beds still keep the water around and the crops stay healthy for
up to a week and I go to look at them with smile on my face.
Methodology Interviewed 33 users, 40 non-users
Demographics (region, age, schooling) Bags produced in 2008 and 2009
Millet Maize Beans Ground nuts
Changes in practices Human labor Farm animals Pesticide and fertilizer use Amount of land
Application of new guidance (users only)
QUANTITATIVE STUDY
Users produced 48% more crops than nonusers (7.22 bags) after controlling for other factors (p=.008)
1 bag = ~120 lbs, 50 gallons
HARVEST RESULTS: ALL CROPS
1 ½ bags
CROP YIELDS BY GROUP
Control2008
Control2009
TalkingBook 2008
TalkingBook 2009
10
20
30
Yie
ld (
Bag
s)
HARVEST RESULTS: PER CROP
Users produced: .76 more bags of millet; a 25% increase
(p=.022)
4.4 more bags of groundnuts; a 48% increase (p=.008)
Market value: $136
Users did not produce significantly more maize or beans Possible reasons :
Messages were not relatively as valuable
Improper application Other unmeasured factors negatively
impacted these specific crops
Village-wide exposure ~360 people from 37 farms
(~40% of village)
Testing of Guidance Partially applied – 52% Applied to entire plot – 21% Did not apply – 27%
RESULTS: EXPOSURE AND APPLICATION
Reasons for decreased yields (non users) Over flooding The land lost its fertility Planted at wrong time
What will users do with surplus? 75% of farmers will sell to:
Pay for health insurance Purchase seeds, labor, animals Improve their houses Pay for their children's school fees
33% will use to properly feed their families
RESULTS: FARMER FEEDBACK
LIMITATIONS OF STUDY
QUALITATIVE Not a random sample Foreigners are distracting
QUANTITATIVE Not a RCT evaluation Subjectivity bias -2 to 2 scale had flaws Small sample size weakened analysis
Committee Buy-in from strong local leaders is key Diversify committee leaders Improve ongoing training
Allocation Household rotation Gender issues Leverage the device
Record “rules” on the devices Improve feedback loop
Behavior Change Improving practices alone made a significant difference Access to inputs inhibited some
LESSONS LEARNED
Behavior Change In home vs. outside vs. word of mouth Peer recognition
Usability Experiment with audio instructions Upload/download using mobile phone
Ownership Value to farmers
Do not provide batteries Talking Book microloan
FUTURE WORK