Data needs for Gender Research Cheryl Doss Consultant, Research Program on Policies, Institutions and Markets, CGIAR Senior Lecturer, Yale University July 8 th , 2013 Presentation given as part of IFPRI’s Gender Methods Seminar Series
Nov 22, 2014
Data needs for Gender Research
Cheryl DossConsultant, Research Program on Policies, Institutions and Markets, CGIAR
Senior Lecturer, Yale University
July 8th, 2013Presentation given as part of IFPRI’s Gender Methods Seminar Series
Two sets of questions for gender research for agriculture:
1) How to improve agricultural productivity? And how does gender fit in?
2) How do changes in agricultural production affect women and girls and men and boys?
These fit into broader questions about the role of agriculture in development .
Data needs1) Survey data needs to be disaggregated; more data at the individual level2) Information on how institutions and structures -- markets for inputs, outputs, credit, and labor, and legal systems -- are experienced differently by men and women and how this impacts the well-being of individuals and communities and the processes of agricultural development and economic growth.
There are many issues related to data collection for gender. For a detailed discussion of the approaches and variables needed, see
Doss, C. (2013). Data Needs for Gender Analysis in Agriculture. IFPRI Discussion Paper 01261.
http://www.ifpri.org/sites/default/files/publications/ifpridp01261.pdf
This presentation focuses on two issues:
• 1) Who to interview• 2) How to handle data from multiple people
within a household.
Collecting Sex Disaggregated Survey Data
Challenge 1: Who should be interviewed? • Does sex disaggregated data mean we have to
interview multiple people per household? • What do we mean by data at the individual
level?
Many possible units of analysis• Individual (farmer, worker, etc)• Household• Intrahousehold (dynamics within household)• Community• Regional/National• Land area or plot• Resource Unit (e.g. a forest or water source)• Institution or Management Unit• Value Chain
Who to interview?
Many household surveys interview one person – often the household head.
If interviewing one person, identify the respondent based on roles and responsibilities, such as the primary farmer.
For analysis, compare male headed, female headed, and couple headed. Include measures of household structure.
Options for interviewing multiple people in household
• Principal couple• One man and one woman• One or two randomly chosen people • Everyone that is relevant for a specific
module. Ask each person about their own…
What is the goal for interviewing multiple people within a household?
• Obtaining a more complete picture of the household. One person doesn’t have all of the information. – Gender differences in roles and responsibilities– Information hidden from spouses
• To learn where perceptions within the household differ.
Gender Asset Gap Project surveys:
• Individual level asset data collected in Ecuador, Ghana, and Karnataka, India
• Two respondents • Household inventory and individual
questionnaire.
Major Assets added by interviewing a second respondent (% added to inventory)
Asset Ecuador Ghana India
Principal dwelling 0.40 Na na
Agricultural parcels 1.36 0.35 0.15
Other real estate 3.91 0.41 0.35
Non-farm businesses
0.52 1.04 0.12
Ecuador Ghana
Asset N (assets)
% who disagree
N (assets) % who disagree
Dwelling 450 35.1 510 7.7
Agricultural land 94 30.9 873 3.3
Other real estate 164 20.1 413 7.8
Non-farm business 534 22.3 641 1.6
Disagreements among couples over who owns the asset.
Benefits to interviewing two people:
• In this case, few additional household assets were identified by interviewing a second person.
• But disagreements over ownership were identified; these may be important for understanding outcomes within household.
Challenge 2: How to analyze data with conflicting answers?
When you need one consistent answer – such as identifying the owner of a parcel of land. Create a decision rule:1. Use one answer (from the primary farmer,
head, primary income earner, oldest adult)2. Use the broadest definitions – include everyone
as an owner who claims to be an owner. 3. Use other information, such as marital property
rules
How to analyze data with conflicting answers?
• For some issues, the disagreement may be the important issue.
• May want to code households where there are major disagreements.
• For women’s empowerment, what may be important is her perception.
• Gender Disaggregated Survey, Kenya Agricultural Productivity and Agribusiness Project
• Interviewed Primary farmer and one other household member
Respondents by Sex and Relationship to Head of Household, Kenya
Primary Farmer Individual Respondent
Relationship to Head Male Female All Male Female AllHead 98% 36% 65% 79% 3% 32%
Spouse 0% 62% 33% 4% 91% 58%
Child 1% 1% 1% 15% 4% 8%
Other 0% 1% 1% 2% 2% 2%
N=farmers 1,160 1,369 2,529 566 957 1,523
Who is the household head?
• 35 both respondents claimed headship• 14 both respondents claimed to be the spouse• In 2 households, the two respondents
identified different men identified as head
1: Primary farmer (PF) does not list a parcel joint w/Individual Respondent (IR); IR lists parcel joint with PF2: PF lists parcel joint w/ IR, but IR does not list any parcel joint w/PF3: PF lists self as an owner on all land parcels; IR lists parcel owned individually.
Distribution of Responses among household members about land ownership
Type of inconsistency
Couples Non-couples
N= hhs Female primary farmer
Male primary farmer
Both males
Both females
Male primary/ Female ind.resp
Female primary/
Male ind.resp
None 52% 51% 90% 86% 62% 62% 53%1 5% 32% 0% 0% 5% 3% 20%2 32% 2% 0% 0% 0% 2% 11%3 6% 4% 10% 11% 33% 33% 7%Other 5% 13% 0% 3% 0% 0% 9%N=households 455 860 10 37 39 99 1,500
Distribution of Households by Percentage of HH Livestock Owned by Spouse of Primary Farmer.
None Some1 Most to All2 More than 100%
Conditional mean N=hhs
Male Primary Farmerimproved cattle 89% 2% 7% 2% 3.79 394local cattle 87% 6% 5% 2% 4.18 573Sheep 72% 11% 12% 6% 2.53 206Goat 60% 21% 12% 8% 4.46 223chicken (indigenous) 15% 42% 20% 22% 6.27 614chicken (improved) 67% 0% 6% 28% 24.17 18Female Primary Farmerimproved cattle 22% 7% 54% 18% 5.42 354local cattle 30% 13% 40% 17% 6.67 349Sheep 31% 8% 35% 26% 5.40 134Goat 27% 18% 32% 23% 6.32 159chicken (indigenous) 74% 11% 6% 9% 8.00 359chicken (improved) 89% 0% 0% 11% 10.00 9
Primary farmer identified total # of animals owned in the household; Spouse identified own animals. 1 Some includes at least one animal to 95% of household animals2 Most to all is more than 95% to 100% of household animals.
Percentage of Couples who Agree on Responsibilities for Livestock, by Sex of Primary Farmer.
Responsible Decide to sell Keep Income
Husband Wife Husband Wife Husband Wife
Improved Cow 19% 42% 48% 54% 52% 60%
Goat 23% 36% 20% 42% 23% 43%
Indigenous Chicken 68% 43% 52% 34% 51% 34%
Note: The sex of the person responsible is not identified in this table.
Conclusions
• Whether you interview one or multiple people will depend on the research question.
• Interview multiple people if needed for full information or when different perceptions within household will affect outcomes.
• Depending on who you interview, you may get very different answers.
• Need to consider whose answer you need. • Be transparent in how you choose.
References:
• The Gender Asset Gap Project: http://genderassetgap.iimb.ernet.in• Cheryl Doss, “The Gender Asset Gap in
Agricultural Assets in Kenya.” Draft.