Presentation for the webinar on Collecting sex-disaggregated agricultural data through surveys that took place on April 21, 2016. Learn more about the webinar here: http://bit.ly/1SkWcSx
PIM Gender team members Cheryl Doss and Caitlin Kieran invited participants to discuss how the "Standards for collecting sex-disaggregated data for gender analysis" drafted by PIM in 2014 have been used to date, with a specific focus on lessons learned by CGIAR centers and external partners.
The webinar was co-organized by the CGIAR Gender and Agriculture Research Network and the CGIAR Research Program on Policies, Institutions, and Markets (PIM).
Collecting sex disaggregated agricultural data through surveys
1. Collecting sex-disaggregated
agricultural data through surveys
On gender and agriculture webinar series
April 21, 2016
Cheryl Doss, Yale University
Caitlin Kieran, Policies, Institutions, and Markets (PIM)
2. The presenters
Cheryl Doss is a development economist whose
research focuses on issues related to agriculture,
assets, and gender, with a regional focus on Africa
south of the Sahara. As a Senior Lecturer at Yale
University, she has taught courses on research
methods, the economics of Africa, and agricultural
development and food security. She also serves as
PIM’s gender advisor.
Caitlin Kieran is the Senior Research Assistant for PIM,
where she monitors and supports the incorporation
of gender analysis within PIM’s research portfolio
and researches issues related to gender, agriculture,
assets, intrahousehold dynamics, and survey
methodology.
3. Definitions:
Sex-disaggregated data are information that are
collected about males and females
Gender-disaggregated data are analytical indicators
derived from sex-disaggregated data on social and
economic attributes
Intrahousehold data – data with specific information
regarding individuals living within the same household
Gender analysis is used to understand the relationships
between men and women, their access to resources,
their activities, and the constraints they face. It generally
requires sex-disaggregated data
4. Research areas
Baseline or descriptive research
What are the constraints facing farmers?
Where are the opportunities for increased production
and livelihoods?
How do farmers respond to living in a risky environment?
What is the impact of projects, programs, and policies?
How can projects be designed and monitored to be
gender transformative?
5. Research questions that require
gender analysis
Are there productivity differences between male and female
farmers? If so, what causes these disparities?
In joint production systems, to what extent does it matter who
receives the extension information, purchases the inputs, provides
the labor, makes the decisions, and controls the outputs?
Who decides what food is consumed? Who grows food for
consumption? Who purchases food for consumption?
Who is most vulnerable to shocks? What are the sources of
vulnerability? How do households and individuals cope with risk and
vulnerability?
Who works within each node along a value chain? What constraints
do they face? How can these constraints be removed?
6. Key issues to address
Unit of analysis
Who to interview
Asking the “who” questions
Context
7. What data are needed?
What is the appropriate unit of analysis?
Individual
Household
Intrahousehold
Community
Region or nation
Land area
Asset
Resource unit
Formal or informal organization
Value chain
Photos credits (left to right, top down): World Agroforestry Centre; Akram Ali/CARE Bangladesh Strengthening the Dairy Value Chain (SDVC) project; S.
Mojumder/Drik/CIMMYT; Alternative Futures/CCAFS; Gwendolyn Stansbury/ IFPRI (Flickr)
8. Individual
A farmer or a worker along a value
chain.
What is the average daily wage for an
agricultural worker? Does it differ by
gender?
Explanation
Sample
Research
Question
9. Household
May consider the household as the
production or consumption unit.
How much of their income do households
spend on food?
Explanation
Sample
Research
Question
10. Intrahousehold
Focus is on what happens within the
household. This does not treat the
household as a single unit, but seeks to
understand how multiple individuals
within the household interact and
affect outcomes.
How is food allocated among
household members?
Explanation
Sample
Research
Question
11. Community
Communities may be the focus of
policies or interventions.
How do we improve local
governance?
Explanation
Sample
Research
Question
12. Region or Nation
For cross-country comparisons,
including those of trade or policies,
national or regional analysis is
appropriate.
How do trade policies affect
agricultural exports?
Explanation
Sample
Research
Question
13. Land area
Spatial analyses of how various policies
affect land use may use land area as
the appropriate unit of analysis. Land
parcels may be the unit of analysis for
analyses of agricultural production.
What proportion of land is irrigated?
Explanation
Sample
Research
Question
14. Asset
An asset such as livestock, houses, or
savings may be the focus of policy or
program interventions.
What proportion of cattle are
vaccinated?
Explanation
Sample
Research
Question
15. Resource unit
A forest, watershed, or lake may be
the appropriate unit of analysis for
questions about natural resource
management.
What proportion of forests have active
management groups?
Explanation
Sample
Research
Question
16. Formal or informal
organization
Including farmer cooperatives,
extension service providers, credit
banks, water user groups, micro-
finance groups, self-help groups, etc.
What proportion of cooperatives have
women in leadership positions?
Explanation
Sample
Research
Question
17. Value chain
The value chain, or a node along it,
may be the unit of research on the
sustainability, inclusivity, or efficiency of
value chains.
What is the value added at each
node along a value chain?
Explanation
Sample
Research
Question
18. Which is the appropriate unit of
analysis?
What is the research question?
Are we interested in individuals, households, or
resources?
Are we doing baseline research on people? Or testing
the impacts of interventions?
19. Questions on landownership:
Individual as unit of analysis - To what extent do
men and women own land?
0
10
20
30
40
50
60
70
Percentage of Individuals who Own Land, By Sex
Women Men
21. Landownership:
Intrahousehold Analysis
How does the distribution of
landownership within the
household affect
outcomes, such as
household expenditures or
children’s health?
Photo credit: Georgina Smith / CIAT. Flickr
22. Studies of livestock
May want to focus on the men
and women who own the
livestock: identify the average
number of livestock per owner
Or focus on the livestock: how
many have received prevention
services?
Photo credit: ILRI/Susan MacMillan. Flickr
23. Key issues to address
Unit of analysis
Who to interview
Asking the “who” questions
Context
24. Who to Interview?
Again, depends on the
research question!
Considerations:
Respondent answering
for her/himself vs. proxy
respondent
Interview multiple
respondents within a
household?
Photo credit: ILRI/Collins (Flickr)
25. Who provides information:
Proxy respondents
For some questions, we typically rely on proxy
respondents:
We rely on adults to answer questions about children’s
education
Many surveys interview “the most knowledgeable person”
– one person in the household – about agricultural
production, landownership, or household enterprises
This works best when information is more objective and
we would expect fewer information asymmetries
26. Interview respondents, no proxies
When questions are about preferences, attitudes, or
knowledge
Why don’t you use fertilizer, improved seeds, or
particular management practices?
WEAI asks women about their roles and responsibilities.
Cannot proxy with responses from other household
members
27. But large gray areas?
Who to interview?
Gender Asset Gap project finds that men
and women report different monetary
values when asked the potential sales
price for their dwelling
Labor Force Participation rates among
household members vary by up to 10%
depending on who within the household
is asked (Bardasi et al.)
Methodological Experiment on Measuring
Asset Ownership from a Gender
Perspective (MEXA) finds different levels of
asset ownership for men and women,
depending on the structure of the survey
28. Intrahousehold analyses
Some intrahousehold analyses rely on data collected
from one household member:
Studies considering the bargaining power of women in the
household and the impact on expenditure may rely on
data collected from one person
Other studies compare responses among household
members. Gender Parity Index of the WEAI compares
empowerment indicators of men and women within the
same household
29. What if multiple responses
provide different answers?
For analysis, may want one answer.
Thus need to reconcile the differences
Or include both in the analysis – men’s
response and women’s response to a
question
Or create a variable indicating
whether there is consensus
30. Developing a research plan
Identify key research questions
Appropriate unit of analysis
Who will be interviewed
How responses from different individuals will be
combined, as appropriate
31. Key issues to address
Unit of analysis
Who to interview
Asking the “who” questions
Context
32. Asking the “who” questions
Key to gender analysis is knowing who is involved in
various activities and whether they are men or women
Multiple approaches to soliciting this information,
depending on selection of respondent(s):
Ask about men and women, but not specific individuals
Ask respondent(s) about him/herself (the “do you…” questions)
o Only facilitates gender analysis if interviewing men and
women
Ask questions about specific household member(s) or all
household members
Ask respondent who has various rights and responsibilities
33. Coding responses to the “who”
questions
Record sex of individual(s)
Example:
o Who provides the labor on this plot?
o Response: 1=Men; 2=Women; 3=Men and Women; 4=Children;
5=Men, Women, and Children
Identify individuals by relationship to respondent
Example from Abbreviated Women’s Empowerment in Agriculture
Index
o Who made the decision to borrow from [SOURCE] most of the
time?
o Response: (Circle all applicable) 1=Self; 2=Spouse; 3=Other
household member; 4=Other non-household member; 96=Not
applicable
34. Coding responses to the “who”
questions (continued)
Record ID code (facilitates analysis of additional individual
characteristics)
Example from Living Standards Measurement Study-Integrated
Surveys on Agriculture (LSMS-ISA) in Malawi
o Who in your household kept/decided what to do with earnings
from crop sales?
o Response: List up to 2 ID codes from network roster
In some contexts, it may be appropriate to leave space for more
than 2 ID codes and identify sex of non-household members
35. Key issues to address
Unit of analysis
Who to interview
Asking the “who” questions
Context
36. The importance of context
Gender relations are heterogeneous
Adapt all data collection methods to the context (social
dynamics, gender roles, other dimensions of identity)
Interview setting
o Consider whether it is appropriate for enumerators to be the same or
opposite sex of the respondents
o Consider whether it is preferable to conduct interviews with men and
women separately or together, depending on social norms and topic
Formulation of and training on survey questions and response options
o Ensure that questions are culturally sensitive and relevant
o Ensure that researchers, enumerators, and respondents share
understanding of terms
37. Tips for learning context
Read existing literature on gender issues before embarking on new
study
Collaborate with local researchers
Avoid making assumptions about topics of inquiry
Use mixed methods
Qualitative methods such as participant observation, key informant
interviews, focus group discussions, etc. can play important role in
developing questionnaires and interpreting findings
Cognitive testing, qualitative method paired with a (quantitative)
survey
o Purpose: systematically identify and analyze sources of response error in
surveys, and use that information to improve the quality and accuracy of
survey instruments (Johnson & Diego-Rosell 2015)
38. Review: key issues to address
Unit of analysis
Appropriate unit of analysis depends on
research question
Who to interview
Collect information from men and
women
Asking the “who” questions
Collect information about men and
women
Context
Adapt data collection methods to
the context
39. References and resources
Bardasi, E., Beegle, K., Dillon, A., & Serneels, P. 2010. Do Labor Statistics Depend on How
and to Whom the Questions Are Asked: Results from a Survey Experiment in Tanzania.
Policy Research working paper; no. WPS 5192. Washington, DC: World Bank.
Johnson, K. B., and P. Diego-Rosell. 2015. Assessing the Cognitive Validity of the Women’s
Empowerment in Agriculture Index Instrument in the Haiti Multi-Sectoral Baseline Survey.
Survey Practice. 8 (2). ISSN: 2168-0094
Rubin, D. Forthcoming. Qualitative Methods for Gender Research in Agricultural
Development. IFPRI Discussion Paper.
Women’s Empowerment in Agriculture (WEAI) Resource Center
The Gender Asset Gap Project: Collecting Sex-Disaggregated Asset Data
The Gender, Agriculture & Assets Project
EnGendering Data Blog
“Good practices” on CGIAR Gender webiste
40. Gender and economics workshop:
Advanced techniques for incorporating gender in research design, data
collection, and analysis for economists and other quantitative social scientists
September 14-16, 2016 at IFPRI, Washington, DC
Led by senior economists Dr. Cheryl Doss (Yale University) and Dr. Agnes
Quisumbing (IFPRI)
Workshop Objectives:
Participants will acquire and apply latest advances in knowledge of how to
incorporate gender into research design, data collection, and analysis in
quantitative research studies
Content specifically designed for economists and other quantitative social
scientists across CGIAR who are not gender specialists and are seeking to
include gender analyses into aspects of their work
More information
41. Q&A
Have you used the guidelines? What insights can you share from trying to
implement them? What challenges have you faced?
What have you learned from collecting sex-disaggregated data that you
would not have learned with just household-level data?
How would you use this information to report on CGIAR’s cross-cutting IDO
on equity and inclusion achieved and the related sub-IDOs including (1)
gender equitable control of productive assets and resources, (2)
technologies that reduce women’s labor and energy expenditure
developed and disseminated, and (3) improved capacity of women and
young people to participate in decision making?
How do we make sure we’re collecting data that allow us to do some
comparisons across projects (and answer some larger research questions)
without losing the context specificity?