Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Views amidst violence: George Varughese
1. Surveys as a tool to
improve aid in fragile states
The Asia Foundation’s Afghanistan
experience
2. A fragile state environment
Highly dynamic environment; uncertain &unstable
Strained state-society relations; distrust & suspicion
Multiple interests and actors at play; high costs of navigation
Insufficient and unreliable information
Weak local capabilities
Weak physical and transportation infrastructure
3. Some ways that surveys can help
Capture trends in a dynamic environment, if done over time
Monitor state-society relations over time
What matters? Public opinion vs. expert opinion
Help capture key issues that are normally sensitive and deeply political
Specific issues unique to the state can be studied in depth
Can be a public good and used by multiple actors, where
coordination risks and fund flows are high
Better inform program design
Better inform spending (targeting and resourcing)
Test assumptions on key issues
As a common baseline to track progress
Build local capacity to provide reliable information
4. Varying objectives of TAF surveys
Objective Survey
1) Inform host government policy- Survey of Public Perceptions of the Mindanao Peace Process (2008)
makers and local policy debates Local Economic Governance Index (EGI) in 5 countries
Voter surveys in pre-election periods (Bangladesh, Cambodia,
Indonesia, Pakistan, among others)
2) Articulate/reflect key concerns Democracy & Conflict in Southern Thailand
of conflict-affected populations Survey of the Afghan People (2006-11)
3) Inform program design Public perceptions of police (Timor-Leste, Bangladesh, Sri Lanka,
Indonesia)
Leaders of Influence surveys (Maldives, Bangladesh)
4) Inform donor strategies or Survey of the Afghan People (2006-11)
international policy Survey of the Sri Lankan People (2010)
5) Measure impact of programs Community policing pilot program in Sri Lanka, Timor-Leste (2009-11)
5. Case in point: Survey of the Afghan People
Nationwide quantitative survey conducted annually using a
structured questionnaire among more than 6200 Afghans
6 surveys conducted since baseline in 2006
Findings at http://www.asiafoundation.org/publications/pdf/989
8. Some Pre-survey Challenges
Objective (s) of the survey
Partnerships
For what and with whom
‘Localisation’ (Afghanisation) of the survey
Ownership and accountability
Objectivity
Capacity
9. Some Challenges during Survey
In general, minimal secondary information available for design
In Afghanistan, last census in 1979, partial only
No electoral rolls
Questionnaire development
Compatibility to other surveys within & outside the context
Collaborative inputs
Curiosity questions vs. actionable findings
Scale of the fieldwork
Weak to nonexistent local capacity
Challenges with female interviewers
10. …Some Challenges during Survey
Weak infrastructure and harsh weather conditions
Mountainous terrain & lack of (good) roads
Long and harsh winters
Ethno-linguistic sensitivities/closed society
Conservatism
Insularity
Respondent selection and gender balance
Balanced gender coverage at each sample point
Travel restrictions for female interviewers
11. …Some Challenges during Survey
Quality control
Added oversight and supervision of field team and data cleaning
Backchecks and accompanied interviews
Managing sampling and error
Increasing insecurity and inaccessibility
Greater replacement of sampling points
Error margin can increase
12. Some Post-Survey and Other Challenges
Analysis and reporting considerations
Overstating the findings
Interpreting every finding
Emphasizing magnitude vs. direction
Looking beyond topline findings to in-depth analysis, when to move
along?
When to integrate with comparisons to other worldwide surveys
Dissemination plans and the ethics of research
Selective data dissemination, suppression of findings
Duplicative efforts, sharing of information
13. …Some Post-Survey and Other Challenges
Risk management
Antigovernment elements, local strongmen, competitors
Host Government
Donors
Way forward: one-off exercise vs. longitudinal – time series
Changing objective over time of conducting large scale surveys of
Afghans
Disciplined to team, timing, and, quality control costs
Choice between parachute survey firm and in-house management
in early years
37. Objectives of the Afghan Survey
Use strong social science research to provide credible, policy-
relevant information about public opinion on recent changes and
trends in government performance, public policy, politics and
political processes, and development progress
Build social research capacity in the country
38. Dealing with outdated population data
Rural sampling -- Combination of PPS and SRS
Urban sampling -- use of maps/locality lists in the absence of
database such as electoral rolls
Innovative field movements for selection of starting points and
households
39. Building the field team
Extensive training of interviewers & supervisors
Questionnaire
Sampling procedure- field movements, selection of households
Kish grid
‘Train the trainer’ sessions
Mock interviews
Female interviewer recruitment in each district of selected sample
point
40. Innovations
Team of women interviewers recruited locally
‘Mahrams’ for women interviewers
Men and women interviewers in each sampling area
Modifications in Kish selections
Women interviewers obtained details of female HH members; men
obtained details of males