1. 1
Consultation Workshop for Developing a Nutrition
Training Roadmap for Administrators across India
Dr. Divya Nair & Madhav Seth
2. 2Objectives of this presentation
Discuss Goals of Capacity
Building
Discuss some Principles on
Content
3. 3Objectives of this presentation
Discuss Goals of Capacity
Building
Discuss some Principles on
content presented
4. 4The basics of Capacity Building
What are possible
channels?
• Technical assistance
• Virtual and in-person training
sessions
• In-depth consultations
• Online learning
• Guidance materials/ knowledge
products
• Coaching and mentoring
What are Goals of
Building?
• Develop knowledge and/or skills of
practitioners
• Improve commitment, systems and
leadership for effective actions
towards outcomes
5. 5
Elements of successful capacity building
Demand-
driven
Regular
Long-termSubtractive
Systems-
focused
• People you want to train want
you to train them
• CB efforts occur multiple
times at a regular cadence
• Include follow-up
• Efforts are embedded in a
long-term relationship
• Reduce the amount of
effort beneficiaries put in
to do same work
• Factor in staff shortages,
demands from superiors,
other demands on time
• Align incentives for data
quality
6. 6Selected IDinsight Capacity Building Experience
Workshops at LBSNAA
• Activity and discussion-based session on using
data to identify problems in the nutrition sector
• Presentation-based session on data universe in
education
Monitoring workshops with
Government of Malawi
• High-level workshops on fundamental M&E
concepts
• Regular trainings on technical skills (Excel, etc.)
Evaluation workshops with
UNICEF Philippines & Kenya
• Technical workshops on evaluation design and
implementation
Data use Needs Assessment
• Field observations of data use needs and
constraints in nutrition sector
7. 7
Needs assessment methodology
14 development partners
Data use needs
assessment
7 district and block Health
Department officials
Semi-structured interviews
8 district and block ICDS officials
3 states, 5 districts
25 front line workers and Lady
Supervisors
Districts: Bihar – Sitamarhi, East Champaran; Chhattisgarh – Raipur, Rajnandgaon; Uttar Pradesh – Kannauj
8. 8
Data capacity
• Large numbers of vacancies across blocks and districts. Lead to
overwhelmed departments.
• Lack of basic monitoring equipment to measure fundamental nutrition outcomes.
Unclear how officials are keeping track of nutritional status of beneficiaries.
• Outcome and output data analysis and reporting mostly driven by development
partners. Skills need to be transferred at some point
Emerging recommendation: Key vacancies need to be filled and appropriate hardware provided
before capacity-building interventions become a focus.
9. 9
Data quality
• Admin data is generally unreliable. Large discrepancies between
official figures and third-party sources.
• Development partners perceive low interest at senior levels in improving data quality
• Many officials seem to be aware data quality is a problem and are aware of
common techniques to check for quality. They identify overburdened staff as
the main cause of poor data quality.
Emerging recommendation: National and state review meetings should include discussions on
data quality and, where applicable, the reasons behind differences between administrative data and
third-party data.
10. 10
Data prioritization
• Program and review priorities are determined at the state and
national level. Officials in districts and blocks said they structured their
work around requests from senior officials.
• Most of the data discussed in review meetings is activity or input-level data.
• Individuals have a large impact on how seriously their subordinates take data
use. All stakeholders noted that a motivated DPO or DM could substantially
alter the way entire departments function.
Emerging recommendation: Data use workshops for district and block officials should ideally be
conducted drawing from existing data use cases, they should start by aiming to change the
philosophy around data use.
11. 11Objectives of this presentation
Discuss Goals of Capacity
Building
Discuss some Principles on
content presented
12. 12Today’s discussion
Workshops at LBSNAA
• Activity and discussion-based session on using
data to identify problems in the nutrition sector
•
Data use Needs Assessment
•
13. 13
Key Ingredient of Successful Capacity Building Work
S
P
I
C
E
Simple
Digestible, modular, jargon-avoidant
Practical
Implementable, applicable to day-to-day responsibilities
Interactive
Audience participation, activities, discussion
Contextualised
Sector-specific content
Engaging
Concise, visual presentations, tightly timed
14. 14
Key messages at LBSNAA POSHAN session
Data can help diagnose the problem in the chain from inputs to
outcomes and impact
Comparisons, checks and time trends can help you validate the data
that you see
MIS is the main source of data at the district. National and third-party
surveys can sometimes complement this data
15. 15
Diagnosing the problem - example
The World Health Organization recommends all pregnant women have at
least 4 antenatal care checkups. However, in District A the percentage of
pregnant women who had at least 4 antenatal visits is only 16% (HMIS). What
is going on?
16. 16
What are potential reasons for this problem?
Lack of awareness of importance of 4 antenatal check-ups
Lack of ability (time, support, money) for pregnant women to get check-ups
Lack of functional health centres
What are the main problems?
Other?
17. 17
Some relevant data for the problem
Categories Indicators Data Source Value for District A
Inputs
Percentage of sub-centres out of total needed
Percentage of ANMs out of total needed
Interventions
Percentage of pregnant women registered for antenatal care
Percentage of pregnant women registered for antenatal care
within first trimester
Percentage of pregnant women who had at least 4 antenatal
visits by delivery
Outcomes
Percentage of mothers who consumed iron folic acid for 100
days or more when pregnant
Percentage of institutional deliveries to total reported
deliveries
ANCs can also improve calcium and deworming consumption, protection against neonatal tetanus, anemia testing, dietary
diversity, increase child birth weight, and decrease the Maternal Mortality Rate
Where do we get this
data from?
18. 18
How would you normally get
information on nutrition
activities and outcomes in your
district?
Discussion Question (data sources)
19. 19
Admin Data (MIS)
● National scale
● Self-reported through
program staff
● Available at fairly high
frequencies
● E.g. HMIS, ICDS-CAS
Third Party Surveys
● Scale depends on demand
● Collected by surveyors from
a sample of the population
● Frequency depends on scale
and operational efficiency of
surveys
● Eg: TSU CBTS, Aspirational
Districts Survey
National Surveys
Let us look at three district data sources
● National scale
● Collected by surveyors from
a sample of the population
● Frequency typically every few
years
● Eg: NFHS, NSS, Multiple
Indicator Cluster Surveys
20. 20
Admin Data (MIS)
● National scale
● Self-reported through
program staff
● Available at fairly high
frequencies
● E.g. HMIS, ICDS-CAS
Third Party Surveys
● Scale depends on demand
● Collected by surveyors from
a sample of the population
● Frequency depends on scale
and operational efficiency of
surveys
● Eg: TSU CBTS, Aspirational
Districts Survey
National Surveys
Let us look at three district data sources
● National scale
● Collected by surveyors from
a sample of the population
● Frequency typically every few
years
● Eg: NFHS, NSS, Multiple
Indicator Cluster Surveys
We will focus mostly
on admin data
21. 21
Some features of administrative (MIS) data
Who collects the data?
Who is represented in
the data?
People who access government
services
Data is mostly reported by front
line workers
What indicators do we
normally see?
Data on inputs and interventions
What indicators are
commonly not seen?
Data on knowledge, attitudes,
practices
ANMs and ASHAs
Pregnant women on ANMs and
ASHAs lists
ANC registration, 4+ ANC, IFA
tablet distribution
Knowledge of ANC, consumption
of IFA
Example: Antenatal check-upsFeatures
22. 22
However, we have seen differences between MIS,
national and third-party survey data
IDinsight survey
sample sizes
District # Children
<5 years
District 1 244
District 2 299
District 3 369
District 4 334
District 5 414
*MIS data is from June 2018; IDinsight data was collected from May - August 2018; NFHS-4 was collected in 2015-16
23. 23
How do we explain these differences?
Different coverage
Different
reference periods
Sample surveys cover entire population, MIS data limited to those who access
government services
Survey questions may ask about different reference periods than MIS data
Outdated
denominators
Denominators in MIS data may need updating, e.g. estimates of pregnant women in
the district
Inaccurate reporting
Key Takeaway: MIS data is very useful, but be aware of its limitations!
Self-reported MIS data may be inflated to display desirable results
24. 24
Checks
Facility inventory, spotchecks, personal
interviews, backchecks, phone audits
So how do we know if the MIS data is accurate?
Comparisons
Different data sources, neighbouring districts,
investigate extreme values
Time trends
Does the rate of change seem sensible?
25. 25
Data on the percentage of pregnant women completing 4+ ANC check-ups varies by
source. For District A, HMIS data indicates 16%, ICDS-CAS indicates 20% and NFHS-4
indicates 7%. Why do you think this difference exists?
Discussion Question (Data Validation)
26. 26
Supportive culture around accurate reporting
can lead to better program performance
How would you encourage more truthful reporting by district and block officials during
the district review meetings?
27. 27
Admin Data (MIS)
● National scale
● Self-reported through
program staff
● Available at fairly high
frequencies
● E.g. HMIS, ICDS-CAS
Third Party SurveysNational Surveys
Let us briefly talk about the other two sources of
data at the district
● National scale
● Collected by surveyors from
a sample of the population
● Frequency typically every few
years
● Eg: NFHS, NSS, Multiple
Indicator Cluster Surveys
● Scale depends on demand
● Collected by surveyors from
a sample of the population
● Frequency depends on scale
and operational efficiency of
surveys
● Eg: TSU CBTS, Aspirational
Districts Survey
28. 29
If you would like to conduct a survey in your
district, these resources can be helpful
Local universities
Third party survey organizations
29. 30
Diagnosing the problem - example
The World Health Organization recommends all pregnant women have at
least 4 antenatal care checkups. However, in District A the percentage of
pregnant women who had at least 4 antenatal visits is only 16% (HMIS). What
is going on?
30. 31
Categories Indicators Data Source Value for District A
Inputs
Percentage of sub-centres out of total needed
Percentage of ANMs out of total needed
Interventions
Percentage of pregnant women registered for antenatal care
Percentage of pregnant women registered for antenatal care
within first trimester
Percentage of pregnant women who had at least 4 antenatal
visits by delivery
Outcomes
Percentage of mothers who consumed iron folic acid for 100
days or more when pregnant
Percentage of institutional deliveries to total reported
deliveries
Relevant data for the problem
Now we know the
sources and how to
validate them. What
next?
31. 32
Categories Indicators Data Source Value for District A
Inputs
Percentage of sub-centres out of total needed HMIS 88%
Percentage of ANMs out of total needed HMIS 90%
Interventions
Percentage of pregnant women registered for antenatal care HMIS 75%
Percentage of pregnant women registered for antenatal care
within first trimester
HMIS 50%
Percentage of pregnant women who had at least 4 antenatal
visits by delivery
HMIS 16%
Outcomes
Percentage of mothers who consumed iron folic acid for 100
days or more when pregnant
NFHS 5%
Percentage of institutional deliveries to total reported
deliveries
HMIS 78%
Relevant data for the problem
33. 35
3
Scenario Table 1
Participants received three documents
Instructions
1 2
A chart paper will also be provided later to help you create a presentation
34. 36
Steps
1
(10 minutes) Read through the scenarios individually. Your scenario may be about either
IFA tablet consumption, Exclusive breastfeeding, or Growth monitoring
2
(15 minutes) Please discuss the given questions in your group to identify the source of the
problem. The data presented in Table 1 may be useful. Is there any additional data that
would be helpful to know?
3
(15 minutes) What actions can you take to alleviate the problem? Are there some actions
that are not in your control, but would be helpful?
4
(20 minutes) Use the chart paper to create a 3-minute presentation on the following:
a) Causes of the problem
b) Actions that can be taken
Please remember to choose a presenter for your group
35. 37
Key messages reiterated to participants
Data can help diagnose the problem in the chain from inputs to
outcomes and impact
Comparisons, checks and time trends can help you validate the data
that you see
MIS is the main source of data at the district. National and third-party
surveys can sometimes complement this data
36. 38
Future IDinsight Capacity Building modules
Data quality
Data visualization
• Why quality data is important
• How to identify suspect data
• How to improve quality of data
• Why visualisation is helpful
• Easy ways to visualise data recorded on paper
• Most appropriate visualisation given set of data
37. 39
Discussion: How can this be operationalized?
• People you want to train want
you to train them
• CB efforts occur multiple
times at a regular cadence
• Include follow-up
• Efforts are embedded in a
long-term relationship
• Reduce the amount of
effort beneficiaries put in
to do same work
• Factor in staff shortages,
demands from superiors,
other demands on time
39. 4141
Our collaborations deploy a large analytical toolkit
to help clients design better policies, rigorously
test what works, and use evidence to implement
effectively at scale.
We place special emphasis on using the right tool
for the right question, and tailor our rigorous
methods to the real-world constraints of decision-
makers.
IDINSIGHT USES DATA AND
EVIDENCE TO HELP
LEADERS COMBAT
POVERTY WORLDWIDE.
41
41. 43
WE HAVE WORKED ACROSS INDIA ACROSS DIVERSE SECTORS
We have worked in 13 countries globally, and 14 Indian states including Delhi
Financial Inclusion | Sanitation | Agriculture | Education | Digital Identity
42. 44
High-frequency tracking of key
socioeconomic indicators at the
district level
Learnings to help improve
socioeconomic wellbeing using
responsive research services
Measurement:
Track district performance
Learning:
Support performance
improvements
IDinsight Objectives IDinsight approach
Data on Demand
High frequency data collection
infrastructure to support
measurement and learning
IDinsight works with NITI Aayog on the Aspirational
Districts program with a three-pronged approach
43. 45
Scope
• Data collection was in November 2018, after POSHAN Maah in September
Timeline
• Stratified random sample using ASHA & Anganwadi lists
• Robust automated data quality checks, audio audits and backchecks
Sampling & Data Quality
Background on Social & Behaviour Change Communication Survey
• Sample covered 6,208 pregnant and lactating women and 1,365
Anganwadi workers
• Representative of 27 NITI Aspirational Districts in 8 states, and ~25 lakhs
pregnant and lactating women
44. 46
Which messages are recalled by pregnant and lactating women?
Source: POSHAN Abhiyaan Social and Behavior Change Communication Survey for MoWCD. November 2018
45. 47
What are women’s knowledge levels?
*Prevention of anemia - Green veg, deworming, IFA or nutritious foods prevents anemia
*Washing hands at critical times - Wash hands with soap before feeding child & after defecation
Source: POSHAN Abhiyaan Social and Behavior Change Communication Survey for MoWCD. November 2018
46. 48
Principles Description
Accurate Data should be unbiased and representative of relevant population
Relevant The metric(s) measured should be linked to the final outcome(s) of interest
Traceable Data should be verifiable
Timely Data is collected and reported in time to inform decision making
Light Data should be collected in a way that is not burdensome on the organization
Useful Data should be used to inform programmatic decisions
.
These principles are important to keep in mind
while developing a monitoring framework
47. 49
Steps Description
Theory of change refinement
Develop a monitoring framework which aligns with the program’s
theory of change and priorities
Metric selection
Determine metrics that accurately measure different aspects of the
program
Data system creation
Develop data collection, management, and reporting systems to
communicate results in an understandable and actionable way
Response codification
Create a ‘response framework’ for monitoring data and make strategic
changes to improve program based on incoming data
There are four key steps for setting up a strong
monitoring framework
48. 50
Additional Concepts:
Sample Size: The total
number of units selected for
survey from the population
Margin of Error
(Confidence Intervals): The
range within which the
sample survey value can
deviate
Typically, if sample size is low,
then margin of error is wider
Sample survey: data is collected from a random
sample of the population to draw inferences
about the entire population
49. 51
27 Districts
8 States
~1000 households per district
~300 surveyors
Example: sample survey for the Aspirational Districts
Program
50. 52
Number of Districts
Health & Nutrition
Indicators
The self-reported data indicators are outside the
survey margin-of-error for many districts
51. 53
This is another example of observed differences
between MIS and survey data
Source for survey data: IDinsight Asiprational District Survey
52. 54
Please discuss the following to the person next to you:
Data on the number of pregnant women at your district varies by source. HMIS data
indicates 4,000 pregnant women and ICDS-CAS indicates 3,500 pregnant women. Why do
you think this difference exists and what are some ways to correct these discrepancies?
Discussion Question (Data Validation)