Page 1
SUCCESSFUL LEADERSHIP FOR
MATERNAL, NEWBORN & CHILD HEALTH
A POLICY ANALYSIS OF FACTORS ASSOCIATED WITH
COUNTRIES’ PROGRESS TOWARDS MDGS 4 & 5
Work in Progress –
Discussion of Methods and Preliminary Findings
Sadia Chowdhury, World Bank ([email protected] )
Shyama Kuruvilla, PMNCH ([email protected] )
Henrik Axelson, PMNCH ([email protected] )
Daniele Caramani, Univ. of St. Gallen ([email protected] )
From Pledges to Action Pre-Forum Technical session, New Delhi, 12 November 2010
Page 2
PART I
BACKGROUND AND
ANALYTICAL FRAMEWORK
Page 3
Background for
policy analysis
•Decade of progress -
declining maternal and
child mortality rates
•Global policy agenda -
G8, African Union, UN
MDG Summit
•Variable progress -
country differences, but
not clear why
Page 4
Defining leadership
Leadership is the
ability to influence,
motivate, and enable
individuals and
organizations toward
achieving agreed
goals and
commitments.
Page 5
Analytical framework
Page 6
PART II
PRELIMINARY FINDINGS OF
COUNTRY CASE STUDIES
Page 7
Case study: Nepal
• Reproductive health
rights in Constitution
• Long-term health plans
and safe motherhood
policies
• Innovations for targeted
groups
• Remote area strategy for
SM
• Community based
newborn care packages and
insurance
• Contraception
• Challenges:
• Last mile difficult to
achieve
• Scaling up
Page 8
Case study: Bolivia
• Address barriers to
access:
• Financial: Maternal
and infant insurance
program
• Geographical:
Extensa progam
• Challenges:
• Initial progress, but
plateau
• Inequity
• Neonatal
Page 9
PART III
DATABASE AND INDICATOR
DEFINITIONS + PRELIMINARY
FINDINGS OF BIVARIATE
REGRESSION ANALYSIS
Page 10
Domains and sub-categories
• Dependent variables
Progress on MDGs 4 and 5 (e.g. average annual rate of
mortality reduction, 1990-2008)
• Independent variables (critical for progress on MDGs)
Governance
Leadership
Entitlements (policies/laws + financing)
• Mediating variables (health-related mechanisms through
which leadership inputs are channelled towards MDGs)
Human resources and infrastructure
Interventions delivered in the health system
Page 11
Domains and sub-categories cont.
• Mediating variables cont.
Interventions delivered in the community
Intersectoral interventions (watsan + nutrition)
Equity
• Moderating variables (contextual factors)
Socioeconomic development
Environment
Sociocultural context (e.g. religious, ethnic and linguistic
diversity)
Gender
Education
Demography
Page 12
Criteria for indicator selection
• Variables hypothesized to influence MNCH outcomes
Informed by literature review
• Consensus on indicator
As evidenced by systematic and standardized use of
indicator in monitoring and evaluation, e.g. indicators
in DHS and MICS, Countdown to 2015
• Regularly collected and publicly available
• Use of indices if possible
E.g. World Governance Index, Global Innovation Index,
Gender-related Development Index
Page 13
Set-up of database
• Set up Excel-file with indicators and info on data source,
# of data points, etc
• Iterative process to reduce number of indicators
Eliminate if adequately captured by indices and/or other
indicators
Current number: 79
• Data collection, entry and review
Some indictors dropped because too many missing
values
• Preparation for Boolean and regression analysis
Determination of cut-off points
Coding: transfer values to dichotomous + ordinal
Page 14
Bivariate regression
• Explore associations between dependent (progress and
MDGs 4 and 5) and independent/mediating variables
To identify variables that could be explored more fully
through:
Boolean analysis and further regression analysis
Country case studies
Page 15
Bivariate regression cont.Variables with association with independent variables
MDG 4 MDG 5
Governance (several indicators) √
Leadership (some of the indicators) √ √
Abortion policy √
Total health spending per capita √ √
ODA per child under five √
Number of doctors and nurses √ √
Malaria treatment √
Child vaccination √
Water and sanitation √
Human Development Index √ √
GDP per capita √
Gender Development Index √ √
Page 16
PART IV
BOOLEAN ANALYSIS:
RESEARCH DESIGN AND
PRELIMINARY RESULTS
Page 17
The Nature of Boolean Analysis
Advantages:
1. Comparative approach: analytical ("why" question:
relationship between variables)
2. Few cases and many variables
3. Qualitative data: dichotomous dependent variable
(presence/absence of phenomenon), non-quantifiable
properties
A different logic than statistical analysis:
1. Multiple causation
2. Combinatorial logic: configurations of factors
3. Analysis of necessary and sufficient conditions: crisp-set
and fuzzy-set.
Page 18
What We Want to Explain
Why are some countries on track and why other are not?
Operational definition depending on MDG4 and 5:
• 20 on-track countries for either/or MDG4 and MDG5. The
countries are all those listed above.
• 6 on-track countries for both MDG4 and MDG5. The
countries are the following: Bolivia, China, Egypt, Eritrea,
Romania, Vietnam.
• 50 countries which are on track for neither MGDs.
Page 19
Distribution of 70 Country Cases
On track vs. not on track:
On track, progressing regressing:
Page 20
Independent Variables with
Explanatory Potential (MDG4)
Stronger results for MDG4 than MDG5.
Being on track:
1. Necessary conditions
Leadership culture (index)(100%), proportion of vaccination above 75%
for measles and for 3dose, proportion of population with access to good
quality water and sanitation above 75%, proportion of women who
attended at least once skilled personnel above 50%.
2. Sufficient conditions
gender development indices above .70 and protective leadership culture
above 4 (on the 1 to 7 scale)(100%) .
Summing up: what would "guarantee" outcome (be sufficient)?
[Bold meaning necessary but not sufficient.]
Page 21
Independent Variables with
Explanatory Potential (MDG5)
As for MDG4: hygiene factors inclusive and protective leadership and
high gender development index: necessary conditions for
a country being on track BUT other factors not helpful
(keep in mind that all countries which are on track for MDG5
are also on track for MDG4 with 1 exception).
MDG5: more prominent role for :
• proportion of women who attended at least once skilled
personnel above 50%;
• attendance of birth by at least 50% of women;
• gender development index above .70.
Model has high "coverage" (0.85) meaning these are necessary
factors BUT low "consistency" (0.50) indicating that it is not
sufficient to lead a country toward being on track for MDG5.
Page 22
Points for Discussion and Open Questions
1. Triangulation and further analysis: (1) logistic regression (2) case
studies (e.g., no information in dataset on "international linkage") (3)
fuzzy-set analysis.
2. Policy advice: "taking action" factors.
3. The nature of the dependent variable:
- On track vs. not on track is one possibility;
-The other possibility being to try to explain absolute rates of
mortalities rather than progression over time.
4. Definition of leadership; relationship between leadership and
policies, expenditure.
5. Plausibility of operationalisation (examples):
- Gender development index and HDI at .70;
- Immunisations, vaccinations, etc. at 50%;
- Leadership culture indices at 4 (on 1 to 7 scale).