Performed quantitative analysis on India Human Development Survey and National Family Health Survey data.
Hypothesis: Women’s autonomy is associated with better health seeking behaviour, which in turn, is expected to lead to improved health outcomes provided community factors such as characteristics of health systems are taken into account.
Methodology: Calculated Empirical Bayes estimates, derived from 2-level hierarchical linear model of district-level women empowerment measures.
District-level Data Analysis: Women's Autonomy and Health Outcomes
1. Women’s Autonomy
and Health Outcomes
District-level Data Analysis
Pankaj Gautam
Malavika Subramanyam
IIT Gandhinagar | Fall 2014
2. What is Autonomy? Why Autonomy?
• Self-government or self-direction: acting on motives, reasons, or values that are one's own.
• Women’s autonomy can be measured in a variety of ways[1], but women’s access to and
control over resources is a fundamental aspect of autonomy.[2]
• How does women’s autonomy operate and lead to improvements in health?
• Policy decisions: Does autonomy have an interactive relationship with health service
availability in influencing health behaviour or outcomes?
• Hypothesis: Women’s autonomy is associated with better health seeking behaviour, which
in turn, is expected to lead to improved health outcomes provided community factors such
as characteristics of health systems are taken into account.[3] [4] [5]
3. Methodology
• Quantitative research using
• India Human Development Survey [6]
• National Family Health Survey [7]
• Relevant question/variable identification from IHDS questionnaire to calculate
autonomy score.
• District-level women empowerment measures’ reconstruction (Desai, 2010).
• Data modelling and estimation using Empirical Bayes method
4. Relevant Questions/Variables for
Autonomy Score Calculation
• Who is the primary decision maker about farm matters in your house?
• Who is the primary person responsible for taking care of the animals in your household?
• Besides work on the household farm or in any of the household's businesses, what work for pay or
goods did [NAME] do last year?
• Who in the household worked in household nonfarm business last year? Please include women and
children.
• Education of each member of the family.
• Exposure to Mass Media: How often do people in the household listen to the radio, read newspapers
and watch TV.
5. Relevant Questions/Variables for
Autonomy Score Calculation
• How much did you pay as school fees for NAME in last year? [in addition to govt. support]
• How much did you spend on [NAME]'s books, uniform transportation, and other materials last year?
• How much did you pay for private tuition last year?
• Has [NAME] ever been enrolled in school?
• At what age did [NAME] start school?
• Please tell me who in your family decides the following things?
• Who chose your husband?
• Did you have any say in choosing him?
6. Control Over Family
Resources
Number of districts: 357
Acknowledgement: Haider Ali, PhD Scholar, Department of Civil
Engineering, IIT Gandhinagar
8. Participation in
Household Decisions
Number of districts: 357
Average score of women who have any say in household
decisions about
• what to cook,
• whether to buy an expensive consumer durable item,
• how many children the respondent and her husband
should have,
• when to take a sick child to the doctor, and
• marriage arrangements for the children. [8]
Acknowledgement: Haider Ali, PhD Scholar, Department of Civil
Engineering, IIT Gandhinagar
9. Multilevel Model
Observations within districts will vary
about the district mean
District means will vary about
the grand mean
𝑌𝑖𝑗 = 𝛽0 + 𝑢0𝑗 + 𝑒0𝑖𝑗
𝑌𝑖𝑗 = ith observation of jth district
𝛽0 = Model estimated crude mean
𝑢0𝑗 = Predicted random effect of jth district
𝑒0𝑖𝑗 = Residual term from ith observation of jth district
Excluded the question on cooking and created a dummy
variable that reflects whether women have a final say in
any of the other four decisions.
Representation:
• Control over family resources
• Access to resources
• Participation in household decisions
10. Women’s Autonomy
Empirical Bayes estimates, derived from 2-level hierarchical
linear model, which are equal to the fixed-portion linear
predictor plus contributions based on predicted random
effects.
The empirical Bayes estimate is a weighted average of the
crude mean for each district k and the crude mean across all
districts in the data. The weights are proportional to the
reliability of the neighborhood measure, estimated by
Legend
𝛾00𝑘 =
𝜏 𝜂
𝜏 𝜂 + 1
𝑗
𝜏 𝑏 +
𝜎2
𝑛𝑗𝑘
−1 −1
𝜏 𝜂 = Variance of random effect
𝑛 = Sample size
12. References/Citations
1. Narayan, D., ed. 2006. Measuring Empowerment: Cross-Disciplinary Perspectives. New Delhi: Oxford University Press.
2. Andrist, L. (2008). Social Capital’s Dark Side and Patriarchy in India. India Human Development Survey, Working Paper No. 7. NCAER, University of
Maryland College Park.
3. Caldwell, J. C. (1986). Routes to Low Mortality in Poor Countries. Population and Development Review, 12, 171-220.
4. Basu, A. M. (1992). Culture, the Status of Women and Demographic Behavior. Oxford: Clarendon.
5. Desai, S. (1998). Maternal Education and Child Health: Is There a Strong Causal Relationship? Demography, 35, 71-81.
6. Desai, S., Dubey, A., Joshi, B.L., Sen, M., Abusaleh, S., and Vanneman, S. India Human Development Survey (IHDS) [Computer file]. ICPSR22626-
v1. University of Maryland and National Council of Applied Economic Research, New Delhi [producers], 2007. Ann Arbor, MI: Inter-university
Consortium for Political and Social Research [distributor], 2008-07-30.
7. International Institute for Population Sciences (IIPS) and Macro International. 2007. National Family Health Survey (NFHS), Mumbai: IIPS
8. Desai, S. (2010). Gender scripts and age at marriage in India. Demography, Volume 47, Issue 3, pp 667-687, 2010-08-01. Springer-Verlag.
Editor's Notes
If autonomy has a substitutive effect on health care availability: women with high levels of autonomy are able to seek care, regardless of whether service availability is high or low. In such a situation, more investments will be needed for improving women’s autonomy than for improving access to care, since autonomy will be able to compensate for deficiencies in service availability.
Complementary effect: both autonomy and service availability should be enhanced together to improve health care utilization rates among women.
In South Asia, social norms dictate that young married women defer to mothers-in-law, husbands and household elders in all matters, including those related to their own health and well-being. Efforts to improve women’s health care utilization, as a result, are hindered by women’s restricted abilities to seek care and act upon the information received from health programs.
Since cooking is generally regarded as one of women’s essential responsibilities within the household, we excluded this type of decision-making and created a dummy variable that reflects whether women have a final say in any of the other four decisions