Mothers’ health knowledge for children with diarrhea: who you are or who you know?

Abstract

This paper explores the relationship among Indian mothers’ health knowledge regarding proper treatment of diarrhea in children and caste, religion and other mediators, estimating causal impacts through a combination of instrumental variables and matching methods. The results indicate that both caste/religious affiliation and access to health networks (measured across three different dimensions) are important determinants of health knowledge. High caste and non-Muslim women are more likely to have correct health knowledge, but not when accounting for greater female education within these groups. Women with access to health networks are also more likely to have correct knowledge. Additionally, high caste women benefit more in terms of health knowledge from having health networks than women from other groups; except if the health worker is of the same caste/religion, in which case low caste and Muslim women sometimes benefit by twice as much or even more. If caste and religion-related differences in health knowledge are to be reduced it may not be enough to improve access to high quality health networks—such networks also have to be homophilous. Improved treatment from and confidence in the medical profession is found to be part of the mechanism linking health network access with improved health knowledge.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3

Notes

  1. 1.

    Diarrhea is particularly dangerous for very small children, due to their immune systems not being fully developed yet, such that the onset of diarrhea is much more likely to lead to death here (as compared to older children and adults).

  2. 2.

    In a descriptive study based on Indian NFHS data (3rd wave, 2005–06), only 26% of children with diarrhea received oral rehydration treatment or increased fluids, as recommended, and 26% received no treatment at all. Sixteen percent received antibiotics, which are not recommended for treating most childhood diarrhea (USAID 2010).

  3. 3.

    See Desai et al. (2010) for more details.

  4. 4.

    Arguably, the household head may not always be the most knowledgeable about health networks, particularly if the mother cares for the children and may not be the head.

  5. 5.

    However, after the main analysis, below, we still pursue a sensitivity analysis where water quality/water access is incorporated. It turns out not be important, neither in substantive nor statistical terms.

  6. 6.

    Occupational sub-caste (see Section 3 for more details).

  7. 7.

    The interpretation of “jati” here is thus not literally as jati in terms of (occupational) sub-caste but rather as main caste or religious community. This interpretation was supported by Sonalde Desai, one of the PIs of the IHDS examined here, in remarks given to the audience at the 2014 IHDS User Conference in Delhi (at which the authors were both participants).

  8. 8.

    These items include medical expenditures, so that households that acquired health knowledge through medical expenditures may appear to have higher expenditures. Given that our primary motivation for including household consumption is simply to capture/proxy household income, we do not think medical expenditures should be excluded from household consumption for this analysis.

  9. 9.

    See Desai et al. (2010) for more details.

  10. 10.

    These are asked both for women and men in the household overall. We use the information pertaining to the women of the household here.

  11. 11.

    Indigenous tribal peoples living in remote areas, also denoted Scheduled Tribes.

  12. 12.

    Considered the untouchables, also denoted Scheduled Castes.

  13. 13.

    With the caveat that there might be biases in the reporting of diarrhea, so that mothers across caste/religion groups do not necessarily have the same perception about what constitutes child diarrhea. Specifically, it seems likely that less educated women—which are disproportionately represented among lower caste and Muslim women—are less likely to correctly diagnose child diarrhea. With the low sample sizes here, however, we disregard this issue and simply treat these findings as indicative.

  14. 14.

    Reflecting also the low overall sample size here, there were no children from Sikh, Jain, or Christian households in this sub-sample.

  15. 15.

    The possible selection into conversion, if anything, would therefore seem to strengthen our subsequent results. That is, these results can be seen as conservative, lower bound estimates.

  16. 16.

    Altindag et al. (2011) using two waves of the NLSY79 find evidence consistent with Kenkel (1991), but do not find that allocative efficiency is the primary reason for why education affects health knowledge.

  17. 17.

    The lower castes do not have rights to drink from common water sources such as public wells and taps out a fear of impurity and pollution, although this practice is on the decline. On the other hand, the donation of water is seen as a virtue in Hinduism, so a dichotomy exists. While the lower castes may be donated clean water, they do not have property rights to it (Pradhan and Meinzen-Dick 2010).

  18. 18.

    Health knowledge in their setting is an index composed of 7 health knowledge questions in the survey covering knowledge of water purity (2 questions), knowledge of the dangers of a wood-burning stove, knowledge of the benefits of breastfeeding during early pregnancy, knowledge of hand washing after defecation and knowledge on hydration following diarrhea.

  19. 19.

    This model and the associated discussion is an adapted version of the expanded Grossman (1972) model developed in Blunch (2006); see Blunch (2006) for more details, including the specification of utility and production functions, etc.

  20. 20.

    For example due to differing norms and traditions related to health and health practice (Das 2013).

  21. 21.

    Again, see Blunch (2006) for more details.

  22. 22.

    As we will also discuss later, we do, however, still attempt to endogenize schooling, though unsuccessfully so: the proposed instruments turn out to be extremely weak, leading to huge standard errors. We, therefore, endogenize only network access and information exposure and resort to the predeterminedness argument for schooling.

  23. 23.

    At a minimum, if these factors are not included, one may systematically over- or underestimate the strength of the caste/religion-health knowledge relationship.

  24. 24.

    While there may be some concern about using the linear probability model (LPM) (for example, predicted probabilities may fall outside the (0,1)-range and heteroskedasticity also is present by default), it can be argued that the LPM is a more robust alternative to a logit or probit model and also approximates the response probability well, especially if (1) the main purpose is to estimate the partial effect of a given regressor on the response probability, averaged across the distribution of the other regressors, (2) most of the regressors are discrete and take on only a few values and/or (3) heteroskedasticity-robust standard errors are used in place of regular standard errors (Wooldridge 2010). All three factors seem to work in favor of the LPM for the purposes of the application here.

  25. 25.

    If measurement error were random, then the estimates would be biased towards zero—so-called “classical” measurement error. Since we have no evidence that measurement error may be non-random, we assume in the following that this is indeed the type that is relevant for this application.

  26. 26.

    Technically, individual i should be left out when constructing the district shares for the IV strategy. Since the number of individuals per district is quite large, however, whether or not individual i is omitted or not does not matter in practice: the correlation between the IV with and without individual i is almost identically one, so that the results for the two alternative specifications are virtually identical.

  27. 27.

    With the caveat that areas with more knowledge may also demand more medical staff, which would increase the likelihood of having an acquaintance/relative in the medical profession.

  28. 28.

    Sensitivity analysis (see Results section) reveals that choosing the (potentially (more) endogenous) village level instead of the district level yields even stronger results, in substantive as well as statistical terms.

  29. 29.

    Since human capital accumulation and skills acquisition arguably depend on the availability of educational institutions, as well as their quality, this has led researchers to follow two main IV strategies in recent years: either using as IVs (1) various combinations of time of year, birth cohort, and/or geographical area of birth dummies to capture variation in institutional factors relevant for human capital accumulations such as compulsory schooling laws or expansion of educational programs (Angrist and Krueger 1991; Duflo 2001) or (2) variables for proximity or exposure to educational institutions in the local area (Card 2001). Since we do not have available the geographical area of birth, we first explored the second of these approaches for the case of educational attainment. It turned out, however, that the instruments were quite weak, thus leading to the so-called weak instruments problem (Staiger and Stock 1997). Additionally, it can be argued that education is at least pre-determined, thus at least addressing the simultaneity-part of the potential endogeneity issue.

  30. 30.

    This allows for a larger estimation sample but since the IV-estimations will be under-identified in this case, only OLS-estimations are pursued for this case.

  31. 31.

    It here seems prudent to exclude TV watching from the observables used in the matching procedure due to the endogeneity concerns of that variable, so that is also the approach taken here. Sensitivity analysis reveals that this choice is not crucial to the obtained matching results.

  32. 32.

    For details on Mahalanobis matching, see for example Rosenbaum and Rubin (1985).

  33. 33.

    The complete set of results is available upon request.

  34. 34.

    As discussed in the Estimation Strategy and Related Issues section endogeneity concerns lead us to use identifying instruments calculated at the district level rather than the (preferable) village level (to allow for inter-village-intra-district migration). Sensitivity analyses using village level based instruments reveal even stronger results than what is reported below (not shown here but available upon request).

  35. 35.

    It should be noted that since the first-stage regression is exactly identified, Hansen’s (1982) J-test for over-identification is not relevant for this application.

  36. 36.

    As based on the results of the Wu–Hausman tests.

  37. 37.

    This is similar to the literature on the returns to education, which typically finds that IV estimates instrumenting for education are greater than the corresponding OLS estimates (e.g., the studies reviewed in Card 2001). Based on earlier studies, including Grilliches (1977), Card (2001) and the studies reviewed therein, the OLS bias could be either positive or negative. Ability bias would tend to generate positive OLS bias, while measurement error would tend to generate negative OLS bias. Due to these potentially opposing effects, it is therefore often difficult to sign the OLS bias a priori. Also, it is possible that supply type instruments will affect individuals who will otherwise have relatively low levels of education more than other individuals. The latter is also related to the so-called Local Average Treatment Effect (LATE) interpretation of IV, whereby IV may identify only the average returns of those who comply with the assignment-to-treatment mechanism implied by the instrument (Imbens and Angrist 1994). In turn, this may help explain why IV estimates sometimes are (seemingly implausibly) high.

  38. 38.

    Since in addition to one identifying instrument for (each of) the health network access main effect(s) we would now additionally need one identifying instrument for each of the caste/religion-health network access interactions, as well.

  39. 39.

    Full results not reported here but available upon request.

References

  1. Abadie, A., & Imbens, G. (2006). Large sample properties of matching estimators for average treatment effects. Econometrica, 74(1), 235–267.

    Article  Google Scholar 

  2. Altindag, D., Cannonier, C., & Mocan, N. (2011). The impact of education on health knowledge. Economics of Education Review, 30(5), 792–812.

  3. Angrist, J. D., & Krueger, A. B. (1991). Does Compulsory School Attendance Affect Schooling and Earnings?. Quarterly Journal of Economics, 106, 979–1014.

    Article  Google Scholar 

  4. Banerjee, R., Cohen-Cole, E., & Zanella, G. (2007). Demonstration effects in preventive care (Working paper no. QAU07-7). Mayo Clinic College of Medicine. Boston: The Boston Federal Reserve Bank.

  5. Borooah, V., & Iyer, S. (2005). Vidya, veda, and varna: the influence of religion and caste on education in rural India. The Journal of Development Studies, 41(8), 1369–1404.

    Article  Google Scholar 

  6. Blunch, N.-H. (2006). Skills, schooling and household well-being in Ghana, unpublished PhD dissertation. Washington, DC: The George Washington University.

    Google Scholar 

  7. Borooah, V. K. (2012). Inequality in health outcomes in india: the role of caste and religion. In Sukhadeo Thorat & Katherine S. Newman (Eds.), Blocked by caste: economic discrimination in modern India. New Delhi: Oxford University Press.

    Google Scholar 

  8. Borooah, V. K, Sabharwal, N. S., & Thora, S. (2012). Gender and caste-based inequality in health outcomes in india, (Working paper, vol. 6, no. 3). New Delhi: Indian Institute of Dalit Studies.

  9. Card, D. (2001). Estimating the return to schooling: progress on some persistent econometric problems. Econometrica, 69(5), 1127–1160.

    Article  Google Scholar 

  10. Chattopadhyay, R., & Duflo, E. (2014). Women as policy makers: evidence from a randomized policy experiment in India. Econometrica, 72(5), 1409–1443.

    Article  Google Scholar 

  11. Chin, A., & Prakash, N. (2011). The redistributive effects of political reservation for minorities: evidence from India. Journal of Development Economics, 96(2), 265–277.

    Article  Google Scholar 

  12. Christakis, N. A., & Fowler, J. H. (2007). The spread of obesity in a large social network over 32 years. New England Journal of Medicine, 357, 370–379.

    Article  Google Scholar 

  13. Cutler, D.M. & Glaeser, E.L. (2007). Social interactions and smoking (NBER working paper no. 13477). NBER.

  14. Das, S. (2012). Caste, ethnicity, and religion: linkages with unemployment and poverty. In Sukhadeo Thorat & Katherine S. Newman (Eds.), Blocked by caste: economic discrimination in modern India. New Delhi: Oxford University Press.

    Google Scholar 

  15. Das, S. (2013). Health culture of scheduled caste: a case study of patni in cachar district of Assam. International Research Journal of Social Sciences, 2(12), 35–41.

    Google Scholar 

  16. Desai, S., & Vanneman, R. National Council of Applied Economic Research, New Delhi, 2010). India human development survey (IHDS), 2005 [computer file]. ICPSR22626-v8. Ann Arbor, MI: Inter-University Consortium for Political and Social Research [distributor]. 10.3886/ICPSR22626.v8.

    Google Scholar 

  17. Duflo, E. (2001). Schooling and Labor Market Consequences of School Construction in Indonesia: Evidence from an Unusual Policy Experiment. American Economic Review, 91(4), 795–813.

  18. Dustmann, C., & Preston, I. (2001). Attitudes to ethnic minorities, ethnic context and location decisions. Economic Journal, 111(470), 353–373.

    Article  Google Scholar 

  19. Glewwe, P. (1999). Why does mother’s schooling raise child health in developing countries? Evidence from Morocco. Journal of Human Resources, 34(1), 124–159.

    Article  Google Scholar 

  20. Griliches, Z. (1977). Estimating the returns to schooling: some econometric problems. Econometrica, 45(1), 1–22.

    Article  Google Scholar 

  21. Grossman, M. (1972). On the concept of health capital and the demand for health. Journal of Political Economy, 80(2), 223–55.

    Article  Google Scholar 

  22. Gruber, J. H. (2005). Religious market structure, religious participation, and outcomes: is religion good for you? B.E. Journal of Economic Analysis and Policy: Advances in Economic Analysis and Policy, 5(1), 1–30.

    Google Scholar 

  23. Hansen, L. P. (1982). Large Sample Properties of Generalized Method of Moments Estimators. Econometrica, 50, 1029–1054.

  24. Huber, P. J. (1967). The behavior of maximum likelihood estimates under nonstandard conditions, In Lucien M. Le Cam and Jerzy Neyman (Eds.), Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability. Vol. 1, Berkeley, CA: University of California Press.

  25. Imbens, G. W. (2004). Nonparametric estimation of average treatment effects under exogeneity: a review. The Review of Economics and Statistics, 86(1), 4–29.

    Article  Google Scholar 

  26. Imbens, G., & Angrist, J. D. (1994). Identification and estimation of local average treatment effects. Econometrica, 62, 467–475.

    Article  Google Scholar 

  27. Jensen, P., & Rasmussen, A. W. (2011). The effect of immigrant concentration in schools on native and immigrant children’s reading and math skills. Economics of Education Review, 30(6), 1503–1515.

    Article  Google Scholar 

  28. Kenkel, D. S. (1991). Health behavior, health knowledge, and schooling. Journal of Political Economy, 99(2), 287–305.

    Article  Google Scholar 

  29. Kovsted, J., Pörtner, C. C., & Tarp, F. (2003). Child health and mortality: does health knowledge matter? Journal of African Economies, 11(4), 542–560.

    Article  Google Scholar 

  30. Patra, S., Perianayagam, A. & Goli, S. (2013). Mother’s health knowledge and practice and their linkage with childhood morbidity, medical care and medical care expenditure in India. In Paper Presented at the 27th IUSSP International Population Conference, Busan, Korea: BEXCO.

  31. Pradhan, R., & Meinzen-Dick, R. (2010). Which rights are right? Water rights, culture and underlying values. In PeterG. Brown & JeremyJ. Schmidt (Eds.), Water ethics: foundational readings for students and professionals. Washington D.C.: Island Press.

    Google Scholar 

  32. Ramachandran, V., & Naorem, T. (2013). What it means to be a dalit or tribal child in our schools? a synthesis of a six-state qualitative study. Economic and Political WEEKLY, XLVIII(44), 43–52. November 2, 2013.

    Google Scholar 

  33. Rao, N., Mobius, M. M. and Rosenblat, T. (2007). Social networks and vaccination decisions (Federal reserve band of Boston working paper no. 07–12).

  34. Rosenbaum, P., & Rubin, D. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70, 41–55.

    Article  Google Scholar 

  35. Rosenbaum, P., & Rubin, D. (1985). Constructing a control group using multivariate matched sampling methods that incorporate the propensity. American Statistician, 39, 33–38.

    Google Scholar 

  36. Staiger, D., & Stock, J. H. (1997). Instrumental variables regression with weak instruments. Econometrica, 65(3), 557–586.

  37. Subramanian, S. V., Nandy, S., Irving, M., Gordon, D., Lambert, H., & Smith, G. D. (2006). The mortality divide in India: the differential contributions of gender, caste, and standard of living across the life course. Am J Public Health, 96(5), 818–825.

    Article  Google Scholar 

  38. UNICEF. (2012). Committing to child survival—a promise renewed. progress report 2012. New York, NY: UNICEF.

    Google Scholar 

  39. USAID. (2010). Treating Childhood Diarrhea in India with ORT and Zinc: Engaging the Pharmaceutical Industry and Private Providers--Lessons Learnedfrom the POUZN/AED Project. Washington, DC: The United States Agency for International Development.

  40. White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48(4), 817–830.

    Article  Google Scholar 

  41. WHO. (2013). Diarrheal disease,” fact sheet N°330. Geneva, Switzerland: World Health Organization. http://www.who.int/mediacentre/factsheets/fs330/en/. Accessed 11 July 2014.

  42. Wooldridge, J. M. (2010). Econometric analysis of cross-section and panel data. 2nd edition Cambridge, Massachusetts: The MIT Press.

    Google Scholar 

Download references

Acknowledgements

We thank Keera Allendorf, Ritwik Banerjee, Christian Bjørnskov, Joyce Chen, Hope Corman, Tor Eriksson, Dylan Fitz, Art Goldsmith, Shoshana Grossbard, Joseph Guse, Tim Lubin, Pushkar Maitra, Martin Paldam, Howard Pickett, Claus Pörtner, Nishith Prakash, Samuel Raisanen, Gautam Rao, and David Ribar; and conference participants at the American Economic Association Annual Meetings, the Population Association of America Annual Meetings, the Economic Growth and Development Annual Conference, the European Society for Population Economics Annual Conference, the Public Choice Society Annual Meetings, the Danish Academic Economists in North America Annual Meetings, the IHDS User Conference, the Liberal Arts Colleges Development Conference, the Southern Economic Association Annual Meetings, and the Society of Economics of the Household Annual Meetings; and seminar participants at Aarhus University, Hokkaido University, Konkuk University, Korea University, the Melbourne Institute of Applied Economic & Social Research, Monash University, Seattle University, University of Southampton, and Washington and Lee University for helpful comments and suggestions. The helpful comments and suggestions from the Editors and two anonymous referees are gratefully acknowledged. Remaining errors and omissions are our own. This research was partially funded by Aarhus University Research Foundation and Danish Council for Independent Research|Social Sciences and Washington and Lee University’s Lenfest Summer Research Grant. The data were kindly provided by the Inter-university Consortium for Political and Social Research, Ann Arbor, MI, on behalf of Sonalde Desai, Reeve Vanneman, and the National Council of Applied Economic Research, New Delhi. The findings and interpretations are those of the authors and should not be attributed to any of the aforementioned individuals or organizations, however.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Niels-Hugo Blunch.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Appendix. Descriptive statistics

Appendix. Descriptive statistics

Table 7 Means and standard deviations for estimation samples

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Blunch, NH., Datta Gupta, N. Mothers’ health knowledge for children with diarrhea: who you are or who you know?. Rev Econ Household 18, 1131–1164 (2020). https://doi.org/10.1007/s11150-020-09478-y

Download citation

Keywords

  • Health knowledge
  • Diarrhea
  • Caste
  • Religion
  • Health networks
  • India

JEL classifications

  • I12
  • I14
  • I15