Why should this posting be reviewed?
See also Guidelines for Comments and Corrections.
Thank you for taking the time to flag this posting; we review flagged postings on a regular basis.close
Post Your Discussion Comment
Please follow our guidelines for comments and review our competing interests policy. Comments that do not conform to our guidelines will be promptly removed and the user account disabled. The following must be avoided:
- Remarks that could be interpreted as allegations of misconduct
- Unsupported assertions or statements
- Inflammatory or insulting language
Reader Comments (2)
Post a new comment on this article
A Case of Misleading Numbers?
Posted by plosmedicine on 28 Mar 2011 at 16:39 GMT
The following comment has been posted on behalf of Dr. Andrew Deonarine, PGY-3 Resident, Public Health and Preventative Medicine at The University of British Columbia:
In the important investigation by Subramanyam et. al, it was asserted that there is no significant relationship between economic growth and the rates of malnutrition in India from the years 1992-2005. One potential concern with this investigation is that not all economic growth is equal: growing at 2000 INR / year, or perhaps 200% in a poor state from 1992 – 2005 is not the same as growing at 2000 INR / year or 200% in a richer state in India. This is because the poorer state may have had an extremely low per capita income to begin with, and even with high growth rates, the resulting average income is still very low. In Bihar, for example, per capita income started at 4657 INR, grew 267% from 1992-2005, and correlated with a 6% reduction in malnutrition. In contrast, Maharashtra started with 12010 INR in 1992, grew a comparable 274% and reduced malnutrition by 14.71% over the same period. Similar arguments can be made concerning rates of change of per capita income. Since initial income level is important, quantifying economic growth using rates of change (percentage or INR/year) may not be appropriate, and the majority of the models developed by Subramanyam et. al, measured economic growth as a rate. This may be why most of these models did not show a relationship between economic growth and reductions in malnutrition. If one were to assume that achieving minimum wage is important in reducing malnutrition, and using a proposed minimum wage cutoff from the government of India (http://labour.nic.in/wage...) of 80 INR / day (or ~ 20000 INR / year), it can be seen that states with high growth and final per capita incomes below this proposed minimum wage level still have high levels of malnutrition in 2005. This type of economic growth can be defined as “inadequate” growth, because it is not adequate enough to reduce malnutrition, being below the minimum wage line. In contrast, states with “adequate” economic growth, which results in final per capita incomes well above the minimum wage line (irrespective of the economic growth measured by rate of change), experience comparatively high reductions in malnutrition. The correlation for both state level reductions in malnutrition and underweight with absolute per capita state domestic product for 1992-2005 is moderate (r = 0.42). This indicates, at least ecologically, that growth (quantified by absolute per capita levels, not rates of change) may have a significant effect in reducing malnutrition. In order to confirm the findings of Subramanyam et. al, it may be informative to divide economic growth into “adequate” and “inadequate” categories using minimum wage criteria (or a similar cutoff). Such an analysis may reveal the potentially powerful effects that economic growth has in reducing malnutrition.
Dr. Andrew Deonarine
PGY-3 Resident, Public Health and Preventative Medicine
The University of British Columbia
Nothing misleading about the lack of an association between economic growth and child undernutrition in India: A reply to Deonarine
plosmedicine replied to plosmedicine on 20 Apr 2011 at 11:52 GMT
This response has been posted on behalf of Malavika Subramanyam and S V Subramanian, two of the authors of the original Policy Forum:
We appreciate Deonarine’s interest in our paper, and thank him for his comment. At the outset, Deonarine’s comment necessitates an important correction whereby it is claimed that the analyses we presented in our above referenced paper are “misleading”. Our paper presented, in an extremely transparent manner, the findings from models which Deonarine seems to view as most relevant, i.e., the association of level of economic growth (see Models 9 -14 in Table 4 of our paper) with undernutrition. It is therefore, unfortunate, that Deonarine chose to incorrectly characterize our study as a case of “misleading numbers”.
That aside, the essence of Deonarine’s critique is that we did not consider the possibility that states might not have attained “adequate” levels of economic development that might then enable them to effectively leverage the anticipated benefits of economic growth to reduce child undernutrition. We found that the level of a state’s economic development was associated with decreased odds of being underweight only in one of six instances (see Models 9 – 14 in Table 4 of our paper), and even that association lost statistical significance when covariates were added to the model. Further, the fact that we observe a decrease in child undernutrition prevalence between 1992 and 1999 among many states even though all states would fall in the category of “inadequate” economic development in 1992, as per Deonarine’s criterion, seems to suggest that the assertion about “adequacy” is not matched by data on its face value. We nevertheless took this opportunity to formally test Deonarine’s “adequacy” hypothesis.
We divided states into “adequate” and” inadequate” development categories, using the cut-off suggested by Deonarine; a state with a Per Capita net State Domestic Product (PCSDP) less than 20,000 INR was classified as “inadequate” and the rest as having “adequate” development. We were able to do this only in data from 1999 and 2006 because all states in 1992 fell in the “inadequate” category. We then re-fit, separately for 1999 and 2006, Model 14 from Table 4 but added the indicator variable for inadequate development to this model. There was no statistically significant association between living in a state with “inadequate” economic development and the risk of child underweight. The mean difference in the log odds of reporting underweight associated with living in a state with inadequate development was -0.202 (SE = 0.175) in 1999 and 0.053 (SE = 0.151) in 2006. Additionally, the coefficient for level of economic growth remained statistically insignificant (Table 1; see link to table below). We additionally tested whether the association between economic growth and underweight varied depending on whether the state had inadequate versus adequate development, by including an interaction term of economic growth and the indicator for inadequate development. We did not find any support for this interaction (Table 1). To the extent one might wish to interpret statistically non-significant results, these additional analyses appears to suggest that at adequate levels of economic development increased economic growth is associated with an increase in the risk of child undernutrition!
In summary, the results of this additional analysis do not support Deonarine’s hypothesis that economic growth can reduce undernutrition only in the presence of certain level of adequate economic development. These results further strengthen our original conclusion that there appears to be no consistent association between state level economic growth and child undernutrition in India, and that there is nothing misleading about this. Indeed, the whole motivation behind our paper was to base the discussion of the role of economic growth and child undernutrition on empirical facts; an aspect sorely missing on this debate as exemplified by Deonarine’s inaccurate comment on our original paper.
Table 1: Mean differences in the log odds (SE) of reporting underweight associated with adequacy of state economic development in the Indian National Family Health Surveys of 1999 and 2006 from multilevel logistic models stratified by survey year.
Please use the following link to open Table 1 (PDF): http://www.plosmedicine.o...
1. Subramanyam MA, Kawachi I, Berkman LF, Subramanian SV (2011) Is economic growth associated with reduction in child undernutrition in India? PLoS Med 8: e1000424.
Authors of this comment: Malavika Subramanyam, S V Subramanian
Author, Degrees and Affiliation:
- Malavika A. Subramanyam, MD, ScD, Research Fellow, Center for Integrative Approaches to Health Disparities, School of Public Health, University of Michigan
- S V Subramanian, PhD, Associate Professor, Department of Society, Human Development and Health, Harvard School of Public Health
Corresponding Author: S V Subramanian, Email: firstname.lastname@example.org
Conflicts of Interest: None