The National Family Health Surveys (NFHS), brought out by the International Institute for Population Sciences (IIPS), provide some of the most critical information on demographic, health, nutrition and socio-economic status of people in the country. The surveys draw on the experience of experts from national and international organisations, besides that of the Government of India. Apart from the expertise at its command, NFHS’s strength lies in technical innovation in data collection — it relies on high-end organisational machinery and technically trained personnel to conduct interviews and supervise fieldwork. NFHS uses a bio-marker questionnaire in which entries are recorded after actual clinical, anthropometric and biochemical testing. There is no other comparable survey that provides such a vast range of information and the NFHS results are important not just for monitoring the health and family welfare schemes, including the SDGs, but also in situating the country’s development globally.
The recent publication of the key results of NFHS-5, covering the period 2019-21, has evoked the attention of researchers, government agencies and the media. This was understandable considering the range of topics covered in the summary factsheets now released and the perceived drought of official data in recent times. The pandemic-forced restrictions have also disturbed several data collection efforts, including the decennial census.
The NFHS was conducted in two phases, one before the pandemic and the other from January 2020 to April 2021, when the pandemic was active in one state or the other. We do not have any information on how the surveyors overcame the problems related to travel and establishing personal contact. Are there systematic differences in the data collection practice between the two phases?
One result has evoked particular attention, overshadowing most other indicators: The sex ratio that shows there are 1,020 females (985 in rural and 1,037 in urban) for 1,000 males in the country. This ratio was 991 in the NFHS-4 and 1,000 in NFHS-3. The 2011 Census recorded the country’s sex ratio as 940 — much lower than the NFHS. Clearly the latter has been reporting a higher sex ratio for the population even in the past. Interestingly, the sex ratio for Kerala in the 2011 Census was the highest at 1,084 while NFHS-5 has estimated it to be 1,121.
Such variations cannot be explained in terms of sampling error margins. NFHS was clearly not designed to estimate a key population characteristic like sex ratio or even the total population of the country that is necessary to estimate the national sex ratio. The survey concentrates on collecting data from specific age groups of men, women and children.
Unfortunately, the organisers of NFHS are not very forthcoming in explaining how households are selected, unlike the NSS, where even instructions to field staff are in the public domain. The NSS definition of a household includes single members living in institutions and theoretically excludes only a section of armed forces personnel and jail inmates. The household mapping for selecting sample households in NFHS possibly excludes those living in hostels, workers’ camps or places that predominantly house men, resulting in a high sex ratio.
In NFHS-4, 14.6 per cent of the households were headed by females (14 per cent in NFHS-3). In the latest round, 24 per cent of households in Kerala were headed by women. For most states, these figures are high (Bihar 23 per cent, Maharashtra 15 per cent, West Bengal 16 per cent). In 2011-12, the NSS found only 12 per cent of Indian households were headed by women. This could mean that male members who stay away from home for work are not possibly included in the NFHS’s household count.
Questions are bound to come up when the survey agencies publish just fact sheets with no metadata. The statistical validity of the indicators needs to be supported by the appropriateness of the survey design. In the present case, NFHS could have avoided the unnecessary media diversion had they focused on other key findings. It also highlights the risk in rushing to announce the results without putting up background information in the public domain. One may, however, understand this in the context of what Tiger Woods once remarked, “Sensationalism sells: Don’t let the facts get in the way of a good story”.
We wish the IIPS had issued a caveat, at least in a footnote, that its sex ratio figures are unlikely to be valid for the total population. Such an explanation is necessary given that a large number of researchers and policymakers hold its publications in high esteem. A news story may die but some messages and images remain in popular mind. The impression given by NFHS-5, of women outnumbering men, could adversely affect ongoing programmes against sex selective abortions, projects that try to remove the neglect of women in education and health and initiatives that try to remove discrimination in access to property rights — this could have been avoided.
The real good news from the NFHS is a secular increase in the sex ratio at birth (SRB) for children born in the last five years (below six years of age, SRB-5), where household mapping would not create a major problem. Importantly, SRB-5 takes into consideration children who die before they attain the age of six. The number of girls per thousand boys has gone up — from 919 in 2015-16 to 929 in 2019-21. An important point is a significant rise in rural areas against a marginal increase in towns and cities. The latter can possibly be attributed to the availability of medical facilities in urban areas which, though reduces maternal mortality and death rate for girl children, increases gender selective termination of pregnancies. The other interesting point is that the sex ratio of children up to the age of six in NFHS-3, for the year 2005-06, is much lower than that of the Population Census of 2001 – this is opposite a similar comparison between the Census and NFHS-5. Furthermore, while the child sex ratio has gone down from 927 to 919 during 2001-11 as per the Census, SRB-5 shows a continuous rise. Finally, the high increase in SRB-5 in rural areas in NFHS-5, noted above, is due to a significant fall in this ratio between the period of the third and fourth NFHS. The NFHS-4 figure is, thus, out of the trend and hence needs to be investigated. It is difficult to resolve these anomalies, unless we have access to metadata.
This column first appeared in the print edition on December 7, 2021 under the title ‘When numbers hide’. Kundu is research advisor, Oxfam India. Mohanan is former member National Statistical Commission
The recent publication of the key results of NFHS-5, covering the period 2019-21, has evoked the attention of researchers, government agencies and the media. 
Share
