ArticlePDF Available

'Provincialising' Vegetarianism Putting Indian Food Habits in Their Place


Abstract and Figures

Large-scale survey data are used to question the most public claims about food habits in India. It is found that the extent of overall vegetarianism is much less—and the extent of overall beef-eating much more—than suggested by common claims and stereotypes. The generalised characterisations of “India” are deepened by showing the immense variation of food habits across scale, space, group, class, and gender. Additionally, it is argued that the existence of considerable intra-group variation in almost every social group (caste, religious) makes essentialised group identities based on food practices deeply problematic. Finally, in a social climate where claims about food practices rationalise violence, cultural–political pressures shape reported and actual food habits. Indian food habits do not fit into neatly identifiable boxes.
Content may be subject to copyright.
march 3, 2018 vol lIiI no 9 EPW E conomic & Political Weekly
‘Provincialising’ Vegetarianism
Putting Indian Food Habits in Their Place
Balmur li Natrajan, Sur aj Jacob
All views expressed are the authors’ own and should not be attributed to
these institutions. The authors wish to t hank the anonymous reviewer
for their comments.
Balmurli Natrajan ( is an anthropologist at
William Paterson University of New Jersey, United States. Suraj Jacob
( is a political economist at Vidya Bhawan,
Udaipur. Both are also visiting faculty, Azim Premji University,
Large-scale survey data are used to question the most
pub lic clai ms abou t food habit s in India. I t is fo und that
the extent of overall vegetarianism is much less—and
the extent of overall beef-eating much more—than
suggested by common claims and stereotypes. The
generalised characterisations of “India” are deepened by
showing the immense variation of food habits across
scale, space, group, class, and gender. Additionally, it is
argued that the existence of considerable intra-group
variation in almost every social group (caste, religious)
makes essentialised group identities based on food
practices deeply problematic. Finally, in a social climate
where claims about foo d practices rationalise violence,
cultural–political pressures shape reported and actual
food habits. Indian food habits do not fit into neatly
identifiable boxes.
In the continuing saga to craft a national policy that satisfi es
the current regime’s urge to control what people eat while
not running afoul of the laws of the land, it is useful to take
a step back and think about a right acknowledged by our courts
and yet not substantively invoked in policy or political decisions.
In a signifi cant judgment in May 2016, the Bombay High Court
overturned an amendment (passed in 1995, receiving presi-
dential assent in 2015) to the Maharashtra A nimal Preservation
Act of 1976. That amendment banned the possession of beef
even from outside the state. In overturning Section 5D of the
amendment and lifting the ban, the judges made it clear that
if the State tells the citizens not to eat a particular t ype of food or
prevents the citizens from possessing and consuming a particular
type of food, it will certainly be an infringement of a right to privacy
as it violates the right to be let alone. (Shaik Zahid Mukhtar v State of
Maharashtra & Ors 2016, emphasis ours)
This paper attempts to see what people in India eat when “let
alone” and what empirical facts such as intra- and inter-group
variation do to claims about group food habits.1
Although the above ruling implicitly has the individual as
the bearer of the “right to be let alone,” “individual choice,” in-
cluding shaping of desires, preferences, beliefs and intentions,
exists uneasily with social pressures to conform. The right to
be let alone is, therefore, scarcely available to individuals who
are routinely subjected to the power and hegemony of a group
or community “culture” and increasingly a “national” culture
th at bar ely sp eak s to their own e xpe rie nce s. Suc h hegemo ny is
sustained within society through the power of the media,
community associations, and self-styled culture police such as
vigilante gau rakshaks aided by legislators making the laws of
the land: all of who regularly make public claims and repre-
sentations about food practices (for example, valorisation of
vegetarianism, and stigmatisation and criminalisation of
beef-eating), seek social affi rmation for their claims, demand
conformity from others, and impose the same upon all with
force or the threat of it. In this context, the right to be “let
alone” needs to be viewed as a function of power and social
position, part of the fears, threats and attempts at hegemony.
Such a situation of not being “let alone” produces a wide range
of responses from the long-suffering individual, ranging from
(reluctant) acceptance of hegemony, to surreptitious transgres sion
of norms, and open resista nce to domi nation. Given the si mul-
taneous hegemony of vegetarianism and stigmatisation of beef, a
cautionary note is needed when fi guring out what India eats: any
self-reported data on food habits are likely overestimations of
Economic & Political Weekly EPW march 3, 2018 vol lIiI no 9 55
vegetarianism and underestimation of beef (and probably meat)
in the diet. Indeed, the widely used and peculiarly Indian term
“non-vegetarian” itself attests to the historical hegemony of
vegetarianism in India, a status that is belied by the facts on the
ground about its prevalence, thereby increasing the use of force
to maintain the hegemony. Hence, we prefer to use the term
“meat-eaters” to refer to those populations who are usually re-
ferred to as non-vegetarians. However, we have retained the latter
term as well since it is an offi cial category in a survey. It is neces-
sary to acknowledge that the category vegetarian is actualised in
everyday practices only through the explicit avoidance of meat,
whereas the category non-vegetarian does not depend on any
such proscription. Consequently, it is more logical to view these
categories as “meat-avoiding” and “meat-eating,” respectively.
This paper presents descriptive data on food habits from
large surveys such as the National Sample Survey Offi ce
(NSSO) as well as the interpretive context for such data. Its main
aims are (i) to establish a baseline of compelling evidence for
evaluating claims about food habits of individuals and practices
of groups in the light of the “beef bans” and resulting atrocities;
and (ii) to raise the intellectual level of public discourse around
beef-eating by complicating some of the key claims around
vegetarianism, meat-eating, and beef-eating in particular.
Thus, we ask: how widespread is the vegetarian diet in India,
and among particular religious and caste groups? Relatedly,
how widespread is meat-eating in India, and especially among
Hindus and across caste groups? And fi nally, can we estimate
the extent of beef-eating in India?
Each of these questions brings into relief the importance of
attending to variation, both within and across social groups,
regions, and gender. Indeed, we fi nd variation to hold the key
to any explanations of social phenomena such as food habits,
and to negotiate the cultural politics around food in India
today. We are, consequently, interested in asking: What does
in-group variation do to the frequent claims about the cultural
practices of social groups? What does regional variation mean
for claims about “national” food practices?
We begin with a synoptic representation of our key fi ndings
which, in our view, seriously question many public claims
about food habits.
(i) The extent of overall vegetarianism is much less than com-
mon claims and stereotypes suggest (no more than 30% and
more realistically closer to 20% of the population).
(ii) The extent of overall beef-eating is much more than com-
mon claims and stereotypes suggest (at least about 7% but
more realistically closer to 15% of the population).
(iii) There exists considerable variation of food habits across
scale, region, group, class, gender; each complicating general-
ised characterisations of India based on meaningless averages.
(iv) The considerable spatial variations within social groups
ensure that almost no group-specifi c claims about food prac-
tices can really pass muster.
(v) The signifi cant gender gap w ithin social groups and regions
makes claims of group and regional “traditions” problematic.
(vi) There is evidence of cultural–political pressures affecting
reported and actual food habits, so that any reported data
need to account for the bias of under-reporting of meat and
beef and over-reporting of vegetarian diet (hence, the need to
provincialise vegetarianism).2
Wha t Do t he Da ta Say ?
Three large-scale surveys are available as potential sources of
estimates of vegetarianism: the NSSO, the National Family
Health Survey (NFHS), and the India Human Development
Survey (IHDS). They are all based on multistage stratifi ed
designs with random household selection.
The NSSO survey is conducted by the Government of India’s
Ministry of Statistics and Programme Implementation, and is
generally considered a high-quality data source. The latest
NSSO data for consumption is the 68th round of the survey
conducted during 2011–12 and it consists of a sample of 1,01,651
households in 7,469 villages and 5,268 urban blocks (NSSO
2013). The survey asks detailed questions regarding consump-
tion of a very wide range of items, and among the three data
sources, it alone asks separate questions regarding speci c
types of non-vegetarian items (eggs; fi sh and prawn; goat meat
and mutton; beef and buffalo meat; pork; chicken; and other,
such as birds or crab). For these and other food items, respond-
ents were asked about quantity of consumption of items in the
last seven or 30 days (separ ately) of t he su r vey.
The NFHS is analogous to the Demographic and Health
Surveys (DHS) conducted in over a hundred countries (IIPS
and Macro International 2007). The NFHS data from the third
round (2005–06), consists of separate large samples of wom-
en aged 15–49 years and men aged 15–54 years (1,24,385 and
74,369 observations, respectively). For food consumption, the
survey focuses on specifi c items—milk or curd; pulses or
beans; dark green leafy vegetables; fruits; eggs; fi sh; and
chicken or meat—and asks respondents about how often the
item was consumed, with four possible options (daily, weekly,
occasionally, or never).
The IHDS was conducted jointly by India’s National Council
of Applied Economic Research and the University of Maryland.
The sample consists of 27,010 rural and 13,126 urban house-
holds spanning 382 of 612 districts in 2001 in all states. Note
th at the sample s ize is less t han half of the NSSO survey, besides
covering a smaller range of consumption items (47 for IHDS
compared to 400 for NSS; see Dang and Lanjouw 2015). Although
the focus is broadly on education, health, and community-
related indicators (Desai et al 2010: 12), the second round of
IHDS, conducted in 2011–12, also asks the question: “Does anyone
in your household eat non-vegetarian food?” Note, however,
that the IHDS statewise estimates, being from relatively smaller
samples, have less robust validity (Drèze and K hera 2015).
Below, we present estimates of meat consumption and veg-
etarianism from each of these three sur veys. All estimates are
generated after accounting for relevant sampling weights and
household sizes.
Table 1 (p 56) presents estimates of vegetarianism from the
three surveys. Given the speci c questions related to consump-
tion asked in each of the surveys, the estimates are not readily
comparable. Nevertheless, we note that none of the estimates
march 3, 2018 vol lIiI no 9 EPW E conomic & Political Weekly
is substantially above a third, and that the NFHS and IHDS
estimates place this fi gure at a little less than a quarter of
the population.3 Further, these estimates are likely to be over-
estimates to the extent that some households may be reluctant
to report meat-eating to a surveyor—especially those from
castes or groups that may feel pressures to mask meat-eating.
More over, in the case of the NSSO, the estimate is only for those
who were vegetarian for the 30 days prior to the survey;4 it is
likely that several meat-eaters, especially among poorer
households, may have reported being vegetarian in the last 30
days, in which case, the NSSO estimate for vegetarianism in
Table 1 is almost certainly an overcount. Such estimates are
also echoed in a popularly cited survey, the Hindu–CNNIBN
State of the Nation Survey, which concludes that “the popular
image of vegetarian India is off the mark, as only 31 percent of
Indians and 21 percent of families (with all the members) were
found vegetarian” (Yadav and Kumar 2006).
Wherever possible, we present estimates for vegetarianism
defi ned as absence of meat, fi sh, and eggs. We note here that the
patterns we report, and underlying arguments, will also hold
if vegetarianism were defi ned as the absence of meat and fi sh
(ignoring eggs). Figure 1 which plots statewise NSSO estimates
for both defi nitions shows that both estimates are fairly close.
The dashed line is the line of equality, and by defi nition all data
points are above it; the dotted line is the linear regression
line. The two lines are fairly close and somewhat parallel,
suggesting that there is little difference in the distributions of
the two de nitions for vegetarianism.
Regional variation: Table 2a shows estimates based on a
dichotomous location classifi cation (rural/urban). Interestingly,
in all three surveys, there is little substantive difference in
rural and urban locations. However, further parsing out for
urban areas shows a more variegated picture. Speci cally, as
Tab le 2 b sh ows , me ga cit ies h ave l owe r inc idenc e; i nde ed, IHDS
estimates the incidence in the six largest metros to be only a
little more than half of the overall incidence in rural or other
urban areas. Any explanation
for this would need to include
the fact of working-class mi-
grations from different parts
of India and across castes and
religions to mega cities. Com-
plicating this picture, perhaps
counter-intuitively, if the mega
cities are kept apart, the NFHS
estimates (Table 2b) show that
vegetarianism increases with
urban size, from small towns
to large towns to small cities to
large cities. While we cannot
say what may be driving this correlation, it does serve to estab-
lish the considerable variation in incidence of vegetarianism, in
this case for urban scale.
Figure 2 presents an even more striking spatial variation across
the country regarding incidence of vegetarianism. Note that
although there are some differences across the three surveys,
with NSSO typically producing higher estimates of vegetarian-
ism than NFHS or IHDS, the difference in estimates for most
states is fairly similar—as shown by the fact that the dotted
quadratic fi t line is somewhat parallel to the dashed line of
equality—suggesting that differences in defi nition (described
previously) are the likely explanation of these differences,
further adding to the validity of the statewise variation (if not
the absolute numbers) produced by each data set. For the
remainder of this subsection, we focus on the NSSO estimates.
The substantive variations across states are nothing short
of stupendous. In the same fi gure, six states have less than 2%
Table 1: Incidenc e of Vegetarian ism in India
NSSO (2011–12) NFHS (2005– 06) IHDS (2011–12)
Vegetariani sm (%) 36.88 24.72 23.48
Obs er va ti ons 1,01, 651 1,98,5 85 41,991
For NSSO, esti mates are for th ose who did not ea t fish, meat or e ggs in the 30 days p rior
to the surv ey; for NFHS, es timates are fo r those who ans wer “never” to th e question o f
frequen cy of eating f ish, meat or egg s; for IHDS, tabl e gives estim ate of those who a nswer
“no” to the ques tion of having a t least one “non- vegetaria n member in the ho usehold.”
The NFHS est imate is the aver age of separa te estimates f or women and men . All the above
notes hold f or estimate s in all subsequ ent tables.
Table 2a: Incid ence of Vegeta rianism
by Location (%)
(2011–12) (2005–0 6) (2011–12)
Rural 36.82 25.30 23.91
Urban 37.04 23.64 22.56
Table 2b: Incid ence of Vegeta rianism
by City Type (%)
Mega cities 19.83
Large cities 29.05
Small cities 26.86
Large towns 22.59
Small towns 18.11
Rural 25.30
Figure 1: Statewi se Estimates o f Vegetarianis m with and with out Includi ng
Eggs (NSSO)
No meat and fish
0 20 4 0 60 80
No meat, f ish and eggs
The dashed line is the line of equality and the dotted line is the linear regression line.
Figure 2: Incidence of Vegetarianism, Comparing Statewise Estimates of
NFHS (2005–06)
0 20 40 60 80
NSSO (2011–12)
IHDS (2011–12)
0 20 4 0 60 80
NSSO (2011–12)
In ea ch gr aph , the d ash ed li ne is the l ine o f equ alit y an d the dot ted l ine i s the q uad rati c
regressi on line. Names ar e abbreviate d in the case of th e 17 “major states” ide ntified b y the
NSSO based on population.
Economic & Political Weekly EPW m arch 3, 2018 vol lI iI no 9 57
incidence of vegetarianism; these are all north-eastern states.5
Even among the 17 major states identi ed by the NSSO based
on population, there are three with incidence less than 5%
(Assam, West Bengal, and Kerala). In sharp contrast, at the
other end of the spectrum, three states have incidence of over
75% (Haryana, Rajasthan and Punjab). In fact, only seven of
the 17 major states have incidence over 50%, and six have less
than 20% incidence of vegetarianism. The interstate variations
form a distinct regional pattern, as is evident from the map in
Figure 3. States in the West and North have relatively higher
incidence, while states in the East and South have relatively
lower incidence. This overturns the stereotype of Chennai’s
“South Indian vegetarian meal” and Delhi’s kebabs and Punjabi
chicken tikka. Such stereotypes may be more a function of the
discursive making of national and regional cuisines, refl ecting
the hegemony of particular classes and castes of social actors
in this production. Interestingly, a recent qualitative study
notes that “[o]verall, however, it is clear that vegetarianism is
the exception and not the rule, even in non-coastal states like
Uttar Pradesh and Madhya Pradesh, where non-vegetarianism
means the eating of goat and sheep meat” (Robbins 1999).6 At
this point, given the paucity of historical data, we are only
able to speculate that this regional pattern could be due to a
complex of factors including agroecological availability of
foods, cultural politics related to locally dominant social
groups (castes, religions), and gendered differentiation in
food habits (more on which below).
Religious and caste group variation: We now turn to varia-
tions by social groupings categorised by religion and caste.
Table 3a focuses on religious categories. There are differences
between NSSO and NFHS estimates along lines that have al-
ready been discussed, but the broad pattern of inter-group
variation is similar in both sets of estimates. Apart from Jains
(overwhelmingly vegetarian) and Sikhs (majorit y v eg et aria n),
no other religious category is majority vegetarian. Hindus—by far
the largest group in the population—are majority meat-eaters
with a little over two-fi fths being vegetarian in the NSSO estimate
and less than a third in the IHDS estimate. Christians and Mus-
lims are overwhelmingly meat-eating populations. Interestingly,
some earlier small-scale ethnographic studies too pointed in the
above direction. In a pioneering paper, A K Chakravarti (1974:
403) suggested that “approximately 65 percent of the Hindus
in India maybe assumed to be non-vegetarians.”
The interesting question is: how do patterns of food habits
form within social groups? Here, quantitative studies are limited
in their ability to uncover the social mechanisms that produce
group-level patterns. We need rich ethnographic accounts that
explore the notion of “community” and the ways in which cul-
tural identity produces norms (for example, food practices)
which are ideologically rationalised and institutionally enforced
to produce distinctive group-level practices. For example, in
the paradigmatic case of Jains who show a remarkable homo-
geneity of food practices, it is very likely that religious concepts
(such as ahimsa), ideology (dogma, doctrine) and precepts, and
an ecclesiastical structure and institutions such as endogamous
marital practices, organise the socio cultural lives of members
and play a crucial role in shaping the food practices inter-
generationally as meat-avoiding. For other groups, it is not as
clear. As argued earlier, since the category of non-vegetarian
does not depend on proscription (in general), the fact of remark-
able homogeneity of meat-eating within Muslim and Christian
populations is not a puzzle. What could be researched, however,
is how and to what extent, Muslims, like Jains, may have
evolved a social capacit y to produce a pattern of avoidance of
taboo foods (for example, pork). Suc h data do not exist as far as
we know. On the other hand, it is not clear whether there exists a
pattern at the group level for the
Hindu populations which show
remarkable heterogeneity with
respect to food practices. The Sikh
case is puzzling, since Sikhism
does not have injunctions against
meat-eating (Guru Nanak having
explicitly rejected vegetarianism)
in shaping food practices.
Figure 3: Inc idence of Vegeta rianism, by State (N SSO Data)
The choro pleth map show s incidence of ve getariani sm in states in ra nges of 10 percent age
points between 0% and 90%.
Table 3a: Incide nce of
Vegetarianism by Religious
Categor ies (NSSO and NFHS) (%)
Hindu 41.88 28.49
Muslim 6.73 1.83
Christian 6.71 0.86
Sikh 79.39 52.96
Buddhist 21.82 6.96
Jain 98.23 94.87
Table 3b: Incide nce of Vegetaria nism by Mega- caste Categor ies (%)
All Hindu
(1) (2) (3) (4)
SC 31.44 15.72 30.70 15.76
ST 27.94 14.89 31.03 16.97
OBC 38.21 27.86 44.24 31.47
Non-SC/ST/OBC 41.35 30.96 51.73 37.67
Table 3c: Incid ence of Vegetar ianism by Soci al Categorie s (IHDS) (%)
Brahmin Fo rward Caste OBC SC ST Muslim Christia n
65.86 32.39 30.51 13.37 8.07 1.02 28.25
march 3, 2018 vol lIiI no 9 EPW E conomic & Political Weekly
To explore further, Table 3b focuses on mega-caste categories.
As before, despite differences in the NSSO and NFHS estimates,
the broad pattern of inter-group variation is similar. In the
overall NSSO estimate (column (1)), for the four categories,
incidence is least among Scheduled Tribes (STs) but closely
followed by Scheduled Castes (SCs), and it is higher among
Other Backward Classes (OBCs) and highest in the residual
category (non-SCs/STs/OBCs). The overall NFHS estimate
(column (2)) gives the same rank ordering among categories.
Further, while the differences among mega-caste groups are
statistically signifi cant,7 the size of the largest gap (namely,
gap between STs and the residual category) is only 13.4 to 16.1
percentage points (NSSO and NFHS, respectively): far smaller
than the differences among states or among religious categories.
Mega-caste Categories
However, the mega-caste categories are cross-cutting religious
categories. To see whether the pattern between STs, SCs, OBCs,
and others holds for Hindus in particular, columns (3) and (4)
of Table 3b present corresponding estimates. The inter-group
pattern is largely similar to the overall estimates (columns (1)
and (2)) except that incidence among SCs is now a little lower
than STs (as opposed to a little higher as before).8
The IHDS is the only survey which presents estimates sepa-
rately for “Brahmin” and “Forward Caste” categories.9 Tab le 3c
shows that only two-thirds of Brahmins are vegetarian: much
lower than stereotypes would have it, although expectedly
hig her t ha n ot her g roup s in t he tab le. Again, suc h a chara cter i-
sation is corroborated by ethnographic studies that have
documented the existence of meat-eating among Brahmin
communities. Apart from the commonly known meat-eating
and fi sh-eating practices of Kashmiri, Bengali and Konkani
Brahmins, an early hint of the variations within Brahmins is
Khare’s (1966) study of meat-eating by one gotra of Kanya–
Kubja Brahmins in Uttar Pradesh, a group that is generally
thought to be strict vegetarians. Another study documents
that it is very common for Brahmins in Garhwal to regularly
consume meat (Joshi et al 1994).
Further, only one-third of forward castes are vegetarian, a
gure that is not very different for the OBCs. One inference
from the above is that the ideological weight of vegetarianism
is sustained largely by Brahmins, rather than the category
forward castes where a majority are meat-eaters suggesting
that the category vegetarian is intimately shaped by caste and
Brahminism. On the other hand, the IHDS estimates of
vegetarianism for SCs and STs are lower than even the NFHS
estimates. Again, while the estimates for Muslims is low and
comparable with NFHS, the estimates for Christians is intrigu-
ingly much higher than in NSSO or NFHS, and in fact not much
lower than that for forward castes and OBCs in IHDS.
Finally, we note that there is considerable spatial variation
in incidence of vegetarianism even within a specifi c social
group. Figure 4 provides statewise incidence of vegetarianism
(NSSO estimates) for 17 major states identifi ed by the NSSO for SCs
(vertical axis) and Hindu non-SCs/STs/OBCs (horizontal axis).
In the case of SCs, although 31% are vegetarian according to
NSSO estimates (Table 3b, column 1), Figure 3 shows that four
of 17 m aj or sta te s have incidence of less tha n 5% (West Be ngal,
Assam, Andhra Pradesh and Kerala) and nine out of 17 have
less than 20%, whereas three states have incidence of over
70% vegetarianism (Punjab, Haryana and Rajasthan). This
spatial variation within the SC category is indeed striking.
The range is exemplifi ed by West Bengal and Punjab, both of
which have large SC populations (24% and 32% of total popu-
lation, respectively, according to Census 2011), and yet their
SCs are at opposite ends of the spectrum of vegetarianism
(1% and 74%, respectively). Similar is the case with Hindu
non-SCs/STs/OBCs. Although 41% are vegetarian according to
NSSO estimates (Table 3b, column 1), Figure 3 shows that two
of 17 major states have incidence of less than 5% (Assam and
West Bengal) while two have incidence of over as high as 90%
(Rajasthan and Haryana).
In Figure 4, the generally positive association across states
between vegetarianism among the SCs and forward castes
(that is, Hindu non-SCs/STs/OBCs) suggests that reported vege-
tar ia ni sm cou ld be shap ed by region al as much as caste factors.
Given the signi cant gap between SCs and OBCs on the one
hand, and the far narrower gap between OBCs and forward
castes on the other (Tables 3b and 3c), one could hypothesise
the continuing, yet differential hold of vegetarianism as a
culturally articulated form of ideological power in inter-caste
relations shaping OBC behaviour far more persistently than SC
behaviour. However, this factor will depend upon the degree
of dominant status and power of forward castes in a state to
impose food norms. Wherever the latter group dominates
sociopolitically, the incidence of OBC (and to a lesser extent,
SC) vegetarianism shows marked increases. This is explored in
Figure 5 (p 59). The left graph plots, for major states, the dif-
ference in vegetarianism between Hindu “forward castes” and
Hindu OBCs, against OBC share of population.10 There is a clear
positive association, implying that as OBC share increases across
states, the vegetarianism gap with Hindu “forward castes”
also increases: indicating ideological “breaking free” by OBCs.
By contrast, there is no clear relationship in the right graph of
Figure 5, which is the analogous plot for SCs. The absence of a
Figure 4: Spatial Variation of Vegetarianism within Specific Mega-caste
Categories (NSSO Data)
Vegetarianism among SCs
0 20 40 60 80 100
Vegetarianism among Hindu non-SC/ST/OBCs
Economic & Political Weekly EPW m arch 3, 2018 vol lI iI no 9 59
positive association for SCs in Figure 5, combined with the
fact of a consistent gap across states between SC and forward
caste vegetarianism (Tables 3b and 3c) suggests that vegetari-
anism in SCs as a group depends on other factors, possibly the
vibrancy of Dalit movements in states and the prominence (or
lack thereof) of food habits as symbolic elements within the
formation of identity.
Consumption and wealth: Does incidence of vegetarianism
vary by economic level of households? Table 4a suggests that
vegetarian households have higher consumption and income
relative to non-vegetarian households, in the NSSO and IHDS
data.11 This is consistent with incidence by economic status
categories reported in Table 4b: the incidence of vegetarianism
increases with higher economic class status. Such fi ndings
reso nate wit h those of Va idya natha n and Nair (198 0: 381) who
had argued long ago that “the variation in total animal protein
intake is closely related to variation in per capita real income
… and the relative costs of different protein sources.” Although
they do not trace the actual social processes through which the
availability of cheap protein translates into buying and
consuming particular meats (including beef), such correlations
point to the plausibility that “cultural” factors (such as religion
or caste identity)even if they were assumed to be impor-
tant—are always part of a larger set of factors that impact
human food preferences.
Gender and intersectionality: Separate estimates of incidence
of vegetarianism for gender are available only from NFHS.
Table 5a presents the basic estimate: incidence of vegetarianism
is higher among women than men, and the gap is substantial
(almost 10 percentage points, with incidence among women
almost 50% higher than incidence among men). Further, the size
of the gender gap is similar
between rural and urban
areas, and across different
city types. Figure 6a presents
incidence for women and men
across states: while there is a
very strong positive correla-
tion across states, the dotted
quadratic fi t line (which is
fairly linear) pivots down from
the dashed line of equality,
Figure 5: Veget arianism amo ng OBCs and SCs i n Relation to Popu lation
Shares (NSSO Data)
Gap betw een Hindu non -SC/ST/OBCs and H indu OBCs
0 20 4 0 60 80
Pop ulat ion o f OBC s (% o f tot al)
gap betw een Hindu non-S C/ST/OBCs and H indu SCs
0 20 4 0 60 80
Population of SCs (% of total)
Population data from: Indian Human Development Report, 2011—Towards Social Inclusion.
The graph s show the quadra tic line of bes t fit and the s haded 95% conf idence inter val.
Figure 6a: I ncidence of Vege tarianism , by Gender and Sta te (NFHS)
0 20 40 60 80
The dashed l ine is the line o f equalit y and the dot ted line is the q uadratic re gression li ne.
Figure 6b: In cidence of Ill iteracy by Gen der and State (Censu s 2011)
men (illiter acy rate)
20 30 40 5 0 6 0
Women (illi teracy rat e)
The dashed l ine is the line o f equalit y and the dot ted line is the q uadratic re gression li ne.
Table 4a: Inci dence of Vegetar ianism, by Hou sehold Expe nditure an d
Encome, NFH S and IHDS (`)
Vegetari an 1,715.34 1,678 .76
Non -v egeta ri an 1,575 .6 0 1,552 .45
hh Expendi ture/ hh Income/ Tot hh C onsumptio n Tot hh Income
Capita Capita Expenditure
Vegetarian 29,365 30,725 1,25,736 1,31,699
Non-vegetarian 24,566 25,191 1,04,152 1,07,624
Table 4b: Incid ence of Vegetari anism, by Econom ic Status,
NFHS and IHDS (%)
Poorest Poorer Middle Richer Richest
NFHS 24.56 28.05 27.855 29.44 40.285
Poor Middle Class Comfortable
IHDS 19.23 26.71 30.49
MPCE stan ds for househo ld monthly pe r capita consu mption expe nditure. MRP a nd URP
stand for m ixed referen ce period and u niform refe rence period , respect ively. IHDS fig ures
are on an annual basis.
Table 5a: Incidence of Vegetarianism,
by G ende r an d Cit y Typ e (NFH S) (%)
Women Men
All 29.43 20.01
(1,24,266 (74,319
Observations) Observations)
Rural 29.90 20.70
Urban 28.46 18.81
Mega cities 24.39 15.26
Large citi es 34.40 23.69
Small cities 32.26 21.46
Large towns 26. 80 18.38
Small towns 2 2.36 13.86
march 3, 2018 vol lIiI no 9 EPW E conomic & Political Weekly
suggesting that as men’s (and women’s) incidence increases,
the gap also increases.
The persistence of such a gender gap, and its widening with
overall incidence of vegetarianism, may be shaped by patriarchal
structures and practices and their regional variations. We can offer
only a very preliminary set of hypotheses to think with at this
point. A portion of the gap may be related to the fact that men eat
outside of the household a lot more than women do, and with
greater moral impunity than women. This allows men to enjoy
greater “fl exibility” from norms in a patriarchal context. The
other side of the same coin is that the burden of maintaining a
“tradition” of vegetarianism falls disproportionately on women.
However, eating out by itself does not result in eating meat. For
this, a link has to be made between meat-eating and ideas of
“masculinity” (Michelutti 2008). Such “gender ideology” may
partially explain why the vegetarianism gender gap is relatively
higher in states where politico –ideological Hindutva is more prev-
alent, a movement that is also masculinist (Banerjee 2012) and
uses vegetarianism in cynical ways (Ghassem-Fachandi 2009).
Figure 6a is consistent with the idea that women in these states
mark the adherence to vegetarianism in far more visible ways
than the men whose actions are often at odds with their claims.
These ideas are mirrored in trends in the illiteracy gender
ga p. Fig ure 6b, c onst ructed from Cen sus 2 011 es timat es, shows
that there is both a persistent gender gap in illiteracy (there is
greater incidence of illiteracy among women than men in all
major states) and that the gap widens with overall illiteracy
(the dot ted qua d rati c t line pivots down from the dashed li ne
of equality). The graph for illiteracy (Figure 6b) bears a
striking resemblance to the graph for vegetarianism (Figure
6a), suggesting that the (variable) strength of patriarchy (as
documented richly in scholarship on literacy and gender) may
also be at the heart of the variation in vegetarianism.
Table 5b shows the vegetarianism gender gap by social
categories. In all cases, incidence continues to be higher
among women than men. In the case of caste categories, the
approximately 10 percentage point gender gap continues to
hold. However, there are substantial differences in religious
categories: the approximately 10 percentage point gap holds
for Hindus (unsurprising since this is by far the largest popula-
tion) but not for other religious categories: it is only half that
for Jains and Buddhists, almost non-existent for Muslims and
Christians, and a whopping 34 percentage points for Sikhs.
The Sikh case is interesting since it throws up questions of the
intersectionality of masculinity and religious precepts. Further,
given the high incidence of Hindu–Sikh households in Punjab,
ethnographic work is needed to illuminate the terms of rela-
tions (or the “pacts of/for conversion”) as it relates to food
practices. However, we are unable to say much more on this
without robust qualitative studies.
Table 5c shows the gender gap by wealth. The gap is lowest
for the poorest quintile, and increases with each quintile until
the richest, with the richest quintile showing almost double the
gender gap of the poorest. Note that overall incidence of vege-
tarianism also increases with wealth quintile, again consistent
with the point made earlier that as men’s (and women’s) inci-
dence of vegetarianism increases, the gap also increases. Again,
we are only able to speculate that patriarchal relations could
be at play in s ome manne r he re. As a r obu stne ss c heck, w e al so
looked at the NFHS couples data set; this is a data set of almost
40,000 households with separate information from both male
and female partners. Couples are meat-eaters in about 65% of
these households, and vegetarians in only about 20% (Table 5d).
Interestingly, in 12% of cases the husband was a meat-eater
while the wife was a vegetarian, and in only 3% of cases was it
the reverse; the difference, almost 10 percentage points, is
similar to the gender gap estimated in Table 5a.
The NSSO i s the only household-level dat a s ou rce t hat prov ides
estimates for beef consumption (the estimates are for beef and
buffalo meat combined). Table 6a shows that the overall
incidence in India is 7.5%, and only slightly more in urban than
in rural areas. As we will argue later, this is clearly an underes-
timation. Table 6b shows that among religious categories,
Muslims and Christians are most likely to eat beef (42% and
27%, respectively), and that the largest religious population
(Hindus) has very small incidence of be ef-e at ing. A mong Hindus,
SCs have 4% incidence and other
mega-caste categories (OBCs and
non-SCs/STs/OBCs) have even
smaller beef consumption (less
than 1%). However, not all OBCs
or non-SCs/STs/OBCs are Hindu,
Table 5b: Incide nce of Vegetari anism by Gende r and Social G roup (NFHS) (%)
Religious Categories
Hindu Muslim Christia n Sikh Buddhis t Jain
Women 34.03 2.22 1.18 70.01 9.25 97.39
Men 22.95 1.44 0.5 4 35.90 4.66 92.34
Mega-caste Categories
SC ST OBC Other s
All (Women) 20. 66 18.09 33.10 35.51
Hindu (Women) 20.68 20.73 37.57 43.59
Buddhist (Women) 9.75
All (Men) 10.78 11.68 22.62 26.40
Hindu (Men) 10.84 13.21 25.37 31.75
Buddhist (Men) 5.00
Table 5c: Incidence of Vegetarianism by Gender and Wealth Quintile (NFHS) (%)
Poorest Poorer Middle Richer Richest
Women 22.99 26.57 27.23 28.92 39.32
Men 16.54 18.65 17.75 18.19 27.04
Table 5d: Couples D ata (NFHS) (%)
Meat Eater Not Meat Ea ter
Man Meat eater 65.22 12.03
Not meat eater 3.17 19.59
Table 6a: Incidence of Beef-eating
All 7.53
Rural 6.97
Urban 8.92
Table 6b: Incid ence of Beef- eating, by Soci al Group (%)
Hindu Musl im Christia n Sikh Budd hist Jain
1.39 41.97 26.51 0 9.31 0
SC (Hindu) ST (Hindu) OBC (Hindu) No n-SC/ST/OBC (Hi ndu)
4.21 0.83 0.68 0.41
4.5 7 5.12 6.23 12. 36
Economic & Political Weekly EPW m arch 3, 2018 vol lI iI no 9 61
and the non-Hindus from these categories report more beef
consumption, which is why the overall incidence in these
categories is higher in Table 6b.
Underestimating beef-eating: However, there are reasons to
argue that the NSSO gures for incidence of some types of
meat-eating, particularly beef, are considerably under-estimated.
To explore this, we compare the NSSO estimates derived through
the consumption survey with fi gures derived from aggregate
estimates through the production side. Specifi ca lly, we tur n to
National Accounts Statistics (NAS) fi gures for production less net
exports compiled by the Organisation for Economic Co-operation
and Development (OECD) and Food and Agriculture Organiza-
tion (FAO) of the United Nations Agricultural Outlook and pre-
sented in column (1) of Table 7.12 These fi gures for aggregate
consumption, derived as a residual from the production side,
are juxtaposed with aggregate consumption fi gures estimates
from the NSSO household consumption survey (column (2)). For
beef, note that the consumption survey produ ces a considerably
smaller estimate, by an order of magnitude of 2.7 (column (3)),
while the order of magnitude difference for chicken and mut-
ton are only half this (1.4).
Admittedly, there are well-recognised, more general problems
in comparing the production-derived (NAS) consumption fi gures
with the direct survey-based consumption fi gures (NSSO), with
the former typically overestimating consumption compared to
the latter (Datt et al 2016; Sundaram and Tendulkar 2003);
however, even in such estimates, the NAS/NSSO ratio of overall
consumption is a little less than 1.5 (Deaton and Kozel 2005),
that is, close to the 1.4 fi gure for chicken in Table 7. This fi ts
well with the fact that unlike beef (or pork), the other two
meats—chicken and mutton—are far less caught within cul-
tural political and group identitarian struggles in India. There-
fore, if we restrict our analysis to a comparison of different
meats, and take the ratio of 1.4 as the “natural” discrepancy
between the two estimates, then any additional discrepancy
can arguably be attributed to under-reporting of a specifi c
meat i n relat ion to chic ken . If this is t he c ase, the n for be ef, the
additional discrepancy is 96%, that is, comparing the difference
2.74–1.40=1.34 with 1.40 from the ratio for chicken. In short,
the NAS estimate is almost double the NSSO estimate after
accounting for natural discrepancy, placing the fi gure of beef-
eaters in India at 14.7% of the population.
Note that the aforementioned estimates are for aggregate
consumption. How do the results translate into incidence of
beef-eating? Suppose the beef consumption of beef-eaters who
re por t not e ati ng b eef i n th e NSSO, is on average similar to con-
sumption of those who do report beef-eating in the NSSO. Then
this would straightforwardly imply that actual incidence is
about 15%, that is, 96% more than the 7.5% estimated in NSSO.
Figure 7: Beef- eating amon g Muslims and SC s, by State (NSSO)
The choro pleth maps sho w incidence of be ef-eating a mong Muslims an d SCs in state s. For the lef t map, sample s ize for CH, HP and PJ was to o small to produce r eliable est imates.
Table 7: Aggregat e Consumption E stimates for D ifferent M eats
(1) (2) (3)
Beef 1,204 440 2.74
Mutton 743 528 1.41
Pork 358 77 4.64
Chicken 2,304 1,651 1.40
Consumpti on estimates a re in 1,000 metr ic tonnes. E stimates for N SS are derived f rom
original d ata using appr opriate samp ling weight s for the 68th r ound (2011¬2); figures
for FAO are taken f rom OECD– FAO Agricultural Outlook (Edi tion 2016). “ Bee f” is be ef an d
buf fal o mea t in N SSO a nd be ef an d vea l in FA O; “mu tto n” is g oat meat and m utto n in N SSO
and sheep me at in FAO; “chicken” is chick en in NSSO and poul try meat in FAO.
march 3, 2018 vol lIiI no 9 EPW E conomic & Political Weekly
Such a fi gure need not be too surprising since the estimates of
beef-eating among Muslims and Christians as reported above
ar e f ar le ss t han what we may a rguably ex pect i n com munities
that do not have cultural–ideological proscriptions against
beef. Note that if some of those who do report beef-eating in
the NSSO still under-report the extent of beef consumption,
then the estimated actual incidence would be concomitantly
lower than 15%. However, we feel that given the nature of
cultural–political pressures on beef-eating and its reporting,
such pressure will be manifested more in denial of beef-eating
rather than under-reporting the extent of beef-eating. Conse-
quently, in our judgment, the fi gure of 15% beef eater s in In dia
is a reasonable estimate in a sociopolitical climate that makes
declarations of beef-eating a hazardous act.
Spatial variation and cultural politics: Below, we explore an
alternative approach to gauge under-reporting of beef-eating.
We begin with the idea that cultural–political pressures against
beef-eating vary by region, and that, arguably, this would be
refl ected in spatial variations in the incidence of reported beef-
eating in NSSO, and therefore also the extent of under-reporting.
Consider, specifi cally, the case of Muslims and SCs (Dalits), two
social groups on whom cultural–political pressures have been
particularly strong: all instances of recent lynchings and beat-
ings of people accused (typically falsely) of killing or eating
cows being Muslims and Dalits thus far. Although Muslims are
the most likely to eat beef among the NSSO categories in Table 6b,
there is considerable spatial variation in reported beef-eating
among Muslims, as shown in Figure 7 (left map) (p 61). Beef-
eating among Muslims is over 50% in three states (West Bengal,
Assam and Kerala), all of which have relatively high Muslim
populations; and yet beef-eating among Muslims is only 7% in
Rajasthan and 10% in Jharkhand, states with lower Muslim
populations. Similarly, there is considerable spatial variation in
reported beef-eating among SCs as well, as shown in Figure 7
(right map) (p 61). Beef-eating among SCs is 22% i n (combi ned)
Andhra Pradesh, 19% in Tamil Nadu, and 17% in Kerala. By
contrast, it is less than 1% in several states: Chhattisgarh,
Jharkhand, Odisha, Rajasthan, Punjab, Madhya Pradesh, and
Haryana (and between 1% and 2% in West Bengal, Uttar
Pradesh, and Assam).
Why might such large spatial variations exist in the incidence
of reported beef-eating in NSSO among Muslims and SCs? There
is some evidence to suggest that at least a part of this spatial
variation may be due to cultural–political pressures. In the case
of Muslims, it turns out that the larger the Muslim share of a
state’s population, the greater the incidence in reported beef-
eating among the state’s Muslims. This is shown in Figure 8,
which plots incidence of beef-eating among Muslims against
share of Muslim population across 17 major states (data from
NSSO and Census 2011, respectively) as well as the quadratic fi t
with 95% confi dence i nter val s. Note that despite only 17 obs er-
vations, there is a clear, substantive, signi cant and fairly linear
positive association between Muslim beef-eating and Muslim
population proportion.13 Given the much higher risks and stigma
that are involved with beef-eating, such a positive correlation
with the shares of population are, if anything, even stronger
than those of OBC populations and meat-eating mentioned in
the previous section. The adage of “strength in numbers” may
be best tested in real-life situations. Thus, when the Haryana
chief minister made his (in)famous comment on beef-eating
and what Muslims needed to do in order to continue to live in
India, he simultaneously said that the “Muslim brothers of
Mewat district have voluntarily given up beef” (NDTV 2015).
The dubious accounting of a chief minister’s barely veiled
threat to Muslims as resulting in a “voluntary” act by a minority
community, was hopefully not lost on the general public.
Similarly, in the case of SCs, the spatial variation in Figure 7
(right map) may be at least partially explained by spatial
variation in cultural–political pressures on SCs. It is striking
that the four southern states top the list of beef-eating among
SCs in the major states; these are precisely the states with a
relatively longer and stronger history of Dalit liberation move-
ments (Jaffrelot 2003; Shah 2004), and, with the exception of
Karnataka, these are also states where Hindutva-fuelled cultural
politics pressures have been relatively less impactful: com-
pared to, say, Chhattisgarh, Rajasthan, Madhya Pradesh, and
Uttar Pradesh, all of which have very low incidence of beef-
eating among SCs. The case of Karnataka, which is a relative
exception among southern states, is particularly interesting.
Although Karnataka is placed at rank number four among states
for SC beef-eating (7%), the other three southern states have
over double or triple this incidence (22% in combined Andhra
Pradesh, 19% in Tamil Nadu, and 17% in Kerala). The Hindutva
movement in Karnataka is far stronger than in Andhra Pradesh,
and the Dalit liberation movement is arguably less articulated,
so that although the two states have somewhat similar SC popu-
lation shares (18% in Karnataka, 20% in Andhra Pradesh), they
have starkly different incidence of SC beef-eating.14
To further examine the relation of beef-eating among Muslims
and SCs to Hindutva politics, Figure 9 (p 63) plots incidence of
beef-eating against average vote share for the BJP in the major
states for the last three Lok Sabha elections (2004, 2009, 2014)
using data from the Election Commission of India (the linear
regression line with 95% confi dence intervals is also shown). We
acknowledge that average BJP vote share across these three
Figure 8: Bee f-eating and M uslim Popula tion (NSSO and Cens us)
beef-ea ting among Mus lims (%)
0 10 20 30 40
Muslims in population (%)
The graph i ncludes the qu adratic reg ression line an d the shaded 95% co nfidence int erval.
Economic & Political Weekly EPW m arch 3, 2018 vol lI iI no 9 63
elections is only a partial indicator of the hold of Hindutva
pressures, especially since there is temporal variation in the
role of Hindutva factors in BJP electoral performance. Never-
theless, as Figu re 9 r eveals, states with greater BJP vote share
tend to have lower incidence of beef-eating among both
Muslims and SCs.15
Finally, we address a potential alternative explanation of
spatial variation in NSSO beef-eating estimates, namely availa-
bility of beef. As a proxy for availability, we take data on number
of bovine livestock from the 2012 Livestock Census (GoI 2012)
and compute statewise fi gures for the number of bovine live-
stock per Muslim and per SC. Figure 10 plots incidence of beef-
eating reported in NSSO against these fi gures. In both cases,
reported beef-eating is signifi cantly negatively associated with
livestock availability, suggesting that availability per se is not
the issu e. Furthe r, sta tes wit h mor e cu ltu ralpo lit ica l pr ess ure s
(as argued earlier) and with weaker (or less radical) Dalit
emancipatory movements tend to be below the linear regres-
sion line, implying that they tend to have lower incidence of
beef-eating relative to what would be predicted from the aver-
age relationship between incidence and livestock availability.
This difference, we argue, is suggestive of culturalpolitical
pressures being a key factor in suppressing the incidence of
beef-eating. In a very insightful commentary on the politics
around beef, we are reminded of the existence of many recipes
for beef dishes that are camoufl aged in public discourse and
on menus in restaurants (Anveshi 2012). These are cultural
artefacts for the ways that data on beef are surely underesti-
mated in these dangerous times.
It should be clear from the above empirical exploration that
characterising India as a vegetar ian land is a gross misrepresenta-
tion of reality: the vegetarian population of India is at best 31%,
and realistically less than 20%. A majority of Indians, clearly,
eat some form of meat on a regular or occasional basis, and
eating only a vegetarian meal is not the cultural practice of a n
overwhelming majority of the countr y. This too could be
changing in the direction of more people eating meat: not a
surprising possibilit y given that sc holars have viewed cultures
as changing with the times and shaped by political, economic,
and ecological pressures on populations. It also implies that
“ po li ci ng of fo o d c ho ic es ” n ee ds to be pa rt of sc ho la rly attempts
to represent cultural groups in a multicultural society. What is
claimed as group or national tradition is not innocent of power
and struggles over hegemony. Our attempt here has been
therefore to initially “provincialise vegetarianism” which has
exerted a far greater infl uence on representations of India and
Indians than merited by its empirical existence.
The evidence presented in this paper also makes clear that
eating beef is the cultural practice of signifi cant numbers of
Indians (at least about 15% or about 180 million people).
Again, estimations of actual numbers of beef-eaters in the
country requires sober accounting for factors such as cultural
politics and deep-seated fears that skew the numbers in the
direction of underestimation. We have tried to stay as close to
the data as possible in this case. Our estimates err on the side
of conservative fi gures that are defensible in interdisciplinary
scholarly debates.
A key way in which we have complicated most pictures of
food habits in India is by insisting on attending to variations
(across different dimensions of location type, region/state,
social group, gender and class, and within social groups).
Attention to variations allows scholars to initiate preliminary
inquiries into underlying explanatory possibilities and social
mechanisms or processes that sustain social phenomena such
as food habits of populations. It allows us to fruitfully engage
with problems of generalisability to populations, which requires
awareness of the need to view culture as a production and not
simply a given, of being critical of public claims about cultural
practices and of viewing social groups as constructed catego-
ri es rat her tha n as act ual mobi lise d gr oups. If (fo od) habit s ar e
indeed formed within cultural spaces where learning and
transmission of meanings of food takes place, then scholars
ne ed t o look at how foo d ha bits be com e ma rke rs o f gr oup ide n-
tities in everyday life struggles and interactions.
Much of these data reveal the need to not assume cultural
homogeneity within socially constructed groups. Scholarship
Figure 10: Bee f-eating an d Bovine Livest ock (NSSO and Live stock Censu s)
Beef-e ating among Mus lims (%)
0 5 10 15 20
Bovine st ock per Muslim
Beef-e ating among SC s (%)
0 1 2 3 4
Bovine st ock per SC
The graph s include linea r regression l ines and the sha ded 95% confi dence interv als.
Muslims SCs
Figure 9: Bee f-eating a nd BJP Vote Share (NSS O and Elect ion Commiss ion Data)
Beef-e ating among Mus lims (%)
0 10 2 0 30 40 50
BJP vote share, avg o f last 3
Lok Sabha el ections (%)
Beef-e ating amonga SC s (%)
0 10 20 30 40 50
BJP vote share, avg o f last 3
Lok Sabha el ections (%)
The graph s include linea r regression l ines and the sha ded 95% confi dence interv als.
Muslims SCs
march 3, 2018 vol lIiI no 9 EPW E conomic & Political Weekly
needs to attend to the very interesting intra-group variations.
Any elision of such variation needlessly reproduces, in scholar-
ship, what social actors mobilising group identities claim in de
rigueur fashion as part of their political objectives. Social anal-
ysis that confl ates social categories with actually existing
groups and their practices only runs the risk of reifying the
very c ateg or ies w hic h need to be questioned . In thi s sense, it i s
better to view India as agroecological and cultural–political
zones rather than as a conglomeration of social groups. Basic
questions such as “what is food” and “who decides who can eat
what” get determined in the registers of power, desire, identity,
and preferences. It requires engaged scholarship that takes
refl ective public positions on ostensibly “private issues” when
they are not “let alone.”
1 A key development since then has been the
upholding of the “right to privacy” as a fun-
damental right by the Supreme Court of
India in August 2017.
2 In spired by Dipesh Cha krabarty ’s (2000) work,
we use the ter m “provincial ise” as a n attempt
to decentre a categor y (vegetarian ism) that has
been hegemonic in discourse and thinking
about food in Ind ia.
3 For N FHS, Table 1 gives estimate for those who
answer “never” to the question of frequency of
eating fi sh, meat, or eggs; for fi sh and meat
alone, the estimate goes up slightly, to 32.61%
for women and 24.30% for men.
4 For NSSO, Table 1 gives estimate for those who
did not eat fi sh, meat, or eggs in the 30 days
prior to the survey; for fi sh and meat alone, the
estimate goes up slightly, to 40.08%.
5 There is some degree of var iation even among
the Nort h East states, wit h Nagala nd, Mizoram
and Tripura hav ing less than 1% incidence
while Sikkim has 12% and Arunachal Pradesh
6 According to the Anthropological Survey of India’s
survey (1993), 88% of the 4,635 communities in
India were meat-eating, suggesting that vegetari-
anism is far less a cultural practice of communi-
ties than it is a preference of individuals.
7 O ur focus is on descr iptive statistics and infer-
ential (statistical test) results are not reported
8 We also separately c hecked i ncidence among
Buddhist SC s, given the long histor y of Dalit
conversion led by Ambedkar. (Note: 90% of
Buddhist women and men are SC, and 90% of
SCs are Hi ndu.) The NSSO est imate (24.63)
and the NFHS estimate (7.38) differ consider-
ably, perhaps because of the relatively smaller
sample size and method of sample selection
involved in each survey, so we do not analyse
these results further, or present them in the
main table.
9 In IH DS, “Forward Ca stes” are those o ther than
SCs, STs, OBCs and Brahmins.
10 Note that we do not have accurate OBC data
b ecause censuses do not compute these. For
Fi gur e 5, we r ely o n da ta pr ov ided in t he Indian
Human Development Report , 2011–Towards So -
cial Inclusion (Planning Commission 2011); the
report calculates these from the 64th round of
the NSSO.
11 The dif ference s are stat istically sig nifi cant,
alth ough stat istical test results are not reported
12 put/meat-consu-
m p m or OECD (2016), Meat consumption
(indicator); doi: 10.1787/fa290fd0-en (viewed
on 1 December 2016).
13 In fact, the estimated coeffi cient for the bivari-
ate linear regr ession of incidence of beef-eatin g
among Muslims against Musl im share of popu-
lation (for 17 major states) is 1.74 (standard
error 0.30), implying through extrapolation
that a hypothetical state with 100% Muslim
population will have 74 percentage points
more incidence of beef-eating among Muslims
than a st ate with no Muslim population..
14 Another factor that could par tially explain the
higher than expected vegetarianism is the
infl uence of religious reform movement s that
in sist on vegeta rian ism as part o f thei r ident ity
15 Despite a small number of data poi nts (17), the
relationship is statistically signifi cant at conve n-
tional levels for SCs (p-value 0.034) although
this is not the case with Muslims (p-value
0.131). In the case of Musl ims, t he relationship
does become signifi cant at conventional levels
(p-value 0.006) if all states and union territo -
ries are included. Note that the e stimated
coeffi cient sizes are fairly large: in t he case of
Muslims, for the reg ression w ith all states and
union ter ritorie s, the est imated coeffi cient is
-0.74, implying t hrough extrapolation that a
hypothetical state with 100% BJP vote share
wi ll h ave 74 perce ntag e poi nts le ss in cide nce o f
beef-eat ing among Mu slims t han a state with
no BJP voteshare. Coincidentally, this fi gure is
identica l to that obtained for t he relationship
of beef-eating with Muslim population (see
note 13), as well as somewhat close to the
gure of 96% y ielded by the earlier compari-
son o f NAS and NSSO esti mates. Note, of course,
that Figure 9 is insuf cient for causal attribu-
tion, and we a re not arg uing that BJP electoral
performa nce directly “causes” meat-eating or
meat avoidance, although we spec ulate that a
“third factor”—namely the cultural politics of
Hindutva—separately causes both better BJP
performa nce and more meat avoidance.
Anvesh i (2012): “What’s the Menu? Food Politics
and Hegemony,” Broadsheet on Contemporary
Politics, Vol 1, No 4, pp 2–3.
Banerjee, S (2012): Make Me a Man!: Masculinity,
Hinduism, and Nationalism in India, SUNY Press.
Chakrabarty, D (2000): Provincializing Europe:
Postcolonial Thought and Hi storical Difference,
Princeton University Press.
Chakravar ti, A K (1974): “Regional Preference for
Food: Some Aspects of Food Habit Patter ns in
India,” Canadian Geographer, Vol 18, No 4,
pp 395–410.
Dang, H A and P F L anjouw (2015): “Povert y
D yna mics in India between 2004 and 2012:
I nsights from Longitudinal A nalysis Using Syn-
thetic Pa nel Data,” World Bank Polic y Research
Working P aper, (7270).
Dat t, G, M Raval lion, and R Murgai (2016): “Growt h,
Urbanization and Povert y Reduction in India,”
No 21983, National Bureau of Economic Research .
Deaton, A, a nd V Kozel (2005): “Data and Dogma:
The Great Indian Poverty D ebate,” Worl d Bank
Research Observer, Vol 20, No 2, pp 177–99.
Desai, S B , A Dubey, B L Joshi, M Sen, A Shar iff and
R Vanneman (2010): Human Development in
India, New Del hi: Oxford University Press.
Drèze, J and R Khera (2015): “Understanding
Leakages in the Public Distribution System,
Economic & Political Weekly, Vol 50, No 7,
pp 39–42.
Ghassem-Fachandi, P (2009): “The Hyperbolic
Vegetarian: Notes on a Fragile Subject in Gujarat,”
Being T here: Fieldwo rk Experi ence and the Mak ing
of Truth, John Bornema n and Abdellah Ham-
moudi (eds), Berkeley: University of California
Press, pp 77–112.
GoI (2012): 19th Livestoc k Census – 2012, Depa rtment
of Animal Husbandry, Dairying and Fisheries,
Mi nist ry o f Agr icu ltur e, Go vern ment of In dia.
IIPS and Macro International (2007): National
Family Health Sur vey (NFHS–3 ), 2005-06: India,
Internat ional Institute for Population Sciences,
Jaffrelot, C (2003): India’s Silent Revolution: The
Rise of the Lower Ca stes in North Indi a, Delhi:
Permanent Black .
Joshi, P C, K K Bangwal, B S Patwal, B Saklani and
B Sakia ni (1994): “In Pursuit of Protein: On
Meat Eating in Garhwal,” Vol 24, No 2, India n
Anthropologist, pp 39–44.
Khare, R S (1966): “A Case of Anomalous Values
in India n Civilization: Meat-eating among the
Kanya– Kubja Brahmans of Katyayan Gotra,”
Journal of Asian Studie s, Vol 25, No 2,
pp 229–40.
Michelut ti, L (2008): The Vernaculari sation of
Democra cy: Politic s, Caste and Re ligion in India,
New Delh i: Routledge India.
NDTV (2015): “‘Never Said Musl ims Must Go to
Pak istan,’ Clarifi es Haryana Chief Minister
Manohar L al Kh attar,” 16 October, viewed on
24 December 2016,
NSSO (2013): “Note on Sample Design a nd Esti ma-
tion Procedure of NSS 68th Round,” National
Sample Survey Offi ce, Ministry of Statistics
and Prog ramme Implementat ion, New Delhi.
Planning Comm ission (2011): India Human De vel-
opment Report 2011: Towards So cial Inclusion,
New Delhi: Instit ute of Applied Manpower
Researc h, Planning Com mission, Government
of India.
Robbins, P (1999): “Meat Matters: Cultu ral Politics
along the Commodit y Chain in India,” Cultural
Geographies, Vol 6, No 4, pp 399–423.
Shah, G (2004): Social Movements in India: A Review
of Literat ure, Sage Publications India.
Shaik Zahid Mukhtar v State of Mahara shtra & O rs
(2016): viewed on 24 December 2016, https://
indian Ka
Singh, K S (1993): People of India, Anthropological
Survey of India.
Sundara m, K and S D Tendulk ar (2003): “NAS
NSS Esti mates of Pr ivate Consumption for Pov-
erty E stimation: A Further Compa rative Exam-
ination,Economic & Political Weekly, Vol 38,
No 4 , pp 37 6–8 4.
Vaidyanat han, A and K N Nair (1980): “On the
Sacred-crow Controversy,Current Anthro-
pology, Vol 21, No 3, pp 380–84.
Yadav, Y and S Kumar (2006): “The Food Habits
of a Nation,” Hindu, 14 August, viewed on
26 December 2016,
todays-paper/the-food-habits-of-a-nation /arti-
... Also, Haryana has the highest incidence of vegetarianism where more than 80% of the population is vegetarian, which adds up to the risk of iron deficiency. 26,27 A high prevalence of H. pylori infection can be seen in around 80% population of rural India, which can also be a leading cause of anemia as it has been observed that H. pylori infection causes iron deficiency. 28,29 Another probable risk factor for anemia can be worm infestation and poor water sanitation. ...
Full-text available
Background: Anemia is the most common micronutrient deficiency. Despite repeated efforts taken under the national program, high prevalence of anemia still observed across various life stages of females in India over the past two decades. Methods: This cross-sectional study aimed to discern the prevalence of anemia in the non-pregnant and non-lactating females of Ambala, Haryana, between the ages of 6 months to 49 years using the gold standard cyan-met-hemoglobin method. Results: A total of 1, 21, 202 study subjects, with a mean age of 28.9±19.3 years were tested for hemoglobin across 524 villages, out of which 97,305 (80.1%) females were found to be anemic with mean hemoglobin level of 10.8±1.4 gm% with 53.2% of all females being moderately anemic. Conclusions: This study reported that anemia is extremely prevalent in all walks of life of females. Multipronged strategy is to be followed to curb anemia from various age groups of females.
... Provisioning of milk products or fruits was also low. These preferences can be partly associated with availability concerns and in part because of household preferences even as only less than one-third of the Indian population are found to strictly vegetarian [70,71]. However, some of these practices for children can be improved as noted in the post-intervention phase findings of this study where an increase in the consumption of eggs and fruits among the exposed group is apparent. ...
Full-text available
Objectives: Child dietary diversity is very low across rural communities in Bihar. Based on the experience of behavior change communication (BCC) module roll out in self-help group (SHG) sessions in rural Bihar, this study aims to assess the impact of the intervention on child dietary diversity levels in the beneficiary groups. Methods: The study is based on a pre-post study design whereby child dietary diversity is examined for a sample of 300 children (6-23 months old from 60 village organizations) during both pre-intervention as well as post-intervention phase. The latter consists of two types of group viz. a) children whose mothers were directly exposed to BCC module in SHGs sessions and b) those who were non-participants but may have indirect exposure through spillovers of BCC activities. Econometric analysis including logistic regression as well as propensity score matching techniques are applied for estimating the changes in dietary diversity in the post-intervention phase. Results: During the pre-intervention phase, 19% of the children (6-23 months) had adequate dietary diversity (eating from at least 4 out of 7 different food groups) and this increased to 49% among the exposed group and to 28% among the non-exposed group in the post-intervention phase. The exposed group have an odds ratio of 3.81 (95% CI: 2.03, 7.15) for consuming diverse diet when compared to the pre-intervention group. The propensity score matching analysis finds a 33% average treatment effect on the treated (ATT) for the group participating in BCC sessions at SHG events. Conclusion: BCC roll out among SHG members is an effective mode to increase dietary diversity among infants and young children. The impact on child dietary diversity was significantly higher among mothers directly exposed to BCC modules. The BCC module also improved knowledge and awareness levels on complementary feeding and child dietary diversity.
... While our respondents consisted of Christians, Hindus and Muslims, the dominance of Hindu values applied to food is apparent in the 'veg'/'non-veg' dichotomy, phrases which are used as both a personal identifier as well as a common way to group foods in India. The use of the term non-vegetarian "itself attests to the historical hegemony of vegetarianism in India" [75] (p. 55). ...
Full-text available
The middle class in India is estimated to be roughly half of the population and, as such, holds considerable sway in influencing consumption trends. We explored food consumption practices and indicators for food transitions among middle-class households in the South Indian megacity of Bengaluru. Through 38 qualitative interviews, we asked respondents about their perceptions of food safety and how they navigate food safety risks in their daily food practices. The COVID-19 pandemic brought the topic of food safety into sharp relief, and consumers were keen on maintaining good health through food consumption. We engaged social practice theory to understand food shopping practices, the rise in immune-boosting foods and the consumer demand for safe, healthy food as this relates to wider sustainable food transitions. We found that middle-class consumers mitigate food safety risks through careful selection of where food is purchased. A rise in immune-boosting foods, traditional herbs and spices part of the regional diet are being revitalized. Demand for organically grown foods is hampered by a lack of trust in verification systems. We argued that government investment in building consumer confidence in both food safety, and organic labeling increases the willingness to pay a premium price among middle-class consumers.
... Some researchers cited health as the primary reason for adopting a vegan diet (Coelho 2019, Corrin & Papadopoulos 2017, Dyett et al. 2013. In many countries, such as India, vegetarianism is a widespread diet and even a religion-based lifestyle (Jayanthi 2001, Natrajan & Jacob 2018, Wunsch 2021. ...
Full-text available
In Hungary, very little data is available on vegetarianism, even though the phenomenon affects many sectors of the economy. Initially, we intended to examine the impact of vegetarianism on the agricultural sector when we realized that there was no adequate data available. Therefore, to collect adequate data on this issue, we compiled a questionnaire on September 9, 2021. We weighted our data based on the distribution of gender/age and gender/education level of the country. In our experiment, involving 1642 Hungarian participants over age 7, 86% of the respondents were regular meat consumers, 9 % flexitarians, 2 % pescatarians, 2 % vegetarians, and 1% vegans. The proportion of people consuming reduced amounts of meat was the highest between ages 46- 60. The proportion of the most radical vegetarian forms is the highest in the 19–25 age group. The proportion of vegetarians in the Hungarian population increases with the education level, as it is the highest among Ph.D. graduates. In the case of the older generation, health motivation, while in the case of the younger generation, environmental and animal welfare motivation is crucial in choosing the form of nutrition. The love of meat's taste and the idea that it is impossible to live a healthy life without it are the two most important reasons Hungarians consume meat. Current costs do not influence the choice of meat versus plant nutrition, but Hungarian society would be sensitive to significant increases in meat prices. Omnivores would largely give up eating meat due to health problems, but they are open to laboratory-produced meat. If artificial meat were offered in supermarkets at affordable prices with the right taste and texture, 43% of respondents would stop eating meat.
... Dietary habits vary dramatically across India. Some regions within the country are predominantly vegetarian and in other parts up to 90% of the households consume meat 28 . The protein needs of people are met by inexpensive and easily accessible domestic protein options in most cases. ...
Due to rapid digitization in India, a large section of the Indian population has gained access to fast and affordable internet services such as online food ordering. It offers a fascinating opportunity to research the way Indian people utilize food ordering apps. This chapter provides an interesting study based on contextual inquiry of Indian food ordering app users. The collected data, in interviews and other minute observations, have been documented into five work models, which explain various facets of the food ordering and delivery process and related user experience of Indian users. The identified user-types reveal insights into the unique mindset of Indian users when using and navigating these apps. The analysis of the work models and various top food ordering apps have also highlighted various subtle but important cultural implications of ordering food.
Full-text available
Never before in human history have vegetarianism and a plant-based economy been so closely associated with sustainability and the promise of tackling climate change. Nowhere is this phenomenon more visible than in India, which is home to the largest number of vegetarians globally and where vegetarianism is intrinsic to Hinduism. India is often considered a global model for vegetarianism. However, in this book, which is the outcome of eight months of fieldwork conducted among vegetarian and non-vegetarian producers, traders, regulators and consumers, I show that the reality in India is quite different, with large sections of communities being meat-eaters. In 2011, vegetarian/veg/green and non-vegetarian/non-veg/brown labels on all packaged foods/drinks were introduced in India. Paradoxically, this grand scheme was implemented at a time when meat and non-vegetarian food production, trade and consumption were booming. The overarching argument of the book is that a systematic study of the complex and changing relationship between vegetarian and non-vegetarian understandings and practices illuminates broader transformations and challenges that relate to markets, the state, religion, politics and identities in India and beyond. The book’s empirical focus is on the changing relationship between vegetarian/non-vegetarian as understood, practiced and contested in middle-class India, while remaining attentive to the vegetarian/non-vegetarian modernities that are at the forefront of global sustainability debates. Through the application of this approach, the book provides a novel theory of human values and markets in a global middle-class perspective.
Full-text available
While authoritarian populism and its relationship to the rural world have gained analytical prominence recently, few have attempted a systematic exploration of how various authoritarian populisms emerge from, and are embedded within, dynamics of capital accumulation, state, and class struggle. Drawing on Poulantzas' approach to "state contradictions," we focus on the ways by which bovine meat figures in Narendra Modi's authoritarian populist project in contemporary India. On the one hand, violent author-itarianism in the country uses beef eating as a powerful tool for subjugating subaltern groups to Hindutva rule. On the other hand, the country houses a rapidly expanding beef meat agro-industry, accounting for as much as 20% of global exports and based on corporate concentration around dominant class interests. We argue that this points to state contradictions in Modi's India witnessing strained accumulation patterns. These contradictions, we emphasize, have distinct ramifications for India's classes of labour in the countryside, as certain groups experience what we describe as a process of "double victimization."
This paper explores Hindu religious concepts in their broadest form in general, and the philosophy and teachings of Radha Soami Satsang Beas (a religious sect with its headquarters in Punjab, India) in particular, to access whether and in what ways the Indic spiritual value system is dovetailed with ecological ethics to make a case for climate action. Considering the influence of this sect in India and the size of its faith community, its role in propagating ecological ethics through the notion of karma, practice of vegetarianism, and service are examined here. The paper concludes that although the philosophical teachings of Radha Soami Satsang Beas like living a lighter life by practising vegetarianism and service do have great ecological values, the ethnographic account of the Radha Soami Satsang Beas shows that the devotees do not show any proclivity toward climate action due to the lack of direct teaching by their guru on the subject and therefore do not establish any relation between philosophical underpinnings of the sect and ecological stewardship. Nevertheless, the ecological sensibilities of this community are more aligned with their spiritual goal and are reflected in their behaviour.
Yoga is a spiritual discipline originating from ancient India. Most notably, yoga has links with Hinduism, Buddhism, and Jainism. This research explores UK yoga teachers’ attitudes toward plant-based diets and beliefs about the moral status of farmed animals. It uses qualitative interviews as the second phase of a mixed-methodology study. In the first phase, an online questionnaire found that over two thirds of UK yoga teachers believed a plant-based diet was most aligned with their yoga practice due to the teaching of ahimsa, or non-harm. Nevertheless, over 70 % were not following a plant-based diet. Interviews revealed this dissonance was related to views about the necessity of consuming animal products for health, especially when pregnant, breastfeeding, or nourishing growing children. The communal nature of eating also meant that family moderated dietary ideals. Interviews further revealed an alternative conception of ahimsa, permitting the consumption of animals in order to prioritize personal non-harm.
Full-text available
Recent National Sample Surveys point to significant poverty reduction in India since 2004–5, with a marked acceleration between 2009–10 and 2011–12. We employ new statistical methods to convert the three pertinent National Sample Survey rounds into synthetic panels, which help reveal original findings on important aspects of income mobility between 2004–5 and 2011–12. In particular, we document a strong pattern of upward mobility out of poverty, with a noticeable acceleration between 2009–10 and 2011–12. We examine some of the key household characteristics of those who have managed to escape poverty and contrast these with those who have fallen into an undesirable welfare status during these periods. We further undertake a careful investigation into the comparability of the survey rounds, prompted by the observation that fairly significant modifications had been made to survey questionnaires. Our findings suggest that changes in questionnaire design have not compromised the comparability of the data.
Full-text available
In a period when the global market is generally supposed to be erasing local meaning and applying uniform significance to objects and consumers, cultural commodity politics are in fact proliferating. Drawing on the case of the Indian meat economy, I argue here that the further that commodities travel in economic life, the more they affect and are affected by non-economic systems of signification. Specifically, the expanding livestock commodity chain in India has increased the political and social visibility of a long-standing carnivorous tradition, and meat is increasingly leveraged for social and political power in surprising ways.
This chapter explicates how the affect of disgust relates to violence by focusing on a case study of an upwardly mobile member of what is generally conceived of as a “lower” social category, a proponent of Hindu nationalism who was complicit in the 2002 anti-Muslim pogrom in the city of Ahmedabad. It sheds light on the unique communicative aspect of meat, which can pose a threat to central Gujarat. It focuses on the role disgust plays in creating new forms of identification by enabling a successful externalization of those aspects of the self that have to be denied. Furthermore, it extends the current paradigms by focusing on processes of identification found to be prevalent in Gujarat during the events of 2002. These processes are, in the literature, insufficiently understood and inadequately taken into account. Finally, it attempts to open up new approaches by shifting the focus onto different intellectual and theoretical grounds.
Make Me a Man! argues that ideas about manhood play a key role in building and sustaining the modern nation. It examines a particular expression of nation and manliness: masculine Hinduism. This ideal, which emerged from India's experience of British imperialism, is characterized by martial prowess, muscular strength, moral fortitude, and a readiness to go to battle. Embodied in the images of the Hindu soldier and the warrior monk, masculine Hinduism is rooted in a rigid "us versus them" view of nation that becomes implicated in violence and intolerance. Masculine Hinduism also has important connotations for women, whose roles in this environment consist of the heroic mother, chaste wife, and celibate, masculinized warrior. All of these roles shore up the "us versus them" dichotomy and constrict women's lives by imposing particular norms and encouraging limits on women's freedom. Sikata Banerjee notes that the nationalism defined by masculine Hinduism draws on a more general narrative of nation found in many cultures. If the outcomes of this narrative are to be resisted, the logic of masculinity, armed manhood, and nation need to be examined in diverse contexts.
What happened to poverty in India in the 1990s has been fiercely debated, both politically and statistically. The debate has run parallel to the wider debate about globalization and poverty in the 1990s and is also an important part of that debate. The economic reforms of the early 1990s in India were followed by rates of economic growth that were high by historical standards. The effects on poverty remain controversial, however. The official numbers published by the government of India, showing an acceleration in the rate of poverty reduction from 36 percent of the population in 1993/94 to 26 percent in 1999/2000, have been challenged for showing both too little and too much poverty reduction. The various claims have often been frankly political, but there are also many important statistical issues. The debate, reviewed in this article, provides an excellent example of how politics and statistics interact in an important, largely domestic debate. Although there is no consensus on what happened to poverty in India in the 1990s, there is good evidence both that poverty fell and that the official estimates of poverty reduction are too optimistic, particularly for rural India. The issues covered in this article, although concerned with the measurement of poverty in India, have wide international relevance - discrepancies between surveys and national accounts, the effects of questionnaire design, reporting periods, survey nonresponse, repair of imperfect data, choice of poverty lines, and interplay between statistics and politics. © The Author 2005. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development/The World Bank. All rights reserved.
The fact that NSS estimates of aggregate household consumer expenditure (HCE) tend to be lower than NAS estimates of private final consumption expenditure (PFCE) has spurred suggestions to pro-rata adjust the NSS-based size distribution on the basis of this difference. Drawing on a joint CSO-NSSO exercise at cross-validation of NAS and NSS, this paper details weaknesses of both types of estimates, but shows that NAS cannot be accepted as a more reliable yardstick for either aggregate consumption expenditure or that on specific commodity groups. This is due to the inherent 'fluidity' of NAS estimates, weaknesses in their underlying database, and the fragility of the rates, ratios and norms used in the commodity-flow balance underpinning the PFCE. Despite its shortcomings, it is argued that NSS is preferable because it is based on direct observations relating to the survey period and because, unlike NAS, it avoids recourse to adjustments based on arbitrary assumptions.
Indian philosophical thought has always attempted to deal with opposite or binary values. However, when faced with contradictions, social scientists try to analyze the form, meaning and function of opposed values in real social situations. The problem of opposed values becomes difficult and anomalous when the culture, at different levels, not only permits oppositions, but simultaneously sanctions them. The data on meat-eating among the Kanya-Kubja brahmans of Katyayan gotra (primarily an exogamous group composed of several lineages) present this type of problem to the social anthropologist.