Deepening Divides

Our earlier article (Natrajan and Jacob 2018) 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. We showed that myths of Indians’ meat-avoidance (vegetarianism) stand exposed when we unpack India in different ways, through the lens of caste, gender, class, and especi­ally region. We also presented evidence to suggest the influence of cultural-political pressures (valorising vegetarianism and stigmatising meat by proscribing and punishing beef-eating in particular, but also meat-eating more generally) on reported food habits. The present article follows up our earlier work by analysing changes in the incidence of vegetarianism over time.

The earlier article used data from three different large-scale, representative surveys. Of these, the National Family Health Survey (NFHS) released a new data set (round 4) after our earlier analysis was completed. This allows for a comparison of vegetarianism across the two NFHS rounds, bookending a ­decade of potential change (2005–06 in round 3 to 2015–16 in round 4). The NFHS is analogous to the Demographic and Health Surveys (DHS) conducted in over a hundred countries. Surveys are conducted for separate large samples of women aged 15–49 years and men aged 15–54 years. Data for round 3 are from 1,24,385 women and 74,369 men. Data for round 4 are from 6,99,686 women and 1,12,122 men. NFHS looks at specific items of food consumption, including eggs, fish and “chicken or meat,” asking respondents about how often the item was consumed. For our analysis and consistent with our previous article, we consider those who answered “never” to all three (eggs, fish, chicken/meat) as “vegetarian.” Appropriate sampling wei­ghts were used to construct estimates of vegetarianism within different aggregates (states and social groups).

Decadal Change

From the data, one interesting finding is that there was little change in the overall incidence of vegetarianism in the decade 2005–15 for women and men: while vegetarianism among women changed marginally from 30.22% in 2005–06 to 30.97% in 2015–16, for men it was 20.60% to 20.73%. This amounts to an increase of 0.75 and 0.13 percentage points for
women and men, respectively (equivalent to 2.5% and 0.6%, respectively). In our earlier article we showed that there exists a significant gender gap in reported vegetarianism—about 10 percentage points higher among women (equivalent to ­almost 50% more among women compared to men). This gap of 10 percentage points, we showed, was persistent across location (rural–urban), class and caste categories. One interesting puzzle we raised was the existence of the gap only among Hindus (10 percentage points) and Sikhs (a whopping 34 percentage points), much less among Jains and Buddhists (about 5 percentage points), and almost non-existent among Christians and Muslims. We had submitted that this gap could be shaped by gender ideologies within households and communities that placed undue burden on the woman to uphold a tradition, and gendered practices of eating out (favouring men).

The new data show how this gap is persistent, pointing to the possibility of a rigidification of communitarian ideas shaped by food beliefs and practices, but also the social norms rapidly being put in place (partially by state ideologies, but also partially within society through social actors such as community leaders, ethnic mobilisers who craft community boundaries as markers of distinction). We bring this point up in order to emphasise that this overall result (of no change in gender gap over time) hides interesting temporal dynamics for sub-groups of the population. We now turn to examining the intersectional changes across caste and class categories.

Change across mega-caste and wealth categories: Figure 1 (Graphs 1 and 2) shows vegetarianism for mega-caste cate­gories. For women, there was little change (less than 1 percentage point increase) in the decade 2005–15 for the categories of Scheduled Tribes (STs), Scheduled Castes (SCs) and Other Backward Classes (OBCs). But there was a relatively substantial increase in vegetarianism for the residual (“other”) category, broadly including privileged castes (4.4 percentage points increase from 2005, equivalent to 12.4% increase). In the case of men as well, the “other” category of privileged castes saw a substantial increase in vegetarianism (3.3 percentage points increase from 2005, ­equi­valent to 12.6% increase). This points to an increasing assertiveness among ­privileged castes with respect to ­vegetarianism.

Figure 1 (Graphs 3 and 4) shows vegetarianism across five wealth quintiles. There was little change in all ­except the richest quintile which saw a 3.9 percentage points and 2.5 percentage points increase for women and men in that category, respectively (equivalent to 9.8% and 9.1% increase). This confirms our earlier ­observation (and some other previous studies cited in our earlier article) that vegetarian practices are correlated with socioeconomic status.

Change across states: Figures 2–6 (pp 23–24) turn to reported vegetarianism across states. Figure 2 plots change in 2005–15 against the baseline (2005). It shows that, on ­average, states with higher incidence of vegetarianism in 2005 experienced greater increase in the following decade, and this is true for both women and men separately. This implies that over the decade there is increasing divergence across states. In Figure 2, the graphs on the right (#2 and #4) focus on the 17 states with population of at least 2.5 crore in the last census (2011). Divergence occurs even in this subset. Further, the size of the divergence is substantial: For Graphs 1 and 3, for every 1 percentage point of vegetarianism incidence in 2005, there is an average inc­rease of 0.14 percentage points over the following decade for women and 0.26 percentage points for men, and this relationship is statistically significant at the 99% confidence level. In fact, the relationship continues to hold with similar large size and statistical significance for the higher-population states with only 17 observations.