Geography of online poll campaigns in India

  • In more than 100 districts of the country, at least one-fifth of the population had access to internet in their household by 2015-16
  • Analysis suggests that the impact of different media on poll process is likely to vary significantly across different parts of India

The results of the latest National Family Health Survey 2015-16 (NFHS 2015-16) showed that only a little above 11% of the country’s population had access to internet at home in 2015-16. However, the raw unit-level data from the survey released last year shows that there was wide divergence in internet access based on caste, class and region.

This means that the impact of online campaigns, including propaganda material and fake news, for the 2019 Lok Sabha elections is likely to be much larger in certain pockets and among certain segments of the population than others.

Where does the target audience for online campaigns reside? What does it look like?

A Mint analysis of the NFHS data shows that households with internet access are concentrated in the prosperous North Indian belt spanning Punjab, Haryana, Himachal Pradesh, Chandigarh and Delhi. Jammu and Kashmir, Mizoram and Nagaland also reported high internet access, but the sample sizes for these states are relatively small.

The NFHS surveyed more than 600,000 households across the country in 2015-16. It is a rich source of data on household ownership of assets and amenities, though the database is primarily used to assess health outcomes. The telecom regulator provides some disaggregated data on internet connectivity for more recent years, but that data is skewed because of multiple internet connections in the same household. The NFHS data are thus the latest estimates of actual internet penetration from an official and large-scale, nationally representative survey.

A disaggregated district-level analysis shows that in 123 of the 640 districts from where data was collected, at least one-fifth of the population had household access to internet. In 50 districts, at least one-third of the population had household access to internet. It is in these districts that the internet battle for hearts and minds will be the fiercest. All estimates have been weighted by the number of individuals in a household to account for the differences in household sizes across income classes.

How did these districts vote in 2014? Parliamentary constituencies and districts usually don’t overlap.

However, the Election Commission also provides assembly segment-wise breakup of Lok Sabha results for each constituency and typically the boundaries of the segment tend to lie within one district unlike in the case of parliamentary constituencies. An analysis of district-wise voting patterns based on aggregation of vote shares of Assembly segments that lie in a district shows that the Bharatiya Janata Party (BJP) polled more votes than any other party in the 2014 elections in roughly half (60) of the 123 districts with relatively high internet penetration.

In 39 of the 60 districts, the BJP’s vote-share lead over its nearest rival is more than 15%, the median vote-share difference between leading parties across India’s districts.

The BJP’s electoral record in connected districts was not very different from its overall country-wide record given that it had the highest vote-share in roughly half of the 640 districts nationally in the 2014 elections. In 42 of the 123 high-internet districts, non-BJP non-Congress parties had the lead, while the Congress had the lead in terms of vote-share in 2014 in the remaining 21 districts (see chart 3). The analysis also shows that the reach of the internet was much more among upper castes than among other caste groups. More than one-fifth of the upper castes reported having internet access, while less than one-tenth of other backward class, scheduled caste (SC) and scheduled tribe (ST) households reported having such access. There is a similar divide across levels of educational attainment as well (see chart 4).

There is a stark divide across caste groups when it comes to access to traditional media as well. As much as 36% upper castes reported reading a newspaper every day. Among SCs, the proportion was half that of upper castes at 18%. For STs, it was lower at 12%. Television, however, appears to be a great leveller, with a large segment cutting across caste groups having access to television.

Overall, the analysis suggests that the impact of different media on the poll process is likely to vary significantly across different parts of India and among different segments of the electorate.

Sriharsha Devulapalli and contributed to this piece.