Why Some Exit Polls May Get Uttar Pradesh Completely Wrong
Exit polls for Uttar Pradesh range from 65:13 in favour of NDA to 54:21 in favour of Gathbandhan. Which is right?
While almost all the exit polls may have predicted a second term for Prime Minister Narendra Modi, there are huge variations in the seat predictions for Uttar Pradesh, the state which sends the maximum number of MPs – 80 in all – to the Lok Sabha. The exit poll results are so dramatically different from each other that it seems that the agencies aren’t talking about the same state at all.
The predictions fall broadly in three categories:
- NDA-Mahagathbandhan Evenly Matched
- NDA Sweep
- Mahagathbandhan Sweep
NDA-Mahagathbandhan Evenly Matched
There are three surveys that fall broadly in this category. The India TV-CNX Exit Poll gives 50 seats to the NDA and 28 to the Mahagathbandhan. The CVoter poll gives 40 to the Mahagathbandhan and 38 to the NDA. And the ABP-Nielsen poll gives 45 to the Mahagathbandhan and 33 to the NDA. All three surveys give two seats to the Congress.
In 2014, the combined vote share of the Mahagathbandhan constituents was more than that of the NDA in 42 seats and less than it in 36 seats.
This means that a complete transfer of votes between the Mahagathbandhan parties with no positive or negative swing would reduce the NDA to 36 seats from the 73 it won in 2014.
Most surveys reflect a minor deviation from this position. CVoter’s survey in particular would mean a near perfect vote transfer in the BSP-SP-RLD alliance and no swing against the BJP.
The CNX survey indicates a less than perfect vote transfer in the alliance and a slight positive swing in favour of the NDA. The Nielsen survey is the reverse, reflecting a perfect transfer of Gathbandhan votes and a slight swing away from the NDA. Curiously, Nielsen’s initial prediction on the evening of 19 May was 22 seats for the NDA but it later adjusted it to 33.
NDA Sweep
There are three surveys in this category. The India Today-Axis poll predicts 62-68 seats for the NDA and 10-16 for the Mahagathbandhan. The Today’s Chanakya poll predicts 65 for the NDA and 13 for the Mahagathbandhan. And the News18-Ipsos poll predicts 60 for the NDA and 18 for the Mahagathbandhan.
These exit polls would mean that Uttar Pradesh has seen a Modi wave that cuts across communities, age groups and other social categories.
Mahagathbandhan Sweep
On the other hand, there are two surveys which show the Mahagathbandhan doing decisively well.
The survey by the news channel National Voice predicts that the Mahagathbandhan could win 48 seats and the NDA would be restricted to 28.
Anthro.ai, which makes predictions based on Artificial Intelligence, says that the Mahagathbandhan could win as many as 54 seats and the NDA could fall by over 50 seats to 22. Both these surveys give 4 seats to the Congress.
These surveys would indicate not just a complete consolidation of Muslim, Yadav and Jatav votes behind the Mahagathbandhan but also a swing against the BJP and in favour of both the Gathbandhan and the Congress.
Seat-by-Seat Predictions
Three of the surveys have given detailed seat-by-seat predictions for the 80 Lok Sabha seats in Uttar Pradesh: Axis, Anthro.ai and National Voice. The predictions by Axis got a lot of focus especially as they were allegedly taken down from its website.
The predictions were put up with some changes later on the India Today website but they still seem to be deleted from the Axis website.
Some of the (allegedly deleted) predictions by Axis defy logic. For instance, it is difficult to understand how SP’s Dimple Yadav would lose from Kannauj, even though she won in 2014 in the middle of a Modi wave. And unlike this election, she didn’t even have the support of the BSP five years ago. The same holds for Mainpuri, where Axis is said to have predicted a tough fight for SP patriarch Mulayam Singh Yadav.
Axis seat-wise predictions in other states also faced a lot of flak. For instance, it predicted Congress winning a seat in Tamil Nadu where it hasn’t put up a candidate.
In its predictions, Axis also gave a curious disclaimer that the seat-by-seat predictions are based “purely on the popularity of the political party during exit polls and not individual candidates and we cannot be held responsible for any variations.”
If candidates are not being factored in, it defeats the entire purpose of an exit poll. Therefore it becomes difficult to take the India Today-Axis poll at face value.
The most detailed seat-by-seat predictions are by National Voice.
In fact, the channel has given detailed, seat-wise vote share predictions as well.
The channel has asserted that there is a swing away from BJP cutting across caste groups. But the Mahagathbandhan isn’t the beneficiary in all the categories. For instance among non-Jatav Dalits, the Congress seems to be gaining at the BJP’s expense.
Though slightly different in terms of numbers, Anthro.ai’s seat-by-seat prediction is very similar to National Voice and this is evident in the map.
The seat-by-seat analysis adds most weight to the surveys predicting a decisive lead for the Mahagathbandhan. However, since many seats are being closely contested, the C-Voter and Nielsen surveys could also prove right in the end.
(At The Quint, we are answerable only to our audience. Play an active role in shaping our journalism by becoming a member. Because the truth is worth it.)