Thanks. One issue with these anecdotal "all castes in revolt" reports is that it sounds like nobody voted for NDA. But NDA still got 44% vote from a broad section of the population, and wasn't starting with a 17-18% baseline "green" vote unlike the INDI mofos.
Did one or two particular castes/communities bring down the vote share ? From looking at the BSP vote pattern and also the broader "Jat" region (west UP/HR/north RJ), there is some basis for thinking that these two communities may have been doing the bulk of the vote cutting. If Mayawati could lose 50% of her vote share from 2019 to the INDI parties, it makes sense that BJP could also lose some of the SC voters.
Or did the Hindu population broadly feel a little less enthusiastic about BJP/NDA this time? A lot of people complain about jobs and inflation every time, that is not unique to this election.
Overall, this situation makes me lean towards two main issues:
(1) Election mismanagement: candidates, campaigning, BJP-RSS coordination
(2) On top of that, misinformation and fake videos that further "confirmed" some people's fears, particularly about reservations...again, the SCs come to mind here.
To get definite answers, one needs to know the constituency and sub-constituency vote shares/patterns. People of different castes/communities are not homogeneously dispersed across even one constituency. Only the party organization has enough resources to get this kind of information and work from that.
One question is still bothering me: 8% drop in vote share is a lot more than the 3% error margin of most scientifically conducted exit polls. Opinion polls before the election can be way off the mark since things can change quickly near/on election day. But exit polls are after the fact, they shouldn't be missing such a large drop in vote share in Bharat's largest state - we are not talking about Goa or Tripura here. And furthermore all the 3-4 exit polls got it wrong.
Now some people will just say "all the pollsters are useless", but it is a fact that a properly designed statistical sample is the most reliable way to estimate actual vote shares. There are no reliable "tea leaf readers" who can predict elections just by hanging around the state, astrology etc, unless you have statistically credible information. None of these people have a credible record, they might just be right one day (broken clock effect).
Is it possible that a lot of people coming out of the voting booth said they voted for BJP but actually didn't ?