The Great Indian Political Drama - 1 (Oct 2017 - Mar 2018)

[hanumadu]

Narendra nath mishra‏Verified account
@iamnarendranath
8h8 hours ago
More
Much awaited opinion poll of @surjitbhalla is out.His projecion,
BJP-125
Cong-55
In recent times his track record is 100% correct.

[ashokk]

Himachal exit poll: Seat share (Source: TOI Online-CVoter)
BJP: 41
Congress: 25
Others: 2

Himachal exit poll: Vote share (Source: TOI Online-CVoter)
BJP: 47.6%
Congress: 44.0%
Others: 8.3%

Himachal exit poll: Seat share (Source: India-Today Axis)
BJP: 47-55
Congress: 13-20
Others: 0-2

[abhijit]

Cong balloon burst

But Cong vote share is worrying.

[hanumadu]

Minhaz Merchant‏Verified account
@MinhazMerchant
2m2 minutes ago
More
.@IndiaToday-Axis MyIndia poll projects BJP 106-Cong 75 despite strong anti-incumbency & Cong's Patidar-Dalit coalition

[hanumadu]

Manak Gupta‏Verified account
@manakgupta
6m6 minutes ago
More
BJP winning Himachal Pradesh with 55 seats (+-7)

#TodaysChanakyaOnNews24

[Indrad]

Cong balloon burst

But Cong vote share is worrying.

+1 they can;t be written off yet...

[hanumadu]

Pinky Rajpurohit‏
@Madrassan
43m43 minutes ago
More
#ABPExitPoll
Saurasthtra-Kutch region (54 seats)

BJP- 31-37 (49%)
Cong- 16-22 (41%)
Others- 0-2 (10%) @abpnewstv #GujaratElection

ABP News‏Verified account
@abpnewstv
34m34 minutes ago
More
#ABPExitPoll

South Gujarat region (35 seats)

BJP- 21-27 seats (52% vote share)
Cong- 9-13 seats (40% vote share)
Others- 0-1 (08% vote share)

ABP News‏Verified account
@abpnewstv
15m15 minutes ago
More
#ABPExitPoll
North Gujarat region (53 seats)

BJP- 32-38 seats (49% vote share)
Cong- 16-22 seats (42% vote share)
Others- 0 seats (09% vote share)http://www.abplive.in/live-tv?ref=livet ... in=content …

Going by the vote share, BJP should be getting more votes one would expect. The vote share is more than in 2012.

Only Surjit Bhalla is giving BJP 125 seats. He was bang on in Bihar.

[hanumadu]

VDPAssociates‏
@VDPAssociates
8m8 minutes ago
More
Gujarat Seat Share Forecast #GujaratElection2017
BJP-142
Cong-37
Others-3

This one I like.

But note that VDPAssociates is giving them lesser vote share than others, but higher seats.

VDPAssociates‏
@VDPAssociates
Follow Follow @VDPAssociates
More
Gujarat Vote Share Projection(Margin of Error +/-3%)
BJP-48%
Cong-40%
Others-12%

[Hari Seldon]

Some poll agency called VDP has surfaced. Surefire way to get hajaar retweets and virality is to put out a GJ exit poll right now, seems like. Lol.

[Hari Seldon]

Here's another:

[Hari Seldon]

Finally, TS Chanakya speaks....

Today's Chanakya‏Verified account @TodaysChanakya 17m17 minutes ago
#TCExitPoll Gujarat Elections 2017
Seat Projection
BJP 135 ± 11 (Plus / Minus 11) Seats
Congress+ 47 ± 11 (Plus / Minus 11) Seats
Others 0 ± 3 (Plus / Minus 3) Seats

They missed BH bigtime but in almost every other poll, including the extreme verdicts, they were spot on.

135 is huge! Its well past the 2/3rd mark and nobody, not even the opposition, will be able to say BJP has underperformed coz of DeMo GST etc ('cept Pagalike, Slurdesai types).

[abhijit]

They missed BH and here they are way off from other polls, just like how it was for BH.

[Klnmurthy]

Dr Manmohan Singh suddenly found his voice when it came to defending his disgraceful record as PM wherein he presided over colonial level of natural resource loot and plunder.

Weren't we asking why NM went soft on cases against MMS, Chidu, Chonia and company? I certainly saw a lot of that sentiment expressed widely on socMedia. Well, silently hi sahi, the wheels are moving. See below.

Krishnadas Rajagopal‏ @kdrajagopal 1h1 hour ago
SC gives green signal to set up 12 exclusive fast track courts to prosecute MPs, MLAs. Courts have to be operstional by March 1, 2018. Will operate for 1 year and dispose 1581 criminal cases pending against politicians.

SC: This is the first step @the_hindu @abaruah64

P.S. I'd imagine Subbu Swamy having a ball of a time at that special tribunal only. Heh.

During 2014 election campaign Modi said in a group discussion with students that he would like to see cases against political figures resolved within a year, so that the guilty are punished and the innocent don't have cases hanging over their head, and janata will have confidence in the system.

Now this. In the face of nearly unanimous opposition by the establishment and hostility from the Supreme Court.

I say it again, Modi is the wisest and most effective leader to come along in a very long time.

Criticize him by all means but keep it thoughtful and constructive.

[kvjayan]

Rajdeep & co. are breaking their heads to somehow interpret BJP's (projected) victory is actually a defeat, because NaMo did a dirty "communal" campaign whereas RaGa's was neat and clean, sticking to "serious" issues.

[kittoo]

Lets not count our chickens before they hatch. I will keep my fingers crossed till Sunday. Lets see what happens.

[Hari Seldon]

Lets not count our chickens before they hatch. I will keep my fingers crossed till Sunday. Lets see what happens.

+1.

GJ satta market is giving majority to INC!

[Suraj]

They missed BH and here they are way off from other polls, just like how it was for BH.

Technically, that's not an analytical way to approach this. All the exit polls start with raw data from their poll, which gives them a vote share projection. Upon that, they all apply a proprietary formula to get seat share. This calculation is where things diverge widely.

There are two steps:

a) Take raw exit poll data, attempt to create a wide population from it. Errorprone if their sample size is small, skewed or not normalized to the behavior of the larger population set clearly. In other words, if 1000 people exit poll gave them voting ratio in a 2-way contest, how does a 1 million people voting bloc vote in that election based on the data you have ? This depends on a LOT of things related to the precise manner in which the 1000 people were selected and asked. You could ask 1000 people known to vote for A instead of B and your exit pol will scream 'landslide for A' even though B may win by a mile.

b) Take projected polling estimate over the full population and translate that into polling ratios and therefore seat share per constituency. Now you not only have to create one projection, but divide that into a second set of data that is per constituency.

If this sounds hard, lets go from 2-way contest to 3-way or 4-way contest. Revisit the above and do it all for 3-4 potential major choices.

So the takeaway is that:

1. for a simple 2-way contest, some exit pollers will get this basic bit really accurate.
2. among those who get item 1 correct, some will also get seat share estimate correct.
3. However, being accurate in 2-way estimation does not scale to 3-way or higher. They're very different beasts.
4. Thus there may be some polls that are good at just voteshare breakdown in a 2 way contest, some that are good in seatshare breakdown in 2-way contest, some that are good in voteshare breakdown in 3+ way contest, and some that are good in the seatshare breakdown in 3+ way contest.

Just because a poll was bad in a 4-way BH contest does not mean their GJ exit poll is suspect . What matters is their voteshare and seatshare estimate track record in purely 2-way contests.

So the question really is, given an exit poll here, how did that entity perform in prior exit polls for pure 2-way contests ?

[Chandragupta]

Is JDU not an NDA ally? Yet Pawan Verma like a filthy snake keeps on attacking BJP on rundieTV. Why?

I want BJP to completely finish such sekoolar snakes. Anyone talking about 'minorities' in India must be completely routed electorally & politically so that no rattlesnake like Laloo & Nitiss & Mamta can ever raise their heads. It would take decades and will take a mammoth effort from Hindus but I think there is a chance this might happen soon enough once pissfuls show their ugly heads more publicly.

Re: Congress' vote share. 40% is still shocking! Why are people still voting for this party? In my eyes, anyone voting for Congress is an out & out traitor.

[Indrad]

yes 40% is worrying!
Any time a split in vote share of BJP will mean return of congress.
Vice versa is also possible but less likely cos Muslims, minority and committed leftist vote banks will always drift to cong.
That Patel stir thing's for sure made a dent though not powerful enough to crack.
All these reservation stir have to be neutralised if possible within next 1 yr or push ball in SC.

[Sachin]

Technically, that's not an analytical way to approach this. All the exit polls start with raw data from their poll, which gives them a vote share projection. Upon that, they all apply a proprietary formula to get seat share. This calculation is where things diverge widely.

Have been meddling with this "Analytics" part for quite some time now (mainly for a change of job ). What I understand from your post is that it is the "sampling size" which matters, and the way these "samples" have been identified. I get a feeling that agencies like Today's Chanakya have a better way to decide on their samples. May be organisations like RSS, have a larger number of "samples" to check with so that they get a better feel of the situation. To be frank; if the "samples" are from Kerala, Modi & Co would not even get one seat in Gujrath .

[Suraj]

Technically, that's not an analytical way to approach this. All the exit polls start with raw data from their poll, which gives them a vote share projection. Upon that, they all apply a proprietary formula to get seat share. This calculation is where things diverge widely.

Have been meddling with this "Analytics" part for quite some time now (mainly for a change of job ). What I understand from your post is that it is the "sampling size" which matters, and the way these "samples" have been identified. I get a feeling that agencies like Today's Chanakya have a better way to decide on their samples. May be organisations like RSS, have a larger number of "samples" to check with so that they get a better feel of the situation. To be frank; if the "samples" are from Kerala, Modi & Co would not even get one seat in Gujrath .

Sampling accuracy lets you generate an accurate picture of a full population's voting pattern from a small sample. But that gives you voteshare projection only. You then need to fragment that voteshare projection to get seatshare projection per constituency.

My point is that there are two steps where things could go wrong, just in the methodology for a simple 2-way contest. It gets much worse where it's a 3+ way contest. So what matters is what's the prior record of the agency in exit polls for 2-way contests, particularly in the same state, but also in others ? Just the fact that they got it wrong in a 4-way contest somewhere else is not pertinent data, because their mathematical model for that is completely different.

[Shyamal]

2-way would be Assam, WB, TN - among post 2014 elections in sizable states.
Chanakya was wrong in TN.
They were very close to the actual figures in Assam and WB.

[Primus]

Where is 5forty3 in all of this?

[Shyamal]

Where is 5forty3 in all of this?

Total silence since 23rd Nov.

[Indrad]

not good^. Last time he fell silent on Bihar all of a sudden and result was catastrophe...