Arun Narendranath, one of the most avid followers of elecoral trends, sent me this email. I am taking the liberty of posting it:
Read ur blog just now. Spot on in saying that four states of AP, Maha, Assam & UP have not been analyzed properly. Lets try and explain the flaws here
Assumption: the voteshare estimates by these TV channels have not been fudged. (though this may not be true).
Let for once assume that vote share given by CNN IBN is correct in the state of AP.
TDP+: 25 + 5%
Now converting the same into regional trends.
In Telanagana it would convert into
In Costa region:
East & West Godavari districts:
and finally in Rayalseema:
This would translate into the following number of seats:
TDP & TRS will get their seats mostly from the Telangana region. TDP will get the additional seats in Krishna district (seats like Vijayawada, Srikakulam). PRP will do well in Nasarpur, Kakinada. INC will sweep Rayalseema (by huge margins) & win about 20-30% seats in costa & Telangana. This effect has been clearly captured by vernacular media(read the predictions of MAA & Itv, Shakshi is owned by YSR's Son). The attached file will clear show how the English media has completely mis-interpreted when the vernacular media caught the trend.
Bottomline: Even if Mr. Babu is not inclined towards BJP/NDA most of his MPs would be. Because they would be mostly from the Telangana region and would have won the election because of the T-Sentiment. If TDP does not come to NDA its worth buying them with Mr. Badal's money bag.
a) Though UPA can make marginal gains in Vidharba & Maratwada, they will loose seats in Mumbai & Thane, Western Maharashtra.
b) The MNS has done extremely well in 3 seats Thane, Nashik & Mumbai-North East (NCP will win in Thane & Nashik because MNS played the spoiler). Though they polled very well in Mum NE, it is difficult to upset Kirit. The assumption that MNS has damaged NDA across the state is false. Their effect was restricted to 3-4 seats.
c) The most interesting factor the exit polls have missed in Maharashtra was the internal conflict between NCP & INC. There were several rebel candidate in the fray from both INC & NCP. Though NCP was in alliance with INC, all its cadres were sabotaging the prospects of INC candidates and vis-versa. For instance in Pune, Mr. Sharad Pawar was campaigning hard for Mr. Kalmadi but his nephew Mr. Ajit Pawar was indirectly supporting Mr. Anil Shirole of BJP. This is very similar to bradley effect (http://en.wikipedia.org/wiki/Bradley_effect). The NCP voter pressed the button against the lotus when they claimed they would do so against the hand. The bradley effect will be found in the following seats.
Pune. Latur, Nanded, Osmanabad, Nagpur, Ahmed Nagar (INC sabotaging NCP) and to an extent in Shiradi, Mumbai NC.
Bottomline: Though the voteshare given to UPA is too high, some explaination can be found in bradley effect, that is why Mr. Ashok Chavan & Mr. Vilasrao Deshmukh want to go alone in assembly election and the NCP wants to go with SHS or MNS (reports of Mr. pawar calling Uddav in London & Mr. Jayant Patil meeting Mr. Raj Thackery)
Gallagher effect is missed by the media groups. They did the same mistake in Karnataka assembly elections, they are doing so now in the UP electoral trends.
Bottomline: Expect in Phase I & IV of UP polls (were BJP did extremely well across seats), in all other phases BJP has done well only in pockets depending on the candidates. For instance Phase II ( Ambedkar Nagar, Faizabad), Phase III (Kanpur, Balaun) Phase V (Anola, Philibit & to an extent in Moradabad). Usually BJP's vote share is consistent across the state, but not so in this election.
Finally Mr. Mahesh Rangarajan has converted the voteshare of BSP:29%, Sp:23%, BJP+: 21% & INC: 18% into 28, 22, 17 & 13 seats. The difference is that INC has consistently got 15-18% votes across the state peaking at very few places, whereas BJP has seesawed between 30% (winning seats) & 15%(decimated seats). This leads to Gallagher effect and more seats to BJP.
On Thu, May 14, 2009 at 4:04 PM, Arun Narendhranath wrote:
Two important points ignored by Psephologist
a) Most of the exit polls are converting state wise trends into seats. Expect Tamil Nadu none of the larger states are monolithic in their voting pattern. The classic example is converting higher voteshare of INC in Rayalseema region to seats across AP.
b) Gallagher effect is also missed by the media groups. They did the same mistake in Karnataka assembly elections, they are doing so now in the UP electoral trends.
For more info read about Gallagher Index @ http://en.wikipedia.org/wiki/Gallagher_Index
These are the two points you can use while attending TV debates on Exit Polls.