We have collated and compiled every Opinion Poll for Delhi Assembly Elections 2015, right from the first one in November 2014 till the one yesterday. On public demand, we also updated the Vote Share and Sample size details, making it the most comprehensive collection of Opinion Polls. But all this data can be confusing. I will try to simplify the same.
First I take a simple average of all polls, monthwise. That is, we see what was the average prediction for each month:
| Vote Share AAP % | Vote Share BJP % | Vote Share INC % | Vote Share Others % | Seats AAP | Seats BJP | Seats INC | Seats Others |
|
|---|---|---|---|---|---|---|---|---|
| November 2014 | 33 | 41 | 11 | NA | 23 | 42 | 4 | 1 |
| December 2014 | 34 | 41 | 20 | 11 | 27 | 38 | 4 | 1 |
| January 2015 | 37 | 40 | 15 | 8.5 | 30 | 35 | 5 | 0 |
| February 2015 | 42 | 37 | 13 | 8 | 38 | 28 | 4 | 0 |
From the above table, if you see the Vote Share column, there is a steady minor fall in BJP’s vote share which goes to 37% from 41% in November 2014, while AAP makes much bigger gains by reaching 42% in February from starting point of 33%. In a way one can say BJP has marginally lost vote share, but AAP has gained much more. Now comes the tricky part, projecting the seats.
BJP, which had 42 seats in November with 41% of vote share, drops to just 28 in February thanks to a mere 4% fall in vote share. On the other hand, AAP’s vote share increases by 9% and seats jump to 38 from 23 in November. Does just a 4% loss in vote share mean a loss of 13 seats for BJP, only because AAP has gained 9%?
Also, most of these polls usually come with a margin of error of 1% to 5%. Assuming a 3% error margin, the entire game changes, with BJP getting 40% vote share and AAP on 39%, point being, in such a tight finish, its hard to be accurate, especially when you are converting Vote shares in Seat predictions
Another interesting point is the Congress Vote share. We can ignore the vote share for November which maybe low because of incomplete data, but from December to February, Congress vote share has dropped 7% from 20% to 13%. BUT projected seats of Congress have remained near 4 throughout. Does this mean Congress is losing votes only in certain constituencies? Could it mean Congress is holding its own in pockets where it ash high vote share (and where it is winning) but is losing votes from places where it would have anyway come third? Have the pollsters made their predictions in such detailed manner?
Also let me pull up the Opinion Polls predictions for Delhi, during Lok Sabha 2014, the most recent elections Delhi has faced:
| AAP seats | BJP Seats | INC Seats | ||
|---|---|---|---|---|
| Aaj Tak / India Today | January 2014 | 5 | 2 | 0 |
| CNN IBN | January 2014 | 5 | 2 | 0 |
| ABP News | February 2014 | 6 | 1 | 0 |
| NDTV | March 2014 | 4 | 2 | 1 |
The final results were of course a clean sweep of 7 out of 7 seats to BJP. From the above, one thing is clear, Pollsters haven’t got Delhi correct most often. Is it because it is “Delhi”, or did they under-estimate the Modi wave, or is it because Delhi based media gets too swept away in the AAP media frenzy? Whatever might be the reasons, we have enough reasons to cast aspersions of Delhi Opinion Polls atleast.
And finally, we have the Satta Bazaar, India’s illegal betting market. Although not a scientific “Opinion Poll”, it is an “Opinion” and one which was one of the closest to the right answer for Lok Sabha 2014. Punters had then bet that Modi will become PM and NDA will reach 272. The actual result was far better with BJP on its own crossing 272 seats. Even today, the BJP is the favourite of the betting market, in sharp contradiction to all latest Opinion Polls. Do the people whose money is riding on something know better than psephologists? Only time will tell. Till then, you tell us what you think will be the outcome: Take part in our Contest for predicting the Delhi Assembly results and win Flipkart vouchers and @amreekandesi‘s latest book!
