|HOME | ELECTION | SPECIALS|
|October 9, 1999||
The Rediff Election Specials/ Krishna Prasad
An exit policy for pollsters
There is good news and bad news for Sonia Gandhi. The bad news of course, she knows. The good news is she needn't apologise to pollsters.
Crystal-ball gazers on television seem to have performed prayaschit for the disaster of 1998 by making a perfect prediction for Tamil Nadu, but they have gone completely awry in predicting which way the polls would go in most of the states where they conducted exit polls. <
DRS, the agency which did Doordarshan's much-criticised poll, got the state-wise prediction right in just two of the 16 states where it claims to have conducted the exit polls, while Insight, the agency which did Star News's exit poll, got it right in just four of the 16 states.
That's a success rate of just 6.25 per cent for DRS. And 25 per cent for Insight.
Both DRS and Insight were 100 per cent correct in Tamil Nadu which turned out to be a minefield for most pollsters last year. Both predicted 25 and 26 seats respectively for the BJP-DMK and their allies. And 26 was what the NDA partners ended up with in the end.
The two agencies were equally on the ball in Gujarat where they predicted 18 and 15 seats respectively for the BJP; the final tally was 19 seats. Insight also got it right in Orissa where it said the BJP-BJD alliance would get 17 to 19 seats (final score: 19) and in Assam where it predicted two seats for the BJP and the final score was indeed two seats.
But overall, both agencies were "far off or off the mark or misleading" in the cases of at least 12 of the 16 states.
That's a failure rate of 75 to 87.5 per cent with a sampling error of 0.
To start with, DRS, which claims to have conducted an exit poll in 475 of the 510 constituencies in the 16 states, gave the BJP and its allies 249 seats while it got 282 seats. It gave the Congress 159 seats against that party's eventual score 127.
But where DRS really got it wrong was when it came to the "Others". It gave the honourable extras 32 seats; they ended up with 92.
Insight got the overall BJP+allies figure all right, all right; but remember, they gave 284 to 299 seats for the BJP and its allies in the 510 constituencies under review. Remember, too, that 537 seats went to the polls, not 510. Insight gave Congress 132 to 154 (actual: 127); and they too underestimated the role of the "Others" though not as much as DRS. They gave them 74 seats; the others got 92.
Uttar Pradesh, of course, was pure disaster. DRS gave the BJP and its allies 46 and Insight 48 to 52; the NDA ended up with 29. Ditto Karnataka. DRS predicted 17 for the BJP-JD-U alliance and Insight predicted an even more optimistic 19; reality: 9 seats.
But it is the manner in which both agencies were completely off the mark in individual state-wise predictions that is most striking and yet ended up making overall predictions that were not too far off the mark that is most striking. And that is enough empirical evidence for Chief Election Commissioner M S Gill who wants a ban on the pollsters.
For instance, DRS gave the BJP and its allies 43 fewer seats than what the NDA got in 8 of the 13 states and gave them 31 extra seats in the other five states. Likewise, Insight gave the NDA alliance 42 fewer seats in 10 out of 15 states and 31 additional ones in the other three states.
Amazingly, DRS, which came under severe fire from the Congress spokesman Kapil Sibal, gave the Congress 52 more seats than what Sonia's party got in 13 of the 14 states; and Insight gave 24 more seats than what the Congress got in six of the 16 states.
The goof-ups are endless.
Andhra: 22 for BJP said DRS; 33 said Insight. Final figure: 36.
Bihar: 30 for BJP said DRS; 27 to 31 said Insight. Final figure: 40.
Haryana: 7 said DRS; 6 said Insight. Final figure: 10.
Maharashtra: 29 said DRS; 31 said Insight. Final figure: 28.
West Bengal: 16 said DRS; 8 to 11 said Insight. Final figure: 11.
And so on and so forth. Maybe this is the reason the Election Commission should give free and unfettered access to pollsters to conduct exit polls. It's great fun to see their mugs on television as if it were god's gospel; and it's even greater fun massacring them after.
Any takers for the new government's first major policy decision: an exit policy for pollsters?
ELECTION 99 |
SINGLES | BOOK SHOP | MUSIC SHOP | HOTEL RESERVATIONS | WORLD CUP 99
EDUCATION | PERSONAL HOMEPAGES | FREE EMAIL | FEEDBACK