![]() Online edition of India's National Newspaper Friday, May 12, 2006 |
|
|
|
|
|
|
| Opinion |
|
News:
Front Page |
National |
Tamil Nadu |
Andhra Pradesh |
Karnataka |
Kerala |
New Delhi |
Other States |
International |
Opinion |
Business |
Sport |
Miscellaneous |
Engagements |
Advts: Classifieds | Jobs | Obituary |
Opinion
-
News Analysis
Sanjay Kumar, Rajeeva L. Karandikar and Yogendra Yadav
THE POST-POLL and exit poll based projections made by the CSDS team for The Hindu and CNN-IBN have proved fairly accurate. These polls predicted a landslide in favour of the Left Front in West Bengal, the best performance of the LDF in Kerala in the last two decades, a comfortable majority for the DMK-led alliance but not for the DMK by itself in Tamil Nadu, and the Congress falling just short of a majority in Assam. The broad political picture projected by us has proven to be correct in each of these cases. The seats estimate of The Hindu-CNN-IBN polls was bang on target in three States. In Tamil Nadu, our seats estimate was 99 per cent accurate. The seats won by both the leading alliances fell within our estimated range. However, the exit poll appears to have significantly over-estimated the vote share lead for the DMK-led alliance by nearly 5 percentage points. In West Bengal, the seats won by the Left Front and the Congress fell within our range, while the NDA and others are within two seats of our estimate. The overall accuracy of our forecast was thus 97 per cent. The vote-share for major alliances too roughly corresponds to our projections. In this respect the projection for Assam proved quite accurate. The seats secured by all the major alliances fall within our estimated range, giving an overall accuracy of 97 per cent. The vote projection also proved quite robust. Our estimates were less accurate in Kerala, where we over-estimated the seats for the winning coalition by 12, even if we take the lower band of our range. That makes the forecast 80 per cent accurate. This was directly related to an over-estimation of the LDF lead by nearly 5 percentage points. We need to investigate this error and take steps to avoid it in future. The Lokniti research programme at the Centre for the Study of Developing Societies (CSDS) entered into a collaboration with The Hindu-CNN-IBN to carry out a series of pre-poll, exit poll, and post-poll surveys with a view to improving the quality of polling in the country, restoring credibility to survey research, and gathering reliable information on political opinions, attitudes and behaviour. In this process we have set three objectives for ourselves: accuracy, transparency, and accountability. It is a matter of satisfaction for us that our projections in Bihar as well as the four States this time have been fairly accurate. In line with the best global practices, we have started sharing all the methodological and sampling details including the social profile of the respondents with our readers. This report is the first step towards being accountable to our readers. As we offered earlier we would be happy to respond to any queries at election@csdsdelhi.org. The accompanying graphic shows a comparison of our projections and the actual results for the four States where we made projections.
Printer friendly
page
News:
Front Page |
National |
Tamil Nadu |
Andhra Pradesh |
Karnataka |
Kerala |
New Delhi |
Other States |
International |
Opinion |
Business |
Sport |
Miscellaneous |
Engagements |
|
|
|
The Hindu Group: Home | About Us | Copyright | Archives | Contacts | Subscription Group Sites: The Hindu | Business Line | Sportstar | Frontline | Publications | eBooks | Images | Home |
Copyright © 2006, The
Hindu. Republication or redissemination of the contents of
this screen are expressly prohibited without the written consent of
The Hindu
|