Exit polls in India have become increasingly sophisticated, polling much bigger sample sizes and adjusting for different factors. While the Economic Times and Indian Express have come out with data from previous elections, it is worth noting that in 2004 and 2009 when the NDA ended up with far less than the exit polls predicted, we find major discrepancies. Exit polls before 2004 appear to be fairly reliable. 2004 is also the year that EVMs were used for the entire Lok Sabha election across the country. While the Courts have ruled that EVMs can be tampered and that paper audits be installed, the Election Commission has only done a trial run of paper audits for the 2014 elections, though it had over two years to implement this feature.
Exit polls have also been very reliable as recently as the exit polls for Assembly elections of 2013. The Lokiniti-CSDS post-poll survey predicted that the BJP would win 43 per cent of the votes in Rajasthan in 2013, the actual number was 45.1 per cent. In Madhya Pradesh, exit poll was 41 per cent for the BJP, actual was 44.9 per cent. In Chhattisgarh, 42 per cent was the estimate and the BJP actually got 41 per cent. In Delhi, exit polls predicted 33 per cent and BJP got 34 per cent. These points towards the reliability of the exit poll methodology used in recent times. Critics in the past have charged that exit polls may be unreliable because of ‘fear factor’ of people afraid to disclose who they voted for. However, no such factor appears to have played a significant role in the recent Assembly polls and it would be hard to make the case that Lok Sabha election, a few months later, would dramatically change that equation.
The current CNN-IBN-Lokiniti-CSDS survey shows a range of 270-282 for the NDA with a margin of error of about 3 per cent. The APB-Nielson survey is fairly similar as well. Other estimates range from a low of 249 of the Times Now to a high of 340 of News 24. The mean value of 286 is fairly close to the cluster of predictions.
Source: Rediff. For CNN-IBN, midpoint of range has been used
Given the reliability of recent Assembly exit polls and the fairly consistent data from the 2014 Lok Sabha polls, any significant deviation from these numbers would be cause for serious concern about election results manipulation and rigging. The broadest range is 250-340 and anything below or above that needs to be investigated more thoroughly.
Exit polls depend on self-reporting. Rigging factors would not be captured in exit polls self-reporting. What kind of rigging would go undetected by exit polls?
1. Pre-poll EVM rigging, where the EVM is manipulated in such a way that all choices go to a single party. A number of such flawed machines were found.
3. Post-poll EVM capture and counting manipulation. This could happen with technology flaws to manipulate the final results and also it has been alleged that EVMs are not in safe custody.
All these scenarios can result in a significant difference between the final results and the exit polls. Such a difference should not be taken lightly if it occurs in the 2014 Lok Sabha election results. The stakes for free and fair election in India are too high to be brushed away.
(Follow Sankrant Sanu on Twitter @sankrant for questions and comments)