Shailesh Kapoor: Poll Of Polls: When Exit Polls become too many to handle

10 Mar,2017

By Shailesh Kapoor


It’s the big election results day tomorrow (March 11), with counting and results in several states, including two that carry massive interest: UP and Punjab.


Exit Poll predictions were released last evening across news channels. The last phase of UP elections ended on Wednesday, but since some bypolls were scheduled for Thursday, news channels had to wait that one extra day before releasing the Exit Poll findings.


Exit Polls have become a bit of a mathematical maze in recent years. There are 8-10 different polls, co-branded with two entities – one a media house or a media brand, and second a research or psephology firm. The results can vary significantly at times, to the extent of comfortably changing the predicted winner. Some of these polls give a “range”, while others predict an exact number of seats (error margins may be documented, but no one cares about those details, as anyone in the forecasting business will tell you).



The new entity in recent years that adds to the maze is one called the “Poll of Polls”. It’s a simple average of all available polls, taken to find a level that’s generally seen as acceptable, because it presumably reduces the overall error margin by leveling things up. That notion (that the error margin reduces via this method) may not exactly be true, but it’s a viewer-friendly thought nevertheless, riding on simplification a a central thought.


About a decade ago, channels and newspapers fiercely protected their work, not wanting others to share their data, and not sharing theirs in return. But in recent times, this has changed, primarily because this multi-source approach hedges their bets. Some channels have had egg on their face in the past, because they stood by some numbers their Exit Poll predicted, only to be found way off the mark on the results day.


So now, even if a channel has its own poll, they share that as one of the data sources. It may get considerable weightage and attention, because they will have better demographic cuts and detailing available inhouse. But the larger story is still projected based on the Poll of Polls. So, it’s routine to see names of competitive channels being spoken (and on the screen) on such days. Last evening, for example, ‘Times Now’ was very visible on NDTV 24X7.


Some channels have stopped commissioning their own polls, well knowing that there will be enough and more available to bite into. In any case, the entire exercise, such as the one last evening, is laden with confusion. For example, while it is amply clear that BJP-SAD will struggle in Punjab, and may end up with single-digit seats, the polls just could not conclude who will win the state. And while most polls suggested BJP would be comfortably the single-largest party in UP, at least one gave a very different picture, putting SP in front.


Politicians, of course, take a stance of acceptance or denial, depending on how the results suit them. Sometimes, they can take a stance of whole-hearted acceptance on one state and a stance of complete denial on another, within seconds of each other. Very few like Yogendra Yadav actually understand how it’s all done. “Expert” comments on Exit Polls are thus largely political rhetoric, devoid of any statistical or rational view on the data.


Do we really need these polls with only a day-and-a-half for the results? Anything for a day’s viewership. And anything for some pre-election mood build-up.


On a lighter note, I wonder what would happen if all channels decide to stop their own Exit Polls and rely on the Poll of Polls, which would then not exist to begin with?


Post a Comment 

Comments are closed.