Karnataka Elections: Some questions about the exit poll process

With mixed verdicts, six exit polls give a majority to Congress, two to BJP and two indicate a hung assembly.

Exit polls are expected to give clarity about the likely outcome of an election before the actual results. But over the last few years, the manner in which a large number of exit polls are conducted or at least shown on TV tends to confuse us instead of giving clarity about possible electoral outcomes.

What is being flashed on most TV channels is the number of seats parties are likely to win without sharing any details about the exit polls and without even calculating/estimating the vote share for different parties. The exit poll on Karnataka elections shown on various television channels last night was no exception.

As many as ten exit polls were shown on various television channels last evening. One would have expected these exit polls to give us clear indication about what might happen in Karnataka, as for the last few months, there have been a lot of curiosity about this election among political parties, candidates, analysts, and media as well as among a large number of common voters.

But with mixed verdicts, six exit polls giving a majority to Congress, two exit polls indicating BJP crossing the majority mark and two indicating a hung assembly, the suspense about what might be the result would continue till May 13, the day when votes will be counted.

Where Are The Vote Share Estimates?

The suspense remains as the estimates from exit polls are mixed. Several questions have been raised about these exit polls, what could be the overall assessment from these exit polls, and which poll seems to be more accurate than others, etc. If we look at all these exit polls, one trend could be safely inferred, i.e. Congress is ahead in the electoral race over its nearest rival, the BJP, with JD(S) being the number three party in this race.

But there are related but important questions: What is the magnitude of the lead Congress enjoys over the BJP; Is Congress winning the majority of seats; And will Congress get a thin majority or big majority?

I am not in a position to offer any clarity on these questions. Had these exit polls shared details about vote share estimates and other methodological details, one would have got a better sense of which poll may be more accurate compared to others and what could be the likely range for number of seats likely to be won by political parties.

But unfortunately, except two exit polls which shared details of vote share estimates, other pollsters did not share any details about vote share estimates, leave aside sharing details about methodology. I wonder, how these exit polls manage to arrive at the calculation of the number of seats without estimating the vote share for different parties?

Why Poll Of Polls Is Flawed

Given the methodology which is expected to be followed for conducting exit polls, the first step in this exercise is to estimate votes for parties which are then converted into seat share. If these exit polls have estimated the vote share, what prevents them from sharing the vote share estimates for different parties? Had all the exit poll shared estimates of vote share and other methodological details been revealed, one would have got better clarity about the quality of the exit polls and to a great extent which exit poll may be more accurate than others.

Some television channels have devised what they think is an innovative method of arriving at a more accurate assessment by doing a “poll of polls” in which they try to take the averages of all the polls. Those who do “poll of polls” believe they have been able to arrive at more accuracy, but there is an inherent fallacy in this method and average of these exit polls does not help us in arriving at more accuracy.

For example, take a five-match cricket test series where a team loses three matches by narrow margins, but wins the next two matches by a very large margin. If the victory margin of runs in the last two matches  is more than the runs by which it got defeated thrice earlier, can this team be declared as the winner of the series? No. Similarly, if only a few candidates of a political party win the election by very big margins, but its candidates lose the election in a large number of seats with very small margin, can the party be declared the winner in the election without winning a majority of seats?

Sanjay Kumar is a Professor at Centre for the Study of Developing Societies (CSDS). He is also a Political Analyst. Views are personal and do not represent the stand of this publication.