Exit poll

Now that election results are a less than 2 hours away, I am thinking about poll outcome and grand  failure of exit polls in India.  It is a bit funny that even after grand failure of exit poll in 2004 and in UP and Karnataka election, most of the guys are back in business without any remorse.  They have predicted UPA clearly ahead this time. I personally feel otherwise and waiting for NDA to return in power. However, I will be happy with return of NDA or UPA with clear majority.  Nightmare situation for me is a government dependent of CPM for its survival.  Only god can save this country from that godless party.

              Let me assume that exit polls are wrong this time too. Although, I have a hunch that Star-Nielson exit poll is going to be more closer to the reality. The question is why exit poll fails in India so often?  Is it because exit polling is an imperfect science? On the other hand, exit polls are very successful in west. Is it that Indian voters are unpredictable? Or can it be that pollster themselves have hidden agendas and problem in attitude. 

      My answer is all of them are partially true reasonings. Yes, exit poll which is just like any data based modeling is imperfect science. The chance that they will go wrong is always there. One aspect was analyzed nicely recently on Swapan Dasgupta blog.  


It was argued there that one important technical aspect so called Bradley effect (http://en.wikipedia.org/wiki/Bradley_effect) is ignored in Indian exit poll. The idea behind this effect is that unpredictability of  polls are increased due to social desirability. For example, in USA a white voter will not often admit that he has voted for white candidate rather than black one. He will fear that by doing so, he will be branded racist.

This effect should be a big player in India. However, none of the exit poll talk about it. So, one aspect is how competent are the pollsters? I guess not too much. I am not sure that Mr. Yogendra Yadav, a self made star of exit polling, in India is technically competent enough. For sure, he has the first mover advantage. This might be true for lot of other guys too.

   Yes! Indian voters are very unpredictable  too.  Part of the reason is India is collection of different competing groups. Who will feel offended and go massively to vote is not clear in advance. We have two examples BJP defeat in 2004 and in recent Rajasthan election. In the first case it was anger for Gujarat riot and in the second case it was angers of Meena and Gujjar in reservation fiasco.  I guess in few years, India will also evolve like USA, where politicians are never against anything but for something. The trick is avoid anger of  specific  group of voters by not being too direct. For example a USA politician will like to be called pro-life rather than anti-abortion.

     Finally, I guess a major part of error is coming because pollsters are too biased. This time on NDTV anger of presenters Pope Barkha Datt etc, was very visible. They really hate BJP. As a pollster your job is to do the unbiased reporting, the moment data get mixed up with your personal feeling, it is going to be a mess. Sometime you will tend to over-estimate the enemy when they are strong (2004), or you will   underestimate them big time (2009). See this funny video about Indian exit polls

Finally, there is a lot of talk about manipulating the exit polls for money. Well, this is supposed to be very powerful in India. All I can say, in long run cheating will not get you very far. The sooner they learn the better.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: