Narendra Modi & Rahul Gandhi together defeated Exit Polls in 2014 elections!!!

Loksabha 2014 election results are out .. its a clean sweep .... Mr. Narendra Modi and his team has won with unexpected numbers. Congratulations to him and his team. The whole credit goes to Mr. Narendra Modi, whoes relentless efforts has bring this historical win to BJP and NDA. This is also win for first political brand MODI and it successful marketing. This results also underlined the reach and importance of social media. Undoubtedly social media has reached to remote rural areas of India. Social media has proved its strength once again.

Now its time to analyse the performance of exit poll results. In my previous post I wrote it would be interesting to compare actual results with predictions; based on exit poll survey. Just to remind, all research agencies have claimed 3-4% error in their predictions. Now lets see what they predicted and actual result are..

Channel wise % difference in projected and actual results 

Above chart shows channel wise difference (in%) in projected and actual results. From above chart it is clear that none of the channels' projections are near to actual results. Difference is lowest in predictions of 'News 24' while highest in 'Times Now'. 

Comparison of projected (average of all channels) and actual results
Note: Figures are based on results+leads data at 9.30pm on 16/05/2014 

In above chart I compared average of all exit poll results with actual results. It again shows there is huge difference in exit polls results and actual results.

This is third consecutive time exit poll surveys failed to predict results. In 2004 and 2009 they overestimated NDA and underestimated others. The only difference is in order in 2014 election exit polls results. This time they underestimated NDA and overestimated UPA.

What went wrong?

Exit poll surveys scored hattrick of failure. You may wonder if similar exit polls can predict fairly accurate in US, then why not in India?

As I wrote in my previous post, it is sampling method which is causing this blunder. Unfortunately research agencies including Nielson are not ready to appreciate this fact. The huge deviation (positive and/or negative) clearly indicates that whatever samples selected were not truly representing Indian voters. Researching agencies are not ready to share their methodology, hence it is difficulty to comment on their appropriateness. However variation in exit polls results among agencies shows that they are using different methods.

What next?

Researching agencies need to adopt appropriate sampling methods so that the can map attitude of voters accurately (5% error is acceptable). But beyond this; puts question mark on credibility of exit polls. Agencies should be more responsible and committed to provide accurate results other wise people will loose faith on such surveys.